"Subclinical" Hypothyroidism: is restricting treatment based on science or on politics?


Hypothyroidism treatment in adults is undergoing a crisis.  One of medicine's great success stories is the development of diagnostic and treatment tools for all levels of primary hypothyroidism, making the physician's task in treating most cases among the simplest to handle. However, science is shifting away from studying the biochemistry of the disease and its complications and toward justifying not treating the overwhelming majority of those who have it. Screening has been eliminated in many countries even though there is a widespread belief that the disease has few, if any, identifying symptoms and signs except in especially severe cases, a common problem associated with diseases that affect the entire body.  A recent avalanche of publications claiming that there is no substantial scientific evidence supporting the treatment of nearly all hypothyroid individuals with thyroid hormone replacement or even for screening for that disease, threatens to turn the clock back for many.  How much of this backlash is based on sound science, and how much on politically motivated bias in a few influential countries artificially limiting the number of physicians and slashing healthcare expenditures?  Yet a minority of cases, which involve the complicating effects of diseases in other parts of the body, make this typically simple picture harder to sort out and threaten to undermine the argument for treating the simpler majority.

It seems strange for physicians to resist treating a disease that is not contagious, is easy to diagnose and treat, and costs patients very little.  The key problem, according to some endocrinologists, appears to be that a full understanding of hypothyroidism came too late in the healthcare game.  Large goiters, "cretinism", a state characterized by extreme mental retardation and growth deficiencies, and the often fatal "myxedema coma" were its only recognized signs and symptoms until fairly far into the twentieth century, and apparently constitute the basis for the training of most non-endocrinologist physicians.  Most cases were not subject to diagnosis via laboratory tests until very recently because of the technological challenges imposed in measuring small molecules such as hormones with extremely low concentrations in the blood.  The radioimmunoassay, a breakthrough in that respect, was developed in the 1950s but not refined enough to be trusted as a diagnostic tool for both hypothyroidism and hyperthyroidism until about 1990.  Besides, a sophisticated understanding of the interactions of the thyroid with the hypothalamus and with the other endocrine glands and the hallmark patterns of their hormones in different conditions did not come until even later, while studies of the effects of hypothyroidism on vital organs other than the heart (with an emphasis on blood lipid levels) began only very recently, mostly in Asia and Germany.

Clinical biochemists continued to develop new assay methods and settled on thyroid stimulating hormone (TSH), produced by the pituitary gland, as the ideal target for which to develop reference ranges to diagnose uncomplicated primary hypothyroidism.  When the first-generation TSH assays were being used, i.e., before the mid-1970s, a cut-off of 10 mIU/L was the standard threshold value for hypothyroidism diagnosis; it is not clear how this value was chosen.  However, as second- and third-generation assays replaced this earlier method, and the presence or absence of anti-thyroid antibodies was used to define the reference ("normal") patient population, the TSH reference range upper limit decreased (through 2000 or so), thereby increasing the number of patients recognized as having the disease.  A TSH level of 10.0 mIU/L was eventually shown to be very unusual, and that a level of about 1.4 mIU/L to be typical of those in normal health, while TSH values in general followed a continuum; the distribution of these values followed something very similar to a bell-shaped curve, though slightly skewed to the right.  This stood in sharp contrast to the previous perception of hypothyroidism as a small cluster of clearly abnormal cases separated from the normal majority by a wide gulf.

In fact, a TSH level of 5.0 mIU/L was shown to be clearly abnormal, i.e., the point at which anti-thyroid antibodies began to appear, and within the right tail of the curve among patients considered to be in normal health; new reference range upper limits began to be set at 4.0 mIU/L and even lower.  Thyroid hormone levels, too, especially free thyroxine (free T4), became key criteria, although their reference range has been narrower and the "elbow" of their distribution curve at the hyperthyroid extreme of that range.  Anti-thyroid antibody levels might also be informative, although what raises their levels, other than elevated TSH, is not clearly understood.

This was how matters stood in 2005, when a good understanding of how to diagnose primary hypothyroidism was generally accepted.

What, then, led to the backlash threatening to undo sixty years years of progress in this field?

Some E.U. and, later, U.S. endocrinologists and primary practitioners began a process to restore the old diagnostic standard by establishing a new category called "subclinical hypothyroidism", which covered those with TSH levels between this new reference range upper limit and the old one, i.e., 10.0 mIU/L, and having free T4 levels within their reference range.  They claimed that there was no scientific evidence supporting the treatment of those in this category with supplemental thyroid hormones, and published many studies meant to support this claim.  When observational studies did not shore up their suspicions that those in this category were as healthy as those with TSH values in the reference range, they called for clinical trials using thyroxine (L-T4) as the study drug, claiming that these health differences did not imply the need for L-T4 treatment, or even that it would be beneficial in any way.  

Some of these trials (often involving fewer than 100 subjects/patients) challenging current hypothyroidism treatment standards on which revised healthcare policy was based merit more scrutiny than they have received because they appear to be underpowered and took place over a very short period of time relative to the known pattern of occurrences of the adverse events used as criteria.  In addition, clinical trials involving study drugs that are already on the market face special ethical difficulties in regard to recruitment of appropriate subjects in the countries where these drugs are legally available: what would be the motivation of such individuals desiring treatment to seek it via enrollment in a placebo-controlled trial?  Meanwhile, the value of measuring anti-thyroid antibodies, once considered very important, has more recently been dismissed in the struggle to streamline thyroid disease diagnosis for cost-saving purposes.

These measures intensified around 2008, when political pressures to reduce medical treatment costs began to ramp up, largely as a result of the actions of the U.S. medical care insurance industry, which was apparently highly influential, if not mainly so, in crafting the Affordable Care Act (ACA).  Although evidence-based medicine (EBM) was already gaining support as a means to advance the field, certain power players increasingly represented it as a means to save money by trimming down medical care, based mainly on the results of recent clinical trials.  Would money saved by elimination of "low value" medical services not be counterbalanced by the considerable expense of running clinical trials?  There would still be obvious beneficiaries of this approach: 1) insurance companies, which would have to pay less, and 2) primary care physicians on fixed salaries, who would have their workload reduced (and, necessarily being high-energy people, would be unlikely to have hypothyroidism themselves and therefore not have any concerns about losing access to needed treatment).

In general, studies have agreed that women and older adults are overrepresented among those with elevated TSH levels.  An especially troubling development, however, has been the increasingly popular argument that very high TSH levels are normal in older people and that those over 65 years of age should not be treated for hypothyroidism.  However, the NHANES III study illustrated that such older adults without any diagnosed diseases had TSH levels only slightly higher than those of younger adults, and that they averaged much lower than what is now accepted as the upper limit of the TSH reference range, i.e., about 4.5 mIU/L (which in turn is much lower than the traditional threshold TSH level for treatment, i.e., 10.0 mIU/L for most hypothyroidism patients).  On the other hand, that study and many others have shown that complicating non-thyroidal diseases, which can also drive down free T4 levels, sometimes below the reference range, are common causes of out-of-range TSH levels in this older age group; because older adults are more likely to have non-thyroidal diseases than younger ones, that causes misleadingly large differences in the average TSH values for those two groups.  Another consideration is that most older people have had measles and other so-called childhood illnesses when they were young because of the unavailability of vaccines, thereby experiencing lasting harm to their health and perhaps becoming more vulnerable to autoimmune conditions as a result.

An important concern is whether hypothyroidism has complications, and what parts of the body they affect.  The relationship of TSH and thyroid hormone levels to various markers of heart health has been studied extensively, and the weak relationship of blood lipid levels to TSH levels in the "subclinical" range, has been used to justify withholding treatment for those with TSH levels below 10.0 mIU/L.  Jonklaas et al. (2014) put forth a more nuanced view; in discussing the "potential deleterious effects of inadequate levothyroxine", they claim that in one key study (Razvi et al., 2012) treating those with "mild" hypothyroidism showed that it reduced "the risk of cardiac events in patients aged 40-70 years, providing some preliminary evidence for normalizing the serum TSH in this age group."  On the other hand, in that particular study, no benefit was shown for those over 70. Cardiologists have done clinical trials that show that even subclinical hypothyroidism thickens some layers of the carotid arteries and that this can be reversed with levothyroxine therapy. However, even for those with "overt" hypothyroidism, Jonklaas et al. (2014) do not mention diseases affecting any parts of the body other than the heart. 

However, research about possible complications involving other organs has been taking place more recently.  There have been several studies of the effects of these hormones on liver function, which suggests a much stronger relationship, and that knowledge has been used to develop a drug that treats non-alcoholic fatty liver disease by stimulating one type of thyroid hormone receptor in liver cells.  There have also been a few studies that have shown a significant relationship hypothyroidism to kidney disease.  Although hypothyroidism is officially required to be ruled out to establish an Alzheimer's dementia diagnosis, very few publications cover possible causal relationships between these two conditions, and the issue of potentially incorrect diagnoses of Alzheimer's based on higher standards for hypothyroidism diagnosis has apparently not been raised by any researchers.  On the other hand, the possibility of greater vulnerability to acute systemic infections, such as influenza or COVID-19, experienced by those with any level of hypothyroidism has not been explored within the last fifty years. 

What is just beginning to be studied is the occurrence of abnormally high TSH levels (usually below 10.0 mIU/L) and abnormally low free T4 levels in individuals with non-thyroidal diseases (sometimes undiagnosed or difficult to diagnose), which could be mistaken for "overt" ("real") hypothyroidism.  Because such individuals cannot rely on thyroid hormone replacement alone to recover, this might explain existing doubt of the value of such treatment in many who have genuine thyroid failure and no other diseases.  Yet another question remains to complicate the picture further: could those with incurable or severe diseases driving TSH levels up experience palliative relief from L-T4 supplementation?  In fact, should we ever assume that a high TSH level is normal and healthy, even if it commonly, or even typically, occurs in individuals stressed by disease or aging?  Are we safe in assuming that a sustained high TSH level does not cause extra wear and tear on the thyroid?  And should we continue to assume that a high TSH simply affects the thyroid gland and no other parts of the body directly?


This essay is first and foremost about a social. political, economic and legal issue, not about the technical challenges of biomedical science.  This issue appears to be a conflict between 1) the need for those with an unusual (but not completely devastating) degree of thyroid failure to receive the treatment that they need, especially to keep up with an ever more demanding workplace requiring excellent physical health and substantial mental agility, and 2) the wish for many physicians, especially those in fixed-salary primary care positions, to reduce their workload without a concomitant reduction in compensation, a cutback in their own medical care (since hypothyroidism is incompatible with employment in the high-energy field of medicine), or the threat of malpractice suits.  In turn, these physicians have found common cause with insurance companies seeking to reduce their financial burden by minimizing expenditures on "low value" medical care, and with the economizing governments of countries providing socialized medicine.  Of course, physicians and insurance companies have expressed concern over different sets of "low value" medical care: insurance companies have tended to target expensive, risky, and unpleasant procedures that had not been studied scientifically at all, while physicians have preferred to target types of patients whom they cannot relate to.  As a result of these complex forces, treatment of "subclinical" hypothyroidism with thyroid hormone replacement has rapidly come under fire as "low value" care, initially by a few endocrinologists in countries with struggling socialized medicine systems, and then by the medical establishment in several other industrialized countries, with the strong support of internists who face a sudden increase in patient load.

I have made every effort to reference full-length versions of articles; however I have limited myself mostly to those available for free.  Unfortunately, that has meant relying mainly on abstracts in the case of articles published before about 2000.  I have made these distinctions in the References section and therefore cannot evaluate the results of those abstract-only articles critically in any way.  In some cases, I have quoted sections of what I thought was ambiguous text from these abstracts rather than trying to paraphrase it. NOTE: Some articles were moved to another site since I originally cited them; I am in the process of identifying them and of updating their citations.

Unfortunately, because most studies supporting the treatment of "subclinical" hypothyroidism were published before 2009 or so, most published afterward did not, and because of a great increase in those available to the general public after that date, most relevant articles available to the general public discourage the treatment of this condition.  This is especially unfortunate because so many groundbreaking studies illustrating special problems associated with "subclinical" hypothyroidism were made fairly early and deserve the scrutiny that later papers have been getting.

There is an enormous number of recent articles discussing whether "subclinical" hypothyroidism should be treated, and a great deal of repetition in them.  I do not claim to have surveyed the complete field, although I can say with confidence that countless numbers of them open with the declarations that 1) "subclinical" hypothyroidism occurs in 4-20% of the population (perhaps referring to differences among countries, though that is never specified), 2) a link has been found between the condition and cardiovascular disease by observational studies, and 3) only new clinical trials using levothyroxine (L-T4) as the study drug can determine whether those affected should be treated with it at all.  They typically describe the treatment of "subclinical" hypothyroidism with the heated word "controversial".  Nevertheless, I hope there is enough here to persuade others to recognize that there are significant problems affecting research supporting the ambitious social experiment of a global withdrawal of treatment of "subclinical" hypothyroidism, and to launch their own (unbiased) investigations.

There are also many articles by cardiologists that cover the effects of hypothyroidism, "subclinical" and "overt" alike, on heart structure and function, that this paper has not done justice to.  There are abundant articles investigating the relationship between subclinical hypothyroidism and the thickness of the intima-media layer of the carotid artery, as well as to the contractility of the heart. They represent an important challenge to claims based on studies of blood lipid levels, typically made by non-cardiologists, that "subclinical" hypothyroidism has little or no effect on heart function.  

This essay, though heavily researched, does not follow the conventions of research publications by confining the writer's opinions to a discussion section.  I have made an effort, however, to distinguish my perceptions from the published observations and conclusions of others by indicating mine in green, although this has been increasingly difficult to maintain as my recognition of bias in research and its interpretation by the authorities has grown.  In addition, the "Problem," "Neglected Areas of Study" and, of course, "Discussion" sections all express my opinion. 

This paper is intended for an audience familiar with the basic terminology of biology, chemistry, algebra and statistics. However, I have set forth definitions of some concepts that are specifically important to this essay.

I have indicated technical terms in red, and the title of a section giving background information is shown in blue. Another section, covering thyroid physiology in depth, has a purple title; an important model of thyroid function presented in a recent publication by a group of German scientists is described here. The other sections should be accessible to those from a variety of backgrounds. 

Donna Tucker has given this essay an excellent "big picture" critical review, and has given me helpful advice leading to a much-improved rewrite of the introduction.  I do not hold her responsible for any errors in this paper. Otherwise, no one has read it critically. Please let me know of any corrections you think it needs.

Disclaimer: I am not a physician and therefore cannot claim to give individualized recommendations. Please do not regard this paper as a source of advice for self-diagnosis or self-treatment. Most research findings reported in this document are based on statistics gathered from research involving small groups of people (or involving animals) by scientists and therefore might not apply to many individuals. In sum, this essay puts a bigger emphasis on asking questions than on providing answers.

To start out this argument, I have laid out the issues about which there is great disagreement, and which stand in the way of solving this general problem if not understood clearly:

What is it like to have hypothyroidism -- and to treat it?

Symptoms of hypothyroidism being misunderstood or overlooked was a live issue in the late twentieth century.  According to Barnes (1976), Arem (1999), Shomon (2000), and Gold (1987), hypothyroidism persisted in being underdiagnosed because patients were not tested for it and their symptoms misread.  These authors are in general agreement about the clinical manifestations of the disease: "brain fog", depression, pain, disability, increased susceptibility to the "major" diseases, and, most strikingly, low energy and even difficulty staying awake.  Did the written pleas of these authors give impetus to the measures taken to improve the TSH assay enough so that the traditional wild guess of 10 mIU/L no longer ruled as the criterion for treatment?  Alas, no recent voices of protest have taken their place.

A recent study of newly diagnosed hypothyroid patients by the Indian physician Thakur (2019), using a questionnaire, found "Tiredness" to be the most common symptom, affecting 68.50% of those patients, followed by muscle cramps (57.90%) and weight gain (56.20%).  "Mental sluggishness" was reported by 42.00%, and "Memory loss" by 39.30%.

Perhaps the most trustworthy account of how patients experience hypothyroidism psychologically is found in an account of the "temporary hypothyroidism" experienced by thyroid cancer patients beginning after their treatment for the disease (Van Nostrand et al., 2010), since the authors have no motivation to exaggerate these patients' suffering.  To remove the cancer, they destroy their patients' thyroids, either by radiation or surgery, and the patients then receive thyroid hormone replacement.  As they adjust to their new therapy, these patients experience fatigue, a feeling of being "slowed down" and having longer reaction times.  Some describe feeling "mildly sedated."  They make more errors when performing tasks requiring attention to detail.  The authors caution them not to drive or to operate heavy machinery for a week or two after the radiation or surgery treatment (and also before, for reasons that are not clear.)  Older patients have "greater hypothyroidism manifestations."  These descriptions and accompanying cautions suggest an effect similar to that of tranquilizers, sedatives, and sleeping pills.

Some people consider sedation to be a pleasant experience and seek it out, and the addiction problem encountered with certain sedating drugs does not, of course, apply to a natural state of hypothyroidism.  The opioid crisis has shown that some people will go to great lengths to seek this feeling of sedation out, and not necessarily for relief of simple physical pain.  For those whose daily lives make few demands on them, who relish a low level of challenge, a state of constant sedation can be an attractive choice for them.  But many people, including some over the age of 65, are employed in challenging jobs that require high levels of alertness and mental agility, and would therefore be likely to find a sedated state to be a handicap.  In addition, citizens in all but two U.S. states are called to jury service up to age 70 and sometimes beyond; federal jury service often involves traveling over 50 miles to a courthouse and 3-6 months of service there.  In fact, certain jobs depend so greatly on the those who fill them being in a non-sedated state that drug-testing of such employees has become standard practice.  This might explain why, on one hand, many untreated hypothyroid patients report no discomfort in relevant quality-of-life studies while others experience great distress, and, on the other hand, why some of those who begin appropriate treatment feel uncomfortable at the outset.

What challenges physicians who treat the disease the most?

That there are so many studies questioning the validity of treating mild hypothyroidism implies that many physicians are unhappy with having to do so.  Yet there are few direct statements of how physicians feel about their hypothyroid patients.  Jonklaas et al. (2014) give us some hints with their discussion of their problems with patients who still report symptoms of hypothyroidism even after their thyroid hormone levels have been normalized by hormone treatment; they hypothesize that "somatization disorders", which are in turn hypothesized to be caused by a "complex psychological or abuse history".  They further hypothesize that a history of sexual abuse might be a major factor in accounting for what they see as a serious personality disorder.

What they might not be taking into account here are the unanticipated changes in psychological state, and maybe even of personal identity, that levothyroxine therapy causes in these patients.  As handicapping as hypothyroidism might be to patients, they might have learned to adapt to that state in many ways, and having to make substantial changes in response to therapy might in itself cause distress even when it is making their bodies healthier and sharpening their mental function.  Besides, some healthy changes, such as an improvement in their critical faculties and a drive toward more assertive behavior, might bring them into conflict with others, including their physicians, who have enjoyed their past easygoing, passive behavior. 

Counseling about these changes might make this substantial transition easier for patients, while physicians might also benefit from such education.

The problem with conditions that affect every cell in the body

Ironically, it is often harder to recognize the signs and symptoms of a condition that affects the entire body than one that affects only one part, especially in an individual with other diseases or abnormalities, because of wide individual variations and yet the lack of uniqueness of particular signs.  That is the reason diagnostic laboratory tests are so important, and the dispensing with screening for such diseases so risky.

Measuring patient subjective experience: Is it that routine?  Have the tools used been properly validated?

Subjective patient experience has become a prominent criterion for determining the effectiveness of treating hypothyroidism with hormone replacement therapy.  But asking patients to describe their subjective experience of drug treatment in terms that researchers find most meaningful or convenient might not have the intended results. The results of studies attempting to measure patient response to T4 treatment suggest that patients have different perceptions from what those running the studies expect them to have, and that the same patient perception might wind up being measured differently on different measurement tools.  Markman et al. (2020) compared the results of subject perception of pain level (on a scale from one to ten) and on a simple three-point scale, i.e., 1) tolerable pain, 2) intolerable pain, and 3) no pain to the results of such perception of pain on a numerical ten-point severity scale.  They discovered that when patients assigned certain numbers to their perceived pain level, they meant something different to them than they did to the researchers.

It is possible that the Markman et al. (2020) publication results could be used to justify cutting back on treatment of pain-causing conditions.  But the authors were simply considering this issue in the context of deciding whether to prescribe opioids for patients in pain that presumably could not be relieved by any other means.  However, when pain is a key symptom of a disease that can be treated or cured, especially easily, safely, and cheaply, a hands-off response or resorting to use of painkillers might be the wrong decision.  The take-home message here is that the psychology of subjective experience and its representation in experiments still needs to go a long way before it can be depended on to justify drastic changes in patient treatment.

What do veterinarians see in hypothyroid horses that might be missed by physicians in humans?

Veterinarians have to rely on clinical signs rather than symptoms, and since that involves few complicating psychological or social issues, diagnosis can be more objective.  According to one British veterinarian ("Problems of the equine thyroid", 2020), iodine deficiency is the main cause of hypothyroidism in horses, although iodized salt is typically added to feed.  While hypothyroidism in humans is apparently considered to be a disease of aging and otherwise a rare congenital condition easily diagnosed and treated, this author makes note of telltale developmental signs of hypothyroidism in foals, such as respiratory problems and bone problems such as "undershot jaw", i.e., underbite, and "bent legs".  Adults can have a variety of muscular problems, some very severe, covered by the umbrella term "tying up", which are apparently brought on by over-exertion in racehorses; this casts some doubt on our generally accepted belief that humans cannot overexercise. 

This was apparently the only article that addressed the special challenge of diagnosing diseases with systemic effects: "These hormones can have an effect on many parts of the body and on a variety of tissue types, which is why thyroid 'failure' can produce many non-specific symptoms."

What does "subclinical" mean in practice and what should it mean, especially the case of hypothyroidism?

According to Shiel Jr. (2020), "A subclinical disease has no recognizable clinical findings. It is distinct from a clinical disease, which has signs and symptoms that can be recognized."  Therefore, such a disease has no significance and requires no treatment unless it worsens, produces complications in other parts of the body, or is a serious infection that threatens to spread to others.  Whether a disease in "subclinical" stages is treated varies greatly.  Subclinical breast and cervical cancer are taken very seriously, and treated aggressively.  Subclinical cardiovascular disease is often treated with lipid-lowering and anti-hypertensive medications.  Subclinical HIV infection (formally called "chronic HIV infection" or "asymptomatic HIV infection"), a long-lasting stage after an initial short, acute one, is now regarded as the ideal stage for the drug treatment necessary to prevent or at least greatly delay the end-stage of the disease, i.e., the syndrome AIDS (Centers for Disease Control and Prevention, 2020).  The justification for treating truly "subclinical" illnesses apparently lies only in the potential danger of their progression, and if that danger is not significant or likely, the "subclinical" disease remains untreated, albeit monitored in some cases.

But what if the definition of the particular "subclinical" disease depends not on clinical signs or symptoms, but on blood test values?  Suppose it isn't quite proven that there are no signs and symptoms for everyone in this range?  Suppose there might be complications of the disease, but they have not been explored very deeply?  What does it mean when conventions had established the dividing line between "overt" and "subclinical" when the relevant instruments of measurement were much more primitive?

"Subclinical" hypothyroidism is defined by TSH levels above their reference range and by free T4 levels within theirs, but, as Abdelhai et al. (2016) put it, "irrespective of the presence or absence of symptoms".  So this definition is obviously misleading.

Model development: why validation is important and often at odds with customs

A model is an representation of a natural phenomenon, simplified to represent its most important parts, that is used to explain or predict that phenomenon's behavior.  There are two basic types of models in biology: 1) structural, in which several concrete entities and their physical relationships are represented, and 2) numerical, in which measurements are related to one another, as in computer simulations or statistical models.  Considering aspects of both yields the best results; should we, for instance, work with a structural model of a simple tube adjoining the thyroid and the heart?

There are customs of statistics that make sense in the abstract, but cause unnecessary confusion and disagreement in some cases. One is based on the belief that reference range limits for certain lab test results can be calculated very precisely if only they can be "transformed" (you might say, "tamed") into normal, i.e., Gaussian (bell curve) frequency distributions.  It is true that normal and log-normal distributions are frequent in nature, while others come close with some striking, albeit small, anomalies that resist being mathematically pounded into submission.  Applying simple general "transformation" approaches to these anomalous kinds of distributions at best wastes researcher time and at worst results in hard cases being made into bad (scientific) law.  What is sad is that simply looking at a plot of relevant data might be the best approach, but that is apparently against the rules.  This problem has rebounded on hypothyroidism sufferers, who are often deprived of treatment because of unresolved disagreements about where to set the top of the TSH reference range, thereby playing into the hands of those who want to return the TSH treatment threshold to 10.0 mIU/L from about 4.0 mIU/L.

Model development is an art to a large extent; there are no hard-and-fast rules about picking variables and weighting them.  In clinical research, many scientists "round up the usual suspects", e.g., age, sex, race, and so on.  But sometimes key variables particular to the study get overlooked.  For example, in a study to determine whether dietary eggs are good or bad, it would make sense to consider soft-boiled eggs separately from fried eggs because the latter are typically cooked with butter or vegetable oils.  Other variables might be the occurrence of an event, or even better, the time to an event; in this case, the duration of the study needs to be long enough and contain enough subjects for enough of these events to occur for a fair comparison to be done.  The variables in the model have to be well-defined so that different people do not have different perceptions of what they mean.  Variables involving subjective judgments need to be validated by a process that eliminates not just bias but misunderstanding.

Evidence-based medicine and what is deemed to be "low-value" healthcare

The original concept of evidence-based medicine has changed quite a bit since Sackett et al. (1996) described it as the "conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients" and as "integrating individual clinical expertise with the best available external clinical evidence from systematic research". They remove all ambiguity in their conclusion: "Good doctors use both individual clinical expertise and the best available external evidence."  However, today there seems to be a belief that validation via clinical trials should be the first step in determining what should be standard medical practice, even in the case of longstanding treatments that most patients believe that they benefit from.  In sum, medical care in the industrialized world is undergoing a complete makeover, for better or worse.

What does it take for a "low value" service to be identified and dropped from standard medical practice today?  Powers and Shrank (affiliated with Humana Inc., an insurance company) and Jain (affiliated with the SCAN Group and Health Plan and Stanford University) (2020), set forth three basic factors ("forces"): evidence, eminence, and economics.  The first form of evidence comes in is skeptical observation by physicians of its safety and efficacy in their practice when it is first used, while the second is "larger, more rigorous follow-up studies" if problems are missed initially.  Eminence is "broad" acceptance within the profession of the "inefficacy" of the service.  The third factor is economics, typically in play when an insurer refuses to cover such a service. 

Although some studies concluded that evidence supporting treating and screening for "subclinical" hypothyroidism with thyroid hormone replacement was weak, many in the field have considered existing evidence against it also to be weak; information about this "controversial" service abounds.  The "eminence" factor is harder to parse.  Should we take it to be reflected in part by position statements by professionals in the field, some by non-practicing physicians in positions of sometimes hard-to-understand authority, e.g., as "methodologists"?  How well they reflect the professional consensus is left open-ended.  If an individual in an apparent position of authority over a vaguely understood domain puts forth an opinion that diverges radically from the currently accepted one (if only informally) without a supporting argument, and published critical responses are few and half-hearted, what should the result on professional practice be?

Powers et al. (2020) give three examples of cases in which evidence, in the form of a clinical trial, showed that a service was ineffective, but "de-adoption" of the service was small.  Arthroscopic surgery for knee osteoarthritis continued at a slightly slower rate after the publication of its relevant clinical trial, and dropped twice as much proportionately when the Centers for Medicare and Medicaid Services (CMS) withdrew coverage, leading to a total reduction of 33%.  Another case was screening for Vitamin D; although their rationale for rating it "low-value" was vague, common sense would consider such screening unnecessary since this vitamin is readily available to the public; a statement by the American Society for Clinical Pathology had no effect.  On the other hand, Canada shut down Vitamin D screening by cutting off its government plan's reimbursement for it.  In a third case, involving the use of a surgical treatment for spinal compression fractures, surgeons were motivated to "de-adopt" their relevant work but not radiologists because of the different approaches of their professional societies.  They conclude by proposing that national governments invoke the economics "force" to make sure that "de-adopting" of "low-value" services happens when "eminence" does not follow "evidence". 

To be entirely fair, these insurance industry authors do not appear to support the deprivation of benign treatments in the examples of cases they present to support lowering standards of quality.  Their recommendations appear to be humane, and the doubt they express in their introduction about the value of screening women over the age of 65 for cervical cancer would probably resonate with many beleaguered patients who have never had a positive result and are tired of having time cut short for more pressing medical concerns!  But those whom they have trusted to implement their philosophy do not seem to share these values, and they apparently did not give enough consideration to taking the necessary measures to rein in abuses.

It is probably not surprising that insurance company executives would regard the existing consensus opinions of practicing physicians with some suspicion and impatience, and their point of view seems to be that "eminence" is mainly a stumbling block to progress.  It is true that in fee-for-service plans physicians might experience the temptation to overtreat patients, while it is in the financial interest of insurance companies to reduce covered services.  Indeed in the U.S., these recommended steps already curtail services covered by federal government insurance plans via the U.S. Preventive Services Task Force and the Affordable Care Act; it seems that the major beneficiaries in this situation are indeed the insurance companies and physicians on fixed salaries.

How are physicians, medical centers, and hospitals interpreting what science is saying?

One message is coming across clearly: screening for thyroid disease is, in general, something to be avoided.  Even those who are aware of that that there can be substantial risks associated with missing a hypothyroidism diagnosis still assert that screening is to be done only in unusual circumstances.  For instance, the Children's Hospital of Philadelphia (2020) states that (untreated) hypothyroidism in a pregnant woman can have "irreversible" effects on her fetus and, according to relevant studies, children of subjects had "lower IQ and impaired psychomotor (mental and motor) development".  They also acknowledge that there is "evidence" that Hashimoto's thyroiditis can cause "pregnancy loss".  Nevertheless, they imply that testing pregnant women for hypothyroidism should be avoided unless they are "at high risk of hypothyroidism", although they do not specify what constitutes high risk.  They do make a practice of determining whether pregnant women have had a hypothyroidism history or are being treated for it.  Are they concerned that they might miss a hypothyroidism diagnosis by not identifying high risk factors or by assuming that only patients with certain high risk factors can have hypothyroidism? 

But there is an alternative interpretation of this document, i.e., that this hospital is genuinely concerned about the risks run by the children of hypothyroid mothers, and is doing its best to get out the word despite its being required by U.S. executive branch bureaucratic rules to limit screening for hypothyroidism.  Is the physician's role shifting from professional to low-level bureaucrat?

How countries with socialized medicine are leading the charge in recent cutbacks in medical services in part by influencing science

Because governments administering socialized medicine are motivated to constrain costs while making some level of medical care available to all citizens regardless of their race, gender, sexual persuasion, or ability to pay, the logical move is to eliminate treatment for some conditions.  This means we are likely to see the most cutbacks in countries that are financially strapped (especially as shown by the quality of their responses to the COVID-19 pandemic).   Since scientific research is used as the basis for elimination of particular services, researchers from these countries are likely to produce the most publications examining the value of each service, and to set more stringent criteria for retaining that service.  This means in essence setting lower standards for public health in these countries, but what does it mean when more affluent countries adopt these lower standards, too, based on faith in the findings of publications in peer-reviewed journals?

Screening: when is it (not) necessary?

Screening tests are defined as those used to detect evidence of a disease in patients with no apparent clinical signs or symptoms of that particular disease.  What conditions patients should be screened for has undergone some controversy for decades, although higher cost-cutting priorities in countries using (and planning to use) socialized medicine have increased the pressure to eliminate screening for some of them.  These are some useful criteria: 1) the cost of the screening, 2) the danger of the screening, both in itself and in the tests it leads to according to current rules, 3) the difficulty of detecting the disease via clinical signs and symptoms, 4) the proven (or sometimes suspected) benefit of standard treatments, 5) the dangers of such treatments, especially when earlier detection does not imply milder treatments, 6) whether the patient is pregnant or is currently suffering from an already identified disease, and 7) the risk of making treatment errors, at least by current standards of such errors.  In addition, common diseases which most patients have not been educated about by the medical profession seem to be logical candidates.  Another consideration is whether the patient is generally healthy, without complaints, or arrives at the emergency room in great distress: perhaps the diagnostic net should be cast wider in the latter case.

Physician empathy: a key missing factor?

Intellectuals have -- on issues ranging across the spectrum from housing policies to laws governing organ transplants -- sought to have decision-making discretion taken from those directly involved, who have personal knowledge and a personal stake, and transferred to third parties who have neither, and who pay no price for being wrong.

Thomas Sowell (2010), Intellectuals and Society

How big a factor is empathy?  It is well-known that physicians' training subjects them to extreme physical stress that requires high energy levels over long periods of time. That stress is likely to weed out all but the healthiest, with the highest energy levels, and, by implication, with not just healthy thyroids, but unusually healthy ones.  Besides, we live in a culture that values high energy levels above nearly everything else in individuals.  Does this make it difficult for physicians to relate to patients in the bottom 5% of the population when it comes to energy level?  Does this predispose physicians to being skeptical about hypothyroid patients' accounts of struggles to get through the day, and even to keep jobs in which they have excellent track records and for which they are well-suited for in terms of intelligence, education, and work ethic, though not energy level? And what happens when physicians are encouraged to name diseases to eliminate screening and treatment for?  It is obviously in their self-interest to indicate those that they do not have and are unlikely to get, and might not even be real to them.

This might explain the notoriety that screening for thyroid disease has gained as an apparent key contributor to bloated healthcare costs.  Ganguli et al. (2021) reported that, contrary to current guidelines, electrocardiograms (ECGs), urinalyses and TSH ("thyrotropin") tests were still being being done at "wellness" visits for a sample of "75,000 fee-for-service Medicare beneficiaries".   In this group, only 8.7% had received TSH tests, which was still considered by the authors to be a major problem because such screening had not been entirely eradicated.  One has to wonder, however, whether this aggressive economizing attitude is carrying over to choices of the tests to perform on very sick patients arriving at emergency rooms.

Here are the supporting facts: the publications driving these policy changes, and knowledge about thyroid physiology:

Clinical Biochemists' Contribution: developing the third-generation TSH assay and the "subclinical" hypothyroidism concept

University of Essen (since 2003, the University of Duisburg-Essen) researchers Saller et al. (1998) published a landmark publication describing the third-generation TSH assay, which increased "functional sensitivity" to 0.01-0.02 mIU/L.  "Functional sensitivity" is a measure of an assay's precision (Armbruster & Pry, 2008).  They determined two very similar estimates of the TSH reference range from two products: "0.30-3.68 mU/l for the ACS:180 TSH-3 assay and as 0.36-3.64 mU/l for the Elecsys TSH assay".  These reference populations were chosen from "healthy" subjects with negligible levels of anti-thyroid antibodies.  These assays made it possible to distinguish overt hyperthyroidism and other conditions causing extremely low TSH values, such as severe non-thyroidal illness, from subclinical hyperthyroidism, theoretically eliminating one source of fear of treating "mild" hypothyroidism.

Kratzsch et al. (2005), affiliated with Institute of Laboratory Medicine, the Institute of Tranfusion Medicines, and Clinics for Nuclear Medicine at the University Hospital Leipzig in Germany, performed a landmark study meant to settle once and for all the normal range for TSH values, having observed that many laboratories derived normal subject populations from small groups of people with widely varying characteristics. They used a large sample of subjects, drawn from an "apparently healthy" blood donor population, and of those, 453 with no non-hormonal evidence of thyroid disease, i.e., anti-thyroid peroxidase antibodies (TPOAb) or thyroglobulin antibodies (TgQAb), a goiter, or a family history of "thyroid dysfunction", and not on any drug except for estrogens, were chosen from that group to form the normal subject population.  Of course, as blood donors they had to have met other health standards.  The TSH value range for this group was 0.12-5.29 mIU/L and that for the 2.5th to the 97.5th percentiles was 0.40-3.77 mIU/L.  They also determined ranges for T4, T3, free T3 (FT3), and free T4 (FT4) by the same method, all in molar units.  Some subset differences were especially interesting: women on oral contraceptives (n=108) had a median TSH level of 1.56 mIU/L, while the other women (n=66) had one of 1.29 mIU/L. The median TSH level for men was 1.35 mIU/L.

The graphs of TSH distribution frequencies presented in this paper make a striking point very clear: when properly measured, TSH levels above 5.0 mIU/L are very unusual in healthy people, and only a small minority have values greater than 2.0 mIU/L.  These graphs also suggest that a TSH level of 5.0 mIU/L or higher is diagnostic of autoimmune thyroid disease, i.e., Hashimoto's thyroiditis.  Of course, there is one problem that Kratzsch et al. (2005) could not solve: even if that paper has properly described the characteristics of the true normal population, there is no guarantee that future TSH assays for individuals will be performed properly.  Besides, labs continue to develop their own normal ("reference") patient populations and TSH normal ranges, and they continue to vary widely.

Clinical biochemists play a crucial role in diagnosing and even in treating thyroid disease, and theirs is arguably the one involving the biggest challenge.  Not all physicians are trained in endocrinology and can recognize the clinical signs of milder cases of hypothyroidism, which increases their dependence on lab tests analyzed by clinical biochemists for substantial guidance.  University of Cambridge (UK) clinical biochemists Koulori et al. (2013) explain the processes and relevant considerations that those in their profession make when performing these analyses.  To measure hormone concentrations in a blood sample, they perform immunoassays, in which specific antibodies are used to identify these hormones. These authors point out that it is important for clinical biochemists to make sure that other kinds of antibodies are kept out of the sample, because, depending on which ones they are, they can erroneously raise or lower the measured TSH concentration of that sample; mistakes along these lines could explain how some labs report chaotically varying TSH levels over time for individual patients. In contrast to what Biondi and Cooper (2008) apparently believe, Koulori et al. (2013) claim that clinical biochemists following proper procedures can ensure that the right antibodies are used in assays.

These authors state that "TSH has been recommended as a frontline screening tests for thyroid dysfunction, as relatively modest changes in TH concentrations are associated with marked excursions in TSH."  They then list seven conditions, including "non-thyroidal illness" that can lead to misleading interpretations of the TSH level; another is "TSH assay interference," which is described above.  They also lay the seven basic combinations high/low/normal of free T4 (FT4) and free T3 (FT3) relative to high/low/normal TSH and the possible interpretations of each.  The take-home message here: physicians still have to use their judgment in some cases even when assays are done correctly.

Spencer (2017) presents a very detailed discussion of the standard methods used in the assays to measure the levels of TSH, the thyroid hormones, and anti-thyroid antibodies. She, an authority in the field, affirms the primary role of TSH levels in diagnosing thyroid disease. According to a Cobas series assay (a Roche product) performed at the University of Southern California Clinical Laboratory, the reference range for TSH is 0.3-4.0 mIU/L for a sample of people less than 60 years of age and not pregnant.

Alexander et al. (2017) discuss the different reference ranges of 14 different TSH assay kits; these were substantial.  Upper reference range limits ranged from 2.15 mIU/L to 4.68 mIU/L for women in the first trimester of pregnancy.  This suggests a substantial accuracy problem remains, even though precision is good. 

The backlash and some alternative voices: Recent physicians' guidelines (recommendations and opinions) for screening, diagnosis and treatment of hypothyroidism

The best lack all conviction, while the worst are full of passionate intensity.

--from "The Second Coming" by William Butler Yeats

Levothyroxine (L-T4) is a prescription drug, not controlled under the Controlled Substances Act (CSA) (Drugs.com, 2019) and is the standard treatment for uncomplicated hypothyroidism.  Some patients might need levothyronine (L-T3) as well, although there is less agreement about that.

Below is a summary of the development of the movement against treating the majority of hypothyroidism cases, as clinical biochemists and endocrinologists (and later on, internists) parted company in their perceptions. As third-generation TSH assays became the norm, this challenged the traditional view of TSH distributions.  Although this changed the approaches some physicians took to treating hypothyroidism, and some influential endocrinologists were open to allowing this initially, there was a backlash on the part of some powerful figures in the field, who were suspicious of a paradigm shift that required treatment of more patients.  Resistance to treatment of "subclinical" hypothyroidism started about ten years later and today consists of what amounts to a movement to restore hypothyroidism treatment to what it was in about the mid-1980s or earlier.  However, there are dissenters who present a model of thyroid function that supports the view that "subclinical" hypothyroidism is neither mild nor inconsequential to health, and who question that the assumption that the combination of abnormally high TSH levels and abnormally low T4 levels reliably represents "real", uncomplicated, hypothyroidism rather than some degree of hypothyroidism complicated by (perhaps undiagnosed) disease in other parts of the body.

These publications rarely discuss L-T4 dose determination methods, which would seem to be closely related to the determination of TSH reference ranges and other important criteria.  It would stand to reason that patients with "mild" thyroid failure would need smaller doses than those whose disease were more advanced, and that patients burdened with diseases afflicting other parts of the body might need lower starting doses than those with uncomplicated hypothyroidism.  But a common apparent assumption here is those with TSH levels below 10.0 mIU/L should be given the lowest available dose, usually no more than 25 mcg daily.

Ladenson et al. (2000): American Thyroid Association guidelines before the backlash

These Johns Hopkins University specialists recommended screening for thyroid disease at age 35 and every five years afterward, and noted that women had a greater incidence of it than men.  Unfortunately, as was the case with so many scientific publications at the time, only the abstract is available to the general public.  It has probably been considered to have been superceded by later ATA guidelines, when the results of recent clinical trials apparently persuaded others in the field that treating most hypothyroid people was useless.

Hamilton et al. (2008): Setting the upper limit of the TSH reference range

These epidemiologists and biostatisticians associated with the Fred Hutchinson Cancer Research Center of the University of Washington developed a reference population based on participants in the Hanford Thyroid Disease Study, a retrospective observational study including 766 subjects that 1) had no clinical signs or symptoms of thyroid disease, 2) had no anti-thyroid antibodies, and 3) had no thyroid nodules as measured by ultrasonography though thyroid enlargement was not assessed, 4) had family history of thyroid disease, and 5) were not on any medications known to affect TSH levels.  Clinical biochemists acknowledged but not listed as coauthors analyzed hormone and antibody samples taken from these subjects.  They concluded that 4.1 mIU/L was at the 97.5th percentile, 3.0 mIU/L at the 90th percentile, and 2.5 mIU/L at the 80th percentile.  They observed a right-skewed log-normal TSH level distribution in this group, which assumes a larger proportional distance between the median and 97.5th percentile than in a normally distributed group. 

Biondi and Cooper (2008) challenge the validity of regarding "subclinical" hypothyroidism as a disease needing treatment

Biondi and Cooper (2008) initiated the movement against treatment of "subclinical" hypothyroidism (which they refer to as "SHypo") by claiming that there is "no consensus on the thyroid hormone and thyrotropin cutoff values at which treatment should be contemplated."  Stating that its definition is "purely a biochemical one," i.e., defined only by blood test results rather than by clinical symptoms and signs, they argued that this condition has no clinical significance.  They also argued that the supposedly inevitable variation in TSH assay results across different labs undermines the claim that TSH levels can be a meaningful method of measuring the degree of illness and make it difficult to establish a reference range for it.  They acknowledge that the presence of anti-TPO antibodies rises as TSH levels do in the NHANES III study and that in the Kratzsch study local iodine deficiency is inversely proportional to TSH levels, but they use this information to argue that because of this, a reference range for TSH alone has an uncertain meaning and suggest no remedy for what they see as a problem rather than an informative pattern. 

They attack the NHANES III study (Hollowell et al., 2002) for calculating the TSH reference range by defining the 2.5th and 97.5th percentiles for the reference population without the latter having a Gaussian ("normal") distribution, although visual inspection of this distribution as shown in Figure 1 of their paper looks very close to just that; even though the curve is very slightly right-skewed, that would tend to make the calculated reference range upper limit higher if the data transformation they implied took place.  They also attack the Kratzsch et al. (2005) study for not having achieved this distribution in their reference population for TSH values.  What caused this distribution shape issue was this population's distribution losing that shape after those subjects positive for anti-thyroid antibodies were excluded, producing a precipitous drop-off after about 2.0 mIU/L.  This is ironic because the "normal" group was so clearly delineated with this discontinuity, while the conventions for calculating the bounds of the reference range for lab parameters in general assume the shape of the reference group values to form a continuous curve, situation much less clear-cut than it was in the Kratzsch et al. study.

Biondi and Cooper summarize a great number of relevant studies, some indicating that health problems such as high blood pressure, unhealthy blood lipid levels, and heart disease are linked to TSH at the upper end of the reference range and just above it in some of those studies, whereas no differences were found in others.  The TSH reference range upper limit values calculated in these subclinical hypothyroidism studies were roughly 2.0 to 4.5 mIU/L, so there is some agreement in that literature about what needs to be treated, though maybe not at the level of precision that Biondi and Cooper apparently felt was necessary.  They also discuss problems with subclinical hyperthyroidism, which they say occurs in 20% of patients taking levothyroxine (L-T4) and has been linked to reductions in bone mineral density and atrial fibrillation in older patients and in post-menopausal women. However, they scruple to add that "Physician education can improve this situation, and inadvertent overtreatment should not be used as an argument against L-T4 replacement therapy in subjects with SHypo."

Gaitonde et al. (2012) interpret the NHANES III study (Hollowell et al., 2002) to mean that hypothyroidism is rare and recommend setting 10.0 mIU/L as the threshold TSH level for hypothyroidism treatment

The NHANES III study concluded that 4.6% of the population had hypothyroidism and that 0.3%, i.e., one in 300, had "overt" hypothyroidism, as defined by a TSH level greater than 4.5 mIU/L and a subnormal T4 level.  Gaitonde et al. (2012) simply say that one in 300 has hypothyroidism, citing that study as the source of that information, and recommend against using TSH levels (and apparently any other measure) as a method of screening for the disease.  Yet they also conclude that "Among patients with subclinical hypothyroidism, those at greater risk of progressing to clinical disease, and who may be considered for therapy, include patients with thyroid-stimulating hormone levels greater than 10 mIU per L and those who have elevated thyroid peroxidase antibody titers."  This might be the earliest publication representing "subclinical" hypothyroidism as being symptom-free and those with the condition not benefiting from treatment.  Unfortunately, Gaitonde et al. (2012) is widely cited as an authoritative source of hypothyroidism information, and Google's search algorithm has ranked it higher than any other hypothyroidism paper.

Garber et al. (2012) are the other early publication to suggest 10.0 mIU/L as the threshold TSH level for hypothyroidism treatment, but many important qualifying subtleties are missed by later studies

Although treatment with L-T4 alone as opposed to that in combination with L-thyronine (L-T3) is not covered in the recommendations, this issue is addressed earlier in the document (p. 1102): "L-thyroxine monotherapy has become the mainstay of treating hypothyroidism" because the body has ways of converting T4 to T3. 

Garber et al., (2012), a group of academic endocrinologists stating the policy of the American Association of Clinical Endocrinologists (AACE), took into consideration several large observational studies to formulate and evaluate hypothyroidism diagnosis and treatment recommendations.   They decided that the reference range to be used for TSH should be 0.45-4.12 mIU/L if local laboratory reference ranges were not available, although they acknowledged some studies that suggested that the top of the normal range be set as low as 2.5 mIU/L and showed respect for another study done by the laboratory chemists Kratzsch et al. (2005), who suggested 3.77 mIU/L for that upper limit. Treatment with levothyroxine should be considered if the TSH is between the top of the normal range (by default, 4.12 mIU/L) and 10 mIU/L and the patient reports certain symptoms, has anti-thyroid peroxidase antibodies, and/or has "atherosclerotic cardiovascular disease, heart failure," or risk factors for those conditions (presumably abnormal blood lipid levels); they assigned this recommendation (#16) a "B" grade (p. 1013).  A TSH level over 10 mIU/L should be a strong indicator for treatment, because of an "increased risk of heart failure or cardiovascular mortality" (p. 1012); this recommendation (#15) was also assigned a "B" grade.  According to Recommendation #17, pregnant women should get similar consideration if their TSH is as low as 2.5 mIU/L (first trimester), 3.0 mIU/L (second trimester) or 3.5 mIU/L (third trimester); this was assigned a "C" grade.   (These grades and associated best evidence levels, i.e., BEL, are explained in Table 3 on page 995.) In their conclusion, they stated that "The decision to treat subclinical hypothyroidism when the serum thyrotropin is less than 10 mIU/L should be tailored to the individual patient."  This soft-pedaled reference to the de facto raising of the upper limit back to 10.0 mIU/L was perhaps the opening of the floodgates; it is not clear how these guidelines were arrived at since their first listed reference, a 1995 document put out by the American Thyroid Association®, is not available to the general public.

They mentioned that Recommendation #16 was downgraded to a "B" from an "A" because relevant clinical trials had not been performed.

Garber et al. rejected some traditional diagnostic measures. Recommendation 5, assigned an "A" grade, says that "clinical scoring systems should not be used to diagnose hypothyroidism." They make an exception to this rule, however, in the case of "nonexperimental, clinically obvious, evidence" such as "myxedema coma" (Table 2).  Recommendation 6 (p. 1011), assigned a "B" grade, says that "clinical assessment of reflex relaxation time, cholesterol, and muscle enzymes should not be used to diagnose hypothyroidism."  However, they grant some apparent validity to "ankle reflex relaxation time," though they dismisses it as "a measure rarely used in current clinical practice" (p. 1011).

The discussion of the diagnostic value of anti-thyroid antibodies (TPO in particular) covers some very complex ground.  There is special concern about the demonstrated link between such antibodies in pregnant women with a treated Graves' disease history, or one of miscarriages, and Graves' disease in their babies, both before and after birth. 

And what about iodine deficiency? Recommendations 32.1 and 32.2 seem to be discouraging its treatment, but because of ambiguous wording, that seems unclear (p. 1016).  There is a clarifying explanation on page 1010 under the header "Excess iodine intake and hypothyroidism," which states much less ambiguously that they discourage treating hypothyroidism with iodine supplements.

They state that "The diagnosis of subclinical or overt hypothyroidism must be considered in every patient with depression" (p. 1008). However, they add that treatment of hypothyroidism does not necessarily cure any coexisting depression.

On p. 1009-10, they discuss why they recommended against using "clinical scoring systems": in the one study that covered this subject, there were considerable overlaps in the three groups studied, i.e., "euthyroid" subjects, "subclinical" subjects and those who were overtly hypothyroid, although there was a monotonic relationship between the percentage in the groups reporting at least four symptoms and the degree of disease. This set the stage for some misinterpretation, apparently: perhaps they meant that those criteria were not necessary or too difficult to design a method to evaluate, while others in the field might have seen this as proof that "subclinical" individuals did not suffer from their condition in any way.

They indicated that, for the most part, hypothyroidism could be treated by internists in most cases.  They provided a list of exceptional cases, including for these types of patients: 1) women who are either pregnant or planning pregnancy, children, and infants, 2) hard-to-manage hormone level cases, 3) those with difficult-to-interpret, i.e., contradictory, test results, and 4) those with certain non-thyroidal endocrine diseases.  The implication here is that test results are generally enough to establish the diagnosis.  Unfortunately, this list might have given some researchers the impression that symptoms do not count because they do not exist.

In general, Garber et al. (2012) offer a very nuanced perspective, giving practitioners a great deal of latitude in diagnosing and treating hypothyroidism.  Unfortunately, many in the field have interpreted this freedom to mean that there is flimsy scientific evidence to support the treatment of nearly all cases of hypothyroidism, and that that treatment should be withheld until more solid proof, especially that provided by clinical trials, was provided.

Garber et al. (p. 1008) discuss the issue of obesity relative to hypothyroidism, stating that the evidence for a relationship is weak, although some early studies were misleading.  They note that hypothyroidism suppresses the appetite, and offer this as an explanation.  They also mentioned that myxedema, an extreme form of hypothyroidism, is characterized by greatly increased fluid retention, which can add a great deal of weight, thereby mimicking obesity.

The U.S. Preventive Services Task Force: re diagnosis and treatment (2004 and 2015)

The U.S. Preventive Services Task Force (USPSTF) made its first recommendation in 2004 regarding (routine) screening for thyroid disease: "I", meaning that it did not have enough information to recommend for or against screening for thyroid disease in "asymptomatic" individuals.  The task force defined "subclinical hypothyroidism" as "an abnormal biochemical measurement of thyroid hormones without any specific clinical signs or symptoms of thyroid disease and no history of thyroid dysfunction or therapy" and added that "Individuals with symptoms of thyroid dysfunction, or those with a history of thyroid disease or treatment, are excluded from this definition and are not the subject of these recommendations."  The statement added that "Clinicians should be aware of subtle signs of thyroid dysfunction, particularly among those at high risk."  Therefore, it is fair to conclude that the later practice on the part of a number of researchers of defining "subclinical hypothyroidism" strictly in terms of lab test results did not draw support for that from this statement.

The USPSTF issued its final statement regarding screening for thyroid disease in March, 2015, reinforcing the previous "I" recommendation and stating that "current evidence is insufficient to assess the balance of benefits and harms of screening for thyroid dysfunction in nonpregnant, asymptomatic adults."  It refers to testing criteria for the diagnosis of "subclinical hypothyroidism" as being "arbitrarily" defined as involving a normal thyroxine level and TSH levels above 4.5 mIU/L, although it also makes a distinction between those with TSH levels below and above 10.0 mIU/L.  They conclude that widespread doubt of the reliability of TSH assays has "led many professional groups to recommend repeating thyroid function tests if the results fall above or below a specified reference interval for confirmation of persistent dysfunction (for example, over 3- to 6-month intervals) in asymptomatic persons before making a diagnosis or considering any treatment strategies, unless the serum TSH level is greater than 10.0 or less than 0.1 mIU/L."  On the other hand, they note that even "overt" hypothyroidism is characterized by "a set of relatively subtle and nonspecific clinical symptoms," if they appear at all.  What is interesting here is that the task force describes the 4.5 mIU/L limit as "arbitrarily defined", despite the major studies that Garber et al. (2012) refer to, but does not see that in the more traditionally determined 10.0 mIU/L limit.

A professor of medicine at the Harvard Medical School and the chief of the Division of Preventive Medicine at Brigham and Women's Hospital, JoAnn Manson, soon afterward expressed her disappointment, attributing their conclusion to "largely to an absence of clear guidance from randomized clinical trials as to the balance of benefits and risks of treatment of subclinical or preclinical disease" (Manson, 2015). She expressed support for "aggressive case-finding and targeting high-risk groups for thyroid dysfunction," consistent with the policies of the physicians' groups mentioned above and the hope that the USPSTF would eventually be able to provide such "clear guidance."

A recent USPSTF recommendation statement regarding depression screening (January 2016) presented new potential problems for those with undiagnosed but symptomatic thyroid disease because it assumes that depression 1) is not associated with any other diseases, 2) can be diagnosed reliably with a questionnaire alone (awarded a "B" grade, therefore recommended), and 3) should be treated only with antidepressants and/or cognitive-behavioral therapy. This puts undiagnosed hypothyroidism patients with depression symptoms at risk for being treated with antidepressants instead of thyroid hormone replacement. I have discussed this matter in detail elsewhere (Pugh, 2016).  NOTE: there is no conflict if depression symptoms observed in a patient are recognized as a justification for testing for thyroid disease.

According to a new report from the USPSTF (December 2016), thyroid screening was not listed among the "high-priority" preventive services which require further research.   However, studies questioning whether subclinical hypothyroidism should be treated have continued to be done.

Hoermann and Midgley (2012) enter the fray, but provide an enlightening graph of a TSH frequency distribution curve for a normal population

These researchers claim that the picture is much more complex and that relying on TSH assay results alone leads to many misdiagnoses.  They point out that TSH varies much less within individuals than across them, which implies that individuals have certain "set points".  This at least attests to the consistency of properly measured TSH levels.  They do display a frequency distribution graph of TSH levels, based on measurements of 150 "euthyroid" individuals illustrating how narrowly concentrated these values are over a small range, which rises sharply to near 1.0 mIU/L, where it peaks and drops to zero between 4.0 and 5.0 mIU/L  (Fig. 1). However, it is sharply skewed to the right, leveling off at a low level between 3.0 and 4.0 mIU/L, making a transformation to a Gaussian distribution difficult, even when the data are log-transformed.  This might explain why for so long there was a debate about whether the upper limit of the TSH reference range should be 2.5 or 4.0 mIU/L.  But the elephant in the room here is that 5.0 mIU/L is very unusual, and 10.0 mIU/L is obviously a strange place to draw the line between "normal" and "abnormal".These researchers cast great doubt on the value of TSH assay measurements by showing the poor correlation between TSH and free T4 levels in a population, as demonstrated by a previously published model (Hoermann et al., 2010).  They acknowledge that thyroid-binding globulin (TBG) has a significant influence on free T4 values, but argue that this poor relationship proves that TSH is a poor predictor of the levels of the hormone that really matters, i.e., free T4. 

Yet they fail to take into consideration that free T4 is not the last word, that later steps, i.e., the deiodination of T4 to produce T3, the active hormone, via enzymes in the kidney, liver, and other organs, as well as the hormonal feedback to the pituitary gland and to the hypothalamus, determine homeostasis, and perhaps the TSH level is the final measure of that.  Of course, when TSH output loses control, i.e., when the thyroid gland is unresponsive or overactive even when the pituitary is producing negligible TSH output, that is when the neat mathematical relationship breaks down and TSH levels go outside a properly determined reference range.  The main question that remains is whether the thyroid itself is diseased or being stressed, perhaps temporarily, by a disease affecting another part of the body.  A new model (Dietrich et al., 2012) took this into consideration and united structure and function.

Biondi (2013): Suggests that both the upper and lower limits of the TSH reference range for recommended treatment should be greatly raised, and that bone problems are bigger than CHD in current reference ranges

In 2013, the next step toward phasing out the treatment threshold TSH level of about 4.0 mIU/L was undertaken by Bernadette Biondi, Associate Professor at the Endocrine Division of the Department of Clinical Medicine, University of Naples Federico II Medical School. She begins by summarizing the findings of Garber et al. (2012) as setting the threshold TSH level for treatment at 10 mIU/L without mentioning that they had also described the subclinical range, i.e., from about 4.0 to 10.0 mIU/L, as a gray area involving patient-physician discussion. She characterized hypothyroidism and hyperthyroidism, as measured by TSH level, as causing a spectrum of effects, pointing out that even though cardiovascular heart disease, i.e., "diastolic dysfunction, dyslipidemia, and vascular alternations in young and middle-aged patients" incidence increased at TSH levels greater than 10 mIU/L, rates of atrial fibrillation and bone fractures increased in the lower part of the reference range.  She admits that "thyroid dysfunction may worsen the prognosis of such associated comorbidities as diabetes, kidney dysfunction, metabolic syndrome and heart failure", and recommends measuring TSH levels if heart failure or type-1 diabetes have already developed. She acknowledges that Taylor et al. (2013) conclude that L-T4 therapy had a "beneficial effect on atherogenic lipid profile and impaired vascular function in patients with TSH levels between 2.5 and 4.5 mIU/L" but claims that "overt and subclinical hyperthyroidism are associated with an increased risk of atrial fibrillation, heart failure, strike, coronary heart disease, and bone fractures, especially in patients with undetectable serum TSH levels (<0.1 mIU/L)" according to two 2012 studies from the same group of researchers, i.e., Gencer et al. and Collet et al., at the University of Lausanne, which had added a Faculty of Biology and Medicine in 2003.

On this basis, she challenged the focus put on treating what patients with what she termed "mildly increased serum TSH (4.5-9.9 mIU/L)," arguing that the hyperthyroid end of the spectrum was more dangerous, and likely to be triggered by supposedly unavoidable medication mistakes.  She also pointed out that upper limits of "normal serum TSH" varied with various racial and ethnic groups, and even more so with age, which she claimed called for establishing different treatment standards for these groups; most striking was her proposed normal range upper limit for those "older than 80 years," i.e., 7.5 mIU/L. She also attacks the Garber et al. (2012) calculation method of the reference range used to establish it at 0.45-4.12 mIU/L on the basis of its not being based on a Gaussian, i.e., normal, distribution; however visual inspection of the TSH level reference population distribution graph in the Hollowell et al. (2002) suggests that TSH population levels come very close to having just that distribution; Biondi makes much of the tiny tail above 5.0 mIU/L.  Another way of looking at these age-related numbers is to see them as indicators of how thyroid gland function degenerates with age, and as a reminder that "typical" and "normal" are not necessarily the same; after all, people over a certain age typically die, although those with good medical care tend to live longer than those who do not. It is also worth noting that the CDC (Hollowell et al., 2002) reported a much lower average value for the TSH levels of people in this age range in the healthy "reference group," suggesting that concomitant non-thyroidal disease in old people often drives up TSH levels by putting extra stress on their thyroids.

Pearce et al. (2013) recommend a moderate approach for subclinical hypothyroidism, leaving the door open to treatment

These mainly U.K. authors recommend a three-month trial on L-T4 for patients aged less than 70 with TSH levels between 4.0 and 10.0 mIU/L if hypothyroidism symptoms exist and a watch-and-wait approach if they do not.  Other factors given some consideration are goiter, dyslipidemia, and diabetes, as well as intent to pursue pregnancy.  Treatment of those older than 70 is discouraged; such patients are required to have a TSH level of at least 10.0 mIU/L to be treated.

Cappola (2013, updated 2017) reinforces the belief in setting the threshold level for hypothyrodism treatment at 10 mIU/L

Cappola mandates treatment of hypothyroidism when TSH levels reach "10 mU/L" (perhaps meaning "10.0 mIU/L") and offers a weaker recommendation for treatment of patients with TSH levels in the 7.0-9.9 mIU/L range.  On the other hand, she states that treatment of those with TSH levels of "5 and 6 mU/L" is "controversial" because those levels are common in people aged 70 and higher.  She bases these recommendations on "evidence from smaller trials and observational studies"; she lists nine references, none of which is the NHANES III study (Hollowell et al., 2002).  She also recommends multiple TSH tests over 1-3 months to rule out acute illness in the case of a test result in the "subclinical" range; it is not clear whether this should be done routinely or only when the patient is known to be ill, and whether treatment should be postponed until after the second test.

Jonklass et al. (2014) re treatment by levothyroxine, but only for "overt" cases

They affirmed levothyroxine (L-T4) as the standard therapy for primary hypothyroidism, and discussed factors to consider in administering the drug to patients that have already been diagnosed and determined to need this treatment.  They provide an extensive discussion of how to treat those with overt hypothyroidism; however, they say, "the topic of subclinical hypothyroidism (SCH) is not addressed, other than in the pediatric population, because of prior extensive reviews of this topic in adults."  Although they do not define "subclinical" hypothyroidism, they direct the reader to four references, which either deal with the issue of hypothyroidism in pregnancy or describe "subclinical" hypothyroidism as a mild condition which might not need to be treated at all. 

The British Thyroid Association Executive Committee (2015)

The British Thyroid Association Executive Committee (Okosieme et al., 2015) put out a statement about best practice for diagnosis and treating hypothyroidism, stating that L-T4 (levothyroxine) treatment worked well most of the time and alternative approaches should be considered in the minority of problem cases; much detail characterized their suggested method for investigating the causes of those problems. Although they mention a normal range of typically 0.4-4.0 mIU/L, they argue that it might have no clinical significance because of the nature of its derivation, and therefore tells us more about what characterizes unusually healthy people rather than about the level at which disease can be recognized, and imply that the 10.0 mIU/L level is a more meaningful dividing point.  They also state that evidence suggests that a high TSH and low T4 are associated with better longevity in people aged 65 and older.  Finally, they say that studies have shown no agreement on whether subclinical hypothyroidism symptoms and signs are useful in diagnosing the disease.

Dickens et al. (2017) recommend allowing pregnant women to receive treatment at lower levels if they are positive for anti-thyroid antibodies and make allowances for HCG effects

In 2017, Dickens et al. took an innovative approach in their "2017 Guidelines of the American Thyroid Association for the Diagnosis and Management of Thyroid Disease during Pregnancy and the Postpartum", making anti-thyroid antibodies a major factor in their recommendations.  Although the prose in this article is rather murky, they seem to be saying that the TSH reference range upper limit for levothyroxine treatment for pregnant women who are positive for TPO antibodies is 2.5 mIU/L, while the limit for those who are negative for antibodies should be 10 mIU/L.  They introduced the concept of the pregnancy-specific (reference) range (PSR), which takes into account the role of placenta-produced HCG in boosting thyroid hormone levels, thereby reducing TSH levels; the PSR is calculated by adjusting the upper limit of the regular TSH reference range downward by 0.5 mIU/L, and the lower limit by 0.4 mIU/L. 

On the other hand, for those women who were already receiving T4 treatment, Dickens et al. recommended increasing the dose by 20-30%.  They did not recommend for or against screening for thyroid disease in pregnant women, so it is not clear how the TSH in their cases would normally become known had they not been previously diagnosed.  The first two authors of this study, Dickens and Cifu, are internists (although Dickens is board-certified only in internal medicine, she lists her specialty as "Endocrinology" on her faculty page), while the third, Cohen, is an endocrinologist (board-certified in endocrinology, diabetes and metabolism, and internal medicine) whose main professional interest is thyroid cancer.  This professional background information was not provided by the paper, although their affiliations with the University of Chicago were given.

Alexander et al. (2017) explain the American Thyroid Association's new threshold TSH level for treatment of pregnant women

Pregnancy causes many endocrine changes that require a different approach to thyroid problems than that for nonpregnant patients. On one hand, the thyroid increases in size as thyroid hormone levels increase, while hCG, excreted by the placenta, performs a stimulating function similar to that of TSH, the levels of which are reduced accordingly in normal pregnant women. Therefore, TSH reference ranges for pregnant women have to be adjusted to a lower and narrower range (with the greater reduction at the high end of the range) than those for nonpregnant women and men to accommodate this difference. Because of this, screening pregnant women for thyroid disease without taking this into consideration can have dangerous consequences, i.e., they can miss some cases of hypothyroidism and incorrectly suspect hyperthyroidism in others. Free T4 levels present an even greater challenge to immunoassays because of the variations in T4-binding molecules such as TBG and albumin.

These authors are the only ones discussed in this essay to cover the differences among TSH assay methods (and associated products); unfortunately, their analysis seemed to be biased in favor of supporting higher values for the threshold TSH for treatment. They list the results for pregnant women's mainly first trimester TSH reference range for many methods (as reported by users in 14 publications from almost as many countries), and these differences were substantial. The lowest first trimester reference range was produced in an Australian study by Architect: 0.02-2.15 mIU/L, with a median of 0.74, using 1817 subjects (weeks 9-13). The highest (really, an outlier) was produced by a Russian study using the Abbott AxSYM: 0.20-4.68 mIU/L, with a median of 2.00 (second trimester). Median TSH values for the other reference ranges averaged a little over 1.0 mIU/L. Free T4 values were much higher than for the nonpregnant population, with medians clustering around 14.0 ng/dL. 

However, the authors chose to weight more greatly the studies showing higher TSH reference range upper limits, recommending 4.0 mIU/L for women in the first trimester of pregnancy, even though only three studies out of fourteen from all evaluated countries, i.e., Russia (4.67 mIU/L), the Netherlands (4.04 mIU/L, which actually covered weeks 8-18) and China (4.34 mIU/L) reported levels above that for the first trimester.  The other first eleven first trimester reference range upper limits ranged from 2.15 mIU/L to 3.67 mIU/L.  The authors did not make their reasoning clear for making this recommendation other than presenting these data.  Subject sample size did not explain it: the apparently heavily weighted Russian and Chinese studies had, respectively, 380 and 640 subjects, though the Netherlands study had 5186.   The biggest study, with 16,334 subjects, was Beswick et al., from the U.K., coming up with a TSH reference range upper limit of 3.50 mIU/L.

Udovcic et al. (2017) explain how thyroid hormones drive the muscle function and underlying biochemistry of the heart, and the conditions hypothyroidism leads to as a result

This is an in-depth description of the mechanisms causing reduced heart contractility and unhealthy blood lipid trends and the way they cause or aggravate heart disease.  Stimulation of the alpha-1 thyroid hormone receptors by T3 increases coronary blood flow and nitric oxide production.  On the other hand, T3 stimulates the renin-angiotensin-aldosterone system, which prevents hypertension.  Hypothyroidism, both subclinical and overt, creates two basic problems: 1) reduced heart output because of lower stroke volume and heart rate, and 2) heart muscle stiffness.  This can predispose patients to heart failure.  However, although the authors affirm the detrimental effects of hypothyroidism on heart function, they express less certainty about the benefits of treating these problems with thyroid hormone, especially in older patients, and caution that only clinical trials can determine this for sure. 

Bekkering et al. (2019) in position paper, set the TSH treatment threshold level at 20 mIU/L, although later in the paper they list many exemptions, including patients already being treated

The latest in clinical guidelines regarding "SCH", i.e., subclinical hypothyroidism, (Bekkering et al., 2019) took a last, extreme step in a position statement (not a systematic analysis), issuing "a strong recommendation against thyroid hormones in adults with SCH (elevated TSH levels and normal T4 (thyroxine) levels)." They add that this "does not apply to women who are trying to become pregnant or patients with TSH >20 mIU/L." Later in their paper, in the "Understanding the Recommendation" section, they added, "This recommendation may not apply to" patients reporting "severe symptoms," "very young adults (such as <= 30 years old)," "women at risk for unplanned pregnancy," and "patients who already take thyroid hormones" (my emphasis) but only because "The evidence presented here looked at the effect of starting medication and only indirectly informs stopping it." So it appears that those who have been previously diagnosed and are under a physician's care for SCH are eligible to be "grandfathered" into treatment, at least for now!  These guidelines were apparently mainly derived from the Feller et al. (2018) meta-analysis, of which the Stott et al. (2017) clinical trial was the most influential, but Bekkering et al. (2019) did not explain how they derived their guidelines from the references that they listed.

The unusual TSH level cutoff of 20 mIU/L might have been derived from the inclusion critera in the Stott et al. (2017) trial, i.e., ages 65 and older, with a TSH range of 4.60 to 19.99 mIU/L, although the authors pointed out in that paper that very few patients in their study had TSH levels greater than 10.0 mIU/L. As discussions of research papers later in this essay show, this criterion TSH level is so high (and not explicitly justified in any way) that it effectively makes T4 levels the sole criterion for hypothyroidism diagnosis. The Stott et al. (2017) study was by far the largest, with 737 subjects, all of whom were at least 65 years of age; the others involved only about 100 subjects or fewer. Of the 22 authors of the statement paper, only two were endocrinologists.  Bekkering's title is "Systematic review expert; Clinical guideline methodologist;" she is now employed by the Academic Centre for General Practice, Department of Public Health and Primary Care at the Katholieke Universiteit Leuven (Catholic University of Leuven), Belgium.

Protests lodged against the Bekkering et al. (2019) systematic review

In a 2019 press release the (British) Society for Endocrinology, together with the British Thyroid Association, responded critically in a press release to the Bekkering et al. (2019) paper, writing that they "disagree with the conclusion that most adults with SCH do not require treatment, as the guideline does not provide sufficient evidence, especially in younger individuals, to support this claim."  They also stated that "due to over-extrapolation from available data, most of which is from older patients with milder symptoms or from small-scale studies, the conclusions drawn are not justified."  This press release made reference to an unnamed meta-analysis that seems to be Feller et al. (2018), which heavily weighted the Stott et al study (2017), on which the Bekkering et al. (2019) recommendations were apparently based.  The press release concluded by emphasizing the importance of future large clinical trials involving symptomatic patients.  However, it did not question the recently accepted belief that patients with a TSH below 10.0 mIU/L do not need treatment.

Peter Taylor, of the Thyroid Research Group, Systems Immunity Research Institute, Cardiff University School of Medicine, and Antonio C. Bianco, of the Section of Endocrinology and Metabolism, University of Chicago, (2019) also protested against the Bekkering et al. (2019) paper in Nature Reviews Endocrinology, arguing that the heavily weighted Stott et al. (2017) study, which involved 34% of all the subjects in the 21 clinical trials reviewed and 75% of those whose quality of life was evaluated, included only subjects older than 65 years, with an average age of 74.4 years, who were generally asymptomatic at baseline. They did, however, concur with some other researchers that high TSH levels are natural in older people rather than a sign of the deteriorating health characterizing the last years of a long life. And, in their zeal to protest the 20 mIU/L threshold level proposed to (re)define those entitled to treatment, they seemed to offer a compromise by proposing drawing the line at 8 or 10 mIU/L, when they simply could have noted that the 20 mIU/L threshold seemed have been based on an arbitrary cut-off point in a single study, and that the authors' not stating the basis of their reasoning further weakened the validity of their stated opinion. This seems strange, since Taylor and Bianco initially claim that 10% of the population has "subclinical" hypothyroidism, while they recognize that only a very small proportion of the population has TSH values greater than that 10 mIU/L threshold level.

Taylor and Bianco (2019) also criticized the quality-of-life measures used in the remaining trial reviewed, spread over a total of only 423 subjects, for being underpowered and covering reports of "heterogeneous measures that assess general health status, and screening tools for minor psychiatric disorders." (The scientific basis for choosing these as measures of thyroid health is indeed not obvious.) They approved of the special caution shown with regard to pregnant women and those planning pregnancy, but also argued that unplanned pregnancies are necessarily part of the picture.

Biondi et al. (2019) set the bar high for treatment, especially for those 65+ years of age, but a little below 10.0 mIU/L, recommend viewing "subclinical" hypothyroidism simply as a risk factor

Biondi et al. (2019) have more nuanced recommendations within their paper, although they reflect a belief that levothyroxine administration is dangerous at TSH levels below that at which they have been proven to be beneficial for heading off two major causes of death.  Apparently on the basis of cutpoints that Gencer et al. (2012) used in their study, they recommended no levothyroxine (T4) treatment for patients over 65 of age with a TSH of less than 7.0 mIU/L, but recommended considering treatment for those with a TSH of 7.0 to 9.9 to "reduce risk of CHD mortality."  For patients <65 years and with a TSH in the 4.5-6.9 mIU/L range with other special conditions (multiple hypothyroidism symptoms, positive TPO antibodies, progressively increasing TSH levels, and a goiter) that treatment with levothyroxine (T4) be "considered."  For those younger patients with a TSH in the range of 7.0 and 9.9, they recommend T4 treatment to  "reduce risk of fatal stroke and coronary heart disease (CHD) mortality."  They also recommend that multiple measurements be done over a period of several months to ensure that the problem is long-term before initiating treatment.  However, they do not have special recommendations for pregnant women.  These guidelines are influenced by the apparent difficulty physicians have avoiding treatment errors, stating that "levothyroxine therapy may be associated with iatrogenic thyrotoxicosis, especially in elderly patients, and there is no evidence that it is beneficial in person aged 65 years or older." (my emphasis) Despite the distinctions made in these recommendations, the "Conclusions and Relevance" section states: "Subclinical hypothyroidism is common and most individuals can be observed without treatment. Treatment might be indicated for patients with subclinical hypothyroidism and serum thyrotropin levels of 10 mIU/L or higher or for young and middle-aged individuals with subclinical hypothyroidism and symptoms consistent with mild hypothyroidism." (my emphasis) Curiously, the corresponding author is not Biondi but Cooper, the third author, an endocrinologist at Johns Hopkins University.  In a nutshell, if Kratzsch et al. (2005) are right about TSH distributions in the "normal" population, this conclusion implies that treatment for hypothyroidism (unless perhaps complicated by an undiagnosed non-thyroidal illness) can be nearly eliminated, and the rest of the article suggests that, except in extraordinary cases, hypothyroidism is merely a risk factor for "real" diseases.

Peeters (2019) thinks the TSH treatment threshold at 10 mIU/L is minimal and perhaps should be higher and regards observational studies with skepticism

Peeters (2019) acknowledges that hypothyroidism has been linked in some studies to heart disease and pregnancy complications, such as miscarriage, but is skeptical that it is a primary disorder when anti-thyroid antibodies are below a certain level and suspects that those with TSH levels below "10-12" mIU/L would not feel better as a result of levothyroxine treatment, although he notes that patients in an observational study with lower levels than 10-12 mIU/L have been shown to have their risk of cardiovascular disease reduced with treatment.  However, he concludes with an often-sounded note by those who sit in judgment of these cases: "data are lacking from randomized, controlled trials to inform the effects of treatment of subclinical hypothyroidism on these long-term clinical outcomes" and that therefore the observational study results should be regarded with skepticism.  He also questions the validity of "surrogate markers of cardiac and vascular function" such as blood lipid levels as predictors of "cardiovascular events and death" although he did not take note of such studies which did not take into consideration whether the subjects were taking lipid-lowering medications such as statins.

Peeters also considers the relationship of TSH levels and measures of excess weight.  He cites the cross-sectional study, Kitahara et al. (2012), pointing out associations between TSH and both BMI and waist circumference.  But then he declares, without citing a supporting reference, "However, substantial weight loss typically results in a decrease in the thyrotropin level, which suggests that subclinical hypothyroidism is an unlikely cause of obesity."  He also lists obesity as a cause of a "persistent increase in the thyrotropin level combined with a normal free T4 level" in Table 1, titled "Causes of Elevated Thyrotropin Levels, Unrelated to Chronic Mild Thyroid Failure."  Peeters is a Professor of Internal Medicine at Erasmus University, has a Ph.D., and has a special interest in the treatment and diagnosis of thyroid disease.

Birtwhistle et al. (2019) recommend against screening for thyroid disease except for pregnant women

Birtwhistle et al. (2019), speaking for the Canadian Task Force on Preventive Health Care, also recommended against screening for thyroid disease, based on the contents of "systematic reviews", one of which was the 2015 USPSTF recommendation.  In their words, "In the judgment of the task force, given the lack of evidence of clinical effectiveness, financial costs of screening asymptomatic adults would represent an undesirable consequence for the health care system." Later in the paper, Birtwhistle et al. (2019) amend that description to exclude pregnant women, who do not need to be "asymptomatic."  However, Birtwhistle et al. (2019) stop short of recommending that the treatment of "subclinical" hypothyroidism end.

A study that played a key role in persuading Birtwhistle et al. (2019) to adopt this viewpoint (Reyes Domingo et al., 2019) concluded that there was no value in screening for thyroid disease, and added, "No studies reporting on the benefits or harms of treatment for subclinical hyperthyroidism or asymptomatic overt hypothyroidism or hyperthyroidism were found."  They found several clinical trials regarding adverse events associated with subclinical hypothyroidism that met their criteria, including the Stott et al. (2019) study; they found the Stott conclusions that there were no differences in mortality, tiredness and quality of life between the treatment and placebo groups to be of "moderate" quality; all others were found to be of low quality.  Reyes Domingo et al. (2019) did state that higher quality studies were needed to make a strong recommendation.  Several other systematic reviews said about the the same thing.

Hoermann et al. (2019) and Midgley et al. (2019) acknowledge the controversy and advocate treatment on a case-by-case basis

German and British researchers Hoermann et al. (2019) express concern in a literature review about the apparent rush to judgment made by those in the field, claiming that studies concluding that treatment of "subclinical" hypothyroidism is risky and impractical should not be hastily applied to individual cases because of the danger to the affected patients.  They point out that the endocrinology involved is very complex and that it is appropriate to supplement LT4 medication with that of LT3 in some cases.  They are reassuring that LT4 is free of side effects if given in the proper dose, which is not difficult to achieve, and that overdoses are easily repaired by withholding the medication for a few days, unless the patient is in imminent danger of having a heart attack or has unusual adrenal gland problems.  They also point out that different LT4 brands often have different potency, which in turn might call for different doses.  In essence, they advocate against letting fear hold a physician back from treating a patient who has a lesser degree of hypothyroidism; some degree of experimentation with LT4 (and maybe supplemental LT3) doses is sometimes necessary, given the weak evidence supporting any particular regimen.  In this way, patients can have input into how they are treated.  The authors acknowledge that there are substantial individual differences in how the disease manifests itself clinically, although they, too, do not discuss the possibility that this is most likely because of the disease's effect on the entire body, some parts of which might already have more problems that those of others.  At any rate, however, their paper is notable for its rationality, humility, and compassion, so sparse in other publications covering the same subject.

Midgley et al. (2019), who were most of the same researchers, acknowledge the pushback against treatment of "subclinical" hypothyroidism, which they believe has occurred because advances in hypothyroidism diagnosis and treatment came rapidly and very late relative to that for other diseases.  They attribute this to the influence of the TSH assay, which generated many new diagnoses in patients previously assumed to be normal as its quality improved and reference ranges narrowed accordingly.  In response, they observe, recent guidelines brought the TSH treatment threshold level back up and in fact the new "subclinical" designation, i.e., defined by an abnormally high TSH but normal free T3 and free T4, was instituted to distinguish between those who needed to be treated and those who did not, greatly reducing the numbers of those eligible for such treatment.  Physicians went from recognizing out-of-reference-range TSH values as an indication for treatment to regarding all values in the reference range as acceptable treatment goals while setting the threshold TSH value for treatment at 10.0 mIU/L; this point of view was indeed reflected in the conclusions expressed in some recently published clinical trials.  These authors claimed that there is an orderly relationship between TSH and free T4 levels that has a different "set point" for different individuals and that both should be considered in diagnosis and treatment, and advocated testing this model in new clinical trials.  Does "set point" mean a new normal?  This raises troubling implications.

Calsolaro et al. (2019) add to the voices of moderation regarding treatment policy for the elderly

Italian researchers Calsolaro et al. (2019) cautiously question the rigid and simplistic accepted standards for treating the elderly, as well as the prevalent assumption that increasing TSH values are a normal part of aging.  They recommend analysis of each case, taking into consideration "age-dependent TSH cutoffs, thyroid autoimmunity, the burden of comorbidities, and the possible presence of frailty."  They report 90% of older people with subclinical hypothyroidism have Hashimoto's thyroiditis, based on assays of anti-thyroid antibodies.  They are concerned about the wide differences in published estimates of the proportion of the population with subclinical hypothyroidism, and discuss the factors influencing that, including increasing treatment of subclinical hypothyroidism, some of which is "suboptimal."  They note, in fact, that TSH and free T3 are low in older patients with "acute and systemic diseases."  They present a graph (Figure 1) which shows that, even in the very old, TSH levels above 7.0 mIU/L are quite rare.  They note that, according to the Framingham study, women with TSH levels less than 1.0 mIU/L and those with TSH levels greater than 2.1 mIU/L were more likely to get Alzheimer's dementia than those with TSH levels between those two values. They recommend treatment of those patients aged 65-75 years if they are "fit" rather than "frail" and meet several other conditions, including anti-thyroid antibodies and concomitant non-thyroid illnesses.  They recommend against treating patients older than 75 unless their TSH levels are at least 10.0 mIU/L (represented as "10 mIU/L" in the paper, but seem to mean the extra significant figure.)

The official U.S. position on screening for thyroid disease

The U.S. government recommends against screening for thyroid disease in adults, even in pregnant women (U.S. Centers for Medicare and Medicaid Services, 2019a, 2019b).

Presentation by Hoenderkamp (2020)

Dr. Hoenderkamp, a G.P. in the U.K., gave a talk titled "Management of Hypothyroidism in Primary Care".  According to her, the distinction between "hypothyroidism" and "subclinical hypothyroidism" is low T4 levels in the former and constitutes a yawning divide.  She has been given special attention because she has (treated overt) hypothyroidism as a result of the surgical removal of her thyroid; as expected, she goes into great detail describing the treatment of this condition in the eleven minutes she devotes to it.  In vivid contrast, however, is the dismissive attention she gives to the diagnosis and treatment of "subclinical hypothyroidism".  She excludes patients aged over 65 with TSH levels under "10 mIU/L" from treatment altogether, and those with higher levels must demonstrate these on two tests at least three months apart.  Younger patients may qualify if their TSH levels (again, two at least three months apart) are above the reference range.  However, if their "symptoms persist" after their TSH levels are brought (presumably anywhere) into the reference range, she says to "consider stopping levothyroxine" and to simply monitor their condition.  Of the four minutes that she takes to discuss "subclinical" hypothyroidism, two are devoted to that monitoring process.  In general, she reassures us that "subclinical hypothyroidism" is a very minor condition, typically without symptoms and not needing treatment, and which very many patients have.

Dr. Hoenderkamp mentioned that, for her, the ideal TSH level was 2.5 mIU/L.


A policy crisis: escalating price of L-T3 in the U.K.

Taylor et al. (2019) reported that L-T3's 28-day price had risen from £4.50 to £258.19 in 2016, and that the number of L-T3 prescriptions dropped from about 2.3 to about 1.5 per 1000 L-T4 prescriptions from 2016 to 2018, and that there was a great variation across clinical commissioning groups (CCGs).  They expressed concern that these wide differences might have been explained by more affluent areas having greater access to private prescribers.

The real question: what does the thyroid do and how does it affect us in health and in disease?

Ask most people what they think the thyroid does, and (if they know anything about it) they are most likely to say it provides energy and that fatigue is the most common sign of a problem. Of course, there are also miscellaneous signs and symptoms that do not seem to be related to one another, e.g., weight gain, cold intolerance, loss of scalp hair and parts of the eyebrows, facial puffiness, thick and dry skin, slow and labored speech, depressed mood and cognitive problems. Trying to identify these patterns in an individual just by referring to a book can be very difficult for someone without related medical training.

Another typical explanation is that the thyroid regulates "metabolism" or "the rate of metabolism." The general impression is that thyroid hormones speed things up, burn calories faster, make thoughts flow faster and of course keep you warm. But these hormones are hardly unique in producing any single one of these effects.

What makes thyroid disease even more difficult to evaluate is that so many hormones and body systems are involved in making sure that this aspect of health is maintained, making definitive diagnosis of the disease  Generally speaking, overt hypothyroidism is diagnosed when TSH levels are elevated, i.e., above the normal range, and T4 and/or T3 levels are below the normal range. Subclinical hypothyroidism (SCH) is a term that suggests an absence of clinical signs and symptoms in spite of abnormal TSH blood test results but is in fact defined by many researchers to describe a particular range of (abnormally high) TSH values diagnostic of hypothyroidism, accompanied by normal T4 and T3 values, without regard to anything clinically observed, with the implication that this condition typically does not occasion treatment; in fact, those being treated with levothyroxine (T4) replacement therapy on the basis of previous "subclinical" lab results and symptoms are typically excluded from clinical trials testing the efficacy of levothyroxine treatment on subclinical hypothyroid subjects.  Its research and individual medical meaning may vary: although the practicing physician Wu (2000) includes clinical signs and symptoms suggesting thyroid disease in the definition of this condition, researchers in general apparently do not. Criterion TSH values vary widely, and there is a good deal of controversy about its medical importance and whether it should be treated.

In fact, thyroid hormones influence (both thermal and chemical) energy production and body tissue maintenance, and their dropping off (via TSH and melatonin) induce sleepiness in individuals running low on energy (which we do at the end of each day). This essay will attempt to explain what cellular processes the thyroid hormones influence, how the body produces these hormones and how problems in these processes manifest themselves as thyroid disease.

What are the scientific underpinnings of this sea change in recommendations from scientists and bureaucrats? A review of relevant scientific concepts and practices follows; to skip to discussions of the recent studies affecting current treatment policy, follow this link.

Explanation of some basic scientific terms and practices: a crash course

NOTE: You do not have to read this section to understand the major points that this paper makes. However, you do have to understand the concepts listed here to read much of it critically.


Metabolism is typically defined as all the chemical reactions going on in the body. In the context of thyroid physiology, this boils down to 1) growth and repair of tissue and 2) energy production. A metabolic pathway is a chain of such reactions leading to results that support life (ideally).

Hormones and neurotransmitters

Hormones are a structurally diverse set of chemical messengers that travel through the blood; only very small amounts of each are necessary to trigger a cascade of metabolic activity, which they do by "docking" with receptors in target cells. Endocrine glands, such as the thyroid, produce some hormones and send them into the blood, in which they can travel anywhere in the body.

In the system controlling thyroid activity, the hypothalamus, a part of the brain, acts a switchboard, receiving input from appetite-controlling hormones located in the stomach, intestines and fat cells and producing hormones that influence the rate of production and destruction of cells throughout the body in concert with available nutrients.

Neurotransmitters (first messengers) link to a receptor in the cell membrane, which activates a second messenger inside the cell  in the setting of this paper, the diacylglycerol metabolic pathway, which leads to the passage of calcium ions through the cell membrane, much as occurs in nervous system function. In contrast to "pure" endocrine hormones, they have a very local function, affecting a series of adjacent cells.   In plain English: picture the neurotransmitter as a person knocking at a door (the cell membrane). Someone inside (the second messenger) accepts the message without opening the door, and goes away to deliver the message to those inside the building, and to those in neighboring buildings. In this case, the message allows other people (calcium ions) to enter those buildings.

However, recent evidence suggests that estradiol, an estrogen, acts as a neurotransmitter in the hypothalamus (Qiu et al., 2003).

Of course, the endocrine hormones, produced by endocrine glands, i.e., TSH, T3 and T4, are discussed here, as well as those in the growth hormone cascade. Several other hormones are produced by tissue cells, i.e., the stomach, intestines, ovaries, pancreas beta cells and fat cells. Estradiol, an estrogen, is examined in many studies referred to below; researchers often refer to it as "E2." These hormones also travel through the blood to their destinations. These relationships are shown in Fig. 1.

Iodine, other halogens and molecules with similar behavior

Iodine (I), bromine (Br), chlorine (Cl) and fluorine (F) (as well as astatine) are members of the halogen group on the periodic table, which have seven electrons in their valence, or outer electron shell; as a result, their chemical behavior is very similar and they have a strong tendency to compete with each other in bonding. The maximum number of electrons in this valence is eight; when the valence is filled, i.e., all eight electrons are present, the particular atom or ion (its state when the halogen takes an electron from another atom) has its lowest tendency to react, i.e., electronegativity. On the other hand, halogens in their elemental rather than ionic form have the highest electronegativity of all groups and as a result are rarely found in nature. Halogens usually exist as ions, known as halides, e.g., I-, or in molecular form, either as bonded pairs of the same halogen, i.e., I2, or bonded with other atoms.  When researchers refer to "inorganic iodine," they usually mean I- (iodide ions) and/or I2 (molecular iodine). "Organic iodine" typically describes iodine atoms that are constituents of hormones and enzymes.

A halide, which is a halogen with a filled valence, i.e., the electrons in its outer shell, has a charge of -1, i.e., its number of protons minus its number of electrons, which causes it to be attracted to ions with a +1 charge by this same arithmetic.  In this respect, it also behaves similarly to ions with the same -1 charge, such as perchlorate ions (ClO4-).  The significance of all of this is that iodide ions can be displaced by certain other such ions, which is rarely if ever desirable and sometimes problematic.

The larger atoms in the same group, in this case, halogens, hold electrons in their valences with less force than smaller atoms because their electrons are farther from their nuclei. Therefore, an iodide ion, for instance, is more likely to be detached from the molecule it is bonded to than are fluoride, chloride or bromide ions. This implies that iodine is readily displaced in molecules such as T4 by these smaller halogens.

Iodide and chloride ions are considered to be essential nutrients; bromide ions are not.  Fluoride ions are not true nutrients, and though their use in public water supplies to prevent cavities is generally accepted, there is some controversy. Iodide ions are highly water-soluble, and they are the most common form that iodine takes in the environment and in the body. Most of the iodine in the body is used in the thyroid gland, but it is also found in other parts of the body.

Water-soluble vs. fat-soluble molecules (or polar vs. nonpolar molecules)

You've heard the old saying, "Oil and water don't mix." What makes certain types of molecules draw closer to one another (the essence of mixing) and others keep their distance? The difference is the presence or absence of mild electrical charges in these molecules, which act sort of like magnetism. Oils, in fact the lipids of which they are a subset, are nonpolar, i.e., they do not have any such electrical charge, so they can lie right next to one another without interference. On the other hand, water is an example of a polar molecule. It carries a small charge because it has a relatively large molecule (oxygen) that pulls electrons a little distance away from the nuclei of the hydrogen atoms that it is bonded to; as a result, the oxygen atom has a slightly negative charge and the hydrogen atoms each a slightly positive charge. So the oxygen atom in one water molecule is attracted to one of the hydrogen atoms in a neighboring water molecule. This is what gives water its surface tension.

The scientific literature typically refers to water-soluble (polar) substances as hydrophilic and fat-soluble (nonpolar) substances as hydrophobic.

Nonpolar molecules easily pass through cell membranes into cells (passive transport) while polar molecules need the help of a transporter molecule to provide active transport into the cell.

Not all types of molecules are completely hydrophilic or hydrophobic; some have a hydrophilic part and a hydrophobic part. Tyrosine is such a case: its phenyl group, i.e., six-carbon ring, is hydrophobic, while its side chain (-OH) is hydrophilic. This means that it can show both types of behavior, apparently: it can be carried along by water molecules attracting its hydrophilic side chain, but the hydrophobic phenyl group allows it to pass through the cell membrane of the thyrocyte through passive transport.

The cell cycle: how the body grows

Living cells each are in one of three basic states at any given time: growth (increase in size), division into two cells (mitosis) and rest.  The cell cycle begins with growth (G1 and S) and ends with mitosis. Growth takes place in two stages, the first (G1) before DNA synthesis (S) and the second (G2). The many steps involved in this cycle are regulated by cyclins, which in turn bind with the enzymes cyclin-dependent kinases (CDK) to form cyclin-dependent kinase complexes, determining which substrates these enzymes will react with. They regulate the cell cycle by determining whether the cell is progressing through the cycle properly, either by giving go-ahead signals if so or by causing programmed destruction (apoptosis) if the process goes wrong. Insulin-like Growth Factor 1 (IGF-1) stimulates the (gene) expression of cyclins D1 and E; cyclin E regulates the G1-to-S phase transition in the cell while D1 regulates G1 (Mairet-Coello et al., 2009, Borowiec et al., 2011). Cyclin A2, on the other hand, responds to DNA damage of the cell in either of two ways based on its timing: 1) before gene transcription/DNA synthesis, to trigger apoptosis, or 2) after gene transcription/DNA synthesis, to repair the DNA damage instead (Finkielstein et al., 2002).

The take-home message here: IGF-1 stimulates cell repair and proliferation, but not apoptosis. (There are certain fail-safe body mechanisms in healthy people that keep IGF-1 from causing cancer, but this paper will not explore this.)  It works closely with thyroid hormones, as shown below.

Transport of substances into the cell

There are two basic ways a small molecule or ion can get into the cell by passing through the cell membrane: 1) by passive transport, i.e., done only by physical rather than chemical energy and 2) by active transport, which involves the use of chemical energy. Fat-soluble (actually, lipid-soluble) molecules, i.e., those with no electrical charge ("nonpolar"), can enter through passive transport. But the more asymmetrical water-soluble ("polar") molecules and ions need active transport to be pulled through the cell membrane, and this is done with a transmembrane protein (ion transporter) which uses electrochemical energy to perform active transport on the ions. 

In the case of a symporter, which is a type of transporter which brings in two types of ions, the protein's receptors capture one type of ion, pulling these inside the cell, creating an electrical charge which pulls the other ion with it. Other ions which can be pulled into the cell in place of another by the symporter are called competitive inhibitors. The relative strength of this pulling force cannot be easily predicted by an ion's structure, although each symporter seems to operate only on pairs of ions with the same electrical charge.

Very large molecules, such as many proteins, enter and exit cells through endocytosis and exocytosis, which are wonders of topology. As they push up against the cell membrane of a cell, the membrane folds inward and that part breaks away from the cell's membrane to enclose the protein. So in a sense the protein (or other large molecule) remains outside the cell while being enclosed in it; this enclosure is known as a vesicle.  Of course, this limits the types of interactions this large molecule can have with the cell.

The development of the TSH assay and associated reference range changes

The studies discussed here involve tests for concentrations of hormones (especially TSH, T4 and T3) and other substances in the blood and in the tissues.  The concentrations of very small molecules are measured via radioimmunoassay (RIA) or by enzyme-linked immunosorbent assay (ELISA), a new non-radioactive alternative.  These assays have increased steadily greatly in precision since the 1950s, affecting methods used to determine the limits of normal ranges for these hormones greatly, and therefore the values of these limits.  Because the lower limits of normal ranges are smaller than their upper limits, they have required greater precision on the part of these assays to be measured meaningfully, and have been established later than the upper limits.  These upper limits have decreased in part because the margin of error has been reduced, making overdosing with replacement hormone based on measurement errors less likely.

The TSH assay went through three generations as it was refined over several decades, which in turn modified its reference range.  According to Hennessey and Espaillat (2015), the upper limit of this range was originally set at 10 mIU/L when the first generation assay was used.  The first second-generation assay that they cover set this upper limit at 6.0 mIU/L in a 1977 publication; a 1979 publication set it at between 5.0 mIU/L and 10.0 mIU/L.   This upper limit wavered between those two limits until 1988, when it began to settle below 6.0 mIU/L, and finally reached 4.5 mIU/L in 2013.  They did not mention the landmark 2005 Kratzsch study, however, which set the top of the reference range at 3.77 mIU/L.

Hennessey and Espaillat also present reference ranges and medians for eight difference age groups as measured by four different studies.  Lower limits for each are about the same, while medians climb gradually from young adulthood (about 1.25 mIU/L) to old age.  For age range 70-79, the median TSH levels were 1.76, 1.74, 1.66, and 1.56 mIU/L.  Only in age ranges of 80 and over (mostly done by one study) do median TSH levels go above 2.0 mIU/L.  But as age goes over 60 years, the curve is increasingly skewed to the right, with increasingly sharply reference range upper limit increases.  This calls into question 1) whether some older people in the reference population had undiagnosed diseases affecting other parts of the body, since they can drive up TSH levels, and 2) whether a TSH level of 5.0 mIU/L is really normal and healthy for older people.

Hennessey and Espaillat also discuss the predictability of progression to (overt) "hypothyroidism" based on TSH levels.  Their definition of this condition, as opposed to "subclinical hypothyroidism", is a TSH level of 10 mIU/L and of course free T4 levels in the reference range.  So it appears that the old TSH benchmark from first-generation TSH assay days is still very much in the picture.

How cellular data are obtained and analyzed in the laboratory

On the other hand, detection of and measurement of the levels of proteins, especially in tissue cells, involves another approach that draws on epigenetics, the study of gene expression, i.e., the "turning on" or "turning off" certain genes as the result of chemical triggers.  Certain molecules known as transcription factors (in the case of this essay, mainly hormones) cause this turning on by binding to a response element, i.e., an expression target segment of DNA of a molecule, creating messenger RNA, which in turns signals ribosomes to create characteristic proteins known as peptides, strings of amino acids which form tissue, enzymes and various other molecules necessary for life processes.  Transcription factors "unzip" DNA at particular points, allowing nearby matching nucleotides to join those of the DNA, forming the messenger RNA. Isolating and identifying this messenger RNA is the key to determining which proteins are created within the cell.  Chromosomes in some individuals have more copies of a particular gene than those of other individuals, making it possible to produce more of those genes' corresponding proteins.  Research papers refer to "turning on" some or all copies of genes as up-regulating those genes and to "turning off" gene copies as down-regulating that gene.

Inflammation and cytokines, and their experimental stimulation

Inflammation is the body's healing process in response to infection or injury. Inflammation consists of two basic functions: 1) repair of damaged tissue, involving blood clotting and tissue regeneration and 2) identification of and destruction of non-nutrient foreign molecules and damaged body cells.  The latter consists of the immune system, composed of two basic parts: antibodies (sometimes called B-cells) and T-cells (a type of lymphocyte, in turn a type of white blood cell) and of macrophages, drawn to the scene of attack by chemotaxis (as though responding to an irresistable smell) and which "digest" unwanted molecules. Antibodies identify problem substances, usually foreign matter and body cells that have been damaged too much; they do this by examining the shape of molecules and comparing that to their library of acceptable shapes. If the molecule "flunks" the test, the antibody marks it for destruction and T-cells it comes in contact with destroy marked cells by breaching their membranes.

This process is mediated by cytokines, chemical messengers produced by T cells and various other cells that allow the immune system to communicate with the endocrine and nervous systems; the introduction of antigens (large non-nutrient foreign molecules, usually proteins or polysaccharides) trigger their appearance via interaction with antibodies.  Shape recognition (correct or not!) by "blind" antibodies is the basis of this process: antibodies match a particular part of the surface of an antigen called an epitope with a "memorized" shape, and mark the antigen for attack.  Perhaps because this decision is made on the basis of such a small area, error is more likely with larger molecules.  Two large molecules performing key functions inside the thyroid are typical targets for antibody attacks that can lead to hypothyroidism: thyroid peroxidase (TPO) and thyroglobulin (Tg).  On the other hand, Grave's disease is caused by an antibody which mimics TSH, increasing the stimulation of TSH receptors, in turn causing hyperthyroidism.

A hapten is a small molecule that tricks antibodies into making the wrong judgment about whether to attack a particular molecule; it acts by attaching itself to the surface of such a molecule, altering its apparent shape. Haptens are designed and used by experimenters to trigger autoimmune responses in (usually animal) subjects, although they exist in nature, too.

Lipopolysaccharide (LPS), a large molecule that is found on the exterior surface of gram-negative bacteria cell membranes, is commonly used experimentally in animal studies as an antigen to stimulate the creation of cytokines, triggering an immune system response.

Freund's adjuvant, which comes in two forms, is used for a similar purpose; its complete form contains dried, i.e., killed, mycobacteria in a water-and-oil emulsion, while the incomplete form just contains the water and oil. It acts as a hapten; experimenters use it to trigger an autoimmune response in animal subjects to large molecules such as thyroglobulin normally found in these animals' bodies.

Erythrocyte sedimentation rate (ESR) is a blood test used to measure the amount of inflammation in the body. It is a low-budget but less effective alternative to the radioimmunoassay (RIA) as a measure of anti-thyroid antibody presence.

Sometimes the inflammation process can go wrong. Sometimes it can choose to repair when attack is appropriate, as in the case of cancer or atherosclerosis.  When the opposite occurs, it is called autoimmunity.  Sometimes, apparently, both problems can occur together and this might even be typical. Could it be that parts of the body that persistently need repair somehow eventually draw the attack response instead?  If so, could thyroid tissue severely damaged by either too much or too little iodine, draw this attack response by having its shape, as detected by antibodies, change?  And are there more haptens in the environment today because of contamination of food by new, industrially produced chemicals? On the other hand, can autoimmune disease be brought on by infection in neighboring tissue, or even in other parts of the body, by mistakes the immune system makes in fighting that infection?  Can a throat infection cause thyroid tissue be targeted for attack? Is thyroid tissue especially vulnerable to antibody attack, especially its chemicals, TPO and Tg, and if so why?  We do know that large, complex molecules are mostly likely to become antibody targets.

Deep reasoning vs. shallow reasoning: appropriate in different situations

Deep reasoning is used to understand all relevant cause-effect relationships and is essential for the design of complex fail-safe systems.  Shallow reasoning, on the other hand, uses pattern recognition to identify familiar situations by their superficial manifestations. To make a crude generalization, deep reasoning is important to science, while shallow reasoning's efficiency makes it very useful in routine practice.

Negative and positive feedback systems and homeostasis

A positive feedback system is basically a vicious spiral: every stimulus drives the system more out of whack.  A negative feedback system is a self-correcting one; when something goes wrong, the system responds by reversing the effects of any stimulus. Homeostasis is a state of physiological stability necessary for health, maintained by a negative feedback system: it is manifested, for instance, by blood test values that stay inside their respective normal ranges with no overall upward or downward trend.

Significant figures

Our measurement instruments are inevitably limited in their precision, so we need a shorthand to indicate the limits of their measurement.  This is indicated in integers by the number of nonzero digits in a number representing that measurement.  Fractions expressed in decimal form follow a slightly different rule: for example, "5.5," which has two significant figures, indicates that the value it represents to can be anywhere from exactly 5 to exactly 6. On the other hand, "10" has only one significant figure and represents a value known to be somewhere between exactly 5 and exactly 15, while "10.0" has two significant figures.

Statistical models

A model is a usually simplified representation of something in the real world, retaining only its most important features and showing their relationships. It can be static or dynamic. A mathematical model can be a simple formula, e.g., F = ma (force = mass x acceleration); some people refer to that type of model as "plug and crank" or deterministic because there's only one possible value ("the answer") for F given two particular values for m and a. On the other hand, m and a can assume any positive values because that's the way the natural world operates, at least as described by the branch of physics known as "mechanics."

A statistical model, on the other hand, is much more complex: while this is not the place to give this a rigorous definition, the "answer" is a "best" fit rather than a particular number, and you, the model developer, make certain judgment calls about what variables to use and how much weight to give to each one. In the simple formula given above, mass and acceleration always get equal weight, but not in the world of statistics, especially as applied to biology. People live only so long; their TSH values cluster in a narrow range (mostly); they can only be male or female (at least the way clinical research is run today!), and so on. Also, you do not have an infinite number of possibilities; you are often working with a limited sample, the number of subjects in your study. Finally, you might not be able to pick the right variables for the model because you are not in a position to conduct all of the experiments you need to get all of the information you need. So you are in essence gambling, and having to take the best calculated risk.

Continuous vs. discrete variables

A continuous variable has an infinite range of possible values, e.g., temperature or time interval.  A discrete variable can have one of a fixed, usually small, number of values, e.g., a test result that is either positive or negative.  Discrete variables are also known as categorical variables because they often represent assignment to categories.  Continuous variables are typically representations of natural phenomena, such as certain blood test results; discrete variables are typically created by the experimenter and given values on the basis of the values of continuous variables associated with experimental subjects.

It is important to realize that cutpoints, i.e., dividing values between categories, contain very little information in themselves. They are often chosen arbitrarily, often because they are round numbers; 10, 20, 100 and 200 are especially popular (and note that each has only one significant figure).  For example, just because a study concludes that, in general, subjects with variable values above a certain cutpoint are doing better in some way than those below it based on a categorical statistical model does not mean that all of the subjects with values above that cutpoint are doing well or that those below it are all doing badly; it simply means that higher values for that particular variable indicate a better chance of the subject's doing well, and that the cutpoint value is included among the values of the subjects in the study (and is ideally in the middle of them).

Dependent vs. independent variables

In the simplest sense, the dependent variable (in a sense, the "answer") goes on the left side of the equation, and the independent ones on the right side. Generally speaking, the independent variables should represent causes and the dependent variable their effect. Ideally, you can control the values of the independent variables by selecting characteristics in your subjects that follow certain patterns.

"Real" vs. surrogate variables

Sometimes the most meaningful choice of a dependent variable, e.g., time to first heart attack, is not feasible to use in a model, and an easier-to-measure surrogate variable, e.g., a blood lipid measurement or subtle change in heart structure or function, is used instead. It has usually been proven to be a strong predictor of the values in the variable it replaces, but sometimes without a full understanding of the causal relationship between them. Surrogate variables are often referred to as risk factors.

Observational studies, clinical trials, and meta-analyses

Observational studies gather data about human subjects without imposing any changes on their lives, but are still designed to identify factors associated with certain medical problems; they cannot determine causal relationships although they can strongly suggest them and probably rule out others.  Researchers also gather certain types of subject information such as age, gender and health status data such as smoking history, blood pressure and lab test results and represent them as independent variables expressed in a statistical model. These studies are often done under the direction of and with the funding of the federal government in order to formulate recommendations to the public.  Observational studies that work with data gathered over a certain time period, are either retrospective, i.e., analyzing data that is already available, or prospective (also known as longitudinal), i.e., gathering the data after the start of the study. On the other hand, cross-sectional studies work with data that have been gathered at only one timepoint. One disadvantage of retrospective and cross-sectional observational studies with respect to clinical trials and prospective observational studies is that they do not always provide data for the variables necessary to answer the questions the studies seek to answer, and that the variables used are often only those easy to measure and which are less likely to require "soft" thinking. On the other hand, these studies cost less and involve less risk to the subjects than do clinical trials. In addition, observational studies cannot prove cause-effect relationships among variables, although they can suggest those relationships by the magnitude of their correlation: paired highly correlated variables are more likely to indicate that one causes the other, or that they are both effects of the same cause. On the other hand, it is probably safe to assume that poorly correlated variable pairs have no causal relationship, which can be helpful in the design of a later relevant clinical trial.  Observational studies are conducted by epidemiologists, who use special statistical techniques that discover complex correlations among carefully selected factors, making it easier to determine possible causal relationships.

Active surveillance studies are like prospective observational studies in that subjects do not receive interventions in the form of treatment, but do submit to diagnostic tests.

A clinical trial is an experiment involving human subjects designed to determine whether a cause-effect relationship exists between a candidate treatment and the health of these subjects. It involves an intervention, such as administration of a study drug, in order to see whether it reduces certain medical problems in the subjects that the study drug was designed to treat (or causes other problems).  The subjects (typically referred to as "patients"), who all have the condition that the study drug was designed to treat, are divided into treatment groups, made as similar as possible to one another in relevant ways by a (rather misleadingly termed) randomization process to maximize the probability that outcome differences among the treatment groups are caused only by the intervention; at least one group receives the study drug, for example, while one group receives a placebo, which is chemically inactive but has the same appearance as the study drug.  Statisticians use inferential statistics (see below) to assess the study drug's performance with respect to the treatment groups mainly by evaluating the change from baseline in key safety and efficacy parameters, known as clinical endpoints; the identity of the subjects is withheld from those carrying out the trial, making it "blinded."  Usually the results of a clinical trial, if they are positive, result in prompt action, such as a New Drug Application (NDA) submitted to the FDA for permission to market the study drug. Some patients enrolled in these trials drop out, often because of illness; the FDA requires NDAs to provide a patient disposition, showing how many patients made it to each stage of the trial, and their reasons for dropping out.

Ethical considerations sometimes make clinical trials inappropriate; ethical rules forbid the exposure of subjects to known sources of harm in these studies.  For example, clinical trials designed to measure the amount of harm done by smoking on subjects would not be allowed because of strong existing evidence that smoking is harmful.  Another limitation of clinical trials is that they necessarily cover only a brief period of time.  On the other hand, some longitudinal observational studies go on for decades.  Yet another is the expense that they incur, largely because subjects, typically referred to as "patients", receive care from investigator doctors, who also decide whether their health problems necessitate their withdrawal from the trial.

The general consensus is that clinical trials are the ideal method for determining whether a newly developed drug is safe and effective.  But when these trials are performed on drugs that are already available, if only through prescription, this presents a risk for bias in subject selection because so few individuals who meet the inclusion criteria would choose the risk of getting a placebo, especially if it meant ending a relationship with a physician whom they were already seeing for treatment.  In fact, since exposing patients diagnosed with a certain condition with a standard treatment to the risk of not getting that treatment would constitute an ethical violation, so that those conducting the trial have to argue that that treatment is not standard, or that the criteria have changed to make the subject ineligible for treatment.  This is unfortunately allowed apparently because of an increasingly popular belief that every form of treatment for every disease should called into question if it has not been derived via a clinical trial, even if the drug involved has gotten official marketing approval.  This subject selection difficulty was circumvented in the Makena confirmatory trial by using subjects from two foreign countries which had not approved the drug (Rubin, 2020).   However, many clinical trials using an L-T4 product as the study drug have been conducted in countries where such products had had marketing approval for a long time, unfortunately generating only minimal objections on ethical grounds from those in the field.

Meta-analyses bring together data from selected small studies and treat the "final" calculations (effects), e.g., p-values or odds ratios, from each as though they were data points from one large study; a measure of heterogeneity called the I-squared, the variation in the results ("effects") among the studies is the other basic value taken.  The greater the heterogeneity, the less certain the calculation of the effect.  As Serghiou and Goodman (2019) put it in their explanation of how it is used, meta-analysis is "a foundational tool for evidence-based medicine."   However, because of the amount of subjectivity involved, most notably in the selection of the studies and assignment of weights, and the sheer amount of data involved, meta-analyses are also very difficult to validate, and there are apparently no rules in place to reverse the legal and social effects of such a study improperly carried out. Calls for publication retraction sometimes occur, but the journal involved can refuse to do so.

Statistical nitty-gritty: how the p-value is derived and how it is, rightly and (sometimes) wrongly, used

The p-value is a statistically derived entity on which rests the conclusion of a typical study.   In practice, it is the proportion of overlap between the values of a variable in two data sets, and is a function of the averages of the values of each and of their distributions; the greater the difference in their averages and the smaller their distributions, the smaller the calculated p-value will be.  In clinical trials, the one data set represents values of patients in the intervention group, while the other represents those in the placebo group.  In this sense, the p-value is a measure of predictability.  These values can be efficacy measurements, e.g., blood pressure readings for a drug being studied for anti-hypertensive effects, or safety measurements, e.g., blood counts for that drug.  Small p-values are considered to be desirable for efficacy measures because they indicate that the drug is making a (preferably the desired) difference.  On the other hand, large p-values are generally desirable for safety measures because they indicate a small difference in the drug's effect on those types of variables.  Setting an excessively small threshold safety measure p-value might cause a significant side effect to be overlooked.

According to Wasserstein and Lazar (2016) in an official statement representing the views of the American Statistical Association, a p-value "is the probability under a specified statistical model that a statistical summary of the data ... would be equal to or more extreme than its observed value."  In the same paper they wrote, "Scientific conclusions and business or policy decisions should not be based only on whether a p-value passes a specific threshold."  Yet is this what in effect the application of evidence-based medicine is doing to public healthcare policy? Are we changing public medical care policy by using statistics in a way that the statistics profession believes is wrong?

The median is the value at the 50th percentile.  The mode is the most common value. Their average (also known as arithmetic mean or just mean) is a more familiar statistic: the sum of all the data values divided by their count. Their standard deviation (and its square, the variance) is a measure of variation in the data, i.e., how on the average they differ from that mean.  A normal distribution is one in which the median, mode, and mean are very similar.  A normal distribution is not possible if any of the data extremes are bounded, e.g., when no value may go below zero.

If the mean of of a data sample is close to that of the population from which it was collected, e.g., everyone in the U.S. or all the patients who have used a certain lab to get blood tests, the measurement result is said to be accurate.  If the variation in such a measurement is small, this measurement process is said to be precise.  Sometimes we do not have enough information about a population to determine a sample's accuracy, but we can measure its precision.  We see this issue later in the assays of hormone concentrations: it took researchers decades to get these assays' precision to be great enough to be useful in determining normal ranges for thyroid hormones.

Inferential statistics compare two data samples or a data sample and a population by using their means and variation measures (usually a particular measure, standard deviation).  In a nutshell, this comparison is a measure of the proportion of the distribution of their values that overlap. If there is no overlap, their p-value is zero; if there is a complete overlap, their p-value is one. The desired p-value varies greatly according to context.  If one is testing a study drug in a clinical trial, that drug might be considered a success if a comparison of the efficacy of the drug (perhaps cure rate) not only showed the better mean for the group receiving the drug, but also resulted in a small p-value.  But you would also expect many safety measures between a study drug group and a placebo group to be essentially the same, e.g., heart health for a study drug treating a minor cosmetic problem, and that would mean similar averages and a large p-value.

The standard interpretation of the p-value in practical terms is that it is the chance of there being no difference between the two variables being compared.  For example, a p-value of 0.05 in a clinical trial testing the efficacy of a study drug means that there would be a 5% chance that the drug is no different than the placebo.  Today, this is considered an unacceptably low value in such trials, and the new cut-off value is currently under discussion.

It's important to remember that a p-value is a continuous entity, and decisions about cut-off values where statistical significance begins are inevitably somewhat arbitrary. If a p-value turns out to be, for example, 0.07, it might not be deemed "statistically significant," i.e., have enough predictive power to meet the study's particular criterion, but it would not be right to say, "There was no difference" between/among the groups being compared. If there were absolutely no difference between two such groups, i.e., that there would be a perfect overlap, the p-value would be exactly 1.  While a study drug would be expected to have much more predictable efficacy to qualify for marketing approval, the same threshold p-value would probably be too small in experiments involving two variables not expected to be so different, such as measures of that drug's safety. What is unfortunate in some recent studies is the practice of replacing the calculated p-value with "ns" if it falls beyond the threshold value set in the experimental design.

An odds ratio is a comparison of probabilities of an event for two samples across one variable, which in lay terms translates into percentages.  For example, if the odds ratio for variable X is 1.29 for sample A relative to sample B, the probability of X occuring in sample A is 29% greater than of it occurring in sample B.  In studies of the probability of risk of an adverse event, the odds ratio is typically referred to as the hazard ratio.

Normal Ranges (technically referred to as "Reference Ranges")

Why are normal ranges called "reference ranges?" According to Brabant et al (2006), there is a conflict between those who believe that "normal" implies absolute health to some people and others who believe that "abnormal" implies the presence of problems requiring correction, and that therefore "these problems have been circumvented by the definition of 'reference' values, implicating (sic) that absolute health does not exist."  This suggests that the professionals who use these ranges feel uneasy applying them strictly in practice, and regard them more as guidelines than rules.  I think this is a special problem because the limits of reference ranges vary widely relative to the clinical signs and symptoms that are typically found with the disease: it is far more likely for patients feel ill when their values on some tests are close to one of the limits of the reference range for that test than it is for them to feel that way when their values for some other lab tests are well outside the reference range.

Reference ranges ("normal" ranges) are very useful indicators of population patterns of certain key measurements, although their precision as determiners of the dividing line between health and sickness is sometimes controversial, especially when risks of harm associated with getting them wrong are unknown or money, legal considerations, or treatment effort in a non-fee-for-service system is heavily involved.  In the case of TSH distributions or those of other parameters following a smooth curve, one has three basic choices: 1) to include all certainly normal values and some possibly abnormal values in the "normal" range, 2) to exclude all possibly abnormal values from the "normal" range (the general approach to determining the "reference" range), or 3) to find some compromise between the two.  The legal advantages of using the wider reference range are obvious.

In practice, the "reference range" is usually defined to represent test values that reliably define what is normal, while some values outside that range might be either normal or abnormal.  The determination of the reference range of a continuous test value involves both cut-and-dried mathematics and subjective physician judgment.  First, a normal patient population (usually local to the laboratory where a particular patient is tested) is defined, i.e., those who appear to be disease-free, presumably both on the basis of clinical signs and symptoms (indicators of disease either reported to or determined by physicians through direct examination) and (previous) lab test results.  

How scientists minimize harm when they don't know everything: type I and II errors, sensitivity and specificity

Live as if you were to die tomorrow. Learn as if you were to live forever. (attributed to Mohandas Gandhi)

This saying seems to give contradictory advice, but it illustrates a standard approach of experimenters to containing potential problems: when in doubt, make the assumption most likely to do the lesser harm if wrong (or the greater good if right).  Sometimes it makes sense to make different assumptions about the same thing in different contexts, and it's standard practice for scientists for do this both in experimental design and in interpretation of results.  Unfortunately it often involves making difficult judgment calls and sometimes leads to controversial results.

Sensitivity is the rate at which positive results are recognized, i.e., false negatives are minimized. Specificity is the rate at which negative results are recognized, i.e., false positives are minimized.

A Type I error is the rejection of a true null hypothesis. A Type II error is the failure to reject a false null hypothesis. If the null hypothesis is that a patient is healthy, an example of a Type I error is wrongly diagnosing (and treating) a disease in that patient, while an example of a Type II error is missing the diagnosis of an existing disease (and therefore not treating it). Physicians often have to decide in the case of each type of disease which error causes the least harm. When the diagnostic method of a disease is clouded by uncertainty and/or controversy, the decision becomes harder. When treatment of one disease disposes the patient to getting another, and the probability of this happening is disputed, this becomes even harder.

The power of a binary hypothesis test is is the probability of not making a Type II error, on a scale from 0 to 1, where 1 means that probability is 0. Studies that use too few subjects in their samples or set the criterion for statistical significance too high risk failing to detect an effect present in the population that the sample is meant to represent.

What does that L- mean? (and those fancy prefixes with the apostrophes?)

Sometimes a chemical is represented with an L- prefix, which distinguishes it from its R- counterpart.  L-thyroxine, for instance, is the mirror image of the molecule R-thyroxine, although they otherwise have the same structural relationships among their atoms.  However, their (bio)chemical activity is very different!  L-thyroxine is a hormone essential for life, while R-thyroxine is useless and may even be bad for you.  Its synthetic form is referred to as levothyroxine in the context of its use as a commercially available drug.

Actually, the full formal name of this molecule is (S)-2-amino-3-[4-(4-hydroxy-3,5-diiodophenoxy)-3,5-diiodophenyl]propanoic acid, according to International Union of Pure and Applied Chemistry (IUPAC) naming conventions;  L-3,3',5,5'-tetraiodothyronine is a somewhat less formal term also used. These conventions are very useful in describing the structure of every conceivable molecule, showing exactly what branches off from where, but in the context of this essay, it's easier to explain things using common names (and better yet, abbreviations!) So, from here on out, there will be no hyphenated or numerical prefixes or, for that matter, apostrophes mentioned in molecule names in this piece, except for L-DOPA.

For the sake of simplicity, I will refer to these molecules by their common names, e.g., thyroxine or its even simpler name, T4. I will also refer to L-tyrosine as tyrosine.

Evidence-based medicine and its impact on thyroid disease treatment policy

There is no question about the benefit of properly carried-out evidence-based medicine. Its potential to eliminate harmful treatments and to spur research into safer and more effective alternatives has been demonstrated at many times in the past; just one example is the clinical trial testing the efficacy of the now-notorious radical mastectomy, which led to gentler treatments that worked as well if not better. (Granted, sometimes common sense might have worked as well in this case: it was well-known even by the ancients that tumors were fed by a profusion of blood vessels, yet the radical mastectomy was based on the assumption that cancer metastasis took place only through the lymphatic system.)  It promises to provide patients with the care they need rather than what an individual physician is motivated to deliver.  But was there ever a concern that political and economic factors might bias government bureaucrat review of the literature as well as the conduct of new, perhaps unnecessary studies, bringing about the removal of safe, effective, well-accepted, and affordable treatments from standard medical practice?

There has been an increasing impetus in recent years for physicians to draw, as much as possible, on the entire body of knowledge relevant to their profession when making treatment decisions rather than only on their personal experience. Sackett et al. (1996) describe evidence-based medicine as the "conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients" and as "integrating individual clinical expertise with the best available external clinical evidence from systematic research." They are also careful to point out that, in the absence of an applicable "randomised trial," that a physician should seek out "the next best external evidence and work from there." Most fundamentally they say, "Good doctors use both individual clinical expertise and the best available external evidence." 

Political scientists Patashnik et al. (2017) gave strong support to institutionalizing the use of evidence-based medicine via government policy as set forth by the Affordable Care Act (ACA, Office of the Legislative Counsel for the Use of the U.S. House of Representatives, 2010), which funded the Patient-Centered Outcomes Research Institute (PCORI) to conduct Comparative Effectiveness Research (CER) on different medical care interventions.  However, Congress put forth rules keeping these research results from affecting federal government policy regarding healthcare.  PCORI (2018) has continued to receive funding through charges added to medical insurance premiums.  Seventy-nine percent of its funding has been devoted to CER, while 16% went for infrastructure and 6% for methodology research (p. 3).  They state: "Generating evidence on how to better treat high-burden, high-impact conditions is only useful if those results get into the hands of those who need them most" (p. 8); this captures their role, i.e., to provide information for patients to evaluate when making treatment choices, especially in regard to diseases that are difficult to treat.

However, the executive branch found a way around this obstacle, using The U.S. Preventive Services Task Force (USPSTF), which was was instituted in 1984.  This group was directed to give regular reports to Congress regarding their findings regarding "evidence-based" recommendations about which screening (of conditions in asymptomatic patients) and preventive care services are recommended, based on the "balance of their benefits and harms" (USPSTF, 2014); these evaluations were officially to be given without regard to financial considerations.  Although the USPSTF was originally intended to be an advisory group grading the quality of clinical research supporting a wide variety of medical services, the Affordable Care Act now ties its grades to coverage by ACA insurance plans, Medicare, and Medicaid.   This is cause for concern because of the differing goals of the USPSTF and the ACA, especially with regard to fields that have been sparsely covered by clinical trials and because of their discrepant weighting of financial considerations; procedures which the USPSTF has flagged as being inadequately supported by scientific evidence, if only because of sparse research in the area, run the risk of not being covered by ACA insurance plans even if they have been part of standard medical practice for a long time and are safe and affordable.  Oddly, the USPSTF receives only brief treatment by Patashnik et al. (2017), in connection with its recommendation that women begin getting mammograms at age 50 instead of at 40 (later reversed).

Some of these concerns have been addressed by NIH researchers Villani et al. (2018), who broke down by funding source categories (government, industry, nonprofit or university, and "unknown") and disease categories the number of studies that the USPSTF had considered in making its recommendations.  This included only 17 studies of "thyroid dysfunction," 41% of which were financed by the government, 6% by industry, 12% by nonprofit or university, and 47% by unknown sources.  In contrast, the category of "cognitive impairment in older adults" involved 255 studies, 55% of which were financed by the government, 29% by industry, 33% by nonprofit or university, and only 12% by unknown sources.

What are the criteria for deciding on the best treatment decision? Some mentioned in the literature are longevity, quality-adjusted life years (QALYs), occurrence of or time to an event such as a heart attack, and of course the surrogate variables, i.e., risk factors, for the diseases that claim the most lives.  Quality-adjusted life years are a combined measure of longevity and the quality-of-life factors 1) degree of mobility, 2) pain/discomfort level, 3) ability to perform self-care, 4) level of anxiety or depression, and 5) the ability to carry on usual activities (Phillips, 2009).  Sometimes the decision involves difficult trade-offs, which involves consideration of the relative risk of Type I and Type II errors, however.  Livingston and McNutt (2011) point out the pitfalls of applying unvalidated models of ideal medical care to large organizations via rigid, oversimplified rules. 

Physicians diagnose thyroid disease largely on the basis of laboratory blood tests, although they are not required by guidelines to screen any groups of patients for thyroid tests; in this way, clinical signs and symptoms play a role in diagnosis by motivating the physician to order such a test. Approximately 0.7% of physicians are endocrinologists; there is one American endocrinologist for about every 57,000 Americans, based on stats by the (U.S.) Endocrine Society (2014).  Clinical biochemists perform these tests, which are far from routine, using assay kits, of which many are available on the market; they also determine "reference" ranges for their particular labs using physician input about which patients are disease-free to determine their samples.   Internists and family practitioners handle most hypothyroidism cases.  Scientific teams develop the blood test tools these professionals use. And patients play a role by recognizing their symptoms and signs and calling a physician's attention to them.

Certain policy-making groups play roles of varying influence on standard practice in the medical care industry. The American Association of Clinical Endocrinologists, the American Thyroid Association, the American College of Physicians, the American College of Family Physicians, and the American College of Physicians once recommended screening for hypothyroidism, offering varying guidelines, but now have retreated from that position. 

Academic journals have substantial influence on science, not simply by the papers they choose to publish but how they categorize them when they are published.  Research articles are characterized by original research, with references to previous related research to establish credit for all contributions to science in these articles.   Literature reviews do not simply cite many references but explain their contributions to a synthesis that establishes the current state of an area of scientific research.  Scientific policy papers are at the very least statements of new rules to follow, preferably backed up with explanations of the relevance of scientific evidence, explaining the positions of authority the authors hold and those to whom these new rules apply.  Editorials are expressions of opinion by those in charge of a journal.  Opinion pieces express points of view such as the author's interpretation of or criticism of a previous publication and are typically shorter and more specific in focus, preferably with explicit use of references, but carry no authority.  It is important for journals to clarify which categories these articles fall in correctly so that readers have a clear understanding of how to regard their role in influencing the scientific conversation. When opinion pieces are confused with research articles or with statements of policy made by representatives of official rule-making organizations, this can cause great dangers, especially when their unvalidated claims are disseminated to the general public through health newsletters.

Thyroid structure and function basics

The hypothalamus-pituitary-thyroid "axis" (where the thyroid gets its orders) and its partnership with its growth hormone counterpart

Pirahanchi and Jialal (2019) have summarized this brilliantly.

The hypothalamus, the pituitary gland and the thyroid gland are united in a negative feedback system called the HPT axis.The hypothalamus, a part of the brain, controls energy balance (Bolborea and Dale, 2013); its paraventricular nucleus (PVN) contains neurons secreting thyrotropin releasing hormone (TRH), which stimulates the anterior pituitary gland, which in turn stimulates the thyroid gland to increase its output via the hormone thyrotropin, more familiarly known as thyroid-stimulating hormone (TSH), which in turns causes the thyroid gland to produce the hormones triiodothyronine (T3) and thyroxine (T4); Rising levels of these two thyroid hormones cause the hypothalamus to reduce its TRH output. To add to the confusion, Varghese et al. (2008) report that their research indicates that the intestines can also produce TSH when stimulated by a viral infection.

TSH stimulation causes mitosis (cell division) in the thyroid, the amount increasing as TSH increases (Ealey et al. 1985); there are other, lesser, stimulants of thyrocyte proliferation, such as estradiol (Gabriela et al. (2012)), some antibodies to the TSH receptor (Morshed SA, 2013) and, if taken internally, lithium (Rao et al., 2005).  The thyroid would atrophy without TSH stimulation under normal conditions because a certain ideal level maintains the thyroid at a constant desirable mass by balancing natural, inevitable tissue loss with tissue creation; if the TSH level is too high and growth exceeds apoptosis (programmed or "normal" cell death), the number of cells increases and a goiter develops; on the other hand, if the TSH level is too low, provided adequate iodine is available, apoptosis will exceed cell-creating mitosis and involution (the opposite of growth) will reduce the thyroid's size, in turn reducing its hormone output.  This provides a plausible explanation of why individuals with some degree of thyroid failure can have normal T4 and T3 values:  when TSH stimulation causes an increase in the number of thyroid cells, this compensates for underperformance on the part of the individual cells. This is a possible explanation of the dynamics of subclinical hypothyroidism, although it does not furnish complete proof of compensation, i.e., full restoration of overall thyroid function.

It is generally accepted that in iodine-sufficient countries that primary hypothyroidism, i.e., not induced by surgery or by diseases of the pituitary gland or hypothalamus, is caused by Hashimoto's thyroiditis, in which anti-thyroid antibodies, i.e., those attacking 1) thyroglobulin (Tg), a protein which combines with iodine to form the T4 and T3 thyroid hormones, and 2) thyroid peroxidase (TPO), an enzyme which enables this combination. In younger individuals, there is a strong relationship between TSH levels and those of these antibodies, as well as of the size and presence of a goiter. However, after years of disease, this relationship breaks down as antibody levels decrease with age and the thyroid degenerates into a "shrunken, fibrotic gland with little or no function", according to the Merck Manual (2006a), p. 1200. The sharp rise of TSH levels in some people beyond the age of 70 might be due to this process, while many other old people have been shown to experience only a slight rise in TSH levels. Some experts in the field believe that this sharp rise is normal in older people and therefore should not be treated, but the mechanism behind this assumption has not been explored in depth and existing data are open to varying interpretations.

Sometimes, however, environmental stresses, such as Vitamin E deficiency and both iodine deficiency and excess, can lead to thyrocyte necrosis, i.e., abnormal cell death, caused by injury; necrotic cells tend to swell rather than to shrink, as apoptic cells do, maintaining some of the thyroid enlargement caused by a goiter (Mutaku et al. 2002).   Because necrosis causes enlargement rather than shrinking, it may contribute to a goiter's permanence. 

TSH works with insulin-like growth factor 1 (IGF-1) and insulin to stimulate DNA synthesis in thyroid cells (Brenner-Gati et al., 1988). IGF-1, in turn, is produced in the liver in response to stimulation by growth hormone (GH), which is produced by the anterior pituitary gland to perform one of the main functions of inflammation, i.e., to stimulate rebuilding of damaged tissue, its production in turn stimulated by hypothalamus-produced growth hormone releasing hormone (GHRH).  Liu et al. (2011) showed that IGF-1 levels were lower in those with goiters than those with disease-free thyroids while IGF-1 levels were higher in the those with (presumably benign) adenomas and even higher in those with papillary thyroid carcinoma.

This process may be impeded by the anti-inflammatory action of cortisol, which is produced by the adrenal cortex in response to stress, which (as is discussed later) prevents the natural night-time rise in TSH if given in high enough doses.

The hypothalamus acts as a switchboard in this system, putting out hormones in response to other hormones sent from fat cells and digestive system organs. Some stimulate the appetite via the arctuate nucleus; others influence the feedback system that involves the thyroid gland and its hormones by stimulating the paraventricular nucleus (PVN). Endocrine hormones travel through the blood to reach their destinations. Recent research is giving us information about special cells in the hypothalamus, tanycytes, which might be the communication bridge between the central nervous system and the endocrine system.

Figure 1. This gives an idea of the components of the feedback system involving the thyroid, but not the suppressive actions; not represented are the feedback mechanisms of the thyroid hormones on the pituitary gland and on the hypothalamus. Only three hypothalamus nuclei are represented: the paraventricular nucleus (PVN), the ventromedial nucleus, and the lateral hypothalamus. The hormones involved in the appetite regulation part are described by Saladin (2007, 1002-4).

This is a negative feedback system. The hypothalamus, the pituitary gland and the thyroid are referred to as the HPT axis. The hypothalamus performs its regulatory function by sensing T3 and T4 levels and raising or lowering Thyrotropin Releasing Hormone (TRH) accordingly. This diagram shows its interaction with another system regulating appetite and with another regulating glucocorticoid levels (of which cortisol is one, which in turn lower TRH levels as they rise. In response to cortisol stimulation, the liver produces glucose from the glycogen it stores and sends it and free fatty acids into the blood.

How thyroid hormones produce energy in the cells

The thyroid is an endocrine gland which, in partnership with many other parts of the body, determines the pace of metabolism in the body. The thyroid produces two hormones, T3 (in small amounts) and T4 (in large amounts). This production activity takes place inside follicles within the thyroid gland. T3 is the metabolically active hormone, producing a short-lasting effect; it is also produced from T4 by the removal of a particular iodine atom by special enzymes located in various organs and tissues.

Thyroid hormone T3 stimulates receptors in the cells of two vital organs: the heart, which has mainly thyroid hormone receptor alpha (THR-alpha), and the liver, which has mainly thyroid hormone receptor beta (THR-beta).  Yehuda-Schnaidman et al (2005), describe it as a "major modulator of mitochondrial metabolic efficiency."  Harper and Seifert (2008) state that "thyroid hormones activate the uncoupling of oxidative phosphorylation through various mechanisms involving inner membrane proteins and lipids." The mitochondria, of course, produce oxidative phosphorylation, which adds a phosphate molecule to adenosine diphosphate (ADP) to create adenosine triphosphate (ATP), which provides body cells with the chemical energy that they need to function.  

The thyroid's basic unit: the follicle

The basic unit of thyroid tissue is the follicle, a spherical object containing a colloid-filled lumen enclosed in a one-thyrocyte-thick layer; the small spaces between follicles contain capillaries and small lymphatic vessels passing among the follicles.  The thyrocytes are epithelial cells that interface with the blood through their basolateral (outer) surface and with the lumen (inner space) via their apical (inner) surface.  Thyrocytes produce the glycoprotein thyroid hormone precursor thyroglobulin (Tg, a target of one type of antibodies, known as Tg-Ab, which mark Tg for destruction) from tyrosine (a mostly nonpolar amino acid which enters the thyrocyte through passive transport), and secrete that and iodide ions that they receive from the blood into the colloid (Marino and McCluskey, 2000).  An important enzyme, thyroperoxidase (TPO, another target of some types of antibodies, known as TPO-Ab) has been found on both the apical and, less frequently, on the basolateral surfaces of thyrocytes (Zimmer et al, 1997).  TPO works there with hydrogen peroxide (H2O2) to complete the building of T3 and T4 by reshaping and cleaving thyroglobulin and bonding it with iodide ions; these hormone precursors are returned from the colloid to the thyrocyte for minor modifications before being sent into the blood as completely formed T3 and T4. The antioxidant Vitamin E is needed to clean up excess H2O2 created for this hormone-building process; the higher TSH levels are, the more Vitamin E is needed, as Mutaku et al. (2002) conclude.

Thyroglobulin is a very large protein, with a molecular weight of about 660,000 grams per mole (Food and Nutrition Board, Institute of Medicine, 2001); it consists of a peptide backbone with 123 tyrosine molecules branching from that backbone. Tyrosine, as discussed above, is an amino acid consisting largely of one phenyl group (a 6-carbon ring) two bonded. T4 and T3 are very small molecules, which are each constructed from two tyrosine molecules taken from a thyroglobulin molecule by the enzymatic action mentioned above in the colloid; they initially remain attached to the thyroglobulin peptide backbone and enter the thyrocyte via endocytosis on its apical side, where they break away and leave the thyrocyte via its basolateral surface and enter the blood, since they are fat-soluble, through passive transport (Saladin, 2004). 

Both deficient and excessive iodide levels relative to existing thyroglobulin can produce a goiter; excessive iodine levels can lead to the formation of goiters because of interference (congestion) created by too many iodide ions according to the Food and Nutrition Board, Institute of Medicine (2001).  Is it possible that the goiter is created in the case of excess iodine ions instead by the HPT axis increasing thyroid hormone production to keep the iodine ions in the thyroid from increasing in number and therefore creating a chemical as well as physical problem?  An important implication here: the presence of a goiter in an individual is not enough to determine whether iodine excess or deficiency is present, although population studies can give helpful clues about iodine in the local environment.

TSH increases the rate of passage of iodide ions into the thyrocyte by stimulating the increase in the number of its sodium/iodide symporters (NIS) integral proteins in thyrocyte cell membranes, but it needs the presence of IGF-1 to do this (Ock et al., 2013). These mediate the transport of iodide ions from the blood into the thyrocyte via the basolateral cell membrane, maintaining the concentration of iodide ions inside the thyrocyte vs. those outside it, in the blood, at a ratio of 20-40 to1. What I'm not clear on is whether increasing TSH increases the number of NIS in the cell membrane of individual thyrocytes or whether the increase in thyrocytes accounts for the increase in the total number of NIS. An NIS can be carried to and from the cell membrane within the cell in vesicles, with TSH drawing them into that cell membrane, thereby creating more opportunities for iodine ions to enter the thyrocyte. Finally, when iodide ion concentrations rise to a certain threshold level, TPO ceases its binding of these ions in organic molecules such as hormones until the iodide ion concentration drops below that level; this phenomenon, known as the acute Wolff-Chaikoff effect, lasts up to 50 hours after a single (over)dose but this implies that continually excessive iodine intake could result in long-term hypothyroidism (Dohán et al., 2003)However, up to this threshold value, it stands to reason that thyroid mitosis and the production of thyroid hormones will increase as iodide ion input increases according to the system this review describes.

NIS also found in the breast, salivary glands, choriod plexus (tissue which lines the brain's ventricles and produces cerebrospinal fluid) and gastric mucosa, with the lactating breast getting the lion's share of the non-thyroid iodide ions (Institute of Medicine, Food and Nutrition Board, 2001); Tazebay (2000) observed the same about the cancerous breast.  Dohán et al., 2003 point out that the lactating breast has a local NIS to provide iodide ions to the nursing newborn; this would tend to lower available iodide ion levels in pregnant and nursing women. Estradiol, the predominant estrogen in women during their reproductive years, has been shown to inhibit this process by down-regulating the thyrocyte symporter gene (Furlanetto et al., 1999).

Although these symporters are the only means by which iodine enters the thyrocytes, they also allow in other ions with a -1 charge, although the "pull" it exerts on these ions varies greatly. Cl-chlorate (ClO3-) and thiocyanate (SCN-) ions are the most competitive inhibitors, practically, in fact, as competitive as iodide ions, while perchlorate ions (ClO4-) are apparently "blockers," though I do not understand what that implies in terms of chemical reactions. Bromide ions, on the other hand, are "pulled" in by only about a quarter of the force that iodides are, and bromate ions only about half of that (Dohan et al, 2003). Thiocyanate ions are found in cruciferous vegetables, e.g., broccoli and cabbage, but the amount in normal consumption of these vegetables is not considered to be a problem in those getting enough dietary iodine and selenium and not possessing certain rare genetic defects.  Cooking these vegetables helps to mitigate this problem.

It is possible that as breast tissue evolved to handle substantial amounts of iodide ions in order to transfer it to nursing infants, this tissue came to depend on the presence of this form of iodine to function normally on a cellular level.

Pendrin, the sodium-independent chloride/iodide transporter is an integral protein that transports iodine from a thyrocyte to the colloid through the apical cell membrane separating them.

The detected presence of antithyroid antibodies is considered to be the hallmark of Hashimoto's thyroiditis (Amino et al., 2002), regarded as an autoimmune disease and is currently accepted as the cause of nearly all hypothyroidism in the U.S. (Merck, 2006b).  There are two types of anti-thyroid antibodies: they attack 1) thyroid peroxidase (TPO), and 2) thyroglobulin (Tg).  Hashimoto's progresses very gradually, irreversibly destroying the thyroid, and might start very early in life. 

Figure 2. The main functions of the thyroid follicle, with one thyrocyte blown up and the activities in the colloid shown on the left. Three forms of molecular transport into and out of the thyrocyte are shown: passive transport (tyrosine, T3 and T4), active transport (Na+ and I-, via pendrin) and endocytosis/exocytosis (thyroglobulin, i.e., "Tg"). "ER" stands for "endoplasmic reticulum."

An important model which suggests answers to some stubborn questions

The thyroid's secretory capacity (GT), i.e., the "maximum stimulated amount the thyroid can produce in a given time unit" (Dietrich et al., 2012), is a direct measure of thyroid function; its units are picomoles per second (pmol/s).  It challenges the thyroid with high levels of infused TSH, measuring its free T4 output, and relating the two with a complex formula.  This process of measuring it is very complex and stressful for the human subject and not suitable for patients, but useful for modeling the pituitary's nonlinearly decreasing ability to compensate as the GT decreases.  These data suggest that, while TSH is a more sensitive indicator than free T4 when free T4 levels are low, i.e., in hypothyroidism; the reverse is true when the latter are high, i.e., in hyperthyroidism.  These two nonlinear curves have elbows, i.e., places of especially rapid changes from mainly horizontal to mainly vertical, at the top of the reference range for TSH and at the top of the free T4 range. This might explain why a TSH assay might experience special challenges with a sample from someone with a true TSH level of about 5.0 mIU/L and/or borderline-low free T4: a minor glitch might drive the TSH measurement up from 5.0 to 20 mIU/L, and the higher the real TSH level is, the harder it would likely be to measure it correctly.  On the other hand, at the upper end of the free T4 range, the associated TSH measurement might contain very little useful information, while the free T4 measurement might be necessary for distinguishing whether that patient has subclinical or overt hyperthyroidism.  What is suggested by this model for those with uncomplicated hypothyroidism is that TSH and free T4 become abnormal together, rather than the apparently commonly accepted view that free T4 remains stable while TSH varies widely and chaotically.

The table below represents values taken at just four points along smooth nonlinear curves representing a monotonic relationship between TSH and free T4 and their relationship to GT in in a graph in the Dietrich et al. (2012) paper;  the graph information supports observations in past publications that the ideal TSH level is somewhat less than 1.5 mIU/L for those whose thyroids have remained whole.  This table represents individuals who are otherwise healthy; this paper also considers possibilities of different relationships of TSH levels to those of free T4, which vary with the nature of non-thyroidal health problems.  These researchers have gone into a great more detail in their model, explaining that patients have different set points depending on non-thyroidal factors.  This table represents the values for one such example set point.

Level of hypothyroidism GT (pmol/s) TSH (mIU/L) Free T4 (pmol/L) Free T4 (ng/dL)
Euthyroidism 3.4 1.2 15 1.2
Sublatent hypothyroidism 1.4 4.0 11 0.85
Subclinical hypothyroidism 1.2 5.0 10 0.78
Overt hypothyroidism 0.7 12 7.5 0.58

Table 1. Changing TSH and Free T4 levels as thyroid secretory capacity (GT ) decreases, summarized from Dietrich et al. (2012) with converted free T4 values in ng/dL added

Chatzitomaris et al. (2017) expand on this to develop a detailed model which showed how different diseases affect TSH, total T4, free T4, total T3, free T3, and reverse T3 levels.  One broad observation that can be made is neither TSH nor free T4 change very much during physical non-thyroidal diseases, and therefore are especially helpful in distinguishing thyroid disease even in the presence of other health problems.  On the other hand, severe physical non-thyroidal diseases tend to drive free T3 and total T3 down, and reverse T3 up.  They go on to discuss a thyroid "allostasis" model, i.e., one that describes the changes parts of the HPT axis go through to achieve homeostasis in response to two classes of stressor: 1) critical, i.e., life-threatening, illness, causing Low T3 Syndrome, also known as Non-thyroidal Illness Syndrome, and 2) pregnancy, obesity, endurance exercise, adaptation to cold, acute psychosis, and post-traumatic stress disorder (PTSD).  The first group of stressors cause the lowering of TSH, the D1 and D2 deiodinases (causing lower free T4), and protein binding to T4 and T3 (lowering total T4 and total T3), and the increase of D3 (lowering free T3 and raising reverse T3), while the second group does the opposite and drugs can cause both types of responses depending on what they are.  They add that "assay interference" (errors doing the assay) and "pre-analytical issues" can produce results that deviate from this model.  Will treatment with thyroid hormones help with either type of condition?  The authors say that this is still of subject of intense controversy.

This has troubling implications because of its openness to the possibility that a high TSH simply represents a new normal for people with non-thyroidal illnesses (not necessarily suffering from low T3 syndrome) and should therefore be left alone.  What about the cases where the high TSH is the only indication of an undiagnosed non-thyroidal disease, and is accompanied by feelings of malaise unusual for the patient but which seem normal to a physician with a large caseload?  Suppose the non-thyroidal disease is difficult to treat, chronic, and/or progressive, or the patient has to wait for appropriate treatment and meanwhile has disabling symptoms that are caused by the extra stress that the other disease puts on the thyroid?  Is it possible that supplemental T4 could help the patient to cope with the disease even if it does not cure it?  At the very least, the patient should have a say in how this issue is approached.

Other body systems processing thyroid hormones

Thyroid-binding globulin (TBG), transthyretin and albumin, all produced by the liver, together bind most T4 (99.97%) and T3 (99.7%) (Mebis and van der Berghe, 2009) so that they cannot react and transport these hormones in the blood; the unbound portion is known as free T3 (FT3) and free T4 (FT4).  T3 and T4 need transport proteins to travel through the blood rather than being randomly tossed around because T3 and T4 are mostly hydrophobic and need a force to attach them to the blood water molecules. The transport proteins link to these hormones at their hydrophilic/nonpolar ends and link to water molecules with their polar/hydrophobic ends.

In the healthy individual, TSH and FT3 levels follow a similar circadian rhythm, with FT3 slightly delayed (Russell et al., 2008).  Since FT3 is the actual amount of chemically active hormone available to the body, it is in theory the one most closely linked to clinical manifestations of hypothyroidism; however, but it's not clear to me that TSH follows that pattern in a person with thyroid disease and therefore may not accurately reflect the severity of signs and symptoms. 

There are three different iodothyronine deiodinase enzymes (D1, D2, D3) which process T4 and, in one case, T3; as their collective name implies, their role is remove an iodine atom (turning it into an iodide ion) from one of these molecules, sometimes making the latter chemically active, sometimes the opposite, depending on the location of that iodine in that thyroid hormone.  Each of these enzymes is determined by a single gene. D1, which operates mainly in the liver, kidney, anterior pituitary and thyroid; converts T4 to T3 by removing the iodide ion (which is recycled) from T4's outer ring.  D2, found in the brain, skeletal muscle and anterior pituitary, does the same things as D1, only locally, and seems to be activated in hypothyroidism (Mebis and van der Berghe, 2009).  D3, on the other hand, is present and only locally operative in such rapidly growing tissue as the skin, placenta, fetal tissue and intestines and in the brain, especially the hippocampus and the temporal cortex (Santini et al., 2001); it has also been found in malignant tumors which apparently "express" D3. It removes an iodide ion from the center ring of tyrosine in T4, generating reverse T3 (RT3) (Mebis and van den Berghe, 2009); by a similar process it also creates diiodothyronine (T2) from T3 (Bolborea et al., 2012).  D3 is also found in the brain, with especially high levels found in the hippocampus and temporal cortex and lower levels found in the thalamus, hypothalamus, midbrain cerebellum, the frontal and parietal cortices and the brainstem (Santini et al., 2001).  The D1 enzyme contains selenium; however, evidence suggests that this mineral is not found in D2 (Chanoine et al., 2001).

By producing RT3 and T2 instead of T4 and T3, D3 effectively has the opposite effect of D1 and D2: RT3 and T2 have very little of the effect that T3 and T4 have on tissue cells, but they do suppress TSH (to a significant but lesser degree), which in turn lowers T3 and T4 production in the thyroid.

In vitro cell research suggests that tanycytes are instrumental in maintaining the HPT axis negative feedback system by secreting a chemical that "degrades" TRH when they sense the presence of T3 at their locations in the hypothalamus. They also apparently produce enzymes D2 and D3, downregulating the former and upregulating the latter as hours of daylight decrease, which in turn lowers T3 levels (Bolborea et al., 2012).

The endocrine system works closely with the nervous system and immune system. There are two basic TSH "signals," i.e., repetitive wave patterns, contributing to form the shape of the amplitude curve of TSH hormone levels over time: the cyclical sine-wave-shaped one that repeats every 24 hours, and the pulsatile, which spikes about 17 times every 24 hours in healthy individuals (Roelfsema et al. (2009). In order to separate these two "signals" in TSH data, these researchers used a computer algorithm called deconvolution.

Some individuals have special difficulty converting T4 to T3 in the brain and skeletal muscle because a genetic defect causes an enzyme's action to block the activation of D2 (mice study, Werneck de Castro et al., 2015).  U.K. researchers Panicker et al. (2009) discovered that subjects with a less common (16% of the study population) homozygous D2 gene variant did more poorly on the General Health Questionnaire 12, a WHO-derived document. McAninch et al. (2015) followed up on this finding by concluding that this defect affects cellular biochemical activities that are associated with nervous system functions, abnormalities of which have been shown to cause "neurodegenerative disorders such as Huntington's disease."  It was not clear from the abstract of the McAninch article whether this problem was associated with Alzheimer's dementia; perhaps the full version of this article provided this information.  A later publication (Sungro et al., 2019) concluded from a mouse study that the "Thr92Ala polymorphism", in the gene controlling the D2 deiodinase, i.e., resulting in the substitution of the Ala amino acid for the Thr92 amino acid in the resulting D2 protein, causes these problems, which were circumvented by adding L-T3 to treatment of the subjects, greatly improving their neurological function.  Perhaps it was especially difficult to diagnose this problem because blood tests could not detect such a local problem.

The parafollicular cells

These cells are found in the connective tissue among the thyroid follicles and produce calcitonin, which blocks the action of osteoclasts, which remove calcium from bones. They express TRH, thereby having some influence on thyroid function.

Relationship to the nervous system, the immune system and to other endocrine glands

The Merck Manual (2006b, p. 1194) refers to "drugs such as dopamine and corticosteroids, which decrease pituitary secretion of TSH, resulting in low serum TSH levels and subsequent decreased T4 secretion" in connection with diagnosis of euthyroid sick syndrome (abnormally low T3 and T4 levels in a patient without thyroid disease but with a disease affecting another part of the body).  Abbott Labs (2013) says that the use of "Dopamine/Dopamine Agonists" and "Glucocorticoids" may "result in a transient reduction in TSH secretion" when administered at high enough doses; this makes sense because the lowered T3 would eventually trigger a TRH increase from the hypothalamus, stimulating the pituitary to increase its output of TSH.  One has to ask, though, what is the effect of habitual or long-term use of these substances?  Suppose a patient has a condition which keeps the levels of at least one of these substances constantly high?  Besides, if TSH is pulled down into the normal range by dopamine intake or stimulation, could a diagnosis of hypothyroidism be missed?

In a nutshell, the versatile nonessential amino acid tyrosine is a building block of both the catecholamines (the neurotransmitters dopamine, epinephrine and norepinephrine) and of T3 and T4. 


It has long been known that hypothyroidism has been linked to high dopamine levels.  Dopamine is a catecholamine, a neurotransmitter which activates the sympathetic nervous system.   Besses et al. (1975) concluded that a dopamine infusion "inhibited" TSH and prolactin. Scanlon et al. (1979) discovered that a dopamine-blocking agent raised daytime TSH in subjects, more so in women than in men.  Crocker et al. (1986) showed that hypothyroidism in rats (induced experimentally by exposure to radioactive iodine) increased dopamine receptor concentrations in the ventral striatum (which houses the nucleus accumbens, now believed to be the brain's pleasure center.)  It is now generally accepted that dopamine suppresses the HPT axis altogether (Haugen, 2009).

DeZegher et al. (1995) raised the concern that dopamine infusion therapy, commonly applied to infants with heart problems to help their heart contractions, could mask the presence of hypothyroidism in those patients and illustrated this by inducing a dramatic rise in TSH in infants receiving a dopamine infusion by cutting off that infusion.  Diarra et al. (1989) examined tyrosine metabolism in several types of rat tissue and concluded that catecholamine synthesis was increased in the medullas of the adrenal glands (epinephrine and norepinephrine) and the brainstem, i.e., the part of the brain adjoining the spine, in hypothyroid patients.

Caffeine indirectly causes dopamine levels to rise. Benvenga et al. (2008) showed that when human subjects drank coffee and took T4 tablets together, their T4 blood levels were lower and peaked later than those of controls; however, in the in vitro stool studies, coffee was determined to be a weaker sequestrant than dietary fibers, aluminum hydroxide, and sucralfate, an anti-ulcer drug.

What's also interesting is that exercise raises brain dopamine levels (Sutoo and Akiyama, 2003). Exercise, in fact, tends to raise overall TSH levels anyway (citation in a later section).

Smoking, which involves the intake of the dopamine agonist nicotine, has been recognized to be a confounding factor in studies comparing patients on the basis of TSH levels, as is discussed later. According to Norwegian researchers Jorde and Sundsfjord (2006), smokers had statistically significantly lower TSH levels and higher T3 and T4 levels than non-smokers; because of the nature of HPT axis feedback systems, this suggests that smoking raises T3 and T4 levels, which in turn suppress TSH levels.

L-DOPA is converted to dopamine in the basal ganglia and substantia nigra (in the brain), involving an enzyme called L-DOPA decarboxylaseL-DOPA in turn is created from the amino acid tyrosine via the catalyst tyrosine hydroxylase, which is found in the central nervous system, peripheral sympathetic system neurons and adrenal medulla. L-DOPA also stimulates the release of GHRH from the hypothalamus (Chihara et al., 1986). Could this explain why a rise in dopamine lowers TSH levels? Maybe the introduction of dopamine lowers L-DOPA levels because less of it is needed to create dopamine. This in turns lowers IGF-1 levels, which in turn lowers the maximum level of TSH that can be used. This in itself could induce hypothyroidism, but what lowers TSH? Perhaps the HPT axis responds to this situation by producing less TSH so that as little as possible is wasted.

Tyrosine is also a precursor of thyroglobulin, in turn a precursor of the thyroid hormones.  This suggests that some metabolic pathways leading to the production of dopamine and the thyroid hormones compete for tyrosine.


Cleare et al. (1995) concluded that hypothyroidism is associated with reduced "central" serotoninergic activity and with clinical depression (determined by responses to two depression rating scales).  Cortisol and prolactin levels were lower among hypothyroid subjects than in controls. Unfortunately, today the medical establishment appears to view low serotonin levels as diagnostic of depression, to be treated by Selective Serotonin Reuptake Inhibitors (SSRIs).


Patients with Cushing's syndrome, a condition in which the adrenal cortex hormone cortisol is excessively high, have lower TSH levels than healthy subjects, especially during the evening and night, when only the latter experience a substantial rise in TSH (from an average of about 1.0 to 2.0 mIU/L) that starts at about 8 pm, peaking at around 11 pm.  This is explained by less vigorous pituitary pulses, i.e., smaller in amplitude and mass, according to Roelfsema et al. (2009).  The ultimate cause of TSH suppression associated with cortisol use appears to be TRH suppression in the hypothalamus (Alkemade et al., 2005, Haugen, 2009). This suggests that high cortisol levels might mask thyroid failure by lowering TSH levels into the normal range in those whose hypothyroidism might have been flagged otherwise. 

More recently discovered, neuropeptide hormones include Neuropeptide Y, expressed in the PVN of the hypothalamus (Pert, 2004).  It apparently causes release of corticotropin-releasing hormone (CRH), which in turn stimulates the anterior pituitary gland to produce adrenocorticotropic hormone (ACTH), which stimulates the adrenal cortex to produce cortisol.

Can high levels of emotional stress misleadingly lower the TSH in patients whose hypothyroidism would be discovered otherwise through TSH blood levels?  And does the adrenal cortex pinch-hit for for a failing thyroid?  There are a lot of troubling questions here.

The immune system: cytokines

Cytokines, e.g., tumor necrosis factors, interleukins and interferons, are messenger molecules that allow the immune and endocrine systems to communicate, and play a complex role in regulating the deiodinases, according to rat studies.  They have been shown to stimulate D1 (in the short-term, with levels dropping after the first hour, but still always higher than in the controls) and D2 levels (rising through at least the first 24 hours, and higher than the controls after four hours) in the rat anterior pituitary, while lowering liver (always lower than controls but basically steady) and blood D1 blood levels (dropping below controls within four hours and but rising almost to that of the controls in 24 hours);  It isn't clear why this happens; the resulting rise in T3 levels in the pituitary should cause the latter to lower its TSH output, and it appears that it does so briefly (not surprising, since T3 has a brief effect), especially four hours after the cytokine-stimulating infusion.  This also drives down blood TSH levels, but in a nonlinear way: they are lowest at four hours after a lipopolysaccharide (LPS) infusion, experimentally introduced in order to stimulate the production of cytokines, but after a high point at one hour after that infusion, they drop to almost the same levels as the controls at 24 hours after the LPS infusion (Baur et al., 2000).

Disruption of the circadian cycle, i.e., by varying the timing of daily light and dark exposure, can intensify immune cell response (by not increase the number of immune cells) by elevating the blood levels of interleukin 6 (IL-6), a cytokine, triggered by an LPS infusion. 

History: from iodine nutrition management to thyroid hormone replacement

Although iodine deficiency (and, less often, excess) has considerable medical importance, it has been treated almost exclusively as a public health issue, rather than a medical one, by the countries of the industrialized world, although its consequences in terms of thyroid disease are generally treated with hormone replacement rather than be repairing a nutritional deficiency when the latter is diagnosed.  Salt iodization has been the public health strategy adopted by 70 countries, but this has proved difficult because of wide regional variations in environmental iodine.   Although iodization of salt processed "for human and animal consumption" has been pushed in Europe since the 1980s, less than 50% of European households use iodized salt, in contrast with 90% in North and South American counterparts (NIH, 2011), although the use of iodized oil has also been effective in iodine-poor regions in Europe. Andersson et al. (2007) claim that ongoing monitoring is necessary because compliance with salt iodization policies has a tendency to fall off, which suggests to me that discrepancies in reported iodized salt consumption among different countries may be due to variations in monitoring.  Median urine iodine levels range from 115 mcg/L in central Switzerland to 728 mcg/L in coastal Hokkaido, Japan (Zimmerman, 2005). 

It's now understood that those living close to oceans are less susceptible to iodine deficiency than those living inland, especially those living in mountainous areas, since iodide ions are transported by air currents from the ocean inland, where they are absorbed by the soil and are washed down the mountains by rain. Adhvaryu et al (2013) showed that there were even more striking differences between goiter rates in the northern states and the southern ones in the United States in 1917 WWI draft data, and that after about 30 years of salt iodization there were still some areas with high rates of goiters, though only in the extreme north of the country. These researchers theorized that this north-south divide was caused by a glacier during the Ice Age.

However, distributing iodine in salt in ways that reflect those geographical differences has not been practical.  Coping with wide variations in individual dietary iodine needs makes this even more difficult, since pregnant women in particular have higher needs and more than half (56.9%) of pregnant American women "surveyed during 2005-2008" had subnormal iodine blood levels (<150 mcg/L), with the corresponding median only 125 mcg/L (Caldwell et al., 2011).

Goiter, myxedema, cretinism, iodine and understanding of how the thyroid gland works: the early European contribution

Understanding that the thyroid gland had a purpose, that it produced a molecule that the body needed to prevent hypothyroidism syndromes called myxedema and cretinism and that iodine, necessarily obtained through the diet, was necessary to build this molecule, was mostly the work of Swiss, French and German scientists during the 1800s, although resistance from their conservative societies kept iodine from being used therapeutically to treat these diseases on a large scale, according to Carpenter (2005). The Swiss physician J.F. Coindet proposed treating goiter in individuals with iodine administered by and monitored by physicians in an 1820 paper; this was perhaps the only time this option was considered.  He actually treated patients with it, with iodine (the implication from Carpenter being that it was isolated from seaweed) dissolved in distilled alcohol, amounting to a daily dose of 165 mg; he reported no toxic effects, although some influential individuals claimed that their experience was otherwise.

An early clinical trial testing iodine, in the form of potassium iodide pills, as a treatment for goiter, took place in France in 1873, using a daily iodine dose of 7.6 mg for 75 days or until the goiter went away. It apparently was successful, with 74% of the subjects being cured; no adverse events were reported (Carpenter, 2005).

Contributions from the United States and Japan: from salt iodization to T4 replacement

Iodine deficiency and its treatment with salt iodization

Sometimes dramatic improvements in public health require the action of a "hero," a well-credentialed person with an important insight, progressive vision on how to act on it, and who is socially and/or professionally prominent. Such a person was American scientist David Marine, who organized, got acceptance and support for, and performed the first clinical trial (1917-1922) using iodine as the intervention that resulted in a change in official policy.  Having observed many dogs with goiters in Cleveland, Marine experimentally reduced goiters in these animals and suspected that there was a regional iodine deficiency problem that affected humans, too.  Although he was employed as a physician at Johns Hopkins University's medical school in Baltimore, Maryland at the time, he decided to return to the Goiter Belt part of the United States to perform an iodine supplementation clinical trial on schoolgirls in Akron, Ohio when permission to conduct it was denied in Cleveland on the grounds that iodine was dangerous. The trial was a success, proving both safety and efficacy.

Meanwhile, Love and Davenport (1920) estimated the goiter rate by state and discovered dramatic differences: these rates per 1000 ranged from 26.91 in Idaho to 0.25 in Florida (Table 18 in their article). Universal salt iodization, in spite of the regional differences shown in the Love and Davenport study, became an official means in the U.S. in 1924 to the end of eliminating goiters. The Morton Salt Company in the U.S. began to add potassium iodide (KI) particles to its salt and the rate of goiters in the Great Lakes area soon dropped to a small fraction of what its original level.  Unfortunately, there was a surge of thyrotoxicosis cases, resulting in an apparent excess of 102 "deaths in Detroit from goiter" in 1926 although this problem "died away after about 2 years," according to Carpenter (2005), who pointed out that similar problems were reported in Tasmania, Europe, South America and central Africa, and that "it now seems agreed that the problem occurs typically in people whose goiter, over long years of relative iodine deficiency, has become nodular and that this tissue is 'out of control' so that as more iodine becomes available to it the output of hormone(s) climbs to toxic levels" (p. 679).

Salt iodization has always been a voluntary program in the U.S.; processed food typically contains uniodized salt and must by law indicate in its labeling when it does contain added iodine (NIH, 2011).

Apparently, putting iodine in public water supplies was briefly considered. Olesen (1927) was skeptical of delivering iodine either in this form or via salt, implying that it was dangerous to administer iodine to those without goiters.  WHO (2008) explained that maintaining consistent levels of iodine in public water supplies would have been more difficult than doing so with salt, since 1) a major monitoring effort would be needed to cover the great number of different water sources (presumably municipalities) involved and 2) iodine is "unstable" in water (unlike fluoride ions?) On the other hand, because different parts of the country contain greatly varying environmental iodine levels, iodized public water would seem to worth considering for tailoring dietary iodine to local conditions.

However, the pendulum then swung the other way, with iodine excess in the U.S. gaining recognition as being much more common than iodine deficiency.  Since Japan had long been noted for its high iodine intake via dietary seaweed, American policy makers apparently connected the dots and decided that excessive iodine intake was the problem.  The FDA banned the use of iodine as a disinfectant in dairy cows in response to a flurry of studies coming out between 1978 and 1986 and as a dough conditioner in the bread-making industry, and American iodine intake, as measured by urinary iodine, dropped precipitously. 

However, this drop-off in iodine intake continued long after these measures had been instituted, and the CDC expressed concern in the National Health and Nutrition Examination Survey (NHANES) III study (Hollowell et al., 1998) that iodine deficiency was threatening to return and recommended follow-up, pointing out that median U.S. urine iodine levels had dropped from about 320 mcg/L over 1971-74 to about 145 mcg/L over 1988-1994; the percentage of the U.S. population having inadequate iodine levels, i.e., less than 100 mcg/L, according to World Health Organization (WHO) standards, increased from about 7% to about 32% in this same time interval (Fig. 1, p. 3405); both numbers were considered acceptable according to the WHO's standards, i.e., no more than 50% below 100 mcg/L.  These researchers recommended continuing follow-up of iodine levels in both the National Health & Nutrition Examination Survey (NHANES) IV, which is currently in progress and intended to focus on allergic disease, especially that of "common heart, vascular, lung, and blood diseases" (Bachorik, 2018), and in the U.S. Food and Drug Administration's Total Diet Study (2017) and added, "Surveillance of thyroid diseases should be emphasized, but we should not wait for the prevalence of goiter to increase or for changes in thyroid disease patterns to occur due to decreased iodine intake (my italics)." (p. 3408) Later on, the CDC (Blount et al., 2006) determined that the geometric mean of urine iodine in the U.S. was 126 mcg/L for women and girls at least 12 years old. In a later CDC study (Caldwell, 2011), median U.S. urine iodine concentrations were 164 mcg/L for both 2005-2006 and 2007-2008, with 9.8% and 8.8%, respectively for those time periods, having such concentrations below 50 mcg/L.  Pregnant women surveyed over the entire period had a median iodine concentration of 125 mcg/L.   NOTE: these numbers were reported in "mg/L" in this abstract, which I assumed to be a mistake. Lee (2014) covered this ground also, expressing concern about the apparent decrease of iodine intake and the lack of recent monitoring in the US.

Oddly, de Benoist et al. (2004), a European group working for the WHO, reported that Americans were getting too much iodine, with a urine iodine median somewhere between 200-299 mcg/L. However, they also reported that goiter prevalence in the Americas was only 4.6%, the smallest of any geographical region in their survey; they drew their information from undisclosed sources in the WHO Global Database on Iodine Deficiency, which in turns obtains its information from surveys of publications in databases such as MEDLINE. Troublingly absent from the study are Japanese and South Korean data.  However, Zimmerman et al (2005) filled this gap by determining the median urine iodine concentration in coastal Hokkaido in Japan to be 728 mcg/L in a paper concluding that an iodine intake leading to less 500 mcg/L in urine iodine was safe for children aged 6-12 years. Chow et al. (1991) had already observed that an estimated total daily dose of 750 mcg (involving an administered 500 mcg dose) slightly lowered free T4 and slightly raised TSH in Welsh adults, with none in the sample developing "biochemical hypothyroidism," with these numbers appearing to reach steady state by about 28 days.

Adding to the complexity, the American Medical Association was promoting the revocation of salt's Generally Accepted as Safe (GRAS) status and the Center for Science in the Public Interest has filed a petition to that effect so that the FDA can regulate the amount of salt in food for sale (AMA, 2008).  The FDA has examined iodine status in the U.S. population in the Total Diet Study since 1961 (Murray et al., 2008).

The shift to T4 replacement therapy

Sawin (2002) provides an account of the sea changes in the therapeutic approach to hypothyroidism. Until the 1950s, American physicians had treated goiters mainly with surgery. But three significant developments during that decade radically changed their role in the treatment of thyroid disease: 1) A Japanese physician's 1912 paper describing an apparent autoimmune thyroid condition, 2) the synthesis of T4 and its marketing as Synthroid, and 3) the invention of the radioimmunoassay (RIA), an analytical method capable of measuring very small blood concentrations of molecules such as hormones. Today, hypothyroidism is considered to have autoimmune origins, i.e., to originate in Hashimoto's thyroiditis, unless pituitary involvement is suspected (Merck, 2006, p. 1194).

Broda Barnes, a hypothyroidism researcher before his time?

Barnes received his M.D. in 1937, when a complicated process of measuring basal metabolism was the diagnostic standard for thyroid disease and before the radioimmunoassay was developed. He began his thyroid research in 1930 as an undergraduate, having been assigned to this topic as a physiology thesis subject by his advisor at the University of Chicago.  He found in his studies that rabbits whose thyroid glands had been removed developed many infections and a great variety of other physical problems and died at less than half their normal life span.  As a medical student afterwards at Rush Medical Center, he saw in some of his patients signs that reminded him of the problems affecting the rabbits he had studied. Although apparently crude methods existed for measuring T3 and T4 levels, he considered them to be unfeasible.  Instead, he discovered that oral body temperature, taken first thing in the morning, was a better indicator than the currently used measurement of basal metabolism rate, and published his findings in 1942 (Barnes and Galton, 1976, p. 43.)  He later modified this approach by using the armpit temperature, determining the normal range for this temperature to be 97.8-98.2 degrees Fahrenheit for all adults (p. 46).  Unfortunately, although his approach was initially ground-breaking, he apparently continued to use it at least through 1976, the publication date of this book, which contains no mention of TSH assays as diagnostic tools.  To be entirely fair, however, the precision of TSH assays at that time was awaiting great improvement. Another problem was that Barnes continued to use the English system rather than the metric system, vital to making the fine measurements of the small units necessary to determine proper medication doses (although he might not have been working with pills but with animal thyroids).  He prescribed doses in grains, e.g., "two grains of thyroid" in 1967 (p. 150). 

In the end, Barnes' diagnostic method was set aside, not simply because of the increasing improvement of the quality of the TSH assay but because of the apparent widening of the normal range for body temperature; it is generally accepted in the USA today that a temperature below 95(in effect, 95.0) F constitutes a medical emergency and that a temperature above 100.4 F is a fever.  Unfortunately, his studies relating hypothyroidism and susceptibility to infection were not followed up on, perhaps because many of the serious infectious diseases common in the early twentieth century are now preventable and/or curable by vaccines and antimicrobial medications, and some even eradicated or simply confined to what are best described as "invisible" populations. However, the threat posed by the COVID-2019 virus, which has been shown to cause serious illness in those with deficient immune responses, might cause our medical establishment to rethink this.

The Hashimoto Paper

New recognition was given to Dr. Hakaru Hashimoto's 1912 landmark paper (published in a German-language journal, not cited), which established a connection between the known clinical signs and symptoms of hypothyroidism and the histological hallmarks of a new disease, characterized by destruction of thyroid cells by lymphocytes. According to Sawin (2002), The Thyroid, a 1951 textbook by T.H. McGavack, noted that "most cases of spontaneous hypothyroidism have a lymphocytic infiltrate".  By "spontaneous hypothyroidism," they may have meant not related to iodine deficiency.  Several other researchers defined "autoimmune thyroiditis" as a condition involving this infiltrate and not always causing a goiter: in 1956, when urine iodine levels were apparently at their highest, one study induced autoimmune thyroid disease, triggering antibodies against their own thyroglobulin, in rabbits by injecting them with that thyroglobulin, along with Freund's adjuvant to help provoke an immune response, and another detected anti-thyroid antibodies in the blood of patients previously diagnosed with Hashimoto's disease. They found that this "lymphocytic infiltrate" "looked like" that seen in Hashimoto's disease. However, these researchers found it quite difficult to find subjects with "classic" Hashimoto's disease (only 12 in three years). To make a long story short, however, a major paradigm shift took place within 10 years and today Hashimoto's disease is now considered to be the most common cause of hypothyroidism. Yet one question remains unanswered: if it took introduction of a hapten to trigger anti-thyroglobulin antibodies in rabbits, what was causing the problem in humans?

Today, Hashimoto's thyroiditis "is believed to be the most common cause of primary hypothyroidism in North America" and calls for "lifelong" T4 therapy (Merck, 2006).  Today, there is still no consensus regarding the primary cause of this autoimmune condition. Hollowell et al. (1998) suspected that it was caused by temporary iodine excess because Dr. Hashimoto's Japan was noted for its unusually high urine iodine levels, believed to be linked to heavy dietary consumption of seaweed. But other possibilities remain to be explored. As we learn more about the function of the body's natural defenses, we can see how they can attack the thyroid in error, even when other targets, such as invading bacteria or viruses, stimulate their action.  Perhaps childhood illnesses ranging from measles to polio thereby raise the risk of autoimmune thyroid disease later on.

Rogue antibodies were eventually found in other forms of thyroid disease, such as Graves' disease, a condition that causes hyperthyroidism.  These antibodies mimic TSH, and might damage the thyroid itself by driving it too hard as a result..

Although measuring iodine levels and treating iodine deficiency in individual patients seems to be an obvious approach, no measures are being taken to modify treatment with diagnostic urine iodine testing or by the development of iodine pills in the US; autoimmune thyroid disease can be detected with the anti-thyroid perioxidase antibodies (TPO-Ab) test, though this measure is apparently not standard, perhaps because determining antibody presence is not helpful in determining the T4 dose.  

Synthroid and thyroid replacement hormone therapy

Synthroid, a synthetic, bioidentical levothyroxine product, came on the market in 1955 and later came to dominate the market. However, it was not the first synthetic T4 product to be marketed (Hilts, 2001).

The invention of the radioimmunoassay and ELISA assay, and the impact of their dramatic quality improvements on TSH level measurements

The radioimmunoassay (RIA), an analytical method capable of measuring very small blood concentrations of molecules such as hormones was developed in the 1950s, although for decades it was not sensitive enough to measure precisely concentrations near the bottom of the TSH normal range and for a while near its top:  The functional sensitivity (a precision measure) of the first-generation TSH assay was only 1.0 mIU/L. The functional sensitivity of an assay is the lowest concentration that it can measure according to a criterion of a maximum of a 20% coefficient of variance across difference performances of that assay; the coefficient of variance is the ratio of the standard deviation to the mean. Since the average American TSH value in adolescents and adults is apparently around 1.5 mIU/L according to third-generation assays, it is obvious that it would not have been feasible to use first-generation assays to determine the lower limit of the TSH, nor would determination of the mean TSH value have been possible. Only sometime in the 1970s, when the TSH assay's functional sensitivity was reduced to 0.1 mIU/L, could these measurements have had any practical value. That functional sensitivity improved to 0.01 mIU/L by 1989 and to 0.002 mIU/L in recent years, as reported by Arafah (2001). Today, a new non-radioactive assay method, enzyme-linked immunosorbent assay (ELISA), is often used instead. 

These improvements led to a narrowing of the TSH reference range, as researchers were able to make finer distinctions in the results of levothyroxine treatment.  Unfortunately, this made endocrinology practitioners' job more difficult because dosing errors on their part became easier to detect and because the narrowing of the range made it harder for them to keep a patient's levels within the range.  As a result, there was some pushback: one example was a British study, Taylor et al. (2014), noted the dropping of the upper limit of levothyroxine reference range from 8.7 to 7.9 mIU/L from 2001 to 2009, and presented this as a problem in itself, leading perhaps inevitably to the increase in the number of subjects experiencing abnormally low TSH values during the study.  While they had the option of attributing this to human error and expressing hope that greater experience could reduce this problem, their conclusion strongly suggested that the narrowing of the reference range was the sole source of the problem.  This study had eleven authors, only one of which (Panicker V) was affiliated with an endocrinology department (at an Australian hospital), while three (including the senior author) were members of the Thyroid Research Group in the Institute of Molecular and Experimental Medicine of the Cardiff University School of Medicine; other authors' affiliations indicate a variety of disciplines, including "Hygiene and Tropical Medicine," "Epidemiology and Population Health," and "Integrative Neurosciences and Endocrinology."  This suggests that the diagnosis and treatment guidelines set down by endocrinologists are being challenged mainly by those in other fields, and that at least some of these individuals have come to regard hypothyroid symptoms as being driven largely by psychiatric problems rather than by disabling physical health concerns.

U.S. Food and Drug Administration's Total Diet Study (2017)

This study measured the concentration in mg/kg of minerals, including iodine, in a variety of foods.  The mean iodine concentrations were generally highest in dairy products, both in milk and eggs, and the lowest in whole fruits and vegetables, and the difference was very large: for instance, the mean concentration of iodine in cheddar cheese was 0.448 mg/kg, while avocados and squash had zero.  One interesting outlier was "cake with white icing", which had a concentration of 1.500 mg/kg.  This suggests that iodized salt eventually came to be used in preparing processed foods in the U.S., although those who prefer whole foods, with an emphasis on fruits and vegetables, might have to rely on iodized salt or on multivitamins to get their iodine needs met.

Today: What is settled and decided (and what is not) regarding dietary iodine

The World Health Organization (WHO), based in Geneva, Switzerland, has stated that its major concern with respect to the prevention of thyroid disease is that iodine deficiency may cause a reduction in mental function: Andersson et al. (2008, p. vii) emphasize the need to prevent "more subtle degrees of mental impairment associated with iodine deficiency that lead to poor school performance, reduced intellectual ability, and impaired work capacity." With this in mind, the authors of this document did a country-by-country study of iodine levels in the world.

The WHO has set these standards for adult urine iodine levels (in mcg/L): <20: severe iodine deficiency, 20-49: moderate iodine deficiency, 50-99: mild iodine deficiency, 100-199: optimal iodine levels, 200-299: risk of hyperthyroidism in "susceptible individuals," >300: in addition, risk of autoimmune thyroid diseases (Andersson et al., 2007, Table 2.5, p. 14). Carpenter (2005) describes "susceptible" individuals as those who have compensated for long-standing iodine deficiency by developing goiter nodules that are dangerously efficient producers of thyroid hormones. Total goiter prevalence, i.e., the proportion of children in a population with a goiter, is an indirect measure of iodine deficiency that WHO also uses: for, respectively, none, mild, moderate and severe iodine deficiency, the lower limits for these categories are 0, 5.0%, 20.0% and 30.0%.  In addition, the WHO recommends using TSH levels, saying they are the "single best indicator" of neonatal "brain damage and impairment of intellectual development" (Andersson et al., 2007, p. 14).  Andersson et al. (2007) also note that rising TSH and T3 in combination with lowered T4 is typical for iodine-deficient populations, although all may remain within the normal range; I take this to mean that the real information is in the trend rather than in the absolute numbers.

The WHO (2008) recommends a daily iodine dose of 150 mcg, achieved by adding 20-40 mg of iodine (in the form of potassium iodide or potassium iodate) to each kilogram of salt, based on the assumption that most people consume about 10 g of salt a day.  This is about one thousandth of the dose that Dr. Coindet used in 1820, and about 2% of that used in the 1873 trial. According to the Food and Nutrition Board of the Institute of Medicine, the median daily iodine intake in the U.S. is about about 240-300 mcg/day for men and 190-210 for women, and the "tolerable upper level" for adults is 1.1 mg/day. Zimmerman et al (2005) reported that people in coastal Japan typically had a daily iodine intake of more than 10,000 mcg (10 mg).

In general, measuring iodine levels in individuals has never been part of standard medical practice, though it has been measured in research.  However, the Food and Nutrition Board of the Institute of Medicine (2001) points out that iodine deficiency can lower blood T4 and normal and/or raise blood T3, though still within their respective normal ranges.  This is no surprise: the enzymes D1 and D2 respond to this (as well as other conditions reducing T4 levels) by converting a higher proportion of T4 to T3, recycling the excess iodide ions removed in this process.  

At any rate, salt iodization had an unexpected benefit: the apparent rise of intelligence in individuals in the U.S. and its concomitant effect on social achievements.  Nisen (2013) cites a paper by Feyrer et al. (2013) that compared the American IQ-equivalent Army General Classification Test (AGCT) results from the two world wars, which had taken place before and after the U.S. salt iodization program began.  One of its key observations was that in the most iodine-deficient parts of the U.S., test scores increased by one standard deviation (15 IQ points) after salt iodization was begun. Adhvaryu et al. (2013) produced maps of rates of goiter occurrences based on WWI 1917 draft data and studied the impact of iodine supplementation and geography on high school graduation rates and participation in the workforce.  Qian et al (2005) and Bleichrodt et al (1994) also found evidence that IQs were reduced almost a full standard deviation in the most iodine-poor parts of their respective countries, i.e., China and The Netherlands.

Has salt iodization in the U.S. eliminated iodine deficiency? Dasgupta et al. (2008) mounted a convincing argument that this belief is based on false assumptions. Most obviously, fewer than half of American adults use any kind of table salt and, as pointed out earlier, processed food is rarely iodized. There was some variation in the concentration of iodine in iodized salt that they examined in freshly opened containers of salt: most (about 90%) ranged from 35 to 65 mg of iodine per kg of salt, although the range was 12.7-129 mg/kg, although most containers claim to have a concentration of 45 mg/kg. His data on the remaining contents of these containers as they are used shows a greater concentration of iodine over time. This should be no surprise, since it seems logical that the heavier iodine-containing molecules would sink to the bottom of the containers: salt, i.e., sodium chloride, has a molecular weight of 58, while potassium iodide has one of 166.

I recently bought some Morton Iodized Sea Salt. According to its list of ingredients, one-quarter teaspoon contains 25% of the RDA for sodium (590 mg) and 45% of the RDA for iodine (presumably 68 mcg). This means that enough iodized salt to meet the iodine nutrition needs of an average person would also meet about 50% of their need for sodium. Fortunately, this sodium requirement is far below the maximum sodium intake recommended by medical authorities. In any case, it appears that the public's main source of dietary iodine is multivitamin pills, each of which typically provides 150 mcg of iodine (Horn-Ross et al., 2001).

Iodine deficiency is apparently no longer considered to be a cause of hypothyroidism in the U.S., which is officially considered to be an "iodine-sufficient" country.  However, because studies measuring iodine levels are no longer done here, it is not clear that this is the case.  The official position is that, unless a non-thyroidal disease is implicated, hypothyroidism is caused by Hashimoto's thyroiditis; the strong relationship between TSH levels and those of anti-thyroid antibodies in many studies bears this theory out.  What is not known and apparently not explored is what causes this autoimmune condition and others of that type.

Iodine vs. key pollutants

Perchlorate Ions

Perchlorate, (ClO4-), used in the manufacturing of many products, is a contaminant of soil, groundwater and drinking water the levels of which have recently risen in the U.S.; it has come under FDA scrutiny because it can interfere with iodine uptake by the thyroid.  Although no article cited has stated this explicitly, perchlorates in general appear to be mainly water-soluble (as are iodide ions) and are apparently found in higher concentrations in food with high water content.  Accordingly, both perchlorate and iodine are highest in dairy products (probably more likely milk than cheese) and higher in vegetables and fruits than they are in meat (Murray et al., 2008).

Blount et al. (2006) of the CDC sought to nail down the minimum problematic perchlorate level and determine whether iodine level, at least in women, influenced that.  These researchers developed two regression models (one for women with iodine levels < 100 mcg/L, one for the other women) to estimate the increase in TSH and T4 levels relative to the increase in urine perchlorate levels. These models involved many independent variables in addition to TSH and T4 in their model to separate all significant relevant factors.  Perchlorate levels in women in both groups had a statistically significant positive correlation with their TSH levels; in women with abnormally low iodine levels, there was also a statistically significant negative correlation between perchlorate levels and T4 blood levels; T3 levels were not studied.  This may be due to greater conversion of T4 to T3, with recycling of the removed iodide ions; this compensates for the fewer T4 molecules available from reduced synthesis and conserves iodide ions, which are recycled.

The study showed that the more perchlorate in an iodine-deficient woman's blood, the greater the corresponding increase in TSH. In the low iodine group, at one extreme, a perchlorate increase of 0.19 to 0.65 ug/L, i.e., the 5th percentile, predicted a rise of 0.23 mIU/L in those with a baseline TSH of 1.40 mIU/L, i.e., at the 50th percentile.  On the other hand, a perchlorate increase from 0.19 to 13 mcg/L, i.e., the 95th percentile, predicted a rise of 2.12 mIU/L in those with a baseline TSH of 3.11 mIU/L, i.e., at the 90th TSH percentile; results were similar but somewhat less dramatic for T4 levels. 

These regression models included other independent variables, selected for their relevance, as is standard for such models; for women in the low iodine group, those variables that were statistically significantly correlated at the p=0.001 level to T4 levels were urinary perchlorate levels (negative), estrogen use (positive), log(C-reactive protein) (positive) and total kilocalorie intake (negative).  In the high iodine group, on the other hand, perchlorate levels were not correlated (p = 0.5503) with T4 levels, while those test results that were correlated (positively, as it turned out) with perchlorate levels at the p<0.001 significance level were log(C-reactive protein) and a positive pregnancy test.  TSH results for these two groups were different: for the low iodine women, statistically significant correlations at (or near) the 0.001 level were found for urinary perchlorate (positive) and for (p = 0.0016) beta blocker use (positive), while for the high iodine women, only age in years, log(BMI) (both positive) and a "non-Hispanic black" demographic classification (negative) were statistically significantly correlated at that level. 

These results indicate that dietary perchlorate has a bigger impact on the hormones of women with lower iodine levels; this is fairly straightforward. 

Potassium perchlorate has been used with methimazole to treat hyperthyroidism (Reichert and de Rooy, 1989).

The take-home message from this study is that, while it would be difficult to eliminate perchlorate ions from the environment, adequate dietary iodide ions would go a long way to keep it from interfering with thyroid hormone levels. 

Bromine compounds in dough conditioners, flame retardants and soda

Who would like to have a new, very reactive (but not very much fun) substance in his/her diet and/or furniture? Probably no one, but bromine compounds have made remarkable inroads into American life and getting rid of them is quite a project! 

Bromine, like iodine, is a halogen, its "upstairs" neighbor on the periodic table, does not have any proven nutritional value and has been shown to create problems.  It appears to displace iodine in the thyroid and increase the rate of iodine excretion, according to Czech research Pavelka (2004).  Although the FDA banned the use of iodine (in the form of potassium iodate (KIO3)), as a "dough conditioner," bromine, in the form of potassium bromate (KBrO3) has been allowed to stay and may indeed have been brought in to replace the banned iodine.  The FDA permits the use of bromine in baked goods, although it has established maximum amounts allowed and monitors bromine-based products. The Center for Science in the Public Interest filed a complaint with the FDA against its use (1999), and later (2005), in a paragraph within a 21-page document. Pavelka, a Czech researcher, published a literature review (2004) arguing that bromide ions replaced iodide ions in the rat thyroid; this is a matter of special concern in his country, where crop farmers use bromine products to "fumigate" crop fields between plantings.  However, Pavelka did not examine iodine deficiency at the start of the study as a factor, although that is a major problem in many European countries.

What is this "dough conditioning" function, then?  Both of these substances strengthen the dough by forming chemical crosslinks of gluten molecules, though at different times.  This is likely to be most important when a great deal of leavening takes place, enabling the loaf to keep from collapsing.  Noël Haegens (2013) describes the chemistry of bread baking in detail, and implies that only bromates will act at the important relatively late time in the dough-raising process, in contrast with early-acting potassium iodate.  So it is conceivable that before iodine was banned from the bread baking process, it might have been sometimes used in concert with potassium bromate. On the other hand, some Tasmanians replaced some of the bromates they were using in bread-baking with sodium iodate (Carpenter, 2005). I am not aware of any regulations requiring bromine to be listed as an ingredient in food, nor do I know if the listing of "dough conditioners" among baked goods ingredients implies the inclusion of bromine compounds.  There is apparently no law requiring individual "dough conditioners" to be specified in such lists of ingredients.

I personally prefer my bread lightly leavened and rather crumbly, and it is probably a good thing, too!

Another controversial source of dietary bromine in the U.S. is brominated vegetable oil, added to some fruit-flavored sodas to keep the flavor mixed in, according to Israel (2011).  The FDA monitors levels of this substance in food products, too.  On May 5, 2014, Strom (2014) reported that Coca-cola announced that it will replace brominated vegetable oil in its citrus-flavored sodas with sucrose acetate isobutyrate and/or glycerol ester of rosin. This action was apparently prompted by a 2012 petition started by Sarah Kavanaugh, who earlier successfully persuaded Pepsi Co. to remove this ingredient from its Powerade product.

Some brominated flame retardants have been found to have a toxic effect on the thyroid and liver in animal studies, according to a Swedish author (Darnerud, 2003).  The EPA (2013) is currently is working on eliminating flame retardants polybrominated diphenyl ethers (PBDEs) from the environment by discouraging manufacture and importing of such chemicals and combinations thereof because of concerns about "neurobehavioral effects;" unfortunately, there are several chemicals in this class and each requires a separate report. Part of that involves requiring the businesses involved to report any such activity. There are new organophosphate flame retardants, which unfortunately have created their own environmental problems (Salamova et al., 2013).

One general take-home message we can get from this is that banning a poisonous chemical from the market is a red tape ordeal, and that finding a safe and effective one to fill the same need might be even more difficult.

Bromine has been banned from food products in California and in most of Europe.

Fluorine and water fluoridation

The fluoride exposure issue has been around for a long time.  Bobek et al. (1976) showed that fluoride "administration" to rats caused a decrease in FT4, T4 and T3 blood levels in a two-month, two-dose study, although there was an increase in the T3-resin uptake ratio, a formerly used test which measures the proportion of T3 unbound to various proteins in the blood, e.g., TBG, and which might have been replaced by the FT3 test today. 

Chinese researchers have examined this issue extensively. For example, Zhao et al. (1998) discovered that iodine deficiency was associated with "dental fluorosis and increased fluorine content in the bone." More recently, Zeng et al. (2012) of the Tianjin Center for Disease Control and Prevention in China, used much more sophisticated technology to analyze metabolism in the thyroid cell when it was exposed to sodium fluoride (NaF) at four different dose levels: they determined that fluoride-exposed cells 1) were leaking more lactate dehydrogenase (suggesting structural damage), 2) had higher levels of "reactive oxygen species," 3) had a higher apoptosis rate and 4) had certain changes in the cell cycle: more time in the synthesis (DNA replication) phase, less in the growth/resting phases (G0 and G1).  The cell cycle data suggests that thyroid cells exposed to NaF were dividing more rapidly than the unexposed cells in the control group, which in turn suggests thyroid enlargement, which can lead to goiter formation.

The U.S. Environmental Protection Agency (2011) presented the results of an extensive analysis of the health risks associated with fluoride in the public water supply based on the findings of the National Research Council (2007).  The basic conclusions drawn from these documents is that fluoride added to the water supply by various products such as toothpastes has pushed the fluoride content of that water above the safety level in some places.  The focus of the EPA's document was on dental fluorosis, skeletal fluorosis and skeletal fractures; however, it apparently did not cover the influence of iodine deficiency on these conditions.

Europe's approach to combating iodine deficiency

The European Union has continued to recognize the problem of iodine deficiency and to work toward eliminating it.  In 2002, the Scientific Committee on Food determined the "tolerable upper intake of level of iodine" for different age groups; the level determined for adults was about 1700-1800 mcg/day, and that intake patterns in the U.K. (97.5% below 434 mcg/day for men) fell far below that level.  Therefore, the risk of overdose under normal current conditions was deemed to be minimal, although the Committee recommended strongly against using "iodine-rich algal products" because of the risk of overdose.  This set the stage for studies involving the use of iodized salt in breadmaking: Vandevijvere et al. (2012, on Belgian subjects) and Skeaff and Lonsdale-Cooper (2013, on New Zealand subjects) found that doing this increased iodine levels but still fell short of iodine sufficiency goals.  The Belgian study found that eating the fortified bread reached iodine level goals for children but not for their mothers.  The New Zealand study determined that the subjects, all children, had improved iodine levels, but still too low to meet expectations.  What was interesting about that study was that thyroglobulin levels were used to determine the presence of goiters, one of the criteria for iodine deficiency: thyroglobulin combines with iodide ions to form the thyroid hormones, so excess thyroglobulin apparently indicates a shortage of iodine.

What is clear is that iodine supplementation in bread in Europe and some other countries is now standard practice, and other foods are under consideration.

Psychological approaches

Zulewski et al. (1997) refined Billewicz (1969), a clinical scoring system relating "classical" symptoms and signs observed in subjects with and without hypothyroidism by investigators to the subjects' diagnosis via standard tests, i.e., of TSH, T4 and T3 levels. They selected subjects from four groups, classifying them with blood tests: 1) controls, who were determined to be free of thyroid disease, 1) "euthyroid" (acquaintances apparently free of thyroid disease) subjects, 3) those being treated with levothyroxine, 4) those with "subclinical" hypothyroidism, i.e., who had TSH levels above 4 mIU/L and had normal T4 levels, 5) those with fairly severe "overt" hypothyroidism, i.e., who had TSH levels above 20 mIU/L, below normal T4 levels, and normal T3 levels, and 6) those whose T3 levels were below normal. They then selected physicians unaware of these diagnoses and had them determine by exam which of 14 clinical signs and symptoms the subjects had. Selection of these signs and symptoms for the final diagnostic tool was done on the basis of comparison with subjects with severe overt hypothyroidism and the euthyroid (control) subjects on the basis of the predominance of these signs and symptoms in the former group. The most notable of those was ankle reflex relaxation time (long discarded from the standard annual physical), while "cold intolerance" and "pulse rate" were dropped from the final diagnostic tool. Then all of the subjects were evaluated using this tool, given a point each for each sign or symptoms that they had, and cutoffs were determined for the categories "hypothyroid," "intermediate" and "euthyroid."

Then the results of these clinical exams were summarized by the six groups determined via blood tests. The proportions of subjects in the "hypothyroid" category were: 1) controls: near zero, 2) "euthyroid" acquaintances: about 5%, 3) "subclinical": about 25%, 4) overt but with normal T3: about 40%, 5) overt and with subnormal T3: about 85%, and 6) those being treated with levothyroxine (T4): near zero. About 50% of the "subclinical" and overt with normal T3 subjects received "intermediate" classifications. These results strongly implied that "subclinical" hypothyroidism was on a continuum as far as "classical" detectable signs and symptoms went.

Notably absent was observation of the presence of a goiter. One possible reason for this is that hyperthyroidism can also cause goiters; pituitary gland activity typically causes goiters in those with hypothyroidism, while rogue antibodies act directly on the thyroid gland itself in the hyperthyroidism goiter of Graves' disease. Another factor is that goiters do not always go away when hypothyroidism is properly treated. This does not mean, of course, that there is no point in considering the presence of a goiter, but that the context has to be considered, making its interpretation more difficult and time-consuming. This argues well for the emphasis on using simple blood tests, but an effective measurement of clinical signs and symptoms is an important tool in assigning meaning to different blood test levels, and could help put to rest physician concerns that the new burden put on them by the requirement for giving those with TSH levels below 10.0 mIU/L consideration for treatment is not just an empty bureaucratic exercise.

Hypothyroidism (especially "subclinical") and measurement of thyroid parameters

Landmark observational studies

Sawin et al. (1985): The Framingham Study, a longitudinal study

This early study determined that 4.4% of the study sample aged over 60 years had TSH levels greater than 10 "microU/L" (presumably equivalent to "mIU/L") and that another 5.9% had TSH levels between 5 and 10 "microU/L".  They also noted that 39% of those in the former group and that 12.7% of the latter group had low T4 levels.  They expressed a preference for using TSH levels as the diagnostic criterion rather than T4 levels (perhaps total T4 rather than free T4, although they did not elaborate in the abstract.)

It is notable is that Hollowell et al. (2002) reported that about 5.0% of their study sample had any level of hypothyroidism, using a TSH level of greater than 4.5 mIU/L, rather than of 10.0 mIU/L, as the criterion.  Although subject selection differences might have accounted for this, it is possible that TSH assays at the time produced higher measurements for the same hormone levels.

Vanderpump et al. (1995): The twenty-year follow-up to the Whickham Survey, a longitudinal study

This landmark study (Vanderpump et al., 1995) was notable because it showed not only the relationship between presence of anti-thyroid antibodies and TSH levels, but that, given a certain TSH level, the presence of these antibodies greatly increased the chance of the affected individual developing "overt" hypothyroidism, defined as a TSH measurement of at least 10 mU/L and/or obvious clinical signs and symptoms of hypothryoidism.  On the other hand, because the baseline hormone values were measured in 1975, before the TSH assay was fully refined, it is not clear how precise those TSH measurements were.  But apparently overall averages are more meaningful.

But perhaps the most striking finding was the strength of both TSH and presence of anti-thyroid antibodies as predictors of the occurrence of "overt" hypothyroidism over 20 years.  There was an approximately 50% chance of someone with either 1) present antibodies and a TSH of about 3-4 mIU/L or 2) absent antibodies and a TSH of about 9-10 mU/L becoming overtly hypothyroid over that time period.  It's possible that the measurement precision problems presented less of a problem with comparisons than with absolute measurements because the errors in each category cancelled one another out.

Canaris et al. (2000): The Colorado Thyroid Disease Prevalence Study (cross-sectional study)

Unfortunately, the complete version of this widely cited U.S. study is not available to the general public.  But the abstract indicates that it provided information that complemented the CDC study in interesting ways: Canaris et al. (2000) did not attempt to develop a reference range, but took straight statistics from the information it gathered, apparently from a questionnaire and TSH, total cholesterol (TC), and low-density lipoprotein (LDL) level measurements, at a "Colorado statewide health fair," which involved 25,862 subjects.  On the basis of a TSH normal range of 0.3 to 5.1 mIU/L, the study determined that 9.5% of the participants had TSH levels above that range, and 2.2% below it, and that 40% of the subjects receiving thyroid medications had TSH values outside the normal range.  They also found that those with TSH levels in the range of 5.1 to 10 (probably actually 10.0) mIU/L had statistically significantly higher levels of TC and LDL than those with TSH levels in that 0.3 to 5.1 mIU/L normal range.  Finally, they found that, while no particular symptoms reported were closely associated with TSH levels, the total number of such symptoms reported were. 

The study had three important limitations: 1) subjects were ambulatory, not too sick to participate, and 2) subjects were not asked about use of cholesterol-lowering medications, which might have influenced their TC and LDL levels, and 3) subjects were not completely blinded: some had prior knowledge of their thyroid status, although their perceptions were apparently wrong in many cases.

The latter observation supports the conclusion that hypothyroidism affects negatively the experience of those who have it, even at what are generally accepted as "subclinical" levels, but that that experience varies in the way it is observed within a range of commonly understood symptoms. Although it was not a blinded study, it appears from the data that many participants, including those being treated, were not aware of their TSH levels and that their responses about their symptoms could be trusted to some extent. At any rate, any claims made by future studies that there is no evidence to support the presence of symptoms in those diagnosed with "subclinical" hypothyroidism, i.e., that is manifested by a TSH in the range of about 5.1 to 10.0 mIU/L, should be regarded with a generous degree of skepticism.

Hollowell et al. (2002): the NHANES III study (cross-sectional study), TSH summary statistics from the U.S. Centers for Disease Control (CDC)

The NHANES III study done by Centers for Disease Control (CDC) (Hollowell et al., 2002) included 17,353 subjects aged 12 and over (the "total population") selected by this method.  Using the criteria for "subclinical" hypothyroidism (TSH>= 4.5 mIU/L and normal T4) and "clinical" hypothyroidism (TSH>=4.5 mIU/L and low T4), they determined that 4.6% of the sample had subclinical hypothyroidism and 0.3% of it had clinical, i.e., overt, hypothyroidism.  The very small percentage of this population which had clinical hypothyroidism is noteworthy because the TSH cutoff was 4.5 mIU/L rather than 10.0 mIU/L, the new standard today in 2019. 

Although TSH levels are bounded by zero at one end, they in fact assume a normal distribution (described technically as "Gaussian" in the literature) slightly skewed to the right with almost half of these levels falling between 1.1 and 2.0 mIU/L, according to the NHANES III study (Figure 1, Hollowell et al., 2002).  Also according to Figure 1, slightly less than 2% of the subjects had TSH levels falling in the 5.1-10.0 mIU/L interval, and the numbers of subjects' TSH levels falling in the 10.1-20.0 mIU/L and 20.1-50.0 mIU/L intervals were barely visible on the graph.

According to this study, the overall TSH mean was 1.47 mIU/L (Table 3); the TSH median was 1.39 (Table 4).  Within this group, median TSH levels were 1.35 for the 12-19 age group, decreased to 1.26 in the 20-29 age group and then increased monotonically to 1.90 in the 80+ age group. Subjects in this reference group not only reported no signs or symptoms of thyroid disease but were determined by TSH (<4.5 mIU/L) and thyroid peroxidase antibody (TPO-Ab, negative) blood tests to be free of thyroid disease, and they were not 1) pregnant nor were they 2) taking estrogens, androgens, or lithium. 

The biggest surprise here: the "Black non-Hispanic" subset had statistically significantly lower TSH and higher urine iodine values (Hollowell et al., 1998) than the "White non-Hispanic" subset of this group, dwarfing male-female differences. One especially striking difference between these two ethnic groups is that TSH levels soared at an increasing rate with age in the "White non-Hispanic" group, while those in the "Black non-Hispanic" group never went very high and dropped off after the eighth decade of age. This might be due to discrepancies in aggressive medical treatment of old people: "White non-Hispanic" people in failing health might be subjected to life-saving treatment that still does not restore them to health more often than their "Black non-Hispanic" counterparts, who might simply die when their health fails and their TSH levels start to rise. In any case, do we have enough evidence to support the assumption that high TSH levels are normal and healthy at any age?

The "Risk Factors" population followed a more complex pattern.  Among women with risk factors (Figure 2), serum TSH values increased linearly with the age of the subjects from a mean of about 1.2 mIU/L in 20-29 age group to about 3.1 in the 70-79 age group, dropping to a mean of about 2.6 in the 80+ age group.  Although this might be interpreted to mean that the oldest women don't need as much medication, it might in fact indicate that a high TSH, i.e., over 3.0, is life-shortening, reducing the numbers of oldest women in that TSH category.  The mean TSH of "risk factors" men followed a "U" shape over age groups, reaching its maximum at about 2.2 in the 12-19 and 70-79 age groups and its minimum at about 1.6 in the 30-39 age group.  As was the case for women in the "Risk Factors" group, their average TSH dropped in the 80+ age group; unlike them, their average TSH went down to that of the 80+ age group in the reference population. 

Spencer et al. (2007): re the NHANES III study, the influence of anti-thyroid antibodies on the determination of the reference range from the CDC (2007, cross-sectional study)

A 16,088-subject study of the prevalence of anti-thyroid antibodies, i.e., thyroid peroxidase antibodies (TPOAb) and thyroglobulin antibodies (TgAb) relative to TSH levels led Spencer et al. (2007) to revisit their assumption that the presence of such antibodies is a reliable indicator of hypothyroidism.  They found that older subjects tended to have higher TSH levels but more often lacked TPOAb and TgAb.  Subjects with TSH levels higher than 10 mIU/L with median ages of 71 for men (n=45) and 74 for women (n=82), and the proportion of those being negative for these antibodies were, respectively, 31% and 11%.  The authors attributed these anomalous cases to "atrophic thyroiditis".  The implications: these findings would improperly lower the upper limit of the reference range for older people.

It is also possible, as the Merck Manual (2006, p. 1200) implies, that longstanding Hashimoto's thyroiditis eventually destroys most of the thyroid gland, thereby reducing the target of the anti-thyroid antibodies.

They found a strong overall positive correlation between TSH levels and positive TPOAb, but not of TSH levels and positive TgAb, although of course they found that neither antibody was a perfect predictor of TSH levels and that the relationship weakened as subjects' age increased.  They did not consider separate reference ranges for different racial/ethnic groups necessary since there were differences across them in the TSH-antibody relationships that probably had not been identified.

Surks and Hollowell (2007): The TSH means age differences issue analyzed

Surks and Hollowell (2007) studied the different distributions of TSH values and proportion of those positive for TPOAb for eight "disease-free and reference" age groups, ranging from 12-19 to 80+, concluding that higher values for the oldest group especially called into question making the top of the TSH reference range 2.5 mIU/L.  The medians for these groups ranged monotonically upward with age from the 20-29 age group from 1.37 to 2.08 mIU/L.  It is worth mentioning, however, that there was much greater variation within the age groups than across them.

Alevisaki et al. (2009): How subtle differences in abdominal fat are related to big differences in free T4 in normal subjects

Alevisaki et al. (2009) took a very intriguing trip off the beaten path with their cross-sectional study of the relationship of abdominal fat, both subcutaneous (SF) and preperitoneal (PF, previously shown to behave similarly to harder-to-measure visceral fat), to free T4 levels in disease-free individuals.  They discovered a significant correlation between the ratio of SF and PF indicating that a slight decrease in that ratio was associated with a large increase in free T4 levels.  When they replaced free T4 with total T3 in their model, however, the correlation was much less and showed a trend in the opposite direction.  Unfortunately, the authors did not give a rationale for choosing the SF-to-PF ratio nor did they speculate about what these results might mean or offer suggestions for directions of future studies.  But at the very least, their results challenge a common perception that there is very little variation in health parameters among the "normal."

Walsh et al. (2010): A 13-year Australian longitudinal study examining the risk of developing hypothyroidism on the basis of TSH and anti-thyroid antibody levels

Walsh et al (2010) designed a study with a very sophisticated (and difficult to follow) analysis method to determine what levels of TSH measured in 110 healthy subjects were the most likely threshold levels for developing hypothyroidism, and which other factors, including anti-thyroid antibodies, had the biggest association with it. They used as baseline values those from a 1981 study which included (normal) subjects with these characteristics: TSH levels between 0.4 mIU/L and 4.0 mIU/L, with TPOAb levels less than 35 kIU/L and TgAb levels less than 55 kIU/L. They then measured these values for the remaining subjects in 1994. The dependent variables, also known as "outcomes", were 1) "hypothyroidism", defined in subjects as a TSH of greater than 4.0 mIU/L or current treatment with T4, and of 2) "overt hypothyroidism", defined as a TSH of 10.0 mIU/L or higher or current treatment with T4. The study included numerous independent variables, i.e., those examined as possible predictors of the dependent variables and others needed to stratify the model.

They performed a logistic regression that showed that sex (female), TSH levels, and especially anti-thyroid antibody levels (more so for TPOAb than for TgAb) were strong predictors of hypothyroidism defined by TSH level category. This identified the variables for which to develop a receiver operating characteristic (ROC) curve for two levels of baseline TSH, i.e., measured at 1981, in order to determine which one was the better predictor of hypothyroidism when measured 13 years later, i.e., in 1994. This analysis determined for each subject and candidate baseline TSH value (2.5 mIU/L and 4.0 mIU/L?) whether the prediction of eventual hypothyroidism based on the subjects original TSH represented a true positive, false positive, true negative, or a false negative. Then they used these values to construct the ROC curves for TSH, TPOAb, and TgAb for hypothyroidism in general and for overt hypothyroidism. For each, they used this method to build the distribution curves for this data: 1) to measure the sensitivity, i.e., the ratio of the true positive counts to the false positive counts for a series of candidate TSH values, and 2) the specificity, i.e., the ratio of the true negative counts to the false negative counts for a series of candidate TSH values. When the curves were built, they calculated the area under them (the C statistic). The ratio of this area to that of the square containing each curve was the resulting sensitivity and specificity. On the basis of these numbers (the bigger the better), they determined that the ideal TSH threshold values were 2.4 mIU/L for hypothyroidism and 2.6 mIU/L for overt hypothyroidism; 29 kIU and 32 kIU for TgAb for hypothyroidism and overt hypothyroidism, respectively; and 22 kIU for TPOAb for both categories of hypothyroidism.

Thayakarn et al. (2019): A comparison of adverse events for treated patients in eleven different cohorts, separated by TSH levels, mostly within the reference range (retrospective study)

U.K. researchers Thayakaran et al. (2019) compared the incidence of heart disease, bone fractures, and overall mortality across eleven categories (including seven in the reference range of 0.4-4.0 mIU/L, as well as four categories representing overt and subclinical hyperthyroidism and hypothyroidism) of TSH levels in 160,439 hypothyroidism patients selected from the The Health Improvement Network (THIN) database (gathered from primary care providers in the U.K.) who were receiving thyroxine treatment over a 23-year period.  Tthey used data for an average of 5.32 and a median of 6 annual visits per patient.  They included patients 18 years or older with a diagnosis of hypothyroidism at baseline and at least one TSH measurement done at a later visit, and without a diagnosis of heart disease, i.e.,  ischemic heart disease, heart failure, stroke/transient ischemic attack, and atrial fibrillation, or of bone fractures, at baseline.  Notable among the many independent variables used in the analyses were whether the subjects had been given prescriptions for levothyroxine and for lipid-lowering medications.  They were then each assigned to one of eleven categories on the basis of their TSH levels.  The average age of the subjects appeared to be about 57.

Notable results: Overall, using TSH=2.0-2.5 mIU/L as the reference category: 1) the hazard ratio for all-cause mortality jumped to about 1.29 at TSH=4.0-10.0 mIU/L and to 2.21 at TSH>10.0 mIU/L, 2) for heart failure, up to 1.4 at TSH>10.0 mIU/L, but down to about 0.3 for TSH=0.1-0.4 mIU/L.  When broken down by sex and age, results for women and those over 65 were anomalous in these ways: 1) at the lowest TSH categories, their risk for heart failure went down, and 2) in the 3.0-10.0 mIU/L ("subclinical") categories, their risk for stroke and transient ischemic attack went down slightly.  They also found a slight but statistically significant increase in "fragility fracture" risk at the highest TSH levels, especially in women and older people.  And, surprisingly: "TSH below 0.4 mIU/L was protective for atrial fibrillation in women, and TSH 0.1-0.4 mIU/L was protective for atrial fibrillation in patients aged 65 years or under."

Results in the seven reference-range categories were similar.   The authors pointed out that this gave physicians freedom to adjust levothyroxine doses to the values within the reference range that best satisfied their patients' preferences.  They cited a study, i.e., Iverson and Mariash (2008) that, based on the recognition that there is a large difference in the patient experience between the extremes of the generally accepted normal range, i.e., 0.4-4.0 mIU/L, used comparisons with free thyroxine data from a control group of 78 euthyroid subjects to help determine a method for finding the ideal TSH value for each of 58 hypothyroid patients.  They found that when the patients were treated, their TSH values were similar to those of the control group, i.e., 1.60 mIU/L vs. 1.73 mIU/L and not statistically significantly different, but that the free T4 values of the treated patients, i.e., 1.36 ng/mL vs. 1.10 ng/mL, were statistically significantly higher than those of the control group.  They recommended on the basis of these results that physicians adjust dosage of patients so that their free T4 values were in the "upper part" of the reference range.  This suggests to me that hypothyroid patients' bodies might be compensating in some way for their disease by creating free T4 more efficiently, which might (misleadingly) bring more of them into the "subclinical" range from the "overt" range as a result.

The jump in mortality in patients with TSH>10.0 mIU/L, i.e., 121%, might have been caused by the older average age of these patients, although the study did not state this explicitly.  Other studies have shown that most people with these very high TSH values tend to be over 70.  Although some researchers have argued that unusually high TSH levels are a normal part of aging and therefore not a point of concern, those values might also indicate health deterioration causing (possibly preventable) death, and these levels, and maybe even lower ones, might be an ominous sign in younger patients. What is also interesting is these dramatic findings were not matched by any of the other endpoints, even though cardiovascular disease and bone fractures apparently have been widely believed to be the main problems associated with out-of-range TSH levels.  What were the other causes of mortality in the case of very high TSH levels?  And at what point in the subclinical range, i.e., TSH=4.0-10.0 mIU/L, does the big turnaround in overall mortality happen? These issues need to be explored.

Another concern is the effect of physician intervention.  What all of the subjects have in common is that they were under physician care.  This might explain why there were fewer cardiovascular problems at the hyperthyroid end of the spectrum: physicians might have been more cautious about overtreating them with levothyroxine therapy than about undertreating them, with the result that TSH levels that were below the reference range in this study were more likely to be found in healthy patients than in those with heart disease relative to those at the other end of the TSH spectrum.  And if hypothyroidism remains untreated until old age, the heart might not be able to take the stress of initiating therapy at that time.  By this reasoning, it makes better sense to start treating younger people long before their TSH levels rise to 10.0 mIU/L, when signs of thyroid failure appear but heart health is still normal.

Landmark Clinical Trials and Meta-analyses


Taylor et al. (2013): A meta-analysis of observational studies: Disease trends within the TSH reference range

Cardiff University and Newcastle University (U.K.) researchers Taylor et al. (2013) expressed skepticism about the quality of studies of subclinical hypothyroidism. They expressed concern that past studies were underpowered and that selection bias in these studies was quite likely caused by many subjects having been asymptomatic and having had their TSH levels tested for reasons other than suspicion of thyroid disease. This selection bias might have caused these subjects to be healthier as a group than the population they were meant to represent, leading to conclusions downplaying the problems associated with having subclinical hypothyroidism. These researchers coped with this problem in this 2013 study by looking at a much larger sample, in this case, including the overwhelming majority of available subjects with TSH values within the reference range, thereby increasing the power of the study as well as eliminating that selection bias.

Through a meta-analysis of observational studies (40 selected from 985 reviewed), they discovered that there were significant patterns of non-thyroidal disease relative to TSH levels within the reference range for women; they stated that the absence of this pattern for men might have been an artifact of their fewer numbers in the study, i.e., that it was underpowered in that respect. These non-thyroidal diseases were cardiovascular events and risk factors and pregnancy problems, which increased with TSH level, while the reverse was true for bone mineral density and fractures. They stated that the data supporting these conclusions had high quality, but that for neurological and psychiatric associations they were inadequate.

Mutlu et al. (2013) performed a clinical trial examining oxidative parameters

Turkish researchers Mutlu et al. (2013) found that superoxide dismutase (SOD) was reduced statistically significantly in subclinical hypothyroidism subjects when they took L-T4.  Unfortunately, only the abstract is available to the general public.

Stott et al. (2017): A clinical trial (the TRUST study) involving various variables in patients aged 65+ in four European countries:

A very influential study (which was in turn the largest study in the very influential clinical trial meta-analysis Feller et al., 2018), Stott et al. (2017) performed an elaborate randomized, placebo-controlled clinical trial with 737 subjects, all selected from "clinical laboratory and general practice databases and records" in four countries (U.K., France, Germany, and the Netherlands) and aged 65 and older, using levothyroxine (L-T4) as the study drug.  The subjects met these requirements: "persistent subclinical hypothyroidism," i.e., having TSH levels, i.e., in the range of 4.60-19.99 mIU/L determined by at least two measurements made 3 months to 3 years apart, free T4 levels in the reference range, without major heart health problems or recent hospitalizations, and not having a prescription for levothyroxine or any other drugs affecting their thyroid or containing iodine.  The selected subjects averaged 74.4 in age, and 53.7% were women.  The subjects were started on either of two doses of levothyroxine, i.e., 25 and 50 mcg, depending on their state of health, and their later dosage was adjusted as needed to maintain their TSH levels within the reference range (0.40 to 4.59 mIU/L); the TSH levels (in mIU/L) for the treatment group averaged 6.41 at baseline, 3.63 at 12 months, and 3.47 at follow-up.  The primary outcomes were subject responses to two questionnaires: the Hypothyroid Symptoms score and the Tiredness score on the Thyroid-Related Quality-of-Life Patient-Reported Outcome measure (ThyPRO). Subjects rated their feelings on a scale from 0 to 100.  Two "secondary" outcomes were the questionnaires EuroQoL (EQ VAS), which had a scale from -0.59 to 1.00, and the Group 5-Dimension Self-Report Questionnaire (EQ-5D), which had a scale from 0 to 100.  They also added a Hyperthyroid Symptoms score to capture symptoms of overdose.

Also included were many and varied health measures, including those of cognitive acuity, muscular function, blood pressure, BMI, and the occurrence of fatal and non-fatal cardiac events; the results of only a few wound up in the reported results. Although there were no statistically significant differences seen across the treatment groups for atrial fibrillation onset, heart failure, bone fracture, or osteoporosis onset, treated patients experiencing at least one "serious" adverse event outnumbered their counterparts in the placebo group to a statistically significant extent (P=0.049): 201 (21.2%) vs. 142 (27.9%).  Another statistic of concern, although not achieving statistical significance, was the number of deaths from all causes: 10 (2.7%) in the treated group vs. 5 (1.4%) in the placebo group; but the total number of cardiovascular adverse events were about the same.  They also looked at measures of hyperthyroidism caused by overtreatment and found none.

The Tiredness score decreased a (conventionally defined) statistically significant amount over the extended period (P=.05) and the EQ-5D score (a measure of quality of life) increased a statistically significant amount over that extended period (P=0.03), although they did not do so over 12 months.  On the other measures, i.e., hand-grip strength, blood pressure, body-mass index (BMI), and waist circumference, differences were not statistically significant, nor were they close to it.  The authors dismissed the results suggesting a quality-of-life improvement, in part with a reference to a previous study and concluded that "treatment with levothyroxine in older persons with subclinical hypothyroidism provided no symptomatic benefits."   On the other hand, there were no significant indications that levothyroxine treatment was harmful, certainly an important point to consider in recommending against such treatment.

The authors made these observations about the limitations of the study: 1) very few patients in the study had TSH levels above 10.0 mIU/L, 2) they recognized that "some authorities" believed that a lower target TSH range, i.e., 0.40-2.50 mIU/L would have been best although their own target range was apparently simply minimally below the upper limit of the reference range, 3) they did not consider anti-thyroid antibody levels, and 4) symptom levels were "low" at the beginning of the trial, reducing the chance of dramatic improvement thereof occurring in that trial.  Indeed, the fact that these patients were not currently receiving levothyroxine therapy at baseline suggested that neither they nor their physicians considered such treatment to be needed; their motivations for entering the trial were not clear. 

Feller et al. (2018): A meta-analysis of clinical trials (2018)

The definition of "subclinical" hypothyroidism in clinical research is defined by lab test values and, by implication, lack of current treatment for hypothyroidism (whether or not the particular patient could have benefited from it).  Whether "subclinical" hypothyroidism should be treated is a controversial subject, but one meta-analysis (Feller et al., 2018) attempted to answer that question once and for all.  It selected 21 clinical trials from 1650 (non-duplicate) relevant studies with a T3 and/or T4 intervention on the basis of their definition of "subclinical" hypothyroidism, no concomitant medical conditions, the intervention being T3 and/or T4 administration, and of whether at least one outcome was represented in a pre-determined list of candidates.  Perhaps the most prominently featured, and heavily weighted, was the 2017 Stott et al. study mentioned above.  These "endpoints" were many and varied, with the body mass index (BMI) being by far the most frequently represented, i.e., in 15 of those 21 studies, even though the science supporting BMI as having diagnostic value for hypothyroidism is weak at best.  Some others included blood pressure (7 studies), the Beck Depression Inventory (3 studies), the Underactive Thyroid-Dependent Quality of Life (2 studies), the Billewicz score (1 study), and a variety of cognitive functioning measures.  In general, there was a very heavy emphasis on questionnaires about the subjects' feelings.  TSH upper limits in the studies' definitions of "subclinical" hypothyroidism ranged from 3.5-10 mIU/L to 4.6-19.99 mIU/L, with some setting no maximum TSH value.  On the other hand, all made "normal free thyroxine" a requirement for that definition.  These studies were mostly small, with only two involving more than 100 subjects.  Significantly, however, the authors noted that "no study included pain as an outcome."

Feller et al. concluded that subclinical hypothyroidism was not improved by treatment from this meta-analysis.  Indeed, the calculated results indicated that the differences between those who were treated and those receiving placebo was vanishingly small.  A problem with the study design is that, because the subjects were untreated at the study outset in spite of having higher-than-normal TSH values, they were not necessarily representative of all of those with the same level of thyroid failure; what proportion of those in the entire "subclinical" thyroid disease population were currently being treated?  This is a general concern that I have: no study seems to have made an effort to determine, among (untreated) patients who are deemed "subclinical" on the basis of their test results, how many sought treatment and were denied it, as opposed to those who did not seek treatment.  And among the latter, how many felt that their health was normal and how many felt sick without they themselves or their physicians suspecting thyroid disease?  Until these issues are clarified, results finding no benefit in treatment of patients with "subclinical" hypothyroidism might wrongly lead to all patients with test results in the "subclinical" range being denied treatment by government decree regardless of their level of illness.

It is not clear if any of these studies measured change from baseline, which is a standard safety and efficacy measure in formal clinical trials. The authors did admit that 14 of these 21 studies did not cover symptoms at baseline.  Feller et al. did write: "For outcomes on which studies reported treatment effects at different time points, we included only the estimate at the most recent follow-up time point, thereby avoiding counting a study twice in a formal meta-analysis." (p. 1351)  Actually, I was under the impression that the conclusive p-values from the clinical trials were the data points used in the meta-analysis; otherwise, it was unclear how the meta-analysis authors arrived at their conclusion. Nevertheless, it is hard to see the validity in any clinical trial that does not measure change from baseline.

What is interesting is that Nazarpour et al. (2018) did a validation study of the Billewicz scoring system using TSH and free T4 test result ranges as the independent variables.  So the Feller et al. study essentially did the reverse, which seems to be circular reasoning.

What is also interesting is that Garber et al. (2012) in their official AACE/ATA statement say in their Recommendation 5 (assigned a grade of "A"): "Clinical scoring systems should not be used to diagnose hypothyroidism."  The Feller study depends very heavily on them in the form of questionnaires (but not on physicians' clinical exams).  However, this does not necessarily indicate a conflict because the cause-and-effect nature of properly conducted RCT clinical trials allows us to measure improvements over time that can be attributed to the effects of the study drug. However, if there are no change-from-baseline measurements, results would not be meaningful; such calculations were apparently not done by Feller et al.

Salman Razvi, Robin Peeters, and Simon H.S. Pearce (2019) expressed their concerns about this meta-analysis in a letter to the editor.  They pointed out that the largest study (n=737) in the meta-analysis (TRUST, i.e., Stott et al., 2017) recruited mostly "asymptomatic older individuals with subclinical hypothyroidism" (average age 74), gave 75% of them less than or equal to 50 mcg daily, and were concerned that measurement of quality of life did not fully capture their symptoms and therefore the severity of their disease.  On this basis, they argued that because of the meta-analysis' conclusion that subclinical hypothyroidism should not be routinely treated, "treatment may be erroneously denied to young or symptomatic patients with subclinical hypothyroidism." Feller, Rodondi, and Dekkers (2019) replied that 95% the patients did indeed have hypothyroid symptoms in the TRUST study, that in fact their mean "tiredness" score was 25 on a scale of 0 to 100. They also pointed out that the other studies came to similar conclusions.  Finally, they argued that the results of all of the studies showed a lack of improvement of these symptoms with levothyroxine therapy. There are still some questions to be answered: why were the patients in the study not already receiving treatment when they had hypothyroidism symptoms? Was is that their physicians believed they were well or that they were not sick enough, or was it because they did not have a relationship with a regular physician?

A hypothyroidism patient satisfaction study: do problems of a few discourage treatment for most?

Peterson et al. (2018) conducted an online survey of 12,146 patients self-identified as currently being treated for hypothyroidism at the American Thyroid Association site. On a scale from 1 to 10, where 1 represented the lowest level of satisfaction, the overall average was 5. Among those reporting having no complicating conditions, those being treated with levothyroxine (L-T4) reported an average score of 5, while those being treated with an L-T4 and thyronine (L-T3) combination, the average level was 6, and for dessicated thyroid extract (DTE), the average level was 7. The respondents also indicated a maximal need for changes in their therapy: on a scale of 1 (no need) to 10 (strong need), they averaged 10. They also indicated that hypothyroidism had maximally affected their lives, with an average level of 10 on a scale of 1 to 10.

It is easy to see on the basis of these results why practitioners would feel reluctant to treat hypothyroidism in patients who did not have a clear health need for it on the basis of trustworthy scientific studies. This is quite likely the motivation for so many studies of the risk of heart disease on one hand and of bone disease on the other, and the caution in drawing hard conclusions about when to treat hypothyroidism by the standard methods. Granted, we have no way of knowing what these patients' TSH levels were and the bases of their diagnoses. But what could be going on here, given that these patients were clearly ill at diagnosis and needed treatment to restore normal function? Could it be that increasing energy made them more expressive of their dissatisfaction with their situation? Would it not make sense for people with low energy to choose their battles while those with enough energy might complain more and try harder to change their situation? And what are the chances that some of their unhappiness is due to having to fix problems accumulating while they were too ill to keep up with them? Perhaps we underestimate the amount of adjustment that these people have to go through to restore their lives to normal, and the frustration they might feel with such problems that are too late to fix.

The other thyroid diseases: their influence on diagnosis of hypothyroidism

What about hyperthyroidism and thyrotoxicosis? How do views about them influence diagnosis of hypothyroidism?

Hyperthyroidism is actually a subset of thyrotoxicosis.  When TSH levels fall below the lower limit of its normal range, or T3 and/or T4 that rise above the upper limits of theirs for any reason in an individual, s/he has thyrotoxicosis.  Hyperthyroidism, on the other hand, is a type of thyrotoxicosis caused by excessive T3 and/or T4 production by the thyroid gland (Floyd, 2013); the most serious cause is Graves' Disease. This condition, caused by an apparently intractable immune system malfunction in which anti-thyroid antibodies mimic TSH, is justly feared by physicians because of the strain it puts on the heart and because its successful treatment often requires permanent destruction of the thyroid via surgery and radiation.  At the other end of the thyrotoxicosis spectrum is a slight and temporary drop in TSH below the bottom of the normal range that sometimes occurs as a result of treatment for hypothyroidism.

Cleveland Clinic endocrinologists Pantalone and Nasr (2010) respond to what seems to be extreme physician fear of low TSH test results, however briefly present, by explaining that there are many potential causes for "subclinical hyperthyroidism," that they cause different FT4 and FT3 level patterns, and sometimes widely varying hormone levels. For example, one type of problem that temporarily suppresses TSH levels, subacute thyroiditis, a time-limited condition involving injury (and eventual repair) to some thyroid follicles, causes the sudden release of the thyroid hormones in these injured follicles into the blood.  It is one of those conditions producing rapidly changing test results; therefore it can masquerade as hyperthyroidism for a short period of time.  Pantalone's and Nasr's take-home message to physicians: do not take irreversible measures until the very many (and sometimes common) possible and relatively benign causes of these scary numbers are considered and ruled out.

Evidence suggests that one reason hypothyroidism might be undertreated is that the US medical establishment regards thyrotoxicosis as not just a greater danger, but so dangerous that the risk of getting it needs to be eliminated altogether.  One compelling reason is that heart disease is apparently considered to be America's most important health problem because it directly claims the most lives, and hyperthyroidism raises that risk more obviously than hypothyroidism does.  According to Jayaprasad and Johnson (2005), hyperthyroidism raises the risk of atrial fibrillation substantially, although it still affects a small minority of hyperthyroid people, i.e., according to the U.S. Preventive Services Task Force (2004), one in 114 cases of medication-induced "TSH suppression."

CDC researchers Hollowell et al. (2002) turned up evidence that hypothyroidism is under-treated, while hyperthyroidism is not (Figure 3).  To illustrate this, they defined a "disease-free" population, i.e., all of those who self-reported not having thyroid disease (but might in fact have had "biochemical evidence" of thyroid disease).  This "disease-free" population had almost the same proportions of TSH > 4.5 mIU/L, i.e., were hypothyroid, as the total population over all studied age groups, much greater than that of the "reference" population.  On the other hand, the "disease-free" population had the same proportions of TSH < 0.4 mIU/L, i.e., were hyperthyroid, over these age groups as the "reference" population, although significantly less than those of the "total" population. 

Why would this be the case? One factor may be an independent group, the USPSTF, which stated in 2004 that it believes that "In general, values for serum TSH below 0.1 mIU/L are considered low and values above 6.5 mU/L are considered elevated" and that it "found fair evidence that the thyroid stimulating hormone (TSH) test can detect subclinical thyroid disease in people without symptoms of thyroid dysfunction, but poor evidence that treatment improves clinically important outcomes in adults with screen-detected thyroid disease."  Furthermore, "the potential harms of screening and treatment are principally the adverse effects of antithyroid drugs, radioiodine, thyroid surgery, and thyroid replacement therapy if detection and early treatment for subclinical disease are unnecessary."  This suggests that this group believes overdiagnosis of hyperthyroidism is more dangerous than its counterpart for hypothyroidism.

There is another, less obvious, reason for physician fear of accidentally triggering thyrotoxicosis, i.e., problems with TSH assays in the latter half of the twentieth century. Until some time in the 1980s, assays were not able to measure values at the lower end of the normal range well enough for the bottom of that range to be established and patient values to be measured with satisfying precision, long after those assays were able to produce trustworthy measurements of the top of the normal range. One early way to minimize the threat of a patient's TSH going too low was to make sure that it was just below the high end of the normal range. Even today, when TSH assay precision concerns have substantially been eliminated, this concern apparently lingers, as the 2004 recommendation of the USPSTF suggests and which Pantalone and Nasr (2010) try to mitigate so vigorously.

There is hope that hyperthyroidism can be treated humanely and effectively someday.  Antithyroid drugs that inhibit thyroperoxidase (TPO) inhibitor expression are one form of standard treatment, although they have substantial side effects.  Chinese researchers (Cai et al., 2014) have shown that diosgenin limits goiter development in Graves' disease by inhibiting thyrocyte proliferation.

Hyperthyroidism diagnosis and treatment challenges: a Massachusetts General Hospital Case

Chiappa et al. (2019) reported an especially difficult case seen at the hospital emergency department.  The patient, a 39-year-old woman, was obviously seriously and acutely ill, but with no specific symptoms other than gastro-intestinal distress.  The focus of her initial visit was stool and urine exams, which were normal.  On her second visit, they ran blood screens, including the standard blood chemistry panel and tests for substance abuse, but her TSH was not tested.  They also ran a CT scan of her abdomen and performed an electrocardiogram, which showed some arrhythmias.  Of course, they took measures to rein in her runaway pulse with drugs as they worked toward determining the ultimate cause of her problems.  They then considered the following candidate diagnoses: 1) various hemodynamic causes of her arrhythmia, 2) kidney stones, 3) "substance use," 4) "serotonin syndrome" (typically caused by substance abuse), 5) ovarian teratoma, and finally 6) hyperthyroidism, based on her rapid pulse, unusually great recent weight loss, and certain "neuropsychiatric" signs.  Their tentative diagnosis, based on these clinical signs and symptoms, was hyperthyroidism, later confirmed with tests of TSH, free T4, free T3, and thyroid receptor antibody levels, all of which were extreme.  The final diagnosis: Graves' disease, in fact an extreme state of hyperthyroidism known as "thyroid storm."  These physicians prescribed propylthiouracil (a TPO inhibitor) for the patient, but after a few endocrinologist visits, she was "lost to follow-up."

This case illustrates the seriousness of Graves' disease and how frightening an encounter with this disease could be to a physician.  Indeed, as the article mentions, thyroid storm has a mortality rate of "up to 30%."  But arriving at the diagnosis involved a long process, ruling out many other conditions, then tentatively deciding on hyperthyroidism strictly on the basis of the patient's signs and symptoms, and only then testing the patient's relevant hormone levels. Would this not seem risky considering the grave and imminent danger posed by this extreme condition?  Yet this process was very much by the book, avoiding testing for thyroid disease until the characteristic signs and symptoms were recognized, but apparently not seen as sufficiently specific until after a long process of elimination. 

A take-home message here: symptoms of a disease do not have to be subtle, nor does the disease have to be mild, for purely clinical diagnosis to be difficult.  This is likely because diseases which affect the entire body, even if they originate in a single gland or organ, are more difficult to recognize than those affecting a single body part.  One reason might be is that there are typically many symptoms, with individual variations, and none of them might be specific to any particular disease. Yet systemic diseases are often the most serious, disabling, and hardest for a patient to recover from.

What about thyroid cancer, "benign" thyroid disease, iodine, and TSH levels?

There are four types of thyroid cancer; the most common kind, papillary thyroid cancer, which accounts for about 80% of cases, progresses extremely slowly. Although two other forms of thyroid cancer are far more virulent, the overall 10-year thyroid cancer mortality rate is 4.0%, as shown by the National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) Program data (https://seer.cancer.gov/faststats/selections.php?#Output) in Table 1. However, almost half of that mortality takes place within a year of diagnosis, perhaps mainly or entirely in patients suffering from the two most severe forms of the disease. NOTE: SEER Fast Stats have been replaced by https://seer.cancer.gov/explorer/. The image below used to be a link.

Table 1. Relative Survival by Survival Time for Thyroid Cancer (Click on square to see the table.)

There are some claims that thyroid cancer rates are increasing, but this essay will not take on this issue. Overall 10-year mortality rates of those with this disease are very low; in fact, the rate for those aged 20-49 years is only 1.0%, while that for those in the "75+" category is a modest 23.6%. It appears from examining these statistics that almost all thyroid cancer victims (and maybe all papillary thyroid cancer victims) die from other causes. However, since presumably all patients diagnosed with thyroid cancer are treated by having their thyroids completely removed, it is unknown how many would die of papillary thyroid cancer if they were untreated. According to Davies and Welch (2014), a reported increase in thyroid cancer was most likely caused by an increase in the diagnosis rate of existing cancers while the death rate has remained unchanged; in particular, diagnoses of small papillary cancer tumors in women soared to much higher rates than those for men, even though autopsies revealed the disease to be more common in men. This discrepancy is interesting in itself, since it implies a bias toward diagnosing and treating women more aggressively than men for this disease.

Kim et al. (2014) showed that rising TSH levels can cause thyrocyte changes leading to papillary thyroid cancer. It triggers a cascade of reactions that suppress the inhibition of a particular oncogene, i.e., a gene that has a substitution mutation which causes it to produce cancer.

Horn-Ross et al (2001) investigated the relationship of thyroid cancer and systemic iodine in women in northern California. They divided their sample into two basic groups, one with a history of thyroid disease (about 20% of that group) and those without ("healthy"). Their main hypothesis was that systemic iodine levels were positively correlated with thyroid cancer occurrence, but that turned out to be the case only with those with thyroid disease history. Unfortunately, because of the nature of that hypothesis, they did not feel free to conclude that those systemic iodine levels were negatively correlated with papillary thyroid cancer rates in the "healthy" group, but that was what their data implied. At any rate, this study suggested that preventing iodine deficiency in the first place would be helpful in preventing thyroid cancer in women. The problem is that women would need to find out whether they had goiters, nodules, etc. before they knew it was safe to take supplemental iodine and before a physician would be motivated to recommend doing so.

There were two insightful aspects of this study I found especially interesting: the researchers measured iodine levels in the subjects' toenail clippings to get an estimate of long-term iodine intake, and determined that most of their dietary iodine came from multivitamin pills, each of which typically contained 150 mcg of iodine. 

When hypothyroidism is complicated or causes diseases in other parts of the body

Hypothyroidism and anemia

Uncomplicated hypothyroidism is the most common cause of nonmegaloblastic macrocytic anemia (nonmegaloblastic macrocytosis), a red blood cell condition characterized by abnormally large mean corpuscular volume (MCV), provided that the individual has an otherwise normal complete blood count (CBC), a normal peripheral blood smear, and normal liver enzymes (Kaferle and Strzoda, 2009).  

Hypothyroidism (both "subclinical" and "overt") and its effect on non-alcoholic fatty liver disease

There are two types of thyroid hormone receptors: thyroid hormone receptor alpha (THR-alpha) and thyroid hormone receptors beta (THR-beta).  THR-alpha is mainly found in the heart, while THR-beta is mainly found in the brain and liver.  Through binding to the latter receptor, the thyroid affects the distribution of cholesterol and triglycerides in the liver and in the blood and, less directly, the placement of these lipids in the body tissues while limiting stress on the heart.

Development of a drug binding with a thyroid hormone receptor that is found predominantly in the liver

University of California at San Francisco researchers Chiellini et al. (1998) synthesized TRβ-selective compound GC-1, which has thirty times the affinity for liver cells as heart cells; University of California at San Diego researchers Trost et al. (2000) showed in a mouse study that GC-1 had great blood lipid-lowering effects, greater than T3 on triglyceride levels and the same on other blood lipids.  University of Sao Paulo researchers Villicev et al. (2007) showed that GC-1 caused 15-23% fat mass reduction in rats, while their control counterparts had an 80% gain in fat mass; in their study, T3 produced more dramatic results but had the disadvantage stimulating the heart.  Trost et al. (2000) administered GC-1 (and T3 in a different treatment group) to the mice via oral gavage, i.e., by daily delivering the drug directly to the stomach by insertion of a dropper down the esophagus; this suggests that the potentially commercialized drug could be taken orally via a pill that dissolves in the stomach.  It was not clear how Villacev et al. (2007) did this.  However, despite its great promise, GC-1 did not make it into clinical trials although it was the subject of many animal studies.  Grazia Chiellini, the senior author of the original paper, left UCSF and went to the University of Pisa in Italy, where she co-authored a 2017 paper summarizing these GC-1 studies (Columbano et al.)

It seems logical to consider the possibility that hypothyroidism might affect the heart indirectly by influencing liver function.  Only recently has non-alcoholic fatty liver disease (NAFLD) come to be recognized as the most common liver disease, but its observed link to thyroid function and blood lipid problems has proceeded rapidly from in-vitro studies to a new drug which has shown promising Phase 2 results according to its developer, Viking Therapeutics (PR Newswire, 2018); it completed the Phase 2a study later, although it awaits FDA approval to continue with Phase 2b (Viking Therapeutics, 2019).  Later, Duke-National University of Singapore (NUS), Duke University, the University of Chicago, and Viking Therapeutics researchers Zhou et al. (2019) published a paper describing a study demonstrating the effects of the new product, VK2809, on mice bred to be deficient in glucose-6-phosphatase, a condition causing glycogen and triglyceride deposition in the liver.  Because many other health problems cause these two problems, this study illustrated the wide practical use of this drug.

Studies of the thyroid-liver relationship

A 2010 French study (Gauthier et al.) might have provided the theoretical groundwork for that drug by identifying a plausible mechanism for thyroid hormones' blocking of fat production in the liver.  Based on the results both an in vitro and in vivo (mouse) experiment, these researchers determined a quirk of molecular geometry to be the key factor, where the binding of the thyroid hormone receptor beta ("THR-beta") to the molecule stimulating fat production, i.e., the carbohydrate-response element-binding protein (ChREBP), crowded out the binding of a liver receptor ("Liver X Receptor") because of their closely adjacent binding sites.  The starting point of this investigation was a Japanese mouse study (Hashimoto et al., 2007), which refers to "Liver X Receptor-α," apparently the molecule that Gautier et al. had referred to as "Liver X Receptor."  Hashimoto et al. were more specific in their references to thyroid hormone associated with the above thyroid receptor, calling it "T3," which presumably Gautier et al. meant.  Hashimoto et al. discovered the interference between the thyroid hormone receptor and the liver X receptor, but, unlike Gautier et al., did not propose a mechanism.

Other researchers conducted clinical research exploring this thyroid-liver relationship. Eshraghian and Jahromi, an Iranian and an Iranian-American, stated in their 2014 literature review that the majority of existing studies found a strong relationship between subclinical hypothyroidism (SCH) and NAFLD.  This literature review mentioned Chung et al. (2012), a South Korean cross-sectional observational study which involved 4648 subjects, finding a statistically significant monotonic relationship between TSH and alanine aminotransferase (ALT, the liver enzyme generally agreed to be the most specific indicator of liver disease) levels in subjects diagnosed with either subclinical or overt hypothyroidism. 

Chinese researchers Xu et al. (2012) responded to the Chung et al. study in a research letter describing their prospective observational study following subjects diagnosed with three levels of hypothyroidism according to their TSH levels at baseline as well as a same-size group with normal TSH values over an average of 4.92 years, determining how many in each cohort developed NAFLD over this period of time.  This study found a statistically significant (p<0.05) monotonic relationship between the hypothyroidism levels and the proportion of those who developed NAFLD.

Mexican researchers Posadas-Romero et al. (2014) conducted a 753-subject cross-sectional study of initially healthy participants in an unrelated prospective study designed to ferret out relationships among subclinical hypothyroidism, fatty liver, and (as yet undiagnosed) heart disease.  Perhaps their most interesting finding given past reported results suggested that although neither fatty liver disease nor subclinical hypothyroidism alone had a significant link to coronary artery calcification, the presence of both conditions in an individual did (to a highly statistically significant, i.e., P<0.001, extent), in fact being associated with the doubling of such calcification rates. On the other hand NAFLD had a demonstrated significant link to the heart disease surrogate variables insulin resistance and metabolic syndrome even in the absence of hypothyroidism; this makes sense in the case of the latter, since an enlarged liver would contribute to waist girth, one of four problems defining metabolic syndrome.  This suggests that we should not hasten to rule out subclinical hypothyroidism as a risk factor for heart disease, even though this and previous studies have apparently not found a relationship between thyroid disease and those heart disease surrogate variables commonly measured at annual physical exams. 

Chinese researchers Yan et al. (2014) reported that a previously observed monotonic relationship between TSH levels and the prevalence of NAFLD could be explained by the presence of TSH receptors in the liver, as illustrated in a mouse study comparing those with knocked-out genes for the liver TSH receptor with normal mice.

Egyptian researchers Farag et al. (2015) did a cross-sectional observational study of thyroid function parameters in 60 patients with NAFLD and in 60 age- and sex-matched controls without NAFLD  and found highly significant differences (P<=0.01) in TSH, free T4, anti-thyroglobulin antibodies (TgAb), blood lipid parameters LDL, HDL, and triglycerides, liver enzymes ALT and GGT, leptin, and insulin levels, and significant, but not highly so (P<=0.05), differences in HbA1C and anti-thyroid peroxidase antibodies (TPOAb), and free T3.  But correlation statistics, i.e., R, relating TSH to other parameters showed some differences: the greatest associations were shown in ALT, GGT, triglycerides, free T4, insulin resistance, anti-TPO, and HbA1C.  Ten of the NAFLD patients had subclinical hypothyroidism (defined as TSH=4.2-10.0 mIU/L), and three had overt hypothyroidism.  The TSH values for the patients averaged 4.45 mIU/L, while those for the controls averaged 2.31 mIU/L.

Finally in 2017, American researchers tackled this question with a placebo-controlled mouse study.  At the Yale School of Medicine, Ferrandino et al. induced mild hypothyroidism in "wild type" mice with a low-iodine diet and showed that these mice gained more fat and were more like to develop NAFLD than their counterparts in the control group.  On the other hand, mice bred to have a knocked-out gene for the sodium/iodide symporter, "the protein that actively accumulates iodide in the thyroid gland," inducing severe hypothyroidism, did not develop these liver problems when fed the same low-iodine diet.  This was an especially important finding because it casts doubt on the current assumption that "mild" or "subclinical" hypothyroidism does not need treatment.

A promising development was produced by Senese et al. (2017), who report that T2 administered to rats reversed the fat increase in the liver produced by a high-fat diet by 1) reducing fat production and 2) increasing fatty acid oxidation in the liver.  Their results indicated, however, that T3 only brought about the latter change.

Singapore and Indian researchers Rohit et al. (2018) produced a thorough review of the complex effects that thyroid hormones have on the liver.  Although some of these effects, e.g., lipolysis of triglycerides, i.e., fats, in white adipose tissue and the deposit of the resulting free fatty acids in the liver, would in themselves add fat to the liver, many other processes cause a net reduction of fat in the liver.  One of their key points is that one type of thyroid hormone receptor (THRß) is found only in the liver, while another (THRα) is found only in heart and bone.  They suggest that the use of a new drug based on the former receptor could be used to reduce liver fat without putting stress on the heart. 

Hypothyroidism and Type 2 Diabetes

Han et al. (2015) performed a meta-analysis and systematic analysis of the relationships of type 2 diabetes and its complications to subclinical hypothyroidism and concluded that they are statistically linked, although they considered the studies' small size a limitation.  However, these authors did not propose a causal relationship.  They measured the rate of 1) subclinical hypothyroidism in type 2 diabetes (10.2% after adjustment for publication bias), 2) the difference of subclinical hypothyroidism in those with diabetic microangiopathy (odds ratio = 1.93 after adjustment for publication bias, significant), 3) the difference of subclinical hypothyroidism in those with diabetic macroangiopathy (odds ratio = 1.59 after adjustment for publication bias, not significant), and 4) the difference of subclinical hypothyroidism in those with peripheral artery disease (odds ratio = 1.85, no adjustment for publication bias needed, significant).

Hypothyroidism, kidney function, and muscle wasting

Nephrologists Basu and Mohapatra (2012) discussed the effects of thyroid disease on the kidneys, as well as the effects of chronic kidney disease (CKD) and its progression to dialysis and transplant on TSH and thyroid hormone levels, and how to interpret the latter with respect to treatment.  They state that hypothyroidism reduces heart rate, cardiac contractility and output, renal blood flow, the beta-adrenergic receptor in the kidney, renin-angiotensin (II)-aldosterone (RAAS) activity, filtration pressure, and glomerular filtration rate (GFR) and increases "peripheral vascular resistance."  The GFR is reduced in part due to the reduced beta-adrenergic stimulation of the kidney and in part (in the case of longstanding hypothyroidism) due to the limited past growth of the functioning parts of the kidney.  Fortunately, most of these problems are reversible.

Basu and Mohapatra provide valuable information about how a particular type of non-thyroidal disease affects no just the TSH and the thyroid hormones but many other parameters.  In CKD, TSH is generally high while anti-TPO antibodies are low (unlike in primary hypothyroidism.)  Early in CKD, T3 (but not free T3) is low in part because of the kidney's reduced conversion of T4 to T3 and in part because T4 (but not free T4) is low, which in turn is positively correlated to rising levels of the cytokines tumor necrosis factor alpha (TNF-alpha), interleukin 1 (IL-1), as well as the inflammation indicator C-reactive protein (CRP), which inhibit the formation of the Type 1 deiodinase, which converts T4 to T3 by removing an iodine atom.  Unreacted iodine accumulates, triggering a prolonged Wolff-Chaikoff effect, which shuts down the production of the thyroid hormones.  After a kidney transplant, T4 and T3 levels return to normal over a period of three to four months.

There is one point of concern in the Basu and Mohapatra article: they describe TSH levels of less than "20 IU/ml" (which I suspect they meant to be "20 mIU/L") as "mild elevations" and express caution about the dangers of triggering hyperthyroidism with L-T4 treatment.  A more meaningful concern (which they might actually mean) is that the low thyroid hormone levels of CKD are the result of the disease and not its cause, making the benefits of modifying them with thyroid hormone replacement medications unclear.

Kumar et al. (2016) did a cross-sectional analysis of the relationship of hypothyroidism and 1) creatinine, 2) creatinine clearance, and 3) creatine kinase, using categorical and linear regression models.  They defined their three levels of thyroid function thus: normal: 0.34-4.25 mIU/L and normal free T4, subclinical hypothyroidism: 4.25-10 mIU/L and normal free T4, and overt hypothyroidism: 10 mIU/L and low free T4.  Though their sample selection method raised some questions since there were 30 subjects in each group (chosen from all patients seen at the Biochemistry Department at the Rajendra Institute of Medical Sciences over a recent year), and their data might not have followed a linear model, it was clear that overt hypothyroidism had a generally statistically significant relationship to kidney function, associated with higher serum creatinine (a measure of a combination of muscle mass and reduced level of kidney function) and lower creatinine clearance (a measure of how much work the kidneys are doing), as well as higher creatine kinase (a measure of muscle wasting).  Subclinical hypothyroidism had a strong, though not quite "significant" negatively correlated relationship (P=0.078) to creatinine clearance, and altogether there did seem to be a strong negative monotonic relationship between TSH and creatinine clearance.  In sum, the reported data strongly suggest that hypothyroidism has a detrimental effect on kidney function, even at the subclinical level. In addition, both levels of hypothyroidism were found to be statistically significantly related to total cholesterol, low-density lipoprotein (LDL), and triglycerides, but not to high-density lipoprotein (HDL).

Hypothyroidism's influence on sarcopenia, i.e., muscle wasting

South Korean researchers Moon et al. (2010) presented a problematic issue in statistical analyses using non-parametric categorical models and analysis of variance to examine relationships between two variables that would be minimal, not simply somewhat short of a high statistical significance threshold, if they are not causally related.  The authors determined via analysis of variance that the odds ratio comparing sarcopenia the female subjects with subclinical hypothyroidism with those who were euthyroid was 2.083, with a confidence interval bounded by 0.845 and 5.137, and declared it an insignificant relationship because zero was in the confidence interval.  However, since 0.845 is so much closer to it than 5.137, the difference from zero being indeed only 3% of it, that it is easy to see that the difference approaches statistical significance very closely.  Of course, the odds ratio in itself indicates that the chances of sarcopenia occurring in older female subjects with subclinical hypothyroidism, i.e., with TSH levels greater than 4.1 mIU/L (represented as "mU/L" in the text) and normal free T4 levels were twice those with normal TSH and free T4 levels. This indicates a difference that should not be ignored, and would have been recognized as "significant" if the variation in the subjects' TSH or sarcopenia measurements had been a little less.

On the other hand, the differences in sarcopenia across TSH levels in the male subjects were genuinely small, with an odds ratio of only 1.092 and a much more balanced confidence interval.

It is not clear why the authors set such a high threshold for determining the relationship of sarcopenia and subclinical hypothyroidism, since it could be used as an argument that there is no chance of sarcopenia in older women being helped by treatment of their hypothyroidism.  One also has to ask why a categorical model was chosen instead of a nonlinear regression one, which might have made the pattern of relationships clearer.

The effect of hypothyroidism on lung function

Egyptian researchers Sadek et al. (2017) discovered reduced lung function according to some measures in subjects aged 20-50 years whose newly diagnosed TSH levels were greater than or equal to 6 mIU/L in a case-control observational study.  Although their tests showed no anatomical lung damage, they revealed functional problems related to diaphragm muscle weakness leading to low blood oxygen and high blood carbon dioxide.

Hypothyroidism and the risk of heart disease: carotid-intima layer thickness vs. blood lipid levels

The main issue: what questions are we trying to answer?

The AACE guidelines imply that the only justification for treating any but the most severe cases of hypothyroidism, i.e., those involving subnormal T4 and T3 values and very high TSH, is to lower the risk of heart disease. Therefore, there have been many studies of the relationship of "subclinical" hypothyroidism and both arterial wall thickening and the blood lipid values regarded by many as its main risk factors. These studies have not explored the possibility that unhealthy blood lipid values (especially those of triglycerides) and abnormal thyroid parameters may both be caused by thyroid disease.

The most obvious question here would seem to be: at what TSH level does heart disease and these risk factors appear? However, in every study of this relationship, subjects with abnormally high TSH values and normal T3 and T4 values are lumped together in one category and compared to "normal" subjects in another category with respect to blood lipid values or the presence of heart disease. One obvious problem with this analysis is that there are very few subjects with TSH values above approximately 6.0 mIU/L, even with a fairly large total subject population, and yet the range of their TSH values is quite large, with the highest (and sparsest) values typically around 15 mIU/L. One major question remains: where did the AACE's criterion value of 10 mIU/L come from?  These studies were not set up to pinpoint such a value.

Monzani et al. (2001): An early, very complex, placebo-controlled clinical trial studying heart muscle structure and function changes in subjects with subclinical hypothyroidism

University of Pisa researchers found that the isovolumic relaxation time (the time from closure of aortic valve to the opening of the mitral valve to begin filling of the left ventricle), the preejection time-to-ejection time ratio, Peak A and Peak E (the peak flow velocity through the mitral valve in late diastole and in early diastole, respectively) and the derived the cyclic variation index (CVI), a measure of the heart volume change between the systole (when the heart contracts to pump) and diastole (when the heart begins ends contraction) on two heart walls, i.e., a measure of heart ("myocardial") contractility, statistically significantly changed beneficially with L-T4 treatment.  In the treatment group, CVI was significantly increased at six months and even more at a year.  The authors concluded that there was a "subtle, reversible impairment" in heart muscle function, and, tellingly, "Therefore, subclinical hypothyroidism is better considered a condition of minimal tissue hypothyroidism than a compensated state."  In other words, normal thyroid hormone levels are not a guarantee of normal heart function.  The authors recommended L-T4 therapy to prevent the progressive development of "clinically significant" heart function problems.

They used a treatment group and a placebo group matched by age, sex, and body surface area, each with 20 subjects, mostly in their twenties and thirties and 90% female, with minimum TSH levels of 3.6 mIU/L and "normal thyroid hormone levels" stable over the period of a year.  The treatment group received two 0.025 mg tablets a day initially, and returned every 3 months for TSH measurements, at which time their dose was increased by 0.025 mg until their TSH levels went below 3.6 mIU/L, at which time their echocardiographic measurements were taken. 

This study was apparently not replicated because of its complexity, but it might have inspired more elegant studies focusing on a few key measures of heart health such as carotid artery intima-media thickness, which were replicated many times.

Arterial thickness: studies of the relationship of TSH levels and carotid-intima thickness

Samuels et al. (2007) experimentally induced subclinical hypothyroidism, discovering negative health effects in a double-blinded, randomized, two-arm crossover clinical trial

These researchers selected 19 women, aged 20-75, who were currently undergoing L-T4 treatment for hypothyroidism and had for three months before baseline, with "normal" TSH levels.  They gave them either their usual L-T4 dose or a lower one for three weeks. The endpoints were the results of a number of tests of cognitive and motor function, i.e., measures of declarative memory (memory of facts and events, a form of long-term memory), working memory (short-term memory), and motor learning, as well as two measures of general well-being, i.e., the Billewicz scale (a measure of hypothyroid symptoms and signs) and the Short Form-36 (a measure of "health and general well-being").  They concluded that via a measure of working memory, the N-back test, working memory (a measure of prefrontal cortex function) was worsened by subclinical hypothyroidism.  Fatigue was also a notable problem in those with the lower dose.  As a validation measure, they asked their subjects which treatment arm they thought they were in, and the overwhelming majority guessed correctly. 

At baseline, the subjects' TSH averaged 3.53 mIU/L (represented as "mU/L" in the paper), was reduced to 2.19 mIU/L after treatment, and rose to an average of 17.37 mIU/L at the end of the other treatment arm.  Though normal free T4 and free T3 levels were requirements throughout the trial, the average free T4 level did decrease 33% from baseline in the non-treatment arm of the study and increased about 8% from baseline in the treatment arm.  Free T3 levels barely changed from baseline at the end of the non-treatment arm but went up 7.4% from baseline at the end of the treatment arm.

These researchers used a threshold p-value of 0.10 and applied Bonferroni corrections to that because of the multiple tests performed.  These results suggest that a new study using just the working memory and fatigue measures should be performed.

Endocrinologists Zhao et al. (2017): A meta-analysis of clinical trials considered the carotid intima-media thickness (C-IMT) in treated subjects with subclinical hypothyroidism, blood lipid parameters, and possible mechanisms

Zhao et al. (2017) studied the relationship of two measures of carotid artery endothelial dysfunction, i.e., the thickening of the carotid intima-media and flow-mediated dilatation, as well the standard blood lipid levels, to the effects brought about by L-T4 administration to patients diagnosed with "subclinical" hypothyroidism.  They analyzed nine clinical trials chosen from 107 found in a search (with exclusion criteria explicitly stated); these trials set the lower criterion limit for "subclinical" hypothyroidism at TSH levels ranging from 3.6 mIU/L to 5.5 mIU/L, and all subjects were younger than 65 years.  The L-T4 doses varied from 25 to 128 mcg daily.   In five of the trials, the control group did worse on these measures, while in three they did about the same. 

This study concluded that "L-T4 has beneficial effects on the development of the C-IMT."

Aziz et al. (2017): Another meta-analysis of clinical trials measuring change in the carotid intima-media thickness (CIMT) in treated subjects with subclinical hypothyroidism

Cardiologists Aziz et al. (2017) concluded that the the thickness of the two layers of the carotid wall just under the endothelium decreased significantly in subjects with subclinical hypothyroidism in 12 clinical trials lasting for six months to a year, with one exception. They provided details for each of these studies, which had mixed results, with eight of those studies showing a reduction, ranging from 1.6% to 42%, while two showed increases: 2% to 4.5%. One showed no difference and another did not report CIMT at all. Visual inspection of the numbers suggested that the studies that reduced TSH levels the lowest, to those typical of the average healthy person, achieved a greater reduction than those that reduced them to levels above 2.5 mIU/L. More recent trials also reported smaller reductions. Although these measurements apparently have a close correlation with atherosclerosis, which in turn is linked to cardiovascular events such as heart attacks, other methods of measuring this risk are considered to be better according to some researchers. Still in all, it seems that anything that reduces carotid artery wall thickness, at least up to a point, has to increase blood flow and to reduce blood pressure, and is regarded by cardiologists to be a good method for detecting artery wall problems early, before heart symptoms manifest themselves.

Polak et al. (2011) concluded that only maximum CIMT measurements (and the presence of plaque) involving the internal carotid artery produced better predictive results than current methods of heart disease severity classifications.

Aziz et al (2017) also compared blood lipid levels and found them to be significantly higher in those with subclinical hypothyroidism except for HDL.


Blood Lipids Levels and those of TSH


Studies of the possible causal relationship of "subclinical" hypothyroidism (SCH) and ischemic heart disease (warning: your eyes might glaze over here)

Many studies have been done to examine how thyroid hormones and their deficiency affect cardiovascular health; The general consensus is that overt hypothyroidism (high TSH, low T3 or T4) does have an adverse effect on the latter, according to a literature review done by the Greek researchers Rizos et al. (2011), and the underlying biochemical reasons are well-understood. However, the consensus in the case of subclinical hypothyroidism (high TSH, normal T3 and T4) is weaker. Why is this? In order to shed some light on the discrepant results, I examined several relevant papers closely. There were differences in SCH group inclusion criteria and in the choices of independent variables: all studies used age and sex, but otherwise there were few similarities. Although this was not mentioned explicitly, the subjects who completed the studies apparently remained in normal health, requiring no medical intervention, and all had been undiagnosed with any form of thyroid disease at the outset of the studies.

Smoking status and use of "lipid-lowering" drugs presented special problems in these studies. Male-female differences in smoking status across cultures are striking here: among the Japanese (Imaizumi et al., 2004), 70% of the SCH men and 80% of the control group men reported being current or past smokers, while the respective proportions among women were 8% and 11%, while in the United States 20.5% of men and 15.8% of women reported being current smokers (CDC, 2012). Smoking status is an important variable because substances that raise dopamine levels have been shown to lower TSH levels in the short term, and in one study (Jorde and Sundsfjord, 2006) to raise T4 and T3; perhaps because of this, combining current and past smokers in this group might have been ill-advised.

Subject use of drugs such as statins prescribed to control blood lipid levels is also an important factor, since successfully lowered LDL using a statin, for instance, might mask any LDL-raising effect of hypothyroidism. Statins also have been shown to lower triglyceride levels (Miller, 2014). Therefore a study in which a higher proportion of SCH group subjects are using such drugs than that of their control group counterparts would have a reduced chance of finding higher LDL and/or triglyceride levels in the SCH group.  Does simply adding an independent variable to the model indicating whether the particular subject uses one of these drugs (or smokes, for that matter) fix this kind of problem? What seems more obvious, of course, is the down side of leaving such variables out. But let's examine the issue as we look at the studies:

Imaizumi et al. (2004) of Japan used the other independent variables systolic blood pressure, BMI, total cholesterol, smoking status, erythrocyte sedimentation rate (ESR, a test for inflammation in lieu of using the radioimmunoassay to detect anti-thyroid antibodies) and the presence of diabetes mellitus. Their study involved 257 SCH subjects and 2293 controls. The lower TSH limit for the SCH group was 5.0 mIU/L, and mean TSH values for the SCH men and women, respectively, were 7.16 and 6.57 mIU/L at baseline. They used a Cox proportional hazard model to compare the survival rates of the SCH subjects and controls at several different time points in the study; it was much greater for the men, and "nonneoplastic disease" (mostly cardiovascular disease) accounted for a disproportionate amount of male mortality. Could this have been an artifact of the large sex difference in smoking status? This suggests that men were more likely to be wrongly classified as "normal" than women, and that those men who were classified as having SCH were more likely to have a higher de facto TSH (though masked at least at times by the effects of nicotine) and a worse degree of hypothyroidism than SCH women. Yet, because past smokers were lumped together with current smokers The trial concluded that SCH was a risk factor for heart disease, but it was probably impossible to sort out the unfortunate influence of nicotine intake on the results. At any rate, Imaizumi et al. (2004) ended their article saying, "the validity of thyroid hormone replacement therapy in subclinical hypothyroidism is left for future studies."

An earlier Italian study relating SCH to blood lipid values (Caraccio et al., 2002), used a smaller group (49 in the SCH group, 33 in the control group). They recognized these pitfalls and eliminated not only smokers but all potential subjects who had any medical problems other than SCH (defined as TSH > 3.65 mIU/L, normal free T4 levels and the presence of both types of anti-thyroid antibodies, i.e., TPO-Ab and Tg-Ab) from the SCH group; these researchers also eliminated those with unstable TSH values over the six month preceding the study. Average TSH values were 1.36 mIU/L for the "euthyroid", i.e., normal, group and 5.43 mIU/L for the SCH group, with 12 women in the latter group having TSH values > 6.00 mIU/L. Using linear regression, they concluded that TSH was statistically significantly associated with low-density lipoprotein (LDL), total cholesterol and apoliprotein B, although they saw a plausible causal relationship only with LDL, which was statistically significantly higher than in controls; unfortunately they did not produce correlation coefficients.  Although the relationship of TSH and and both total cholesterol and and LDL across patients appeared to be rather weak in the graphs plotting them, there was a stronger one between those blood lipid parameters and TSH reduction during T4 therapy.  The study included a randomized placebo-controlled clinical trial involving only SCH in which the study drug was levothyroxine sodium; there seemed to be a stronger relationship between the decrease in TSH levels and both LDL and total cholesterol in patients receiving the study drug.  There also was a positive correlation with lipoprotein A (Lp(a)), although T4 therapy did not alter its values: they speculated that it represented a "genetic influence" and noted that it was closely correlated with "a positive family history for coronary heart disease and/or diabetes mellitus."

Hueston and Pearson (2004), defined SCH as having a TSH ranging from 6.70 to 14.99 mIU/L and having a T4 greater than 4.5 mcg/dL, using a 215-member sample of subjects (untreated with supplemental T4) meeting these criteria in their part of the NHANES III study; their control group consisted of 8013 "normal" subjects, so the subclinical subjects constituted 2.6% of the total number of subjects. When they compared the two groups using only TSH levels and blood lipid levels in their model, via a Student's t test (using SUDAAN to adjust for over-sampling of certain demographic minority groups), they found their HDL and LDL values to be almost identical, while the triglyceride levels of the SCH group were statistically significantly higher than those of their counterparts in the control group. However, when they altered their model by adding the independent variables sex, age, race and use of "lipid-lowering drugs," they claimed that these differences vanished when calculated as odds ratios in a relative risk model in which the original TSH and lipid variables were converted into categorical normal/abnormal variables. The adjusted relative risk for the SCH patients having high triglyceride levels was 1.83 (183% the chance of members of the control group being so), with a confidence interval of 0.87-3.85; since 1.0 was in this interval, it was deemed to be a statistically insignificant difference. However, it seems that it came very close to being significant, since 0.87 is much closer to 1.0 than 3.85 is. Besides, the defining limits of the TSH range for the SCH group seemed rather arbitrary, as was the case for the other SCH studies.  In fact, these numbers should be interpreted to mean that there is actually a much greater chance of TSH and triglyceride levels having a causal relationship than not, although TSH level trends cannot be claimed to be perfect predictors of those of triglyceride levels.

Perhaps more to the point, the application of such strict standards to a comparison of entities assumed to have no difference at all, and in which any difference should be of interest, seems wrong.  Why is the risk of concluding that there might be a causal relationship between TSH and triglyceride levels considered to be greater than than that of concluding that there is none?  Why must we set a much higher level of proof for the former than for the latter?  Even though these numbers do not prove beyond a shadow of doubt that high TSH levels might cause high triglyceride levels (or vice versa), they offer far less reassurance that there is no correlation at all.  This is important because falsely assuming that TSH levels have no effect on triglyceride levels could have serious consequences, and even more so that this is settled science, could be dangerous since it contributes to the impression that "subclinical" hypothyroidism has no complications and is therefore not a true disease, not needing treatment.

The authors attributed this shift into (borderline) statistical insignificance to adding the variable representing higher "lipid-lowering drug" use to the model, because a higher proportion of the SCH group reported using such drugs. But shouldn't this factor have had the opposite results? Had fewer subjects in the SCH group been using "lipid-lowering drugs," their triglyceride levels (and maybe even more likely, those of LDL) might have been higher, creating an even larger difference. Also, this study did not take into consideration smoking status, which may have been confounded with gender.  But taken at face value this study tells us that HDL and LDL levels are probably not an issue, and gives us a pretty good idea of where TSH levels begin to affect those of triglycerides. To be entirely fair, however, lipid-modifying drug use was an important variable and no other studies seem to have included it.

However, SCH and first heart attacks in postmenopausal women did not show a significant correlation, according to LeGrys et al. (2013).The Whickham study (Vanderpump et al., 1995), which involved 2779 apparently undiagnosed patients, had also determined that there was no relationship between "autoimmune thyroid disease" and "morbidity and mortality from ischemic heart disease. Although they did not reveal all of the independent variables that they used in their model in the abstract (which was all that was available), some others were age, "cholesterol," "mean arterial blood pressure," "smoking history" and "skinfold thickness index."

Razvi et al. (2010) reanalyzed the Vanderpump et al. (1995) data and concluded that there was in fact a significant increase of ischemic heart disease in subjects with subclinical hypothyroidism. Their analysis, using logistic regression, also involved 2279 "euthyroid" patients and 97 with SCH; the criterion range for the latter was a TSH level in the range 6.0-15 mIU/L. The non-hormonal variable values that had statistically significant differences observed across the euthyroid/SCH groups were age, sex, systolic and diastolic blood pressure, total cholesterol, and smoking status ("never," "past," and "current"). On the other hand, differences for social class (five divisions), mean weight, history of cardiovascular disease, and history of diabetes mellitus were not statistically significant. The relative risk for all ischemic heart disease "events" for the SCH group relative to the control group was 1.76, with statistical significance at the P=0.01 level, and 1.79 for fatal such "events," with statistical significance at the P=0.05 level.

Even though there was some disagreement in the conclusions that these studies drew, even the ones that concluded that there was no relationship between TSH levels and cardiovascular disease and certain of its risk factors produced results that were close to statistical significance. On the other hand, because SCH was defined so broadly in all of the studies, we are left wondering at what TSH level the risk of cardiovascular disease becomes significant. Because of this, it is hard to translate these results into therapeutic practice.

Does subclinical hypothyroidism put its sufferers at risk for any other problems? Are they less able to cope with acute infections, for instance? This would seem to be a more urgent problem than the long-term risk of heart disease.

A study of the effects of L-T4 treatment of subjects with subclinical hypothyroidism (SCH) on ischemic heart disease (IHD) rates

More to the point is a U.K. retrospective observational cohort study of recently diagnosed, previously untreated subjects each with a baseline TSH measurement between 5.01 and 10.00 mIU/L (Razvi et al., 2012), which used seven years of data from the United Kingdom General Practitioner Research Database.  It concluded that this treatment was associated with a lower rate of IHD (at least for subjects in the 40-70 year age range).  The USPSTF (LeFevre, 2015) regarded this study with special favor because it gave information on whether treatment actually worked.  On the downside, however, this study did little to establish a threshold TSH at which T4 treatment was helpful, although it challenged the hypothyroidism treatment threshold TSH level of 10.00 mIU/L, apparently currently in effect in the U.K.

I was concerned about his conclusion that treatment did not help the subjects in the 70+ age group, especially given that the subjects in this group selected for treatment had higher baseline values for TSH and BMI than those on placebo at the statistical significance level of P<.001 and might have needed different doses of L-T4. (This was true of the other age group for just TSH.) In addition, no consideration in the experimental design was given to whether any subjects were taking statins to control their total cholesterol levels (although values for the latter were considered during randomization.)  The morbidity/mortality hazard ratios for the seven groups of disease considered across the 70+ group clustered around 1.00, while those for the 40-70 age group averaged 0.60, favoring the treated subset.  What was especially interesting is that treated subjects in both age groups had a much lower rate of "malignant neoplasms."

Case histories reflecting new assumptions of subclinical hypothyroidism, generalizing them to the patient population

... throwing away your umbrella in a rainstorm because you are not getting wet

-Ruth Bader Ginsburg

One indicator of a new hands-off approach to "mild" hypothyroidism is a recent article describing the treatment of a 62-year-old patient in the JAMA Diagnostic Test Interpretation section (Papaleontiou and Cappola, 2016). This patient, who presented with a TSH of 4.88 mIU/L and showed two symptoms of hypothyroidism, i.e., hair thinning (and "itchy scalp") and cold sensitivity, was given an apparently tentative diagnosis of subclinical hypothyroidism on the basis of her mildly elevated TSH. Her T4 was not measured because her below-10 mIU/L TSH was considered sufficient to rule out overt hypothyroidism. The authors did treat her, apparently because of her symptoms, although they expressed skepticism that those symptoms were related to a thyroid problem. When treatment with 25 mcg of levothyroxine (daily) had reduced the patient's TSH level to 2.75 mIU/L, researchers ordered an antibody (TPOAb) test, and on the basis of normal results on the latter and on the patient's reported improvement in her symptoms, the researchers took the patient off the medication. Although the patient's TSH scores rose again on the average, her symptoms reportedly went away. The authors apparently assumed that the patient had no non-thyroidal illness that might have caused a subnormal T4 result, and the continuing absence of symptoms when the patient's TSH went up to 5.14 mIU/L four months later did not appear to jog their curiosity. The authors say that "mild TSH elevations ... may be a normal manifestation of aging." What remains a mystery is how the patient came across in general: Did she manifest a normal energy level? Was she a submissive patient referred for a consultation by a primary care physician? Did she feel financial pressures to go off the medication and make her visits to the physician as infrequent as possible? Another concern is the authors' assumption that TSH levels "fluctuate" wildly and apparently unpredictably, implying that mild hypothyroidism is usually a temporary condition, although the relevant literature does not appear to support this point of view; the time of day appears to be the main factor influencing brief variations in TSH levels, in turn because it is related to short-term stress levels, as studies by Van Reeth et al.(1994), Leproult et al. (1997), Scheen et al. (1998), and Buxton et al.(2000) suggest; this is discussed in a later section.

Another case study which reflects this newly skeptical attitude toward treating "subclinical" hypothyroidism, especially in "elderly" individuals (Portillo-Sanchez et al., 2016), involves a 72-year-old patient with "coronary heart disease" and "type 2 diabetes," with a high (7.2 mIU/L) TSH and a normal (1.3 ng/dL) T4. He was started on 75 mcg of levothyroxine daily, which was high according to the Synthroid dosage page of Drugs.com (4 Nov 2019, a site managed by several registered pharmacists), which recommends the starting dose for hypothyroid individuals who are "elderly" and with "cardiac disease" to be from 12.5 to 25 mcg. Perhaps it is no surprise, therefore, that this patient developed symptoms and signs of heart disease after a month of treatment, and that his TSH dropped to 0.1 mIU/L. At any rate, the authors decided that this patient should have never been treated for hypothyroidism, and expressed their belief that their experience treating him suggested that older patients in general should not be treated for hypothyroidism unless their TSH is higher than 7.5 or 8.5 (presumably "mIU/L," although they actually stated "IU/L"), while they also noted that TSH values above 5.5 mIU/L affected no more than 3% of the population. In fact, they concluded that "treatment of subclinical hypothyroidism in elderly patients is probably not worth the costs and potential downsides." To be entirely fair, the authors might have had patients without reported symptoms or obvious signs in mind, i.e., what "subclinical" seems more logically to mean; they might treat a future patient in obvious distress differently. However, it was apparent that their experience with just one patient sufficed to determine their general treatment strategy.

Two Possible mechanism linking SCH and "bad" blood lipid levels

Rizos et al. (2011) in their literature review cite studies which conclude that T3 1) sensitizes LDL receptors, which limit the production of LDL, 2) stimulates the activity of lipoprotein lipase, which breaks down harmful lipoproteins and 3) upregulates apoliprotein AV, which reduces triglyceride levels. Since T3 levels are normal by definition in SCH patients, the latter would not be expected to have their blood lipid levels affected by these mechanisms. So where could be the underlying cause(s) of worsened blood lipid profiles in SCH patients?

It might be that infection is the causal link: infections of parts of the body other than the thyroid might stimulate TSH production and worsen blood lipid profiles in ways that suggest higher risk for cardiovascular disease. According to Varghese et al. (2008), viral infections (by the T3D reovirus) might cause TSH synthesis in intestinal epithelial cells. On the other hand, Apostolou et al. (2009) showed that acute brucellosis infection  was associated with an "atherogenic lipid profile" up to four months after recovery from the infection, and cited analogous studies associating cytomegalovirus, Chlamidia pneumoniae, and Helicobacter pylori with such a profile. 

But are hypothyroid patients especially susceptible to infection? There is a remarkable deficit of research on this subject, even though it is well-known that hypothyroidism typically causes cold intolerance and it is generally accepted that fevers stimulate the immune response (Nalin, 2005). However, one physician in a landmark book on hypothyroidism devoted a chapter to his dawning realization of a tendency for hypothyroid patients to get infections much more than healthy people, and for treatment of the hypothyroidism to lessen the occurrence of infection in these people (Barnes and Galton, 1976, 86-100).

Kolata (2016), a science writer, discovered some relevant knowledge at the Institute of Diabetes and Digestive and Kidney Diseases that hadn't been shared widely: what causes insulin resistance (and apparently bad blood lipids) was a shortage of subcutaneous fat cells relative to an individual's need to store unused dietary energy. The number of these cells does not automatically increase when one's dietary energy increases, although the individual cells might be able to expand in size to a limited degree. When there is no more room in the subcutaneous fat cells, the fat has to be stored elsewhere in the body, such as the liver and several other vital organs, causing the various manifestations of "metabolic syndrome." Perhaps when people experience the sudden weight gain frequently associated with hypothyroidism, the excess dietary fat goes to these parts of the body, producing the laboratory test results associated with this syndrome.

One reason that SCH might cause higher triglyceride readings and perhaps LDL in blood test results is that people with hypothyroidism digest food more slowly and may not succeed in emptying their stomachs before being tested. There seems to be a lot of controversy today about whether nonfasting triglyceride levels are as meaningful as their fasting counterparts and even about the magnitude of the impact of (measured) triglyceride levels on heart health, but wide differences in digestion rates may be muddying the waters. Since ultrasound technology can be used to determine whether an individual's stomach is indeed empty, the means exist to determine this experimentally and therefore to sort out these issues. I hope this will eventually be done.

Another confounding factor may be the fact that a high TSH in some types of people may not indicate hypothyroidism, subclinical or otherwise: as discussed later, those engaging in heavy exercise and therefore most likely to have good blood lipid values may drive their TSH levels higher, perhaps out of the normal range. Maintaining such high levels of exertion may eventually do permanent damage to the thyroid (and apparently no research has been done on this), but it may take years for this to happen.  In any case, researchers have not explored this issue to my knowledge.

The impact of estrogen levels on thyroid health, and implications for pregnancy, infertility and breast cancer

Clinical research: the relationship of estrogen levels to thyroid performance

Estrogen levels have an impact on thyroid health and even "subclinical" ("mild") hypothyroidism in pregnant women can cause their babies to be born prematurely and have lowered IQs.  Although the relationship of TSH levels and goiters to pregnancy outcomes has long been known, more recent research has shown that iodine transport and metabolism may explain this: do the thyroid and the parts of the body directly involved in pregnancy compete for iodine?  But there are really two separate questions we need to answer: 1) which problems are caused by iodine deficiency? and 2) which diseases can be successfully treated by using iodine as a drug?

An early study (Sawhney et al., 1978) of rhesus monkeys measured TSH increase when 50 mcg/kg of estradiol was administered; over 28 days it increased from an average of 1.4 to 3 mIU/L, where it plateaued.  The authors concluded that a reduced rate of TSH degradation, presumably by the liver, was responsible rather than an increase in TSH production by the pituitary gland. 

Furlanetto et al. (1999) discovered that estradiol, the most potent estrogen, "down-regulates the sodium-iodide symporter gene" in the thyrocytes. Tazebay et al., 2000 added that a local sodium/iodide symporter "mediates active iodide transport" only in the lactating or cancerous breast.  Stoddard et al. (2008) suggested that inorganic iodine may provide protection against the development of breast cancer, reporting that in vitro treatment of a line of human breast cancer cells with a solution containing 5% molecular iodine (I2) and 10% iodide ions (I-) in the form of potassium iodide (KI) caused up-regulation of genes causing estrogen breakdown and down-regulation of genes stimulated by estrogen.  However, it is also possible to draw the conclusion that at least one of these forms of iodine is necessary for breast growth. Is it possible that estrogen acts on these symporters in a way that routes iodide away from the thyroid and especially toward the breasts, at least during lactation and in the case of breast cancer?  If so, can the breasts get too much iodine with treatment for hypothyroidism?  And what effect does iodine have on them?

A finding with important therapeutic implications was the Eskin et al. (1995) rat study, which sought to distinguish between the breast tissue effects of I2 and I-:  According to several measures of breast pathology, i.e., "lobular hyperplasia," "extraductal secretions" and "periductal fibrosis," the presence of I2 was either beneficial or neutral, but harmful or neutral in the presence of I-.  These findings were supported by Garcia-Solis et al. (2005), who found evidence that I2 showed that the magnitude of the in vitro anti-cancer effect was strongly correlated with gene expression of the ion exchanger pendrin, a iodide/chloride (I-/Cl-) "transporter," which transfers I- from the thyrocyte to the colloid in the thyroid and performs an analogous function in other locations in the body, and inversely correlated with gene expression of lipoperoxidation, i.e., reacting of peroxides with fat in the tissue.  Notably, there was no change in the gene expression of the Na+/I- symporter, which brings I- from the blood into the breast and thyroid cells.  Garcia-Solis also drew the conclusion that I2 (but apparently not I- or T4) reduced tumor incidence in rat mammary tissue, and that this was associated with an antioxidant effect.  Funahashi et al. (1996) agreed that "inorganic iodine," perhaps a combination of I2 and I- (but not specified in the abstract) suppressed breast tumor growth in rats, but medroxyprogesterone acetate (MPA) had a stronger effect, and the combination of the two was even stronger.  They concluded this because the degree of tumor suppression in the rat tissue was correlated with the magnitude of iodine concentration (four different dose levels) in that tissue.

These were troubling findings, since the form that the thyroid uses, and which is available to the consumer, is I-.  But I wish they could explore the reasons for the different effects these two forms of inorganic iodine have on the body.

Arafah (2001), who studied estrogen and progestin supplementation (HRT) in women, suggests that increased TSH in pregnancy can be explained by 1) TBG increase, 2) degradation of thyroid hormones by the placenta, 3) transfer of T4 to the fetus and 4) increased maternal excretion of T4.  This paper reported the results of a prospective observational study involving 11 normal women, 18 with benign, i.e., non-cancer, thyroid disease and 7 who had been treated for thyroid cancer who were each given 0.625 mg of "conjugated estrogens" daily and maintained on their baseline dose of T4.  In women with normal thyroid function, TSH varied very little, staying between 1.0 and 2.0 mIU/L, while the TSH in women treated for hypothyroidism jumped from an average of about 1.5 mIU/L at baseline to one of almost 4.0 at 12 weeks, although it quickly dropped down near original levels at around Week 18.  However, these averages disguised a great deal of variation; some of the TSH values for the subjects with benign disease went above 10.0 while others remained below 1.0.  On the other hand, T4, FT4 and TBG values did not vary significantly across these various groups.

Demonstrating that an existing "normal" range is too wide requires taking unfortunate measures which, although within the ethical rules of clinical trials, are perhaps too imprudent to state openly.  Yet some clinical trials have effectively done this for pregnant women with hypothyroidism, though their stated goals were to show that their subjects needed an increase in T4 supplements.  Unfortunately, because of increasing pressure to adhere to official standards, this unfortunate measure has become necessary to reform those standards.

Alexander et al. (2004) concluded that an increase in levothyroxine was necessary for pregnant women with hypothyroidism. They performed a clinical trial involving pregnant women, fourteen of whom had hypothyroidism caused by benign conditions, i.e., either Hashimoto's disease, treatment for Grave's disease or a benign thyroid nodule, and six of whom had been surgically treated for thyroid cancer, using levothyroxine replacement therapy at levels intended to "suppress" TSH levels, i.e., to keep them minimal.  The authors stated that T4 dose was increased for the fourteen when TSH went over 5.0 mIU/L, and over 0.5 mIU/L for the six.  These women had been maintained before entering the study at TSH levels far below this maximum, arriving at the start of the study with a baseline average of 1.53 mIU/L for the  fourteen subjects with "benign" disease and of 0.06 mIU/L for the other six treated for thyroid cancer.

During the study, however, their TSH levels were allowed to rise much higher.  At eight weeks, the average TSH for the entire group (which included six thyroid cancer patients who in theory needed to have their TSH levels kept below 0.5 mU/L), rose to a mean of 4.2 mIU/L, with a standard deviation of 3.8 mIU/L.  Percent increases in thyroxine dose averaged 35% for the fourteen and 48% for the six. Of the fourteen patients, two pregnancies ended abnormally, one with a miscarriage and one with a stillbirth (though neither patient had Hashimoto's disease); one thyroid cancer patient had an induced abortion.  The rise in estradiol levels in pregnant women is very large: in this study, they rose, on the average, from somewhat less than 50 pg/ml to about 10,000. 

Maraka et al. (2017) performed a retrospective cohort study of pregnant women with TSH levels in the range of 2.5-10.0 (my interpretation, although the authors wrote it as "10," improperly according to my understanding) mIU/L, 16% of whom had been treated with supplemental levothyroxine. They divided the group into four cohorts, depending on whether they were treated during the study and whether their TSH levels fell in the range of 2.5-4.0 mIU/L or in the range of 4.1-10.0 mIU/L in order to determine the differences in health outcomes that could be explained by the treatment. They concluded that among those in the two cohorts, pregnancy loss was less in the treated women than in the other cohort at a significance level of P<0.01, although this significance was not present in the lower pre-treatment TSH group. On the other hand, the treated group had more medical problems, most notably preterm delivery at a significance level of P=0.01; this significance was not present in the high pre-treatment TSH group. Only one medical problem affecting the women themselves, gestational diabetes, was significantly increased (marginally) in the higher pre-treatment TSH group, while gestational hypertension and pre-eclampsia were significantly increased in the lower pre-treatment TSH group. One shortcoming of the study is that it did not distinguish the patients' conditions and TSH by trimester; this matters because the unborn child does not have a functional thyroid during the first trimester, putting more stress on the pregnant woman's thyroid.

Estrogen and the hypothalamus: in vitro studies

Estrogens, like all small nonpolar molecules (including other steroids), pass through the cell membrane without resistance and bind to a receptor that triggers gene expression. However, recent research (Qiu et al., 2003) on the effects that estradiol has on the hypothalamus' arcuate nucleus shows that estradiol also has a quicker-acting mechanism, similar to that of the action of the nervous system, by which it binds to receptors cells on the cell membrane, which activate a G protein, via the diacylglycerol metabolic pathway.

Is there a connection between Alzheimer's dementia and hypothyroidism?

The possibility that untreated hypothyroidism could lead to Alzheimer's dementia has not been explored formally.  There are troubling indications that that connection, and possible causal relationship, might exist while being overlooked by researchers assuming that Alzheimer's originates in the brain, especially the female brain (and of course that it is linked to the e4 mutation of the Apoliprotein E gene).  There is an important overlap in symptoms, and ruling out hypothyroidism has been a standard step in Alzheimer's diagnosis.  But tightening standards for hypothyroidism diagnosis in order to prevent cases of atrial fibrillation and bone fractures risks might be causing some people with hypothyroidism to be diagnosed with Alzheimer's instead, eliminating their chance for treatment and improvement or elimination of dementia symptoms.  Besides, do we know for sure that those with free T4 levels in the normal range and TSH levels below 10.00 mIU/L have none of these symptoms?

Shomon (2000, pp. 57-62) reports a relevant suspicious case.  A female patient who had suffered from severe hypothyroidism and whose mother was diagnosed with untreatable dementia tells a chilling story that strongly suggests that the latter fell through the cracks of a medical care system that failed to consider hypothyroidism as a possible cause of her dementia and extreme physical decline.  Her physician delivered the diagnosis very kindly, but without apparent concern that he might have made a mistake.

Harvard and Boston University researchers Tan et al. (2008) analyzed the relationship between TSH levels and Alzheimer's dementia incidence.  Women and men (dementia-free, each group aged an average of 71 at baseline, with a small standard deviation) were analyzed separately; each group was divided by TSH into tertiles.  Separate analyses were done on the sample including and excluding those receiving L-T4 hormone treatment; no difference was found.  The middle tertile for women (N=1108) included the TSH range of 1.01 to 2.10 mIU/L, while that for men (N=756) was 0.90 to 1.80 mIU/L.  At 12.7 years from baseline, 142 women (12.8%) and 67 men (8.9%) developed Alzheimer's according to the text.  The women's rates spread out over time, with the highest rate being for the T1 group (the most hyperthyroid), and the second highest for the T3 group (the most hypothyroid).  The men's differed far less across their tertiles, with the T3 group (the most hypothyroid) having the highest Alzheimer's rate; their differences were not considered to be statistically significant.  The rates of Alzheimer's in the T2 groups (the normal ones) barely differed between men and women at that point according to the data presented in Figures 1 and 2.  To judge from the plot, it seems that the differences in rates of incidence of Alzheimer's were very small before the age of 80, but at that age, if there is a causal relationship, normalizing TSH could stave off Alzheimer's for several years, and that interval increases dramatically with age.  But what would have been the case if the data for those without Alzheimer's at about 70 had not been excluded?  Since there were more women in the study population, that begs the question of where the "matching" men were. 

One observational study of cognitive function of hypothyroidism, Ceresini et al. (2009), using the Mini-Mental State Examination, failed to turn up a relationship between hypothyroidism of every severity level and cognitive impairment.  Differences between those with hypothyroidism and those without thyroid disease (using a reference range of 0.46-4.68 mIU/L) were small, while the scores for the hyperthyroid subjects were a little (though statistically significantly) lower.  Both were dwarfed by the differences between the two age groups (65+ and younger); the older group averaged at the borderline level, while the younger group's scores averaged about 20% higher.

Science writer Pincott (2020) described an ongoing study that Lisa Mosconi, director of the Women's Brain Initiative and associate director of the Alzheimer's Prevent Center at Weill Cornell (University) Medical College was conducting.  Mosconi claimed that women's brain glucose metabolism slowed down a great deal after menopause although "a 2009 study" found that "newly menopausal" women did as well on cognitive tests as they did before menopause.  Yet women get diagnosed with Alzheimer's at an earlier age than men do, and one fifth get it by the time they are 65, according to an unnamed source in the article.  Mosconi hypothesized that reduced estrogen levels explain this.  However, thyroid levels were not under consideration, apparently because detailed available information about the effects of estrogen provided her and apparently people in the field with a satisfying explanation.  However, data from the Women's Health Initiative Study challenges this hypothesis, linking the age of onset of Alzheimer's with use of estrogen/progestin drugs. That finding was challenged, in turn, by the argument that the subjects started taking supplemental hormones at too late an age, i.e., 65 or older, and the earlier use of birth control pills might also be a factor.  Reports such as this might be interesting to the general public, but the validity of this researcher's conclusions still need to be tested via publication and replication attempts before they can be regarded as settled science.

The surprise: subclinical hypothyroidism is not a big factor for depression, but subclinical hyperthyroidism is

South Korean researchers Hong et al. (2018) found that subclinical hyperthyroidism was a much bigger trigger of depression than subclinical hypothyroidism is.  There were large variations in scores on the depression portion of the Patient Health Questionnaire (PHQ-9), but the great differences in the subclinical hyperthyroid and normal subjects' overall values overcame that to produce statistically very significant differences, showing a clear monotonic relationship.  Yet, it's worth noting that the most severely afflicted category, i.e., 15-27 symptoms, had more subjects with subclinical hypothyroidism than the others, but not with a monotonic relationship.  They used a "0.62–6.68 mU/dL" (could they have meant "mIU/L"?) reference range, noting that those values were higher than those used in Europe; the proportions of hyperthyroid and hypothyroid subjects were about the same.  The median TSH for all subjects in the study was 2.23 mIU/L.

The relationship of hypothyroidism and body mass index (BMI)

The relationship of hypothyroidism to weight gain is a very complex one, so much so that analyzing the body systems involved is remarkably unhelpful in developing a predictive model of that relationship.  There are many protective physiological processes that lessen the impact of hypothyroidism on weight, some of which involve actions of the nervous system, which can been modified by psychological and social factors.  What is even more confusing is that primary hypothyroidism rarely causes obesity, but morbid obesity in itself can raise TSH levels, albeit slightly and rarely above the normal range, because of the stress it puts on the body.  Perhaps hypothyroid individuals typically adjust to their lower metabolic rate with a reduced food intake.  But if individuals take some time to limit their food intake accordingly, they might gradually gain weight.  This problem might be aggravated by rigid family or work routines, which might make gradual adjustment difficult.  On the other hand, appetite loss caused by psychological distress might overtake drop in metabolic rate.  Existing scientific data challenge the stereotype of the obese hypothyroid individual, but there subtleties that have prompted a great deal of study.

TSH, like most hormones, has several unrelated effects, only one of which is stimulation of the thyroid.  In a clinical trial, Santini et al. (2010) discovered that administration of TSH via two intramuscular injections a day apart to patients who had had their thyroids removed as treatment for thyroid cancer (and were also receiving replacement T4 therapy) had a rise in leptin levels, known to suppress appetite, suggesting that the TSH administration triggered this rise via stimulation of TSH receptors in adipocytes.  Since this experiment shows that TSH alone has this effect, rather than operating indirectly via stimulating T4 production, this has interesting implications about body fat management and BMI: hypothyroidism, by triggering a rise in TSH, suppresses appetite, thereby containing the weight gain that lowering of free T4 would tend to cause on its own.  On the other hand, Santini et al. (2014) reports that rising leptin levels cause a decrease in TSH levels. 

The complete appetite-regulation mechanism involves two nuclei in the hypothalamus, i.e., the arcuate (ARC) and paraventricular (PVN), as well as hormones other than T4 secreted by the neurons in these parts of the brain and many other parts of the brain.  The arcuate nucleus alone excretes 12 neurotransmitters and neuropeptides as well as propiomelanocortin (POMC), which is broken down into three more hormones, including the alpha-melanocyte-stimulating hormone (α-MSH), which causes appetite suppression by acting on the PVN, which secretes several hormones including thyrotropin-releasing hormone (TRH) and in hard-wired connections to other parts of the brain, i.e., the brainstem, spinal cord, and limbic system via interneurons and centrally-projecting neurons.  In sum, this is a very complex negative feedback system, i.e., one that resists change, helping to maintain homeostasis.

In a cross-sectional analysis of the data provided by Holloway et al. (2002), Kitahara et al. (2012) concluded that there was a statistically significant but not monotonic relationship between TSH (geometric means) and BMI (grouped by Worth Health Organization categories of underweight, normal, overweight, obese class I and obese classes II and III), although the differences between groups it was slight, much smaller than the relationship of TSH to age group; the geometric means of the TSH levels of these groups were, respectively, 1.59, 1.48, 1.63, 1.69, and 1.63 mIU/L.  However, this group of 3,114 people used in this study did not include anyone with out-of-range values for TSH, free T3, or free T4, so that might have decreased the differences among these groups and obscured any existing monotonic relationship.  The authors expressed no assumptions about causal relationships, but wrote, "Experimental and/or longitudinal studies are needed to assess whether weight loss or maintenance among individuals who are euthyroid may help to prevent the development of subclinical or overt hypothyroidism and associated health risks."

Rotondi et al. (2010) compared two groups of subjects with "hypoechoic" thyroid ultrasound results, i.e., indicating an enlarged thyroid usually interpreted to be a goiter.  One group had mostly normal BMIs and some who were overweight but not obese, and another "morbidly obese" group, i.e., with BMIs greater than 40.  They found that 85.5% of the former group and 20.9% of the latter group had anti-thyroid antibodies, and that 11 out of the 105 morbidly obese patients had "isolated hyperthyrotropinemia," although they did not report any TSH summary statistics.  At the end, they found that two out of 105 non-obese subjects had "lack of detectable thyroid abnormality," while that was the case for 68 out of 105 of the obese subjects.  They concluded that, although obesity raises TSH slightly, it was not associated with autoimmune thyroid disease.  This suggests that obesity is a de facto non-thyroidal illness that put stress on the thyroid as a result but did not trigger an autoimmune attack on the thyroid.

Laurberg et al. (2012) discuss methods for distinguishing between those who have experienced weight gain because of hypothyroidism (mainly because of fluid retention) and those who are genuinely obese (because of excess fat).  They discovered that obese individuals without thyroid disease experience a slight rise in both TSH and thyroid hormone levels as an "adaptation".   Therefore, measuring free T4 levels is a crucial part of the hypothyroidism diagnostic process.

An eight-week Italian clinical trial using lifestyle modifications as the intervention, using as subjects 206 children whose BMIs were at the 98th percentile or higher, determined that both TSH and insulin resistance were lowered as a result (Longhi and Radetti, 2013).  The children experienced a mean weight loss of 15.4%, a mean TSH reduction of 10.6%, a free T3 reduction of 13.2%, no change in free T4, a 2.7% reduction in HDL, a 43% reduction in triglycerides, a 38.5% reduction in LDL, a 52% reduction in fasting insulin, and a 76% reduction in leptin.  They also experienced a mean of 2.6% reduction in lean tissue, and a 21% reduction in body fat.  These researchers defined "high normal" TSH levels as 2.5-6.0 mIU/L; the mean TSH values of the subjects were 2.64 mIU/L at baseline and 2.36 mIU/L at the end of the study.  This study implies that obesity can raise TSH levels, but is a relatively minor factor.

The impact of non-thyroidal diseases on thyroid test results and possible HPT axis hormone treatment

A variety of non-thyroidal diseases can apparently drive up TSH levels.  But undiagnosed cases can introduce confusion into studies of the relationship of TSH levels and thyroid disease.

TSH levels seem to follow a general pattern in chronic or progressive non-thyroidal illness: if a disease persists in a part of the body other than the thyroid, the TSH rises, sometimes out of the normal range but apparently rarely up to 10.0 mIU/L.  But as its severity increases, TSH levels go back down, and so do those of T4.  However, a decrease in T3 levels is temporarily staved off by the increasing adaptive conversion of T4 to T3.  If the disease is so severe that net catabolism (breakdown of tissue) becomes established, then more T4 is turned to reverse T3 (rT3) and less to T3, which falls steadily; the only known function rT3 has is to provide feedback to the hypothalamus and to the pituitary gland (in a way similar to that of T3), causing the latter to reduce its output of TSH as a result.  The progress of T3 downward below the normal range is considered to be a strong predictor of imminent mortality (Mebis and van Berghe, 2009).  This means that at some early/milder stages of yet-undiagnosed non-thyroid disease, some patients would be officially considered to have "subclinical" hypothyroidism, if indeed they had their TSH levels tested, while others whose T4 dropped below the reference range would be considered to have "overt" hypothyroidism. It stands to reason to take a second look when TSH levels rise to abnormal levels, whether or not they are caused by disease originating in the thyroid.  In fact, do we really know for sure that in patients whose TSH levels go above the reference range, a drop in free T4 levels below their reference range includes all of those with hypothyroidism and none with any other kind of disease? 

Data supporting this early rise in TSH levels are admittedly sparse because of limited research in the field.  For instance, Varghese et al. (2008) indicate that viral infection can stimulate the production of TSH outside the normal HPT axis.  Creutzfeldt-Jakob disease sufferer with only 4 months to live came to the Massachusetts General Hospital with "rapidly progressing ataxia," i.e., having trouble with walking due to her worsening proprioceptive sense.  Her early blood test results were normal other than high values for both types of anti-thyroid antibodies, especially against TPO (260 IU/mL with a reference range of <9), and her TSH was 4.51 mIU/L, with the corresponding normal range 0.40-5.00 mIU/L, unusually high but not quite enough so to be considered to be a sign of trouble in itself. (Chwalisz et al., 2019)  Another interesting case is that of central hypothyroidism, and to a lesser degree, primary hypothyroidism, are more common in patients with Erdheim-Chester Disease, a kind of blood cancer, according to National Institutes of Health (NIH) researchers Shekhar et al. (2020).

Dietrich et al. (2020) point out in a meta-analysis that patients diagnosed with post-traumatic stress disorder (PTSD) were shown to have higher TSH and free T3 levels than those without PTSD, and also had a greater incidence of cardiovascular disease.  The causal relationship was readily determined by the time sequence of the precipitating event and the rise in TSH levels.  Allostatic load, a measure of stress, was taken.  In sum, the authors determined that the PTSD brought on by the traumatic incident caused both the elevated TSH and the cardiovascular disease, thereby throwing doubt on the common conclusion that a high TSH causes cardiovascular disease. 

Critical illness, i.e., "low T3 syndrome" and stages on the way

Chen et al. (2020) found an unusual pattern in patients with COVID-19 pneumonia (n=50) when compared with matching healthy controls (n=54) and other patients with non-COVID-19 pneumonia of similar severity (n=50); all patients had at least moderate levels of disease.  Although both groups with pneumonia had lower TSH and total T3 levels than the healthy controls, those levels were much lower in the COVID-19 group than in the non-COVID-19 pneumonia group.  Yet, contrary to the general pattern, the total T4 values of those in the non-COVID-19 pneumonia group did not differ from those of the healthy controls (indeed were almost exactly the same), while those in the non-COVID-19 pneumonia group went down about 7%; although the difference was not statistically significant, that was because of the large variance in each group.  On the other hand, blood albumin levels of the two groups with pneumonia were both lower than those of the healthy group by almost identical amounts.  When the COVID-19 group was broken down into moderate, severe, and critical levels of severity, however, the critical level (n=12) had a 15% drop in total T4 on the average; however, the difference was not statistically significant.  This suggests that there was a special situation causing a reduction in the conversion of T4 to T3 at least until critical severity was reached.  The authors note that, based on studies of SARS on the HPT axis, damage to the parts of the pituitary gland responsive to TSH levels kept it from responding to TSH level drops by increasing its output of TSH.  At any rate, after recovery (and apparently all recovered), these values returned to normal, and free T4 and T3, measured for the first time then, were also normal.  It was not clear what kind of pneumonia that the non-COVID-19 group members had or how it was treated.   According to Chatzitomaris et al. (2017), the pattern of lowered TSH, unchanged or lowered total T4, and lowered total T3 is found in sepsis, which has been reported at a special problem with COVID-19 pneumonia.

A pattern of normal (or low) TRH, TSH, low T3 and T4, and high rT3 has long been known as euthyroid sick syndrome or non-thyroidal illness syndrome (NTIS). This might be the body's natural strategy to take lessen the stress on the thyroid caused by the (non-thyroidal) disease, at least until help arrives (and sometimes it does not.) 

TSH levels aren't necessarily the most sensitive indicators of conditions that might benefit by being treated with supplemental hormones in the HPT axis. The exceptions are generally found in "critically ill" patients, i.e., with catabolic illnesses; they are so ill with a disease affecting another part of the body that they experience persistent protein loss.  According to the Adler and Wartofsky (2007) literature review, levels of the thyroid hormones, especially T3, are the best indicators of the severity of a patient's serious, acute non-thyroid disease and its prognosis.  They discuss the D1 iodothyronine deiodonase (though not as such) and its activity in the liver.  They suggest that such a seriously ill patient, i.e., one with "deteriorating clinical status" who has "test results suggestive of hypothyroidism" might benefit from intravenous T3 infusion.  On the other hand, they recognize that many non-thyroidal diseases can induce "so-called subclinical hypothyroidism," i.e., high TSH and normal T4, and believe patients with this pattern should get the standard treatment for hypothyroidism.

Adler and Wartofsky considered infectious disease (HIV and sepsis), diseases of the heart (MIs, unstable angina, acute coronary syndrome and heart failure), kidney (nephrotic syndrome and end stage renal disease) and liver (cirrhosis, acute hepatitis and chronic liver disease).  However, they did not consider non-thyroid cancer or diabetes, two diseases which often make enormous energy demands on patients.  They do consider the thyroid side effects of several classes of drugs when used in "critical illnesses," most notably "stress doses" of corticosteroids and intravenous dopamine; both induce central hypothyroidism, i.e., caused by malfunction in the hypothalamus or pituitary gland, reducing TSH, T4, T3 greatly and free T4 somewhat less.

Pingitore et al. (2008) did a promising small (10 in each of treatment groups) trial in which they administered an infusion dose of L-T3 to patients with "ischemic or nonischemic dilated cardiomyopathy" who had "low-T(3) syndrome."  The subjects' results on a variety of tests of heart function, hormone levels, and a neurotransmitters improved, and no side effects were observed 48 hours after the infusion. 

Belgian researchers Lambert et al (1990) studying AIDS patients found a strong negative correlation between CD4 lymphocyte count and TBG levels, but not with those of cortisol-binding globulin (CBG) or sex hormone-binding globulin (SHBG).

Haugen (2009) did a more exhaustive run-down of the effects of various drugs on TSH and the thyroid hormones. One drug of special concern to those who treat thyroid disease is amiodarone, which is used to treat heart arrhythmias and contains iodine.

Since the liver produces TBG, and D1 to convert FT4 to T3, thyroid parameters in liver disease patients depart from the usual pattern described above. Borzio et al. (1983) tried to answer this puzzling question: why, in spite of the important role the liver plays in thyroid hormones metabolism, and the dramatic reduction in T3 seen in cirrhosis, does liver disease usually not cause clinical hypothyroidism symptoms and signs?  Their answer was very elaborate, but this is how I read the narrative of their data: TSH rises to an average of 3.1 mIU/L in cirrhosis patients and 2.7 mIU/L in hepatitis patients, in contrast to 2.4 mIU/l in controls, which contributes to a high FT4 (an 11.9 mIU/L in cirrhosis and chronic hepatitus patients, respectively) in contrast with that of controls (9.9 ± 0.3 mIU/L).  As a result, the FT3 of cirrhosis patients was 69% of that of controls, a striking improvement over the analogous ratio for total T3 values (56%).  Since the liver's production of TBG would seem to be reduced in liver disease, that would be an obvious suspect; however, TBG was in fact higher in the liver disease patients than in controls.  One possible explanation for this is that a decrease in liver function causes a rise in estrogen levels because one role of the liver is to break down estrogen; estrogen has been shown to cause TBG levels to rise.  In fact, the effects of rising estrogen levels, and by extension pregnancy, on thyroid parameters and overall health is a very heavily studied topic, which I will soon discuss in more detail.

Mebis and van den Berghe (2009) produced a more sophisticated analysis, which involved a close examination of the roles of the three thyronine deiodinases (D1, D2 and D3); Each of these enzymes removes an iodide ion from T4 (not always from the same location in the molecule) to produce what is sometimes a useful new molecule.  In health, the D1 iodothyronine deiodonase, mainly located in the liver (which sends its products to other parts of the body), is dominant, converting T4 into T3; in both acute and prolonged "critical illness," the D3 iodothyronine deiodinase converts T4 into reverse T3 (rT3).  The D2 iodothyronine deiodonase, like its D1 counterpart, converts T4 to T3 but is found in other tissues, i.e., the brain (80%) and skeletal muscle, operates only locally and appears to be upregulated in hypothyroidism.  Understanding of the roles of these enzymes in critical illness was necessary to pin down the body's elusive compensatory mechanisms under these conditions, and to determine, which, if any, hormone supplementation regimen could provide effect support to these patients' recovery processes.

Mebis and van den Berghe (2009) challenged what they saw as a prevalent perception that low T3 syndrome (especially "reduced expression of TRH in the hypothalamus") was "adaptive and beneficial" rather than an aggravating factor or a red flag in "critical illness" calling for medical intervention.  The compelling mystery here is why critically ill patients have low thyroid releasing hormone (TRH) and low blood T3 and T4; iodothyronine (a component of T3 and the Dx enzymes) is likewise very low.  The authors reject these possible theories of how this combination of conditions in critically ill patients mitigates their danger with these arguments: 1) descending T3, typically occurring during progressive protein catabolism (disintegration) in critical illness, is obviously not adaptive, 2) high cytokine levels, because they were shown not to correct defective T3 and T4 levels, while cytokine antagonism alone likewise failed to help and 3) upregulating levels of T3 receptors, found in the "oldest and sickest" patients, is a natural response to declining T3 levels, but not an effective one in the end. 

A key study (Baur et al., 2000) concluded that D1 (mostly in the first hour) and D2 (more slowly, over at least 24 hours) were both upregulated when put in direct contact with some cytokines stimulated by the injection of lipopolysaccharides (LPS) into rat anterior pituitary tissue, whether in vivo or in vitro; however, in vivo, D1 in the liver remained constantly slightly lower and less variable in tissue receiving the LPS injection than in control tissue.  This supported the theory that local hyperthyroidism in the anterior pituitary gland caused by upregulation of D1 and D2 in turn caused by heightened immune system activity in that gland led to systemic hypothyroidism caused by the compensatory downregulation of D1 in the liver, in term lowering its conversion to T4 to T3. This local hyperthyroidism would trigger a drop in TRH production.  Therefore, Mebis and van den Berghe (2009) reasoned in their literature review, a compensatory infusion of TRH could improve such patients' survival chances. They refer to a study (Van den Breghe et al., 1998) in which they explored the use of infusions of different combinations of TRH and at least one of two substances stimulating growth hormone secretion, i.e., GH-releasing peptide-2 (GHRP-2), and growth hormone releasing hormone (GHRH) with acutely ill patients in an intensive care unit (ICU) in a trial with a complex crossover design: they concluded that HPT axis hormone levels were probably thrown off by the illnesses that put the patients in the ICU and could be jump-started by infusions of these hypothalamus hormones.

Wu (2000) states that thyroid disease is much more common in diabetes patients than in others, that in fact 30% of female type 1 diabetes patients have thyroid disease, and that 36% of all diabetes patients have hypothyroidism.

Punekar et al. (2018) performed a case-control study of 100 patients with "decompensated" (symptomatic, implying severe) liver cirrhosis. They found that 41 had low FT3 syndrome (low FT3, normal FT4, and normal TSH), 20 had hypothyroidism (high TSH, regardless of T4 levels), 15 had non-thyroidal illness syndrome (NTIS) with low T4, one had hyperthyroidism, and 23 were euthyroid.

New Relevance of Chronic Fatigue Syndrome?

Ruiz-Nuñez et al. (2018) have concluded that chronic fatigue syndrome (now officially called "myalgic encephalomyelitis") sufferers are more likely to manifest the signs of mild "Low T3 Syndrome."  However, the case is weak in this study involving 36 independent variables, including 17 types of thyroid function tests.  The most notable difference was between Vitamin D levels, which were higher in those chronic fatigue syndrome (CFS) than in those in the control group.  However, it seems clear that none of these variables has predictive value for the occurrence of CFS.

The special issue of older people: should we assume that the hallmarks of "subclinical" hypothyroidism are "normal" in those aged over 65 and should therefore be untreated and maybe ignored altogether?

Harvard Health Publishing (2020) points out that the signs and symptoms of hypothyroidism in older people differ from their younger counterparts and argues that they should not only be guides for diagnosis but have serious implications.  High cholesterol unaccompanied by any other abnormal findings is often an indicator of hypothyroidism in this group.  Hypothyroidism can also cause "reduced blood volume, weaker contractions of the heart muscle, and a slower heart rate," factors in developing heart failure, i.e., inadequate heart pumping effectiveness.  Muscle and joint pain as well as constipation and psychiatric problems are also key symptoms often occurring alone in some individuals.  Finally, hypothyroidism can cause "cognitive decline", which can be mistaken for (irreversible) dementia.

On the other hand, Razvi et al. (2008) determined in their meta-analysis that individuals over age 65 with "subclinical" hypothyroidism were no more likely to suffer from ischemic heart disease than their euthyroid counterparts.

Unresolved questions: differing value systems and problem-solving based on them

Controversies persist in part because the necessary technology has only recently been developed, while many landmark studies used what is now obsolete technology.

What about the stress of exercise (and staying up late) on the thyroids of healthy subjects?

Would strenuous exercise would put stress on the thyroid, perhaps raising TSH levels to an unhealthy extent?  This is important not simply because years of overexercise (which might be becoming more and more popular, especially in some municipalities) might cause permanent damage to the thyroid, but because those engaging in this practice are likely to be seen as especially healthy, thereby unfairly raising the upper limit of the TSH reference range, locally, if not globally.  How can we tell how much is too much, and keep this factor from making it more difficult for others who have genuinely diseased thyroids to be diagnosed?

Van Reeth et al. (1994) first measured the impact of physical exertion on TSH levels over time as part of an ambitious study seeking to develop a regression model involving several hormones to predict circadian cycle phase shift changes caused by exercise undertaken near midnight.  The authors were surprised by the results: "unexpectedly, exercise had robust stimulatory effects on plasma TSH levels."  This study using 17 young, "healthy" men with a clean lifestyle and who had followed normal sleep schedules before the start of data collection.  Although the subjects "had no personal history of ... endocrine illness," Subject 11, one of two subjects whose TSH data were featured in graphs, had TSH values that soared to about 5.0 mIU/L in the baseline (continuous bed rest) study and rose up to almost 8.0 mIU/L in the exercise study at the end of the exercise period.  In contrast, the TSH values of Subject 14 (a much more "normal" subject according to the previously discussed CDC data) soared up to about 2.2 from about 1.8 during the baseline study during the after-bedtime period (and was lower during the day) and ranged from 2.5 to 2.9 during the exercise period in the other study.  Unfortunately, the researchers merely regarded the "unexpected acute effect of exercise" on the TSH curve as a problem for the regression model, one that fortunately could be overcome only because TSH levels were already on the rise before the exercise period; no overall TSH statistics over time were presented).  In fact, as a result, it was hard to sort out TSH values as a separate factor.  Perhaps the most interesting thing to come across clearly and emphatically was that TSH and melatonin levels appeared to dovetail very closely, with the TSH leading.  In fact, these results suggested that TSH levels might govern circadian rhythms, which seem to be the logical outcome of the body's available energy.

Leproult et al. (1997) introduced the psychological factor "sleepiness" and found that, in subjects who stayed up late at night it had a strong statistical relationship to the levels of the hormones that Van Reeth et al. (1994) had studied and body temperature.  Specifically, they observed that increasing "sleepiness," "decreasing body temperature, rapidly rising cortisol concentrations, and maximal levels of melatonin and TSH" occurred together based on a study that simply involved bed-rest without sleep; however, in the part of the study involving a 3-hour period of exercise, they only studied the relationship of the "sleepiness" measurement to melatonin levels.  However, they made mention of the "sleep-facilitating effects of corticosteroids" supported by this study and by others, and speculated that "continuously rising cortisol levels and the persistence of high concentrations of TSH and melatonin may have contributed to increased subjective fatigue."  Unfortunately, the absent TSH data during the exercise study deprived this study of much interest for those interested in this issue.  On the other hand, Fig. 2 showed that the maximum TSH for these subjects was three times that of the minimum during the baseline (bedrest) study.

Scheen et al. (1998) performed a study that focused on the impact of exercise on TSH levels.  They measured the TSH values of each subject over a 24-hour period, during which a 3-hour exercise period was scheduled at one of three different times of the day, i.e., the morning, afternoon and around midnight.  During the time of the day that the subjects logically would be most tired, e.g., at bedtime, their TSH levels increased almost 100% near the beginning of this exercise period over their pre-exercise values, staying at about that level throughout the exercise and dropping gradually afterwards.  TSH rises were much smaller during the morning and afternoon (up to only about 50% of pre-exercise levels) and returned to those pre-exercise values much sooner.  Unfortunately, only relative, not absolute, TSH values are reported in this article, but it does show that TSH can be elevated a certain amount by exercise, and given what Buxton et al. (2000, Fig. 3) have discovered about TSH at rest, i.e., that a TSH of about 1.5 mIU/L is to be expected in the normal and healthy at rest, an exercise-driven TSH under the most stressful conditions might be unlikely to go much above 3.0 in a healthy individual.

Roelfsema and Velthuis (2013) took detailed measurements over a 24-hour period of TSH levels (with a sleep period lasting from about 11:30 pm to about 6:30 am) in a healthy 34-year-old man with a BMI of 23.  His levels stayed almost entirely below 0.5 mIU/L from 9 am to 9 pm, in fact dipping down to about 0.25 mIU/L for about three hours in early afternoon, and rose sharply and irregularly to about 1.7 mIU/L at about 11 pm, then spiked at 2.5 mIU/L at about 1 am.  His TSH levels then descended to about 1.3 mIU/L to his waking time, then dropped slightly below 1.0 mIU/L.  Given this information, and the customary practice in the U.S. of taking blood samples at about 8 am in fasting patients, his level would most likely be recorded as 0.8 mIU/L.  These levels bounced around quite a bit, showing that pulses of TSH were involved, but not a smooth continuum.

The chemistry of muscle damage and its relationship to hypothyroidism

If physical activity brings up TSH levels, muscle damage should also be evident. The enzyme creatine kinase is recognized as a by-product of muscle damage and used to be used to diagnose heart attacks until the use of measurements of levels of the enzyme troponin took its place in that respect. Hekimsoy and Oktem (2005) concluded that creatine kinase levels were abnormally high in 57% of patients with "overt hypothyroidism" and in 10% of those with "subclinical hypothyroidism."

On the other hand, Ness-Abramof et al. (2009) determined that none in a sample of 25 patients with very high TSH levels (30-75 mIU/L) had abnormal troponin levels. Again, because only the abstract was available, it was not clear that this simply meant they were not diagnosed with heart attacks. On the other hand, 14 out of these 25 had abnormally high creatine kinase levels, with a range of 86-1221 "U/L," where 170 U/L was considered to be abnormally high for women and 195 U/L for men.

So this raises these questions: 1) Are hypothyroid people experiencing the same type of muscle damage as those who overexercise? and 2) will overexercise present (itself) in the doctor's office as hypothyroidism?

Iodine in other parts of the body: how their needs impact thyroid function

Anything involving iodine has to be related to the thyroid, so studies of its effect on other parts of the body have to be considered.  Eskin et al. (1995) have come to the conclusion that precancerous and cancerous breast tissue take up iodine, which may trigger apoptosis in that tissue.  In previous studies, they had determined that thyroid hormone had no part in this process.  However, this could have an impact on thyroid function if the iodine that would have otherwise been incorporated into thyroid hormones were taken up by sick breast tissue (or an overactive thyroid might hog the iodine).  Stoddard II et al. (2008) did an in-vitro study of the effects of a combination of ionic iodine, i.e., in potassium iodide (KI), and molecular iodine (I2) on  a line of estrogen-responsive breast cancer cells derived from a woman, i.e., from the MCF-7 cell line.  They used microarray analysis, a method to measure gene expression, to determine the probable tissue changes in these cells, and discovered three genes that up-regulated steroid metabolism (which I think means breakdown of steroids, which include the estrogens) and three genes down-regulating estradiol production.

Neglected areas of study and other factors

The infection connection: can early infections lay the groundwork for autoimmune hypothyroidism?

Previous research has strongly suggested that illness, especially critical illness, might overburden the thyroid by the demands it places on it. But what about the possibility that collateral damage to the thyroid, perhaps permanent, might be inflicted by the action of the immune system as it fights infections that attack other parts of the body? Is it possible that the severe febrile illnesses that we used to call the "childhood illnesses," for instance, attack the thyroid or confuse the immune system in ways that cause it to develop anti-thyroid antibodies? Could these conditions be the major causes of Hashimoto's thyroiditis?

What about increased susceptibility to infection?

No one since Dr. Broda Barnes has investigated whether the risk of infections or worsening chances of recovery from infections increases with untreated hypothyroidism, "subclinical" or otherwise.  Now that the COVID-19 pandemic has arrived, it might be a timely concern.

That other heart/lung issue: oxygen saturation levels

Studying oxygen saturation levels in hypothyroidism sufferers is worth considering because the sedating effects of hypothyroidism might depress breathing, and the only study of lung function in those with hypothyroidism has indicated weak mechanical lung action.  It is very easy and cheap to measure: oximeters cost only about $20, are easy, safe, and painless to use and do not need to be hooked up to anything (although they use batteries).  Yet they only seem to be used in hospitals.  Since oxygen saturation level is a diagnostic criterion for COVID-19, this factor needs to be understood.

Menopause (and adolescence)

There is one great research gap: the study of menopause as timing for the onset of hypothyroidism.  Although menopause is not a disease, it is a time when disease risk increases, maybe because the great change in hormone levels makes homeostasis, that state of ideal physiological balance necessary for health, more difficult to achieve.  As we learned in biology class, long-term upsetting of homeostasis can lead to a great variety of diseases.  Adolescence is another time of life that challenges homeostasis, as higher teenage male TSH levels seem to suggest.

It seems to be popularly believed that any simply apparent, simultaneous occurrence of menopause and hypothyroidism in an individual can be explained by the similarity of symptoms.  A simple online search can turn up articles promoting this point of view from a variety of sources which suggest that the signs and symptoms of menopause are routinely mistaken for those of hypothyroidism and vice versa. 

Lack of personal relevance to physicians

Very few physicians have personally experienced hypothyroidism or menopause.  Because they have entered the field relatively recently, most women physicians are too young to have experienced menopause, while the extremely high energy demands made on medical students, interns, and residents (at least in the U.S.) effectively screens out those with vulnerable thyroids.  They are also unlikely to mingle socially with those with hypothyroidism because their social circles are likely to include mainly other physicians or those at their social level.  Because of this, physicians are more likely to be skeptical of patients' reports of disabling fatigue, and perhaps are simply unable to imagine what such patients are experiencing.  If hypothyroid patients, especially women, come to them pleading for help in maintaining highly responsible jobs, these physicians might be skeptical that they are holding such jobs, or suspicious of whether they belong in them.

Inability of affected individuals to advocate for themselves (and misleading signs of recovery in those who are treated)

Is it possible that many, if not most, patients with even "subclinical" hypothyroidism are too ill to find their way to physicians or to speak out publicly on the general issue of hypothyroidism?  Have they learned to be passive and trusting in order to conserve their energy?  The flip side of this problem is that when hypothyroid individuals receive treatment for their disease, they might become more assertive and set higher standards for their well-being, which might be misinterpreted by their physicians to be a sign of worsening health.  Because very ill hypothyroid patients have to choose their battles, they make "good" patients, especially if they blame themselves for their illness.

Social attitudes about the signs and symptoms of hypothyroidism

Hypothyroidism tends to make its victims sickly and unattractive, and perhaps prone to complicating diseases which in themselves carry severe social stigmas. Our society favors the slim, energetic, adventurous, bright-eyed qualities of those with healthy thyroids.  These attitudes, sometimes shared by physicians, might keep sufferers from getting the help that they need. Many people, including physicians, stereotype those seeking hypothyroidism treatment, even if they have been diagnosed with it by other physicians, as being unusually fat and seeking diet pills, as being stupid and seeking a miraculous remedy for that, or just as being lazy and irresponsible.

We in the U.S. might not have shaken our attraction to eugenics, founded on the notion that nature has made a fairly large proportion of citizens defective in fundamental, unchangeable ways.  Even with the discovery of epigenics and its having standard uses requiring sophisticated knowledge of how it operates, there apparently remains a deeply rooted belief in our culture that diseases with systemic effects are an essential part of an individual's character, perhaps even having psychological origins.  Susan Sontag dealt with these issues in detail in Illness as Metaphor (1978), and although attitudes toward cancer have changed since then, what she wrote about could apply well to hypothyroidism (although the term "cretin" has fortunately disappeared from common parlance.)

Major media silence about hypothyroidism, while it has become tabloid fodder

There are many diseases that get lavish coverage by the major media, most notably breast cancer.  But hypothyroidism is never mentioned by properly informed individuals who have the general public's ear.  Therefore, those who have the disease are unlikely to recognize it as such, and those in their family and community are unlikely to extend a helping hand. Still others with little understanding about the disease and who have been led to believe that TSH levels have no diagnostic value tie up physicians' time and cause them to be impatient with those who are genuinely ill.

This information vacuum is too often filled by unscrupulous individuals without medical training or scientific education, who promote useless and/or harmful remedies, manipulating the public with blame-the-patient messages. Unfortunately, hypothyroidism has become the subject of cynical articles in magazines aimed at the most gullible, while authorities in the field have done very little to reverse this terrible situation.

Cultural value differences affecting government oversight

Some countries have fundamentally different missions in healthcare, which necessarily affects their approach to the prevention and treatment of thyroid disease.  The US places its highest emphasis on longevity, and therefore focuses on the diagnosis, treatment, prevention of (and screening for) the diseases that are the most frequent immediate causes of death, such as ischemic heart disease, diabetes, and cancer, and only secondarily on those diseases that cause significant pain and disability without being fatal in the short run, even though they might set the stage for the killer diseases.  On the other hand, Asia is apparently more concerned about the ability of citizens to function normally in society, to be free of unnecessary handicaps, both mental and physical ones, in essence, quality of life.  Asians also seem to be more interested in understanding the mechanisms of diseases rather than simply relying on the results of clinical trials, and do not put a primary emphasis on minimizing medical care expenditures.  In general, the industrialized world has apparently set the social performance bar higher in recent times as public health has improved, which puts those with untreated hypothyroidism at an increased disadvantage. 

Germany developed the technology necessary to show that most people had TSH levels much lower than was previously thought, and what was previously considered to be normal was sometimes quite unusual, leading to the controversial concept of "subclinical" hypothyroidism. Today, Germany has refrained from performing clinical trials for the main purpose of determining future restraints on treatment.  However, government authorities in northwestern Europe apparently are recognized as the authorities on international standards for hypothyroidism treatment, and currently focused on finding justifications for cutting healthcare costs.

Economic factors

Treatment of hypothyroidism is hardly a cash cow for the medical care industry.  Patients who receive standard treatment, i.e., one T4 pill a day, generally pay $25 or less a month for this (although drug prices in general have shot up from this value, from late 2016) and see a doctor to get a prescription once a year, typically at the annual physical, where they are tested only for TSH levels specifically in connection with this disease. On the other hand, treating hypothyroidism's individual signs and symptoms separately is more lucrative: statins (and their successors, PCSK9 inhibitors) and antidepressants can be very expensive.  Figure in the estrogens and progestins that many women are prescribed at menopause, often to treat psychological pain, and this is quite a bit of money!  In addition, hypothyroidism's complications can also be expensive to treat, and therefore are worth more economically.

On the other hand, an apparent monopoly has driven up the price of an L-T3 drug in the U.K., although only a very small proportion of patients there have received prescriptions for it even before that price increase.

Another factor is the cost of diagnosed disability: a condition that is diagnosed mainly on the basis of symptoms is likely to draw suspicion of attempted fraud on the part of patients who report them because certified disability can bring such a patient financial benefits. Therefore, even if a standard treatment exists for such a condition, a physician's concern about protecting the financial interests of the patient's employer, the government, or of the involved insurance company, might override that physician's feelings of obligation to meet the patient's health needs.

Finally, in the march toward (affordable) socialized medicine, some people will apparently need to be thrown under the bus, as twentieth century history has taught us.

Physicians' increasing workload and concern about malpractice suits

Physicians in the U.S. and many other countries are facing a sharp increase in the number of patients needing and qualifying for treatment. The ACA has added 20 million citizens to the ranks of the insured. Immigration, legal and illegal, is occurring at a substantial rate, much of it caused by a long-standing war in the Middle East, especially Syria. These wars cause an unending stream of American casualties, many serious and leading to lifelong disability. In addition, the burden of new red tape, meant to force physicians to adhere to the latest government-determined rules, allows physicians less time to meet with patients and to treat them. Because the number of physicians is not rising fast enough to keep up with this increasing demand, rationing of healthcare is inevitable, and the temptation offered by the option to end the treatment of some diseases is apparently irresistible.  In addition, unanticipated crises such as the spread of the COVID-19 virus put great stress on an already lean-and-mean medical care system.  While some treatments are obviously unnecessary or prohibitively expensive, other diseases rarely affecting the power brokers of these societies might be at the top of the hit list if no one speaks up for the forgotten people that those diseases afflict.

Physicians have complained about these issues in online forums and many have written about leaving the profession because of worsening work conditions. What will be the impact on patient care?  Badly conceived restructuring of physicians' jobs has created a crisis, according to Hartzband and Groopman (2020).  Burdened by an oppressive interactive system of online red tape associated with the new electronic health records (EHR), physicians lose autonomy, a sense of competence, and relatedness with patients and coworkers; they experience divided loyalties between those they serve and a system that in its own way is both very needy and uncaring.  These authors recommend substituting evidence-based measures of quality for the "meaningless metrics" now in place.

These problems only aggravate physicians' fear of being sued for making mistakes in adjusting the levothyroxine dose for hypothyroidism patients, which became increasingly challenging as reference ranges narrowed.  It was predictable, then, for them to push back by calling for an increase in the reference range, most effectively done by increasing the upper limit of the range.  In addition, the greater danger of hyperthyroidism (at least of Graves' Disease) helped to shore up the logic of this recommendation; a physician is much more likely to be sued for missing a hyperthyroidism diagnosis than one of hypothyroidism.  Of course, casting doubt on the validity of "subclinical" hypothyroidism, i.e., characterized by a TSH in the range between the value that was currently accepted as the top of the reference range and the old value that physicians had gotten used to in the past, as a true disease with identifiable symptoms and potential complications has laid the groundwork for restoring primary care physicians' former easier situation. 

The U.S. eugenics tradition has been lowering the bar for what is deemed to be "normal" (as opposed to "healthy") and therefore unworthy of treatment

Even though the eugenics scandal and its harm to medical care standards has officially ended, there apparently still remains a stubborn belief that about ten percent of our nation's population is naturally defective and that providing medical treatment for these people's problems is necessarily a quixotic attempt to improve on nature; measures currently in place emphasize separating them from the well-adjusted ninety percent.  When researchers argue that hypothyroidism has no unique signs or symptoms, and imply that existing such manifestations have unsavory origins such as poor health habits or bad character, or simply originate in mental deficiency or incurable dementia, this sets the stage for the de facto dehumanization of many if not most with this disease.

Government funding: now a conflict-of-interest concern?

In the past, national governments were generally trusted to make their research funding decisions on the basis of the potential quality of proposed scientific research. However, as the U.S., the U.K., and some countries in the European Union feel pressure to cut costs in the service of making healthcare more affordable, these governments might favor funding studies that promise to show that certain current forms of healthcare have little or no value.

Conflict of interest on the part of the patient: real or imagined

Do physicians typically suspect patients claiming to have hypothyroidism symptoms of having ulterior motives, such a wish for diet pills or for a respectable alternative to a psychiatric diagnosis? Are they suspicious of patients who are open about wanting to avoid taking statins? How often do physicians suspect sufferers of lying about substance abuse?  Is it possible that so many people in a particular region have wanted to see thyroid specialists in the past for problems that turned out to be unrelated to thyroid disease that new patients are routinely turned away or discouraged by requests for mountains of paperwork up front?   What can be done for those with thyroid disease facing this regional medical care problem?  Is a local thyroid specialist shortage a factor in raising the upper limit of the TSH normal range locally?  As the opioid abuse epidemic moves into the spotlight and the proportion of physicians who have grown up in high-crime neighborhoods increases, will patients feel increasing pressure to prove that they are not selling levothyroxine on the black market?  Patients who live in blighted neighborhoods or in cities with high poverty and crime rates might find it especially difficult to convince physicians that their reports of symptoms, now necessary for diagnosis according to the new guidelines, are sincere.

Perception of hypothyroidism and Alzheimer's dementia as women's diseases and reflections of women's stereotypical mental weakness

Social problems, including health problems, that affect women are taken less seriously because of a persistent belief in women's natural psychological inferiority.  Although we in the U.S. take credit for having achieved social equality for the sexes, in practice we treat men with more respect and work harder to meet their needs than we do women's.  This might explain why hypothyroidism is not considered to be an important disease.  Similarly, exploration of possible treatable causes of Alzheimer's (or of diseases that mimic it), such as hypothyroidism, has not taken a constructive path, and Alzheimer's is increasingly being seen as a natural defect of women's brains, with estrogens seen as the only possible endocrine factor.

What about those with low-normal or borderline free T4 and high TSH levels?

This is one area of research which is never touched on, although problems associated with high-normal free T4 have been studied, with positive results reported.  Much research has gone into determining which TSH levels are associated with "real" hypothyroidism, but it stands to reason that since whether free T4 is too low or not is the only universally agreed-upon criterion for diagnosing "overt" hypothyroidism, it should also be studied by clinical trials.  Of course, those who believe that there is no monotonic relationship between TSH and T4 levels, that free T4 levels are completely fine-tuned by TSH secretions, would see no point in pursuing this issue.  But they would need to explain their belief in this TSH-T4 relationship model in order to be convincing to those who insist on a robust model that can be trusted to do right by patients.


A revolution in (not) treating most cases of hypothyroidism

Better a diamond with a flaw than a pebble without. (attributed to Confucius)

The perfect is the enemy of the good. (Anon)

Quis custodiet ipsos custodes? (Juvenal)

This is what high-quality research tells us:  There is a typical TSH value for healthy people, apparently somewhat less than 1.5 mIU/L (if measured in the morning), and values that are higher or lower are associated with worse health in proportion to the distance that they are from that ideal value. As the TSH level varies, so does the free T4 level, but in a continuous nonlinear inverse way: toward the high end of what is typically defined as the reference range for TSH, i.e., where hypothyroidism starts to manifest itself, the TSH level rises very rapidly relative to the decrease of free T4 level.  The opposite happens at the hyperthyroid end of the reference range, when free T4 increases ever more rapidly as the TSH level decreases at an ever smaller rate.  In sum, TSH is a much more sensitive indicator of the degree of hypothyroidism, while free T4 is the more sensitive indicator of the degree of hyperthyroidism.  These curves are all very smooth, so that the dividing lines between "normal" and "abnormal" are not obvious, although places along the elbows of these curves (at the high end of both of the TSH and free T4 reference ranges) are good common-sense candidates.  These are not clinically insignificant numbers; they represent the two points at which the pituitary gland begins to lose its control over the thyroid, i.e., either 1) where the thyroid is functioning too poorly to respond to pituitary input, or 2) where the thyroid is so active that even lack of pituitary input cannot bring its secretions down to a normal rate.  Dietrich et al. (2012) have shown that the thyroid's maximum secretory capacity drops to only about one third of normal before hypothyroidism is even considered to be "subclinical" according to current criteria.

Anti-thyroid antibody levels might be useful in confirming the level of stress on the thyroid.  In individuals with uncomplicated Hashimoto's thyroiditis, such antibody levels rise as TSH levels do, whereas when a non-thyroidal disease alone is what drives up TSH levels, these antibody levels sometimes remain low, though this is hardly an ironclad rule.  Trying to determine whether a patient has thyroid disease or not based simply on TSH and free T4 levels could lead to confusion about the source of the problem as a result.  Even more so, defining thyroid disease formally only on the basis of these hormone levels without including measurements of anti-thyroid antibodies in these diagnostic criteria could lead to overlooking the disease(s) actually causing these abnormal hormone levels.  One thing that is unfortunate about this oversight is that negative experiences treating (undiagnosed) non-thyroidal diseases driving up TSH levels with only thyroid hormone replacement might lead physicians to assume that there is no point in delivering such treatment to patients with similar TSH and T4 levels, even when thyroid failure is indeed the cause of their abnormal thyroid hormone levels.  Unfortunately, another source of confusion is that these non-thyroidal illnesses might cause TSH levels to rise only within the "subclinical" range, while pulling free T4 levels below their reference range; admittedly, this phenomenon has apparently not been researched.

It should be evident that "subclinical" hypothyroidism is not an ideal state of affairs when it comes to general health, and that some serious diseases might be prevented from developing by hormone replacement treatment of thyroid failure.  But the one major area of controversy remaining is how well patients feel when treated.  For some, the mildly sedating effects of early hypothyroidism might provide welcome comfort, while proving to be a handicap for those choosing (or needing) a lifestyle, indeed a career, requiring high alertness levels and the restless state of mind that drives one to solve problems quickly and effectively.  We as a society depend heavily on people who meet the latter description, since many protect us from danger.  So it is important not to oversimplify treatment rules, and it stands to reason that the patient's wishes and circumstances should always be taken into consideration when treatment decisions are made.

How does the model from showing the real relation from TSH and free T4 levels differ have more validity from the one assuming perfect compensation of free T4 levels by those of TSH?

The point of view adopted by policymakers who insist that "subclinical" hypothyroidism does not represent a real problem for patients in general seems to assume that 1) a TSH level of 1.5 mIU/L and one of 10.0 mIU/L differ very little, if at all, in their effects on how an individual feels and functions, 2) TSH levels typically vary widely and chaotically and it takes months to determine a pattern in them, 3) (in the view of some) treatment with L-T4 permanently fixes most high-TSH conditions, and 4) free T4 levels are typically stable except in those rare few with severe hypothyroidism.   The logical conclusion that follows from this point of view is that TSH levels are useless for diagnostic purposes and that only free T4 levels should be used for diagnosis of hypothyroidism.  But these assumptions, which never have been supported by scientific evidence, have recently been questioned and refuted convincingly, even as they continue to guide influential policy decisions in some countries.

It is important to remember that TSH and free T4 alike are surrogate parameters of thyroid health, while the thyroid's maximum secretory capacity is an actual measure of thyroid function.  But we use such surrogate parameters as TSH and free T4 levels to measure thyroid function because they are cheaper, easier, cause less stress to the patient, and because they do a good job of delivering this information.  Their value as diagnostic tools has been validated by the demonstration of their relationship to the thyroid's maximum secretory capacity.  Could it be that the perception of TSH levels as being confusing and unreliable indicators of thyroid health is influenced by lab performance plagued by inaccuracy and imprecision?  Should we be concerned that different labs produce widely varying measurements of the same patient's TSH levels?

What are the dangers research has shown of neglecting "subclinical" hypothyroidism?

Heart disease:  The general consensus is that it has a strong link to heart failure, but a weaker one to ischemic heart disease.  It thickens a part of carotid arteries' walls, and this can be reversed by thyroid hormone therapy.  But it is not a major factor in heart attacks, and does not increase all-cause mortality.  Studies of blood lipid levels in subjects with "subclinical" hypothyroidism have come up with a few conflicting results, perhaps because of the common use of lipid-lowering medications, such as statins, which are not always accounted for in experimental designs.  However, there is strong agreement, at least among Asian researchers, that subclinical hypothyroidism is linked to high LDL and triglyceride levels, although the relationship to HDL levels is weak at most.  Therefore, although subclinical hypothyroidism does not pose an immediate threat to heart health, treatment might nip a progressing problem in the bud, i.e., be a legitimate part of preventive medicine in the way that blood pressure medications and statins are.

Liver disease:  It has a strong link to non-alcoholic fatty liver disease because one of the two thyroid hormone receptors in other organs is concentrated in the liver.  A drug stimulating that receptor in order to reduce liver fat is in clinical trials.  But recognition of this relationship is not been reflected in approaches to fatty liver disease in the U.S., apparently because of concerns that thyroid hormones make the heart work harder.

Respiratory problems:  Although the lungs' anatomy are not affected, lung function is reduced because of diaphagm muscle weakness, resulting in low oxygen saturation and high carbon dioxide concentration in the blood.

Pregnancy problems, such as miscarriages, preterm birth, stillbirths, and birth defects:  Pregnant women with "subclinical" hypothyroidism tend to have more of these problems than women with normal thyroid profiles.

Neurological and psychological problems:  Thyroid surgeons have observed signs of sedation in "temporarily hypothyroid" patients; this has not been studied systematically in those with functional hypothyroidism but needs to be.  Women with TSH levels of 2.1 mIU/L and higher on the average are more likely to be diagnosed with Alzheimer's dementia than those TSH levels in the 1.0-2.0 mIU/L range, but not as likely as those with TSH levels below that.  Many studies claim that subjects with "subclinical" hypothyroidism have no more quality-of-life problems than those who are normal, but it is not clear that the patients in those studies were in a position to compare themselves to those in normal health or to their experiences with normal health in the past, when their TSH levels were normal.  Interpretations of the few, small, clinical trial results agree that treating patients aged over 65 years has no evident benefit.  However, some practitioners in the field disagree, arguing that the signs of symptoms of hypothyroidism in older people are poorly understood and emphasize the danger of mistaking hypothyroidism for Alzheimer's dementia in these patients.

Susceptibility to infection and difficulty recovering from it: This is a neglected field of study currently.  However, the COVID-19 crisis might cause priorities to shift in that direction.

Unfortunately, under pressure from public figures leading the movement to give medicine a complete makeover, researchers reporting these results generally feel compelled to add disclaimers undermining their message by claiming that their results cannot be taken too seriously until they are tested by clinical trials, implying that T4 therapy is very risky.

Are we really ready to end T4 therapy for nearly all people over age 65?  Do we have solid proof that higher TSH levels are normal and healthy for these older people?

One of the most disturbing conclusions from recent research claims that, because older people on the whole have higher TSH levels, these levels should not be interpreted the same way as those in younger people. These researchers claim that we should assume that these higher levels neither indicate some degree of thyroid failure, nor are they are a sign of greater stress on the thyroid, until compelling proof to the contrary emerges.  These claims tend to ignore two important considerations: 1) on the average, disease-free older people have TSH values only slighter higher than those of their younger counterparts, and 2) because older people are more likely to have chronic diseases, and more of them, these diseases might be driving up their TSH levels.  Although low T3 syndrome can drive TSH levels down, patients with this problem do not live very long and probably do not have a major influence on population average TSH levels. 

People over age 65 are a very diverse group.  Even though some are very ill and indeed disabled, some are in excellent health and hold highly responsible jobs.  Many have filled high positions in government, for instance.  These positions require high energy levels and great clarity of thought.  To require them to have TSH levels of at least 10.0 mIU/L and abnormally low free T4 levels to receive treatment would not only subject them to undue hardship, but conceivably put our society at risk.

Issues leading to confusion

Thyroid failure proceeds on a spectrum, while medicine has traditionally insisted on a clear dividing line between "healthy" and "sick": So how should treatment of different levels of thyroid failure proceed?

Because of the smooth curve that TSH levels follow in the population, and uncertainty exists about the precise point at which to draw the line, it stands to reason that doses should be determined in a nuanced manner, with very small doses for very mild cases and larger ones for more advanced ones.  Are dose gradations of 25 mcg small enough to fit patients' needs?  What dose should patients be started at, and when should they be increased or decreased?  This seems to be discussed more often by pharmacists than by physician researchers, whose approaches differ significantly.

Why treated hypothyroid patients are often dissatisfied and therefore why treating hypothyroid patients might be unrewarding

Many patients who are treated for "subclinical" hypothyroidism with L-T4 are dissatisfied for reasons that are just recently being explored.  The most obvious, as pointed out by Midgley et al. (2019), are that physicians often don't aim for the ideal TSH level for individual patients within the reference range, instead settling for values just below the upper limit of that range.  Many population studies have shown that a TSH level of 1.5 mIU/L or even less is typical of adults in normal health, yet some influential clinical trials using L-T4 as the study drug report adjusting subject TSH levels in subjects to over twice that level.  Ironically, it seems that the very dramatic differences that patients experience at different TSH levels within the reference range have led to study conclusions that there are no significant differences in patients' subjective experiences over wide TSH ranges; this appears to be the result of poor experimental design and uninsightful, sometimes even biased, interpretation of results.

There also is a concern that, in a minority of patients with genetic anomalies affecting deiodinases, there is special difficulty in converting T4 to T3.  In some cases, this conversion apparently occurs in the brain (and not just in the liver and kidneys), affecting neurological health, so that problems affecting it might not be detected by the T3 blood test.  Sometimes diseases affecting these three organs are responsible for these problems.

Maybe patients feel better when they are slightly hypothyroid because of their sedated mood, which might or might not correspond to good physical health or to adjustment to a challenging work world.  It seems that the real depression sufferers are hyperthyroid.  Coming back from a sedated state appears to cause distress in many cases.  This simple factor might have led to a great deal of confusion in trying to pin down the quality of life of those with "mild" hypothyroidism: what feels good in the short term might turn out to be very bad in the long term, especially in some situations.  This broadens into a social issue when one considers the risk posed by individuals in the workplace who are "comfortably numb" despite passing standard drug tests and being required to perform tasks that require a high level of alertness, not just to do the job but to protect public safety.

Yet another problem is that elevated TSH levels might arise from undiagnosed diseases that affect other parts of the body, because of the additional stress that they put on the thyroid.  Because L-T4 treatment cannot cure those diseases, patient distress caused by them would continue.  What remains unexplored is whether L-T4 supplementation can improve a patient's experience after such diseases are properly diagnosed and put under appropriate treatment.

There is also a psychological twist to this problem in the case of those who receive proper treatment, who are restored to normal physical health.  The Czech word "litost" might shed some light on this mystery.  As defined by Milan Kundera in The Book of Laughter and Forgetting, it means "a state of torment created by the sudden sight of one's own misery." When the well-documented brain fog of hypothyroidism clears upon treatment, the patient might see clearly for the first time the destruction the disease has caused, not just to his/her body, but to his/her life, e.g., career and relationships.  Perhaps gaining what should have been their natural powers all along, realizing the devastating effects the disease had had on them, and recognizing that much of the social damage done could not be undone is a major psychological blow, especially when they face this terrible reality alone, without understanding or sympathy from others. And then there is the effect of the Maslow Hierarchy of Needs: when some types of needs are met, others invariably take their place.  This could amount to a major personal identity issue because of personality changes.  Physicians add insult to injury when they make light of this. What is quite scary is when this irritating mystery leads to an attempt to (mis)use science to justify withholding treatment from these people.

Yet in contrast to all of this, the treatment of pregnant women with hypothyroidism is given great care according to American Thyroid Association guidelines, which recommend regular testing of thyroid function during pregnancy, although this is not mentioned in U.S. lists of covered services provided to pregnant women.

Iodine sufficiency is still an open issue

The general approach to iodine deficiency in the U.S. and western Europe has apparently evolved to using iodized salt in preparing processed foods.  However, as FDA nutritional data indicate, whole foods, such as fresh fruits and vegetables, tend to be iodine-free, and their consumption is promoted by our medical establishment.  U.S. population iodine levels have not been examined in a long time.  Should we be concerned?

Classification of academic journal articles to avoid misunderstanding of their applicability

Editorials, literature reviews, and research findings should be identified as such and so categorized in academic journals.  If a submission makes statements about diagnosis and treatment standards without explicit references to either original research results or to the findings in other publications, it should be presented as either an editorial or some other statement of opinion.  If it represents official standards, it should indicate the positions of authority of the authors and those to whom their authority are applied.

The Catch-22s that make it difficult for hypothyroidism sufferers to get treatment

1.  If a medical service is considered to be beneficial and low-risk and withholding it is considered to impose risk of harm, it has tended not to be reviewed via clinical trials because of ethical considerations.  On the other hand, if few or no clinical trials have been performed reviewing it, policy-makers now tend to dismiss it, arguing that there is insufficient scientific evidence that it is not "low-quality" and recommend against providing that service until relevant clinical trials have proved its value.

2.  Patients without symptoms, or at least symptoms that suggest hypothyroidism, should not have their TSH levels tested, according to current U.S. government standards.  But since no particular symptoms of hypothyroidism are considered to be unique to the disease, reporting symptoms associated with hypothyroidism is still no guarantee that the patient's physician will feel required to administer the TSH test.  Although one study has shown that the total number of symptoms, rather than any one symptom, is the better predictor of a high TSH, this is apparently hard to translate into standard diagnostic practice.

There is still another hurdle that symptomatic patients face on the way to being tested: if one of their symptoms is depression, and their responses to a questionnaire indicate depression, the USPSTF recommends treatment with antidepressants or with cognitive-behavioral therapy (CBT) and apparently nothing else, regardless of whether that depression is a symptom of the patients' hypothyroidism.

3.  Patients who have "subclinical" hypothyroidism as defined by (not always yet measured) TSH and T4 levels and are too sick to make it to a doctor are invisible to the policy makers.  However, if they produce the necessary (temporary) energy to do so by drinking coffee or another caffeine-containing drink, or simply conserve their energy until they're able to get there, their TSH levels might be too low for them to qualify for treatment. 

4.  Patients over 65 are not supposed to receive levothyroxine therapy according to the latest recommendations.  A TSH level of 10.0 mIU/L or above is considered by many physicians to be the minimum level for younger people to require treatment, but the majority of patients with such a TSH level are over 70 according to at least some studies.

5.  Patients are considered to have "subclinical" rather than "overt" hypothyroidism if their T4 levels are in the reference range, no matter how high their TSH levels are.  But their T4 levels might not go below the reference range unless they have at least one (perhaps undiagnosed) disease affecting another part of their bodies.  How many patients with uncomplicated hypothyroidism actually experience abnormally low T4 levels?  It might not be very many; this has not been investigated and should be.  Of course, patients with TSH levels in the reference range and T4 levels below its corresponding reference range have traditionally been given the "euthyroid sick syndrome" diagnosis and are assumed to have a healthy thyroid and a disease affecting another part of the body and therefore not to benefit from supplemental thyroid hormones.

6.  Most obviously: hypothyroidism is one of the best understood illnesses, and as such, easy to diagnose (with laboratory tests) and treat; although drug prices have risen dramatically in recent years, levothyroxine remains one of the most affordable.  Ironically, this is leading to hypothyroidism treatment being taken away because its earning power for the medical care industry is so slight. Besides, it seems that only publishing papers debunking its established diagnosis and treatment methods offers a path for most researchers in the thyroid disease treatment field to achieve career success.  If our government really cared about making medical care affordable, it would be taking action to remedy this general problem!

Symptoms: a misunderstood diagnostic battleground?

The battleground area seems to be that of symptoms.  If "subclinical" hypothyroidism does indeed have identifiable symptoms, that should make universal screening unnecessary, which seems to be the general professional consensus. The problem seems to be not so much that hypothyroidism does not have symptoms, but that they are not unique to the disease.  Could it be that physicians are resistant to treating "subclinical" hypothyroidism because they are used to reading its symptoms as evidence of bad health habits instead?  Do they see themselves as enablers, giving undeserving patients an unfair advantage?

Perhaps diseases that affect the entire body are hardest to diagnose on the basis of symptoms, even when severe

Unfortunately, the diseases that produce symptoms most likely to be dismissed as "nonspecific" are those that affect the entire body, because that makes diagnosis based simply on clinical signs and symptoms alone very challenging. Yet systemic effects are more likely to be disabling if not outright serious, so this is a major problem.  Thyroid disease affects every cell in the body, thereby producing a great variety of signs and symptoms across as well as within patients.  This point seems to have been missed by nearly all of the researchers in the field, unfortunately, although this observation has been made by a veterinarian.  Perhaps the Chiappa et al. (2019) publication, which describes a thyroid storm case which eluded diagnosis until many other candidate diagnoses were ruled out because of a delay in testing the patient for thyroid disease, makes this point especially powerfully.

What about the threat of concomitant non-thyroid illness (NTI)?  Will "subclinical" hypothyroidism affect the course of that disease? Will more severe NTIs be mistaken for overt hypothyroidism?

Even though those with "subclinical" hypothyroidism might be able to walk the tightwire of health by limiting their energy expenditure in day-to-day life, could they maintain it if another illness, such as influenza or measles, struck?  How well would their thyroids be able to handle these problems?  I have not been able to locate any studies in the last 50 years that address this issue, although the stress of pregnancy is acknowledged.

What proportion of those diagnosed with "overt" hypothyroidism, i.e., those with elevated TSH levels and low T3 and T4 levels, have a non-thyroid illness in addition?   It stands to reason that a serious illness such as cancer or diabetes would put extra stress on the thyroid, and could pull thyroid hormone production down below the normal range, especially in individuals who already are in the "subclinical" range.  Indeed, one study showed that T3 infusions given heart failure patients had promising results and did not appear to cause any harm.

The Alzheimer's dementia connection

Neurological hypothyroidism symptoms can mimic Alzheimer's, and the continual drive to raise threshold TSH levels for treatment with thyroid hormones could cause some with "subclinical" hypothyroidism to be wrongly diagnosed with Alzheimer's.  Women face a special problem according to one study:  both hypothyroidism and hyperthyroidism raise their risk of being diagnosed with Alzheimer's, while such associations are minimal for men.  Do these women actually have Alzheimer's and just seem to?

An elephant in the room: perhaps some scattershot lab results anyway, and a resulting fear of diagnostic mistakes

There needs to be more attention paid to the problem of inconsistent assay values across labs and across different TSH assay kits.  Widely varying distributions of TSH assay results, especially in the "subclinical" range of 4.0-10.0 mIU/L, have led to vastly different perceptions of how common "subclinical" hypothyroidism is and of the nature of its manifestations, as well as common physician distrust of TSH assays. Treatment of those who need it should not be held up because of the apparently inadequate oversight of labs.  The work of Dietrich et al. (2012) illustrates how difficult it is to get an exact measurement of TSH levels when they go above about 5.0 mIU/L, especially when T4 is near the bottom of the reference range or below it, and to get such exactness in measuring free T4 when it goes above the reference range, and perhaps when it is in the upper part of that reference range.

There are many TSH assay kits available, with widely ranging reference ranges.  Accuracy is probably an issue with some. 

Scientific Process Issues

Interdisciplinary studies of the effects of thyroid failure on other parts of the body are rare, if they exist

For example, why are there no studies of how thyroid problems affect the heart, and when treatment is needed, that involve both endocrinologists and cardiologists?  When cardiologists argue that the thickening of artery walls in hypothyroidism, and the reversal of that process with L-T4 treatment is an important indicator of the relationships of thyroid and heart health, what they say is often dismissed by thyroid specialists.  Too often, we have endocrinologists or internists diagnosing a variety of heart diseases, or psychological conditions, without collaboration of professionals in those disciplines.  In fact, some studies rely on questionnaires rather than physical examinations.  Perhaps most striking is the reliance on mathematical analysis by many who have had very little mathematical training, while professional statisticians are often allowed only a very limited role: simple categorical models seem to be the standard, while more innovative continuous models are not given as much recognition.  How trustworthy is this information if we care about author credentials?

Reference ranges for TSH and the thyroid hormones: how should they be interpreted once the method for calculating limits is set?

Some researchers have suggested using separate normal ranges for those of different racial or ethnic groups, but this is dubious science because these groups are poorly defined from a strictly physiological standpoint and their lifestyles, such as traditional diet, are more likely than their apparent genetics to explain their differences.  As for using different reference ranges for different adult age groups, this might discourage treatment of older people because of an implied perception that their fading health is "normal" (unless they have cancer or heart disease).  Although their treatment should be more cautious, should they not have a right to try to have their suffering relieved by means other than treatment with painkillers or poisons?

An important problem: dependence on a few recent studies that use samples not representative of the relevant population

The formally stated definition of "subclinical" hypothyroidism in research is based entirely on TSH and thyroid hormone levels, i.e., that the former is abnormally high but the latter are normal.  However, subjects in recent clinical trials studying the harm done by this condition had one more important characteristic: they were not currently being treated for hypothyroidism at the start of the study, therefore they do not properly represent the entire population of individuals who would have had these hormone levels without treatment.  How many people now being treated for hypothyroidism had these "subclinical" levels before being treated?  Suppose their numbers are much greater than their counterparts who are not being treated?  Is it obvious, then, to recommend withholding treatment from all individuals with "subclinical" hormone levels, regardless of the reason that they are not currently receiving treatment?   How many will lose the treatment they are currently getting on the basis of the new standards?

Problems with pedantic or even misguided interpretations of p-values and of odds ratios are complicating interpretation of relevant studies

The problems with the underpinning science are evident in heaps of seemingly innocuous detail.  For example, interpretations of p-values and odds ratios appropriate (i.e., extremely small and large, respectively) to the efficacy sections of clinical trials of new drugs have been applied to what are essentially safety studies; for instance, what happens, for example, to hypothyroidism treatment standards when a strong but imperfect relationship between TSH and triglyceride levels is dismissed as a nonexistent relationship in a study that busy government bureaucrats assume is sound?   There seems to be a new attitude, reflected in recently published case studies, that "subclinical" hypothyroidism is a temporary condition that often can be cured by a few months of levothyroxine treatment, even when TSH levels return to their original out-of-range values; yet apparently no studies involving multiple patients support this treatment approach. 

Setting down hard-and-fast treatment rulings based on sparse information

Can an "I" recommendation, made mainly on the basis of weak scientific evidence on the basis of recent criteria, i.e., the absence or scarcity of relevant clinical trials, cause insurance coverage to be revoked on government-sponsored plans to be revoked?  Can it remove a diagnostic or treatment measure, long considered to be settled science, from standard medical practice, even if it's not dangerous or expensive?  Can such a bureaucratic ruling be repealed in the way an enacted law could, by a process other than an executive order?

Failing to take into consideration whether subjects were taking lipid-lowering medications such as statins in experimental design

Unfortunately, at least some studies of the relationship of TSH levels to heart disease or blood lipid levels, some very large, did not appear to mention adjustment for subjects taking lipid-lowering medications. Subjects taking them should have been analyzed separately from those not taking them.

As wonderful as the third-generation TSH assay is, and even the free T4 assay, they cannot tell us everything

The TSH assay is a snapshot in time.  It does not tell us whether the patient is bed-ridden or running ten miles a day, to the point of exhaustion.  Nor does it tell us if that patient has another (stressful) disease, or what medications the patient was taking.  It simply tells us the amount of stress that the thyroid was experiencing at the time that the patient's blood was taken.  There is no substitute for taking a medical history!  Granted, an individual's TSH levels vary much more over the day than they do across days at the same time of day.  If a lab is coming back with inconsistent values across days, its quality should be investigated. 

Not all TSH assays are created equal; there are different products on the market, and their reference ranges vary widely.  Although the precision of at least some of these products is excellent, there are obviously accuracy problems with some.  While knowledge about these variations can probably help labs pick the best product, some might choose one with an especially wide reference range in order to minimize the number of diagnoses while still remaining on firm legal ground.  Patients are unlike to be informed about these considerations.

What we know about the mechanism of thyroid function should make us skeptical of claims that TSH levels are simply misleading

The medical establishment justifies its emphasis on TSH levels as the main criterion for hypothyroidism diagnosis by claiming that it is the most trustworthy measurement of the severity of the disease and they are on firm ground in doing so: TSH assays have become extremely precise in recent years (although it appears that some products and labs are better than others), and studies of the effects of exercise intensity and circadian rhythms have shown a close correlation of TSH levels, melatonin levels and feelings of alertness. We also know that TSH levels and goiter growth are strongly related, and that at least some viral infections can drive up TSH production by the intestines as well as causing atherosclerotic changes. Knowledge of these mechanisms should make us wary of conclusions drawn from clinical research that claim that an abnormally high TSH is not really a problem in itself.  Is a large collection of poorly functioning thyroid follicles as desirable as a smaller collection of healthy thyroid follicles?  Is the quality of T4 produced really the same?  Is the timing of its production the same and the route it takes through the body the same? 

Can we really measure the precision of normal range limits?

On the other hand, this knowledge cannot serve as a basis for very precise measurements of the limits of the TSH normal range; in fact, it is not clear how to determine the precision of this measurement at all (at least given that membership in the "healthy patient population" requires subjective judgments). Yet these limits are typically reported using two significant figures, e.g., 4.5 mIU/L for the top of the normal range. In practice, however, a lab test result reported as "4.4" mIU/L (even though greater precision, represented by more significant figures, might be possible) would be probably be interpreted by many physicians to be "normal." How big a fudge factor is necessary in reality? And how should clinical signs and symptoms figure into that? Maybe it's not necessary to find the magic number: if a patient's TSH level is close to the top of the normal range, perhaps it is reasonable to consider treatment, and perhaps rather inhumane to dismiss the possibility out of fear of deviating from perfection.

The elephant in the room: what happened to concern about improving public health?

Recently we came to view health not simply as freedom from disease but as a state of feeling good and free of limitations to fulfilling our full mental and physical potential.  As public health increased, so did the demands of the workplace; excellent health became de rigeur for maintaining a career with the modern corporation. Professional-class employees were expected to put in long hours, handle ever more challenging tasks, and even to travel huge distances as modern medicine made high energy levels ever more possible. The level of health that one could achieve without the help of the medical and (what is best described as) the "health maintenance" industries is no longer adequate for people with certain seemingly minor diseases to get along in a job that pays a living wage. The world is a vastly different place from what it was when only the worst cases of hypothyroidism were treated.

But as the push for affordable socialized medicine gained momentum around 2009, the bar began to be set lower again as good health eluded a precise scientific definition. Our government chose to cut costs by eliminating many of the traditional preventive aspects of healthcare, not just the screening for ordinary diseases and malnutrition, but the treatment of nonfatal but sometimes disabling "subclinical" illnesses that led to major ones if untreated. And a funny thing happened: American longevity started going down (Carroll, 2019). It was blamed on poor economic conditions and the opioid abuse epidemic, but did they really just start in 2016?  What is the whole story behind the effects of the new austerity? How much of this originated in the general lowering of quality standards reflected in so many aspects of American life?  And now that the COVID-19 pandemic has hurt our economy far worse and damaged our healthcare system, what will be the impact?

Basic classifications of thyroid stress conditions

There are three basic conditions that degrade thyroid function: 1) Iodine insufficiency, which is dealt with only via public health measures, 2) Hashimoto's thyroiditis, an auto-immune condition that gradually destroys the thyroid, causing TSH levels to rise, free T4 levels to go down over many years, and anti-thyroid antibody levels to increase, and 3) non-thyroidal diseases, which put enough stress on the thyroid to drive up TSH levels, and, depending on the condition, either drive free T4 down or up.  In non-thyroidal illness syndrome (NTIS), the thyroid, liver, and kidney function are so stressed by an illness outside the thyroid that TSH, free T4, and free T3 levels fall beneath their respective reference ranges; a subnormal free T3 level typically indicates a life-threatening condition.  Will thyroid hormone supplementation help those suffering from some of the conditions in the third group?  Promising research is underway.

Possible solutions for physicians' resistance to treating hypothyroidism sufferers

An easier set of tests better suited to measure a long-term situation

Alternative tests for those who are skeptical of the TSH assay's ability to capture an individual's general state of health exist.  A more slow-moving parameter is the mean corpuscular volume (MCV, high in hypothyroidism); homocysteine levels (normal in hypothyroidism, ruling out folate and vitamin B12 deficiencies), and liver enzyme levels (normal in uncomplicated hypothyroidism, ruling out liver problems, the other main cause of macrocytosis) can be used to eliminate nearly all other causes of high MCV.  It stands to reason that blood cell parameters would change very slowly because only a very small proportion of red blood cells die and are replaced by new ones every day, since the normal lifespan of a red blood cell is 115 days.  If there are other problems hampering red blood cell creation, and hypothyroidism is causing the macrocytosis, it might take even longer for the red blood cells to return to normal size.  Even when the TSH is restored to an ideal value, a still-high MCV might indicate that full recovery is not complete.  By the same token, an untreated individual with an isolated high MCV is likely to have had hypothyroidism for some time.

Does treatment of hypothyroidism reduce MCV?  If so, how long does this take?  This issue calls for further study, and might tell us whether hypothyroidism has a lasting detrimental effect on health, causing problems that last well beyond the time that it takes to restore normal TSH levels.

Increasing awareness of thyroid disease, including the social stigma associated with hypothyroidism, before dispensing with screening

Very little attention is given to increasing patient awareness of the symptoms and signs of thyroid disease.  If screening for thyroid disease is discontinued and patients fail to recognize its manifestations in themselves, they might not think it appropriate to mention them at visits.  The onset of hypothyroidism especially when caused by Hashimoto's thyroiditis, can be very insidious; besides, our stiff-upper-lip-warrior society discourages us from talking about pain, fatigue, and feeling stupid and often gives us the message that these are signs of mental illness or of character defects.  Besides, physicians who no longer do screening might lose their ability to recognize the disease in patients.  Besides, the social stigma associated with many symptoms and signs of this disease calls for physicians to take special care not to cut diagnostic corners.

Greater patient autonomy and responsibility, with concomitant tort reform

Since studies of quality of life and symptoms alike do not seem to be able to capture the hypothyroidism sufferer's subjective experience, it might be best to allow patients to make these judgments while limiting their rights to sue physicians for allowing them to make mistakes.  Official certification of some as "competent patients" might be a first step for skeptics.  Of course, since governments and insurance companies help foot the bill, they might resist taking this step. Unfortunately, this is a reminder of how our "free" society is becoming less free.

Make the process of becoming an endocrinologist shorter and more efficient

Becoming an endocrinologist in the U.S. requires 12 years of training beyond college, yet relatively low pay does not reflect this.  The numbers say it all: we need far more than one endocrinologist for every 57,000+ Americans, especially considering the pressure that diabetes is putting on their services.  

Remember what evidence-based medicine was originally intended to do

Because evidence-based medicine is being used to reform our medical care system with a special emphasis on reducing costs, a rush to judgment is especially dangerous.  We should never abandon a system that applies rigorous publication standards to studies, and remains open to the traditional reliance on replication of studies for validation purposes.  Eliminating a long-standing treatment that has no clear harms if properly administered is really a large-scale experiment in itself and monitoring might not be enough to reverse the damage done if it turns out to be the wrong choice.

I propose these guidelines for all clinical trials used to justify ending such a standard treatment, especially if cost reduction is the major motivation:

1. The sampled subjects should properly represent the population that they are claimed to represent. Subjects' motivation to enter the trial should be a compelling need to try treatment with the study drug and an inability to obtain it in any other way.  In the case of a chronic illness, restricting that sample to those who are currently untreated when many others in that population are being treated does not satisfy this condition.

2. The sample should be large enough not just to achieve sufficient power, but to represent a meaningful proportion of that population: a clinical trial involving a disease that affects a large number of patients should use a large sample.

3. At minimum, every clinical trial (including those used in meta-anlyses) should include change-from-baseline calculations for each subject and dependent variable(s) considered, on which the key statistics should be based.

4. Data should be gathered in ways that make minimal demands on subjects to describe, especially to quantify, their subjective experience, unless of course the primary objective is to study patterns in subjects' expression of that experience.  Observed behavior and lab tests are better measures, and researchers should perform the quantifying based on criteria that is either objective or based on their firsthand examination of patients.  Of course, subjects' feedback about whether they feel that they are benefiting from the study and whether they wish to continue matters; in an important sense, this is behavior rather than a description of their feelings according to unfamiliar parameters because it involves decision-making about how to act.

5. The intervention should, if possible, restore the subject's state of health to what is expected of the reference population.  For example, if the typical TSH value for a subject's age group is 1.6 mIU/L, a clinical trial in which the median resulting value in the treatment group is 3.2 mIU/L does not meet this requirement.

6. Derived statistics such as p-values and hazard ratios should be properly applied and interpreted.  Conclusions should not state that there were no differences between treatment groups unless the calculated p-value is very close to one.  If threshold p-values are used, they should be appropriate to what is measured: for instance, they should be smallest if used in the efficacy portions of clinical trials of a new drug with great risks.  P-values used as criteria for treatment decisions should be stated explicity, rather than simply "NS" if they do not reach the statistical significance standard of that particular study.

Setting small threshold p-values for safety comparisons, i.e., where ideal treatment group values are similar to those of control groups, effectively sets low standards for safety.  In the case of such comparisons, other differences need to be noted: if the means for the two groups differ widely, even if their large variances produce a large p-value, this is worthy of attention.

7. The length of the trial should be appropriate to the endpoints (dependent variables) and the size, age, and general health of the sample. For example, if the endpoint is the number of heart attacks or of onsets of rare diseases, and the sample is small, young and healthy, the length of the trial will need to be much longer than a year or two.

8.  The number of dependent variables, i.e., outcomes, should be limited.  Although the use of interaction variables and Bonferroni corrections can reduce the chance of incorrect conclusions about cause-and-effect relationships, it is better to determine these variables in a previous study (or studies), including long-term (prospective) observational studies and base the trial on a clearly understood and stated mechanism.


Copyright © 2013-2020 by Dorothy E. Pugh.  All rights reserved. 


Abbott Laboratories (2013) Synthroid (levothyroxine sodium) tablet. Retrieved 12 August 2013 from https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=1e11ad30-1041-4520-10b0-8f9d30d30fcc

Abdelhai AR, Salem IM, and Gawish HH (2016) Comorbidity assessment and thyroid hormones in independently living elderly. ZUMJ 22(3):99-107. Retrieved 12 Nov 2020 from https://zumj.journals.ekb.eg/article_4643_696908b78f9a02f9a4d89413846e21a5.pdf

Adams KL, Castanon-Cervantes O, Evans JA, and Davidson AJ (2013) Environmental circadian disruption elevates the IL-6 response to lipopolysaccharide in blood. J Biol Rhythms 28(4):272-7. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/23929554 (abstract only)

Adhvaryu A, Bednar S, Molina T, Nguyen Q, and Nyshadham A (2013) Salt iodization and enfranchisement of the American worker. Retrieved 5 Oct 2013 from https://www.elon.edu/docs/e-web/academics/business/economics/faculty/bednar/MortonSalt_body_aug2013.pdf

Adler, S. and L. Wartofsky (2007) The nonthyroidal illness syndrome. Endocrinol Metab Clin N Am 36:657-672 https://www.temple.edu/imreports/Reading/Endo%20-%20Non-thyroidal%20illness.pdf

Akinci B, Comlekai A, Yener S, Demir T, Bayraktar F, Yuksel F, Yesil S (2007) The alternation of serum soluble CD40 ligand levels in overt and subclinical hypothyroidism. Hormones 6(4):327-33. Retrieved 7 May 2020 from https://www.hormones.gr/199/article/article.html

Alevizaki M, Saltiki K, Voidonikola P, Mantzou E, Papamichael C, Stamatelopoulos K (2009) Free thyroxine is an independent predictor of subcutaneous fat in euthyroid individuals. Eur J Endocrinol 161(3):459-65. Retrieved 5 Oct 2020 from https://eje.bioscientifica.com/view/journals/eje/161/3/459.xml

Alexander, EK, E Marqusee, J Lawrence, P Jarolim, GA Fischer, and PR Larsen (2004) Timing and magnitude of increases in levothyroxine requirements during pregnancy in women with hypothyroidism. N Engl J Med 351:241-9 https://www.nejm.org/doi/full/10.1056/NEJMoa040079#t=article

Alexander EK et al. (2017) 2017 guidelines of the American Thyroid Association for the diagnosis and management of thyroid disease during pregnancy and the postpartum. Thyroid 27(3):315-389. Retrieved 17 Jan 2020 from https://www.liebertpub.com/doi/full/10.1089/thy.2016.0457

Alkemade A, Unmehopa UA, Wiersinga WM, Swaab DF, and Fliers E (2005) Glucocorticoids decrease thyrotropin-releasing hormone messenger ribonucleic acid expression in the paraventricular nucleus of the human hypothalamus. J Clin Endocrinol Metab 90(1):323-7 (abstract only) Retrieved 24 Mar 2014 from https://www.ncbi.nlm.nih.gov/pubmed/15509645

Allan, W.C., J.E. Haddow, G.E. Palomski, J.R. Williams, M.L. Mitchell, R.J. Hermos, J.D. Faix and R.Z. Klein (2000) Maternal thyroid deficiency and pregnancy complications: implications for population screening.  J Med Screen 2000;7:127-130 https://jms.rsmjournals.com/content/7/3/127.long

American Association of Clinical Endocrinologists (2017) Underdiagnosed autoimmune disease the focal point of national campaign. Retrieved 11 Feb 2017 from https://media.aace.com/press-release/underdiagnosed-autoimmune-disease-focal-point-national-campaign

American Medical Association (2008) Time to pinch off the salt. Retrieved 26 Jan 2013 from https://www.ama-assn.org/amednews/2008/01/14/edsa0114.htm

Amino, N, H Tada, Y Hidaka, and K Hashimoto (2002) Hashimoto's Disease and Dr. Hakaru Hashimoto. Endocr J 49(4):393-7. Retrieved 26 Jan 2013 from https://www.jstage.jst.go.jp/article/endocrj/49/4/49_4_393/_pdf

Amouzegar A, Mehran L, Sarvghadi F, Delshad H, Azizi F, Lazarus JH (2014) Comparison of the American Thyroid Association with the Endocrine Society practice guidelines for the screening and treatment of hypothyroidism during pregnancy. Hormones 13(3):307-13. Retrieved 10 Jun 2020 from https://link.springer.com/content/pdf/10.14310/horm.2002.1486.pdf

Andersson, M, B de Benoist, I Darnton-Hill, and F Delange (2007) WHO:Iodine deficiency in Europe: A continuing public health problem. Retrieved 20 Jan 2013 from https://www.who.int/nutrition/publications/VMNIS_Iodine_deficiency_in_Europe.pdf

Angelousi, A., Karageorgopoulos, D., Kapaskelis, A., & Falagas, M. (2011). Association between thyroid function tests at baseline and the outcome of patients with sepsis or septic shock: a systematic review, European Journal of Endocrinology164(2), 147-155. Retrieved Mar 7, 2021, from https://eje.bioscientifica.com/view/journals/eje/164/2/147.xml

Aniello F, Couchie D, Bridoux A-M, Gripois D (1991) Splicing of juvenile and adult tau mRNA variants is regulated by thyroid hormone. Proc Natl Acad Sci 88:4035-9. Retrieved 15 Aug 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC51588/pdf/pnas01059-0535.pdf

Anonymous (editorial) Trouble at the lab. The Economist. Retrieved 22 Oct 2013 from https://www.economist.com/news/briefing/21588057-scientists-think-science-self-correcting-alarming-degree-it-not-trouble/print

Antonica, F, DF Kasprzyk, R Opitz, M Iacovino, X-H Liao, AM Dumitrescu, S Refetoff, K Peremans, M Manto, M Kyba, & S Costagliola (2013) Generation of functional thyroid from embryonic stem cells.  Retrieved 27 Jan 2013 from https://www.nature.com/nature/journal/v491/n7422/pdf/nature11525.pdf

Apostolou E, (2009). Persistence of an atherogenic profile after treatment of acute infection with brucella. J Lipid Res 50(12):2532-9. Retrieved 30 May 2014 from https://www.jlr.org/content/50/12/2532.full.pdf+html

Arafah, BM (2001) Increased need for thyroxine in women with hypothyroidism during estrogen therapy.  N Engl J Med 344:1743-49. Retrieved 17 Jan 2013 from https://www.nejm.org/doi/full/10.1056/NEJM200106073442302#t=article

Arem, R. (1999) The thyroid solution. NY:Ballantine

Armbruster, D. A., & Pry, T. (2008). Limit of blank, limit of detection and limit of quantitation. The Clinical biochemist. Reviews29 Suppl 1(Suppl 1), S49–S52. Retrieved 12 Dec 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2556583/

Asvold BO, Bjoro T, Nilsen TI, Grinnell D, Vatten LJ (2008) Thyrotropin levels and risk of fatal coronary heart disease: the HUNT study. Arch Intern Med 168:855-60. Retrieved 6 Jan 2020 from https://www.ncbi.nlm.nih.gov/pubmed/18443261 (abstract only)

Avadhanula S, Introne WJ, Singyoung A, Soldin SJ, Stolze B, Regier D, Ciccone C, Hannah-Shmouni F, Filie AC, Burman KD, Klubo-Gwiezdzinska (2020) Assessment of thyroid function in patients with alkaptonuria. Diabetes and Endocrinol 3(3):e201357. Retrieved 11 Apr 2020 from https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2763227

Aziz M, Kandimalla Y, Machaparapu A, Saxena A, Das S, Younus A, Nguyen M, Malik R, Anugula D, Adus A, Rasool A, Veledar E, and Nasir K (2017) Effect of thyroxin treatment on carotid intima-media thickness (CIMT) reduction in patients with subclinical hypothyroidism (SCH): a meta-analysis of clinical trials. J Ahteroscler Thromb 24(7):643-59. Retrieved 10 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5517537/pdf/jat-24-643.pdf

Bachorik P (2018) National Health & Nutrition Examination Survey (NHANES). Retrieved 12 Nov 2019 from https://www.niehs.nih.gov/research/clinical/studies/nhanes/index.cfm

Ball SG, Sokolov J, and Chin WW (1997) 3,5-diiodo-L-thyronine (T2) has selective thyromimetic effects in vivo and in vitro. J Mol Endocrinol 19(2):137-47. Retrieved 10 Apr 2014 from https://jme.endocrinology-journals.org/content/19/2/137.full.pdf

Balthazar U and Steiner AZ (2012) Periconceptional changes in thyroid function: a longitudinal study. Reprod Biol Endocrinol 2012 10:20 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3337801/

Barnes BO and Galton L (1976) Hypothyroidism: The unsuspected illness.  NY:Harper & Row

Bassett JHD amd Williams GR (2016) Role of thyroid hormones in skeletal development and bone maintenance. Endocr Rev 37(2):135-87. Retrieved 27 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4823381/

Basu G and Mohapatra A (2012) Interactions between thyroid disorders and kidney disease.  Indian J Endocrinol Metab 16(2):204-13.  Retrieved 10 Feb 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3313737/?report=printable

Baur, A, K Bauer, H Jarry, and J Köhrle (2000) Effects of proinflammatory cytokines on anterior pituitary 5'-deiodinase type I and type II. J Endocrinol 167:505-15. Retrieved 6 April 2013 from https://joe.endocrinology-journals.org/content/167/3/505.full.pdf

Benvenga S, Bartolone L, Pappalardo MA, Russo A, Lapa D, Giorgianni G, Saraceno G, Trimarchi F (2008) Altered intestinal absorption of L-thyroxine caused by coffee. Thyroid 18(3):293-301. Retrieved 27 Dec 2013 from https://www.ncbi.nlm.nih.gov/pubmed/18341376?dopt=Citation (abstract only)

Bekkering GE, Agoritsas T, Lytvyn L, Heen AF, Feller M, Moutzouri E, Abdulazeem, Aertgeerts B, Beecer D, Brito JP, Farhoumand PD, Singh Ospina N, Rodondi N, van Driel M, Wallace E, Snel M, Okwen PM, Siemieniuk R, Vandvik PO, Kuijpers T, Vermandere (2019) Thyroid hormones treatment for subclinical hypothyroidism: a clinical practice guideline. BMJ 365:I2006. Retrieved 25 May 2019 from https://www.bmj.com/content/365/bmj.l2006

Besses GS, Burrow GN, Spaulding SW, Donabedian RK, and Pechinski T (1975) Dopamine infusion acutely inhibits the TSH and prolactin response to TRH. J Clin Endocrinol 10(1):7-15. Retrieved 7 Jul 2013 from https://jcem.endojournals.org/content/41/5/985.abstract

BestPractice (2011). Hypothermia.  https://bestpractice.bmj.com/best-practice/monograph/654/basics/pathophysiology.html

Billewicz WZ, Chapman RS, Crooks J, Day ME, Gossage J, Wayne E, Young JA (1969) Statistical methods applied to the diagnosis of hypothyroidism. Q J Med. 38(150):255-66. PubMed PMID: 4181088. (Article is inaccessible; information from out-of-print journal was apparently not preserved.)

Bianco AC, Anderson G, Forrest D et al. (2014) American Thyroid Association guide to investigating thyroid hormone economy and action in rodent and cell models. Thyroid 24(1):88-168. Retrieved 3 Sep 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3887458/

Biondi B and Cooper D (2008) The clinical significance of subclinical hypothyroidism. Endocrine Reviews 29(1):76-131.  Retrieved 11 Dec 2019 from https://academic.oup.com/edrv/article/29/1/76/2354999

Biondi B (2012) Mechanisms in endocrinology: heart failure and thyroid dysfunction. Eur J Endocrinol 167(5):609-18. Retrieved 21 Nov 2020 from https://eje.bioscientifica.com/view/journals/eje/167/5/609.xml

Biondi B (2013) The normal TSH reference range: what has changed in the last decade? J Clin Endocrinol Metab 98(9):3584-87.  Retrieved 6 Nov 2019 from https://academic.oup.com/jcem/article/98/9/3584/2833082

Biondi B, Cappola AR, Cooper DS (2019) Subclinical hypothyroidism: a review.  JAMA 322(2):153-160.  Retrieved 8 July 2019 from https://www.ncbi.nlm.nih.gov/pubmed/31287527 (abstract only, but I had access to the hard-copy article vai subscription)

Birtwhistle R, Morissette K, Dickinson JA, Reynolds DL, Avey MT, Domingo FR, Rodin R, and Thombs RD (2019) Recommendation on screening adults for asymptomatic thyroid dysfunction in primary care. CMAJ 191(46):E1274-80.  Retrieved 24 Nov 2019 from https://www.cmaj.ca/content/191/46/E1274

Bleichrodt N and Born MP (1994) A metaanalysis of research on iodine and its relationship to cognitive development. In: Stanbury JB, ed. The damaged brain of iodine deficiency: cognitive, behavioral, neuromotor, educative aspects. NY:Cognizant Communication. Pp. 195-200. Retrieved 1 Aug 2013 from https://www.ceecis.org/iodine/04a_consequences/02_int/chapt_on_i_brain.pdf

Blount, BC, JL Pirkle, JD Osterloh, L Valentin-Blasini, and KL Caldwell (2006) Urinary perchlorate and thyroid hormones levels in adolescent and adult men and women living in the United States. Environ Health Perspect 114(12):1865-71. Retrieved 19 May 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1764147/pdf/ehp0114-001865.pdf

Bobek S, Kahl S, Ewy Z (1976) Effect of long-term fluoride administration on thyroid hormones level blood in rats. Endocrinol Exp 10(4):289-95. Retrieved 21 Aug 2013 from https://www.ncbi.nlm.nih.gov/pubmed/1087230

Bolborea, M and N Dale (2013) Hypothalamic tanycytes: potential roles in the control of feeding and energy balance. Cell 36(2):91-100.  Retrieved 23 April 2013 from  https://download.cell.com/trends/neurosciences/pdf/PIIS0166223612002238.pdf?intermediate=true

Bonner, WP (1967) Iodination of public water supplies. U of Florida. Retrieved 29 Sep 2013 from https://ia600404.us.archive.org/6/items/iodinationofpubl00bonn/iodinationofpubl00bonn.pdf (Ph.D. dissertation)

Borowiec AS, Hague F, Gouilleux-Gruart V, Lassoued K, Ouadid-Ahidouch H (2011) Regulation of IGF-1-dependent cyclin D1 and E expression by hEag1 channels in MCF-7 cells: the critical role of hEag1 channels in G1 phase progression. Biochim Biophys Acta 1813(5):723-30. Retrieved 23 May 2014 from https://www.ncbi.nlm.nih.gov/pubmed/21315112

Borzio, M, R Caldara, F Borzio, V Piepoli, P Rampini, & C Ferrari (1983) Thyroid tests in chronic liver disease; evidence for multiple abnormalities despite clinical euthyroidism. Gut 24:631-6.  Retrieved 6 Jan 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1420033/pdf/gut00404-0043.pdf

Bowthorpe, J.A. https://www.stopthethyroidmadness.com/

Brabant G, Beck-Peccoz P, Jarzab B, Laurberg P, Orgiazzi J, Szabolcs I, Weetman AP and Wiersinga WM (2006) Is there a need to redefine the upper normal limit of TSH? Eur J Endocrinol 154:633-7. Retrieved 25 Jul 2013 from  https://eje-online.org/content/154/5/633.long

Brenner-Gati L, Berg KA, and Gershengorn MC (1988) Thyroid-stimulating hormone and insulin-like growth factor-1 synergize to elevate 1,2 diacylglycerol in rat thyroid cells: stimulation of DNA synthesis via interaction between lipid and adenylyl cyclase signal transduction systems. J Clin Invest 82:1144-1148. Retrieved 24 Nov 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC303630/pdf/jcinvest00081-0404.pdf

Buxton, OM (2000) Daytime naps shift the human circadian rhythms of melatonin and thyrotropin secretion. American Journal of Physiology 287(2):R373-R382.  Retrieved 29 Dec 2012 from   https://ajpregu.physiology.org/content/278/2/R373.full

Cai H, Wang Z, Zhang H-q, Wang F-r, Yu C-x, Zhang F-x, Gao L, Zhang J, and Zhao J-j (2014) Diosgenin reliaves goiter via the inhibiton of thyrocyte proliferation in a mouse model of Graves' disease. Acta Pharmacologica Sinica 35:65-73. Retrieved 25 Apr 2014 from https://www.nature.com/aps/journal/v35/n1/pdf/aps2013133a.pdf

Caldwell, KL, Makhmudov A, Ely E, Jones RL and Wang RY (2011) Iodine status of the U.S. population, National Health and Nutrition Examination Survey, 2005-06 and 2007-2008. Thyroid 21(4):419-27. Retrieved 7 May 2013 from https://www.ncbi.nlm.nih.gov/pubmed/21323596 (Abstract only)

Calsolaro V, Niccolai F, Pasqualetti G, Tognini S, Magno S, Riccione T, Bottari M, Caraccio N, and Monzani F (2019) Hypothyroidism in the elderly: who should be treated and how? J Endocr Soc 3(1):146-58. Retrieved 15 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6309133/

Canaris GJ, Manowitz NR, Mayor G, Ridgway EC (2000) The Colorado thyroid disease prevalence study. Arch Intern Med 160(4):526-34. Retrieved 4 Oct 2019 from https://www.ncbi.nlm.nih.gov/pubmed/10695693 (abstract only)

Cappola A (2013, updated in 2017) Subclinical hypothyroidism. Retrieved 11 Dec 2019 from https://www.endocrinologyadvisor.com/home/decision-support-in-medicine/endocrinology-metabolism/subclinical-hypothyroidism/

Caraccio, N, N Ferrannini and F Monzani (2002) Lipoprotein profile in subclinical hypothyroidism: response to levothyroxine replacement, a randomized placebo-controlled study. J Clin Endocrinol Metab 87(4):1533-8. Retrieved 2 Feb 2013 from https://jcem.endojournals.org/content/87/4/1533.full.pdf+html

Carpenter, KJ (2005) David Marine and the problem of goiter. J Nutr 135:675-80.  Retrieved 31 Jan 2013 from https://jn.nutrition.org/content/135/4/675.full.pdf+html

Carroll L (2019) U.S. life expectancy declining due to more deaths in middle age. Retrieved 17 Dec 2019 from https://www.reuters.com/article/us-health-life-expectancy/u-s-life-expectancy-declining-due-to-more-deaths-in-middle-age-idUSKBN1Y02C7

Casey, BM, JS Dashe, E Wells, DD McIntire, W Byrd, KJ Leveno, and FG Cunningham (2005) Subclinical hypothyroidism and pregnancy outcomes. Obstetrics and Gynecology 105(2):239-45. Retrieved 18 Jan 2013 from https://www.rima.org/web/medline_pdf/obstetgynecol_239.pdf

Center for Science in the Public Interest (1999) Consumer group calls for ban on "flour improver:" potassium bromate termed a "cancer threat." Retrieved 19 Aug 2013 from  https://www.cspinet.org/new/bromate.html

Center of Science in the Public Interest (2005) Re: Docket Number 1998N-0359, CFSAN Program Priorities for FY 2006, 70 Fed. Reg. 29328-29329 (May 20, 2005). Retrieved 19 Aug 2013 from https://www.fda.gov/ohrms/dockets/dockets/98n0359/98n-0359-c000120-01-vol9.pdf

Centers for Disease Control and Prevention (2012) Cigarette smoking the United States. Retrieved 19 May 2014 from https://www.cdc.gov/tobacco/campaign/tips/resources/data/cigarette-smoking-in-united-states.html

Centers for Disease Control and Prevention (2020) Stages of HIV infection. Retrieved 10 May 2020 from https://wwwn.cdc.gov/hivrisk/what_is/stages_hiv_infection.html

Ceresini G et al. (2009) Thyroid function abnormalities and cognitive impairement in the elderly. Results of the InCHIANTI study. J Am Geriatr Soc 57(1):89-93. Retrieved 15 Jun 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2631617/

Chandra KC, Kant R, Ekka BR, and Kumar A (2016) Comparative study of lipid profile, renal and muscle damage in subclinical and overt hypothyroidism. Intl J of Contemporary Medical Research 3(10):77-83.  Retrieved 24 Nov 2019 from https://www.ijcmr.com/uploads/7/7/4/6/77464738/ijcmr_984_oct_16.pdf

Chanoine J-P, Nève J, Wu S, Vanderpas J, and Bourdoux P (2001) Selenium decreases thyroglobulin concentrations but does not affect the increased thyroxine-to-triiodothyronine ratio in children with congenital hypothyroidism. J Clin Endocinol Metab 86(3):1160-3. Retrieved 5 Dec 2013 from https://jcem.endojournals.org/content/86/3/1160.full.pdf+html

Chatzitomaris A, Hoermann R, Midgely JE, Hering S, Urban A, Dietrich B, Abood A, Klein HH, Dietrich JW (2017) Thyroid allostasis -- adaptive responses of thyrotropic feedback control to conditions of strain, stress, and developmental programming. Front Endocrinol 8:163. Retrieved 16 Feb 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5517413/

Chemburkar, SR, KC Deming, RE Reddy (2010) Chemistry of thyroxine: an historical perspective and recent progress. Tetrahedron 66:1955-62. Retrieved 25 Jan 2013 from  https://www.sciencedirect.com/science/article/pii/S0040402009019449

Chen AX, Leung, AM, and Korevaar, TIM (2019) Thyroid function and conception. N Engl J Med 381:178-181. Retrieved from https://www.nejm.org/doi/full/10.1056/NEJMclde1902637 (short introduction only; I read the hard-copy version of this publication.)

Chen M, Zhou W, and Xu W (2020) Thyroid function analysis in 50 patients with COVID-19: a retrospective study. Thyroid. ahead of print. Retrieved 18 Jul 2020 from https://www.liebertpub.com/doi/full/10.1089/thy.2020.0363

Chen, I  (2012) Pregnancy and the thyroid. https://www.nytimes.com/ref/health/healthguide/esn-hypothyroidism-expert.html?pagewanted=all

Chen, L, H Chen, M Shen, Z Zhou, A Ma (2010) Analysis of perchlorate in milk powder and milk by hydrophilic interaction chromatography combined with tandem mass spectrometry. Retrieved 29 Jan 2013 from https://www.ncbi.nlm.nih.gov/pubmed/20170169

Chiappa V, Anderson MA, Barrett CD, Stathatos N, Anahtar MN (2019) Case 24-2019: A 39-year-old woman with palpitations, abdominal pain, and vomiting. N Engl J Med 381(6):567-77.  Retrieved 7 Aug 2019 from https://www.nejm.org/doi/full/10.1056/NEJMcpc1900142 (I had access to hard-copy article, no abstract was available)

Chiellini G, Apriletti JW, Yoshihara JW, Baxter JD, Ribeiro RC, and Scanlan TS (1998) A high-affinity subtype-selective agonist ligand for the thyroid hormone receptor. Chem Biol 5(6):299-306. Retrieved 28 Nov 2019 from https://www.ncbi.nlm.nih.gov/pubmed/9653548 (abstract with link to full publication)

Chihara K, Kashio Y, Kita T, Okimura Y, Kaji H, Abe H, and Fujita T (1986) L-dopa stimulates release of hypothalamic growth hormone-releasing hormone in humans. J Clin Endocrinol Metab 62(3):466-73. Retrieved 4 Dec 2013 from https://www.ncbi.nlm.nih.gov/pubmed/3080462

Children's Hospital of Philadelphia (2020) Thyroid disorders and pregnancy. Retrieved 6 Nov 2020 from https://www.chop.edu/pages/thyroid-disorders-and-pregnancy

Chow CC, Phillips DI, Lazarus JH, and Parkes AB (1991) Effect of low dose iodide supplementation on thyroid function in potentially susceptible subjects: are dietary levels in Britain acceptable? Clin Endocrinol 34(5):413-6. Retrieved 7 Dec 2013 from https://www.ncbi.nlm.nih.gov/pubmed/2060151 (abstract only)

Chung GE, Kim D, Kim W, Yim JY, Park MJ, Kim YJ, Yoon JH, Lee HS (2012) Non-alcoholic fatty liver disease across the spectrum of hypothyroidism. J Hepatol 57(1):150-6. Retrieved 22 Feb 2018 from https://www.ncbi.nlm.nih.gov/pubmed/22425701 (abstract only)

Chwalisz BK, Buchbinder BR, Schmahmann JD, Samore WR (2019) Case 32-2019: A 70-year-old woman with rapidly progressive ataxia.  N Engl J Med 38(16):1569-78.  Retrieved 11 Nov 2019 from https://www.nejm.org/doi/full/10.1056/NEJMcpc1909624 (abstract only, had access to the hard copy)

Cleare AJ, McGregor A, and O'Keane V (1995) Neuroendocrine evidence for an association between hypothyroidism, reduced central 5-HT activity and depression. Clin Endocrinol 43(6):713-9. Retrieved 8 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/8736274 (abstract only)

Cleveland Clinic https://www.clevelandclinicmeded.com/medicalpubs/diseasemanagement/endocrinology/hypothyroidism-and-hyperthyroidism/

ClinLab Navigator (2013) Thyroid function tests. Retrieved 10 Jul 2013 from https://www.clinlabnavigator.com/thyroid-function-tests.html

Colett T-H et al. (2012) Subclinical hyperthyroidism and the risk of coronary heart disease and mortality. Arch Int Med 172(10):799-809. Retrieved 16 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3872478/

Columbano A, Chiellini G, and Kowalik MA (2017) GC-1: A thyromimetic with multiple therapeutic applications in liver disease. Gene Expr 17(4):265-75. Retrieved 28 Nov 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5885148/

Comtois, R, J Hébert and JP Soucy (1992) Increase in T3 levels in hypocorticism in patients with chronic adrenocortical insufficiency. Acta Endocrinol 126(4):319-24. Retrieved 17 Jan 2013 from https://www.ncbi.nlm.nih.gov/pubmed/1317631

Crocker, AD, Overstreet DH, and Crocker JM (1986) Hypothyroidism leads to increased dopamine receptor sensitivity and concentration. Pharmacol Biochem Behav 24(6):1593-7. Retrieved 3 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/3090565 (abstract only)

Danese MD, Powe NR, Sawin CT, and Ladenson PW (1996) Screening for mild thyroid failure at the periodic health examination: a decision and cost-effectiveness analysis. JAMA 276(4):285-92. Retrieved 28 Aug 2014 from https://www.ncbi.nlm.nih.gov/pubmed/8656540 (abstract only)

Darnerud PO (2003) Toxic effects of brominated flame retardants in man and in wildlife. Environ Int 6:841-53. Retrieved 4 Oct 2013 from https://www.ncbi.nlm.nih.gov/pubmed/12850100

Dasgupta, PK, Y Liu and JV Dyke (2008) Iodine nutrition: Iodine content of iodized salt in the United States. Environ Sci Technol 42(4):1315-1323. Retrieved 20 Jan 2013 from https://pubs.acs.org/doi/full/10.1021/es0719071

Davies L and Welch G (2014) Current thyroid cancer trends in the United States. JAMA Otolaryngol Head Neck Surg. 2014;140(4):317-322. Retrieved 21 Apr 2014 from https://archotol.jamanetwork.com/article.aspx?articleid=1833060#Methods (abstract only)

de Benoist B, Andersson M, Egli I, Takkouche B, and Allen H (2004) Iodine status worldwide: WHO global database on iodine deficiency. Retrieved 8 Aug 2013 from https://whqlibdoc.who.int/publications/2004/9241592001.pdf 

Demers, L.M. and C.A. Spencer (2002) Laboratory medicine practice guidelines: laboratory support for the diagnosis and monitoring of thyroid disease. National Academy of Clinical Biochemistry.  Retrieved 25 Dec 2012 from   https://www.aacc.org/sitecollectiondocuments/nacb/lmpg/thyroid/thyroid-fullversion.pdf (archived) 

Denoche HC, Glavas MM, Tudurí E, Karunakaran S, Quong WL, Philippe M, Britton HM, Clee SM, and Kieffer TJ (2016) Disrupted leptin signaling in the lateral hypothalamus and ventral premammillary nucleus alters insulin and glucagon secretion and protects against diet-induced obesity.  Retrieved 1 March 2020 from https://academic.oup.com/endo/article/157/7/2671/2422718

deZegher F, Van den Bershe G, Dumoulin M, Gewillig M, and Devlieger H (1995) Dopamine suppresses thyroid-stimulating hormone secretion in neonatal hypothyroidism. Acta Paediatr 84(2):213-4. Retrieved 11 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/7756813 (abstract only)

Diarra A, Lefauconnier JM, Valens M, Georges P and Gripois D (1989) Tyrosine content, influx and accumulation rate, and catecholamine biosynthesis measure in vivo, in the central nervous system and in peripheral organs of the young rat.  Influence of neonatal hypo- and hyperthyroidism. Arch Into Physiol Biochem 97(5):317-32. Retrieved 3 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/2480086 (abstract only)

Dickens LT, Cifu AS, and Cohen RN (2019) Diagnosis and management of thyroid disease during pregnancy and the postpartum period. JAMA 321(19):1928-9. Retrieved 21 Jun 2019 from https://jamanetwork.com/journals/jama/article-abstract/2733078 (abstract only, though I had access to the hard-copy publication)

Dietrich J, Landgrafe G, and Fotiadou EH (2012) TSH and thyrotropic agonists: key actors and thyroid homeostasis. J Thyroid Res 35184. Retrieved 19 Feb 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3544290/

Dietrich JW, Hoermann R, Midgley JEM, Bergen F, and Müller P (2020) The two faces of Janus: why thyrotropin as a cardiovascular risk factor may be an ambiguous target. Front Endocrinol 11:542710. Retrieved 7 Nov 2020 from https://www.frontiersin.org/articles/10.3389/fendo.2020.542710/full

Dimitriadis G, Mitrou P, Lambadiari V, Maratou E, and Raptis SA (2011) Insulin effects in muscle and adipose tissue. Diabetes Res Clin Pract S52-9. Retrieved 17 May 2018 from https://www.ncbi.nlm.nih.gov/pubmed/21864752 (abstract only)

Dohán O, de la Vieja A, Paroder V, Riedel C, Artani M, Reed M, Ginter CS, and Carrasco N (2003) The sodium/iodide symporter (NIS): characterization, regulation, and medical significance. Endocr Rev 24(1):48-77. Retrieved 31 Jul 2013 from https://edrv.endojournals.org/content/24/1/48.full.pdf+html

Drugs.com [Internet] Synthroid dosage. Updated 4 Nov 2019. Retrieved 6 Jan 2020 from https://www.drugs.com/dosage/synthroid.html

Dumont, J., R. Opitz, D Christophe, G Vassart, PP Roger, C Maenhaut (2011) Ontogeny, anatomy, metabolism and physiology of the thyroid. Retrieved 27 March 2013 from https://www.thyroidmanager.org/chapter/ontogeny-anatomy-metabolism-and-physiology-of-the-thyroid/

Ealey PA, Emmerson JM, Spidey SP, Marshall NJ (1985) Thyrotropin stimulation of mitogenesis of the rat thyroid cell strain FRTL -5: a metaphase index assay for the detection of thyroid growth stimulators.  J Endocrinol 106(2):203-10. Retrieved 18 Feb 2013 from https://www.ncbi.nlm.nih.gov/pubmed/4040548

Eastern North Carolina Digital Library (1986) John P. East late a senator from North Carolina memorial addresses delivered in Congress.  https://digital.lib.ecu.edu/historyfiction/fullview.aspx?id=msj

Eisenberg, M, F Santini, A Marsili, A Pinchera, and J DiStefano (2010) TSH regulation in central and extreme primary hypothyroidism. Thyroid 20:2015-1228. Retrieved 10 April 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2974848/pdf/thy.2009.0349.pdf

Eisenberg, M, M Samuels, and J Di Stefano (2008) Extensions, validation, and clinical applications of a feedback control system simulator of the hypothalamo-pituitary-thyroid axis. Thyroid 18:1071-85. Retrieved 12 April 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2962855/

Eldarawy MM, Rashad NM, Hammam AA, Wadea FM, Ibrahim TMH, Alkolaly AK and El-Shal AS (2016) Prevalence of subclinical hypothyroidism in elderly patients with cardiovascular disease. Int J Adv Res 4(12):1954-61. Retrieved 9 July 2020 from https://www.journalijar.com/uploads/259_IJAR-14407.pdf

Endocrine Society (2014) Endocrine clinical workforce: supply and demand projections.  Retrieved 20 Sep 2019 from https://www.endocrine.org/-/media/endosociety/files/advocacy-and-outreach/other-documents/2014-06-white-paper--endocrinology-workforce.pdf?la=en

Enriori PJ, Chen W, et al. (2016) α-melanocyte stimulating hormone promotes muscle glucose uptake via melanocortin 5 receptors. Mol Metab 5(10):807-22. Retrieved 4 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5034615/

Eshraghian A and Jahromi AH (2014) Non-alcoholic fatty liver disease and thyroid dysfunction: a systematic review. World J Gastroenterol 20(25):8102-9. Retrieved 20 Feb 2018 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4081681/pdf/WJG-20-8102.pdf

Eskin BA, Grotkowski CE, Connolly CP, and Ghent WR (1995) Different tissue responses for iodine and iodide in rat thyroid and mammary glands. Biol Trace Elem Res 49(1):9-19. Retrieved 14 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/7577324Feyrer J, Politi D, and Weil D (2013) The cognitive effects of miconutrient deficiency: evidence from salt iodization in the United States. Washington DC: National Bureau of Economic Research. Retrieved 30 Jul 2013 from https://www.nber.org/papers/w19233

Farag K, Arafat K, and Rashem M (2015) Study of thyroid functions in Egyptian patients with non-alcoholic fatty liver disease. Al-Azhar Assiut Medical Journal 13(2):182-93. Retrieved 8 Mar 2020 from https://www.aamj.eg.net/journals/pdf/2680.pdf

Feller M, Snel M, Moutzouri E, Bauer DC, de Montmollin M, Aujesky D, Ford I, Gusselkloo, Kearney PM, Mooljaart S, Quinn T, Stott D, Westerndorp R, Rodondi N, and Dekkers OM (2018) Association of thyroid hormone therapy with quality of life and thyroid-related symptoms in patients with subclinical hypothyroidism: a systematic review and meta-analysis. JAMA 320(13):1349-1359.  Retrieved 14 Oct 2018 from https://jamanetwork.com/journals/jama/article-abstract/2705188 (abstract only, but I had access to the full article)

Feller M, Rodondi N, and Dekkers OM (2019) Response to letter: Thyroid hormone therapy for subclinical hypothyroidism. JAMA 321(8): 804-5.

Ferrandino G, Kaspari RR, Spadaro O, Reyna-Neyra A, Perry RJ, Cardone R, Kibbey RG, Shulman GI, Dixit VD, and Carrasco N (2017) Pathogenesis of hypothyroidism-induced NAFLD is driven by intra- and extrahepatic mechanisms. Proc Natl Acad Sci U.S.A. 2017 Oct 24;  114(43):E9172-80.  Retrieved 9 May 2018 from https://www.pnas.org/content/114/43/E9172

Finkelstein A (2020) A strategy for improving U.S. health care delivery -- conducting more randomized, controlled trials. N Engl J Med 382(16):1485-90.  Retrieved 18 Apr 2020 from https://www.nejm.org/doi/full/10.1056/NEJMp1915762 (introductory text only, but had access to the full article via subscription)

Finkielstein CV, Chen LG, and Maller JL (2002) A role for G1/S cyclin-dependent protein kinases in the apoptotic response to ionizing radiation. J Biol Chem 277(41):38476-85. Retrieved 23 May 2014 from https://www.jbc.org/content/277/41/38476.long

Floyd JL (2013) Thyrotoxicosis imaging. Retrieved 1 Oct 2013 from https://emedicine.medscape.com/article/383062-overview

Food and Nutrition Board, Institute of Medicine (2001) Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc. Washington, DC: National Academy Press: 258-89.

Franco, M., Chávez, E., & Pérez-Méndez, O. (2011). Pleiotropic effects of thyroid hormones: learning from hypothyroidism. Journal of thyroid research2011, 321030. https://doi.org/10.4061/2011/321030. Retrieved 14 Jan 2021 from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3134217/

Fukusumi M, Iidaka T, Mouri A, Hamamoto Y, and Kamimura M (2014) Respiratory failure associated with hypoventilation in a patient with severe hypothyroidism. Respirol Case Rep 2(2):79-80. Accessed 25 Nov 2017 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4184513

Funahashi H, Imai T, Tanaka Y, Tobinaga J, Wada M, Morita T, Yamada F, Tsukamura K, Oiwa M, Kikumori T, Narita T, and Takagi H (1996) Suppressive effect of Iodine on DMBA-induced breast tumor growth in the rat. J Surg Oncol 61(3):209-13. Retrieved 16 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/8637209

Furlanetto, TW, LQ Nguyen, and JL Jameson (1999)  Estradiol increases proliferation and down-regulates the sodium/iodide symporter gene in FRTL-5 cells. Endocrinol 140(12):5705-11. Retrieved 20 May 2013 from https://endo.endojournals.org/content/140/12/5705.full.pdf+html

Gabriela V, Arciuch A and Di Cristofano A (2012) Estrogen signaling and thyrocyte proliferation, thyroid and parathyroid disease - new insights into some old and some new issues, Dr. Laura Ward (ed.), ISBN: 978-953-51-0221-2, In Tech, available from https://www.intechopen.com/books/thyroid-and-parathyroid-diseases-new-insights-into-some-old-and-some-new-issues/estrogen-signaling-and-thyrocyte-proliferation. Retrieved 14 Feb 2014 from https://www.intechopen.com/download/get/type/pdfs/id/31307

Gaitonde DY, Rowley KD, and Sweeney LB (2012) Hypothyroidism: an update. Am Fam Physician 86(3):244-51. Retrieved 26 Mar 2020 from https://www.aafp.org/afp/2012/0801/p244.html

Ganguli I, Lupo C, Mainor AJ, et al. Assessment of Prevalence and Cost of Care Cascades After Routine Testing During the Medicare Annual Wellness Visit. JAMA Netw Open. 2020;3(12):e2029891. doi:10.1001/jamanetworkopen.2020.29891. Retrieved 17 Feb 2021 from https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2774080

Garber JR, Cobin RH, Gharib H, Hennessey JV, Klein I, Mechanick, Pessah-Pollack R, Singer PA, and Woeber KA (2012) Clinical practice guidelines for hypothyroidism in adults: cosponsored by the American Association of Clinical Endocrinologists and the American Thyroid Association. Endocrine Practice 18(6):988-1028.   Retrieved 21 July 2019 from https://journals.aace.com/doi/pdf/10.4158/EP12280.GL

Garcia-Solis P, Alfaro Y, Anguiano B, Delgado G, Guzman RC, Nandi S, Diaz-Munoz M, Vasquez-Martinez O, Aceves C (2005) Inhibition of N-methyl-N-nitrosourea-induced mammary carcinogenesis by molecular iodine (I2) but not by iodide (I-) treatment Evidence that I2 prevents cancer promotion. Mol Cell Endocrinol 236(1-2):49-57. Retrieved 16 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/15922087 (abstract only)

Gauthier K, Billon C, Bissler M, Beylot M, Lobaccaro JM, Vanacker JM, and Samarut J (2010) Thyroid hormones receptor beta (TRBeta) and liver X receptor (LXR) regulate carbohydrate-response element-binding protein (ChREBP) expression in a tissue-selective manner. J Biol Chem 285(36):28156-63.  Retrieved 20 Sep 2018 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2934680/pdf/zbc28156.pdf

Geary N (1999) The effect of estrogen on appetite. MedGenMed 1(3). Retrieved 13 Jan 2014 from https://www.medscape.com/viewarticle/722328_4

Ghassabian A   (2011) Maternal thyroid function during pregnancy and behavioral problems in the offspring: the Generation R Study. Pediatric Research 69:454-9. Retrieved 19 Jan 2013 from  https://www.nature.com/pr/journal/v69/n5-1/full/pr9201184a.html

Gencer B et al. (2012) Subclinical thyroid dysfunction and the risk of heart failure events: an individual participant data analysis from six prospective cohorts. Circulation 126:1040-1049. Retrieved 16 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3884576/

Giorgadze TA, Baloch ZW, Pasha T, Zhang PJ, and LiVolsi VA (2005) Lymphatic and blood vessel density in the follicular pattern lesions of thyroid.  Mod Pathol 18:1424-31. Retrieved 24 Jun 2013 from https://www.nature.com/modpathol/journal/v18/n11/pdf/3800452a.pdf

Gold MS and Morris LB (1987) The good news about depression: cures and treatments in the new age of psychiatry.  NY:Villard Books

Gong N, Gao C, Chen X, Fang Y, and Tian L (2019) Endothelial function in patients with subclinical hypothyroidism: a meta-analysis. Horm Metab Res 51(11):691-702. Retrieved 2 May 2020 from https://www.thieme-connect.com/products/ejournals/html/10.1055/a-1018-9564#TB2018-10-0352-0001

Hadi HAR, Carr CS, and Al Suwaidi J (2005) Endothelial dysfunction, cardiovascular risk factors, therapy, and outcome. Vasc Halth Risk Manag 1(3):183-98. Retrieved 15 May 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1993955/

Haegens N (2013) (Untitled) Retrieved 19 Aug 2013 from https://www.classofoods.com/page2_1.html

Hamilton TE, Davis S, Onstad L, and Kopecky (2008) Thyrotropin levels in a population with no clinical, autoantibody, or ultrasonographic evidence of thyroid disease: implications for the diagnosis of subclinical hypothyroidism. Retrieved 30 Oct 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2729186/

Han C, He X, Xia X, Li Y, Shi X, Shan Z, et al. (2015) Subclinical Hypothyroidism and Type 2 Diabetes: A Systematic Review and Meta-Analysis. PLoS ONE 10(8): e0135233. Retrieved 10 Feb 2021 from https://doi.org/10.1371/journal.pone.0135233

Harper ME and Brand MD (1993) The quantitative contributions of mitochondrial proton leak and ATP turnover reactions to the changed respiration rates of hepatocytes from rats of different thyroid status. J Biol Chem 268(20):14850-60. Retrieved 12 Apr 2014 from https://www.jbc.org/content/268/20/14850.long

Harper ME and Seifert EL (2008) Thyroid hormone effects on mitochondrial energetics. Thyroid 18(2):145-56. Retrieved 27 Oct 2019 from https://www.ncbi.nlm.nih.gov/pubmed/18279015

Hartzband PH and Groopman J (2020) Physician burnout, interrupted. N Engl J Med 382(26):2485-7. Retrieved 29 Jun 2020 from https://www.nejm.org/doi/full/10.1056/NEJMp2003149

Harvard Health Publications (2011) When depression starts in the neck. Retrieved 5 Oct 2016 from https://www.health.harvard.edu/newsletter_article/when-depression-starts-in-the-neck

Harvard Health Publications (2020) Hypothyroidism signs and symptoms in an older person.  Retrieved 25 Jun 2020 from https://www.health.harvard.edu/diseases-and-conditions/hypothyroidism-symptoms-and-signs-in-an-older-person

Hashimoto K, Matsumoto S, Yamada M, Satoh T, and Mori M (2007) Liver X Receptor-α gene expression is positively regulated by thyroid hormone. Endocrinol 148(10):4667-75. Retrieved 29 Sep 2018 from https://academic.oup.com/endo/article/148/10/4667/2501237

Haugen, B.R. (2009) Drugs that suppress TSH or cause central hypothyroidism. Best Pract Res Clin Endocrin Metab 23(6):793-800.  Retrieved 21 Dec 2012 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2784889/

Hekimzoy Z and Oktem IK (2009) Serum creatine kinase levels in overt and subclinical hypothyroidism. Endocr Res 31(3):171-5. Retrieved 8 May 2014 from https://www.ncbi.nlm.nih.gov/pubmed/16392619 (abstract only)

Helenius, T and  Tikanoja S (1986) A sensitive and practical immunoradiometric assay of thyrotropin. Retrieved 5 Oct 2016 from https://www.clinchem.org/content/32/3/514.full.pdf 

Hennessey JV and Espaillat R (2015) Subclinical hypothyroidism: a historical view and shifting prevalence. Int J Clin Pract 69(7):771-82. Retrieved 6 Aug 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6680332/

Hilts PJ (2001) After 46 years of sales, thyroid drug needs F.D.A. approval. The New York Times, 7 Jul 2001. Retrieved 27 Dec 2013 from  https://www.nytimes.com/2001/07/24/science/after-46-years-of-sales-thyroid-drug-needs-fda-approval.html

Hoermann R, Eckl W, Hoermann C, and Larisch R (2010) Complex relationship between free thyroxine and TSH in the regulation of hypothyroidism. Eur J Endocrinol 162(6):1123-9. Retrieved 28 Nov 2020 from https://eje.bioscientifica.com/view/journals/eje/162/6/1123.xml

Hoermann R, Midgley JE. TSH Measurement and Its Implications for Personalised Clinical Decision-Making. J Thyroid Res. 2012;2012:438037. doi: 10.1155/2012/438037. Epub 2012 Dec 9. PMID: 23304636; PMCID: PMC3529417. Retrieved 25 Nov 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3529417/

Hoermann R, Midgley JEM, Larisch R, and Dietrich JW (2018) The role of functional thyroid capacity in pituitary thyroid feedback regulation.  Eur J Invest 48(10):e13003. Retrieved 9 Mar 2020 from https://www.ncbi.nlm.nih.gov/pubmed/30022470 (abstract only)

Hoermann R, Midgely JEM, Larisch R, and Dietrich JW (2019) Individualised requirements for optimum treatment of hypothyroidism: complex needs, limited options.  Drugs Context 8:21297. Retrieved 26 Feb 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6726361/

Hoenderkamp R (2020) Management of hypothyroidism in primary care (video). Retrieved 3 May 2020 from https://vimeo.com/390796470

Hoffmann, K (2008) The effects of anabolic steroids on thyroid function. Retrieved 5 Oct 2016 from https://forums.musculardevelopment.com/showthread.php/43292-The-Effects-of-Anabolic-Steroids-on-Thyroid-Function

Hollowell JG, Staehling NW, Flanders WD, Hannon WH, Gunter EW, Spencer CA, and Braverman LE (2002) Serum TSH, T4, and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III).  The Journal of Clinical Endocrinology & Metabolism 87(2): 489-499. Retrieved 7 Dec 2019 from https://academic.oup.com/jcem/article/87/2/489/2846568

Hollowell JG, Staehling NW, Hannon WH, Flanders DW, Gunter EW, Maberly GF, Bravermann LE, Pino S, Miller DT, Garbe PL, DeLozier DM, and Jackson RJ (1998) Iodine nutrition in the United States. Trends and public health implications: Iodine excretion data from National Health and Nutrition Examination Surveys I and III (1971-1974 and 1988-1994). J Clin Endocrinol Metab 83(10):3401-8. Retrieved 19 Aug 2013 from https://jcem.endojournals.org/content/83/10/3401.full.pdf+html (no longer found there) and later on 7 Dec 2019 from https://academic.oup.com/jcem/article/83/10/3401/2865262 

Hong JW, Noh JH, Kim D-J (2018) Association between thyroid dysfunction and depressive symptoms in the Korean adult population: the 2014 Korean National Health and Nutrition Examination Survey.  PLoS One 13(8). Retrieved 2 Jun 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6091963/

Horn-Ross PL, Morris JS, Lee M, West DW, Whittemore AS, McDougall IR, Nowels K, Stewart SL, Spate VL, Shiau AC, and Krone MR (2001) Iodine and thyroid cancer risk among women in a multiethnic population: the Bay Area Thyroid Cancer Study. Cancer Epidemiol Biomarkers Prev 10(9):979-85. Retrieved 14 Nov 2013 from https://cebp.aacrjournals.org/content/10/9/979.long 

Hueston WJ and Pearson WS (2004) Subclinical hypothyroidism and the risk of hypercholesterolemia, Ann Fam Med 2(4):351-5. Retrieved 11 May 2014 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1466694/pdf/0020351.pdf

Ioannidis, JPA (2016) Evidence-based medicine has been hijacked: a report to David Sackett. J Clin Endocrinol. In press. https://www.sciencedirect.com/science/article/pii/S0895435616001475

Imaizumi M, Akahoshi M, Ichimaru S, Nakashima E, Hida A, Soda M, Usa T, Ashizawa K, Yokoyama N, Maeda R, Nagataki S, and Egushi K (2004) Risk for ischemic heart disease and all-cause mortality in subclinical hypothyroidism. J Clin Endocrinol Metab 89(7):3365-70.  Retrieved 30 Jan 2013 from https://jcem.endojournals.org/content/89/7/3365.full.pdf+html

Iqbal, A., Figenschau, Y. & Jorde, R. Blood pressure in relation to serum thyrotropin: the Tromsø study. J Hum Hypertens 20, 932–936 (2006). https://doi.org/10.1038/sj.jhh.1002091. Retrieved 4 Dec 2020 from https://www.nature.com/articles/1002091

Israel B (2011) Brominated battle: soda chemical has cloudy health history. Retrieved 19 Aug 2013 from https://www.environmentalhealthnews.org/ehs/news/2011/brominated-battle-in-sodas

Iverson JF and Mariash CN (2008) Optimal free thyroxine levels for thyroid hormone replacement in hypothyroidism. Endocr Pract 14(5):550-5. Retrieved 29 Oct 2019 from https://www.ncbi.nlm.nih.gov/pubmed/18753096 (abstract only)

Jayaprasad, N and F Johnson (2005) Atrial fibrillation and hyperthyroidism. Indian Pacing and Electrophysiol J 5(4):305-11. Retrieved 3 Feb 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1431605/

Jo S et al. (2019) Type 2 deiodinase polymorphism causes ER stress and hypothyroidism in the brain. J Clin Invest 129(1):230-45. Retrieved 20 Aug 2020 from https://www.jci.org/articles/view/123176

Jonklaas J, Bianco AC, Bauer AJ, Burman KD, Cappola AR, Celi FS, Cooper DS, Kim BW, Peeters RP, Rosenthal MS, and Sawka AM (2014) Guidelines for the Treatment of Hypothyroidism: Prepared by the American Thyroid Association Task Force on Thyroid Hormone Replacement. Thyroid 24(12): 1670–1751. Retrieved 28 Sep 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4267409/

Jorde R and Sundsfjord J (2006) Serum TSH levels in smokers and non-smokers. The 5th TromsØ study. Exp Clin Endocrinol 114(7):343-7. Retrieved 14 May 2014 at https://www.ncbi.nlm.nih.gov/pubmed/16915535

Joshi, SR (2011) Laboratory evaluation of thyroid function. Supplement to Journal Assoc Physicians India 59. https://www.japi.org/thyroid_special_jan_issue_2011/laboratory%20evaluation.pdf

Juhar SJ (2014) Why doctors are sick of their profession. WSJ August 30-31 C1-C2. Retrieved 31 Aug 2014 from https://online.wsj.com/articles/the-u-s-s-ailing-medical-system-a-doctors-perspective-1409325361

Kaferle J and Strzoda CE (2009) Evaluation of Macrocytosis. Am Fam Physician 79(3):203-8. Retrieved 9 Oct 2019 from https://www.aafp.org/afp/2009/0201/p203.html

Kitahara CM, Platz EA, Ladenson PW, Mondul AM, Menke A, and Berrington de Gonzalez (2012) Body fatness and markers of thyroid function among U.S. men and women. PLoS One:7(4):e34979. Retrieved 11 July 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3325258/

Kolata, G. (2016) Skinny and 119 pounds, but with the health hallmarks of obesity. July 22, 2016. https://www.nytimes.com/2016/07/26/health/skinny-fat.html

Kim YH, Choi YW, Han JH, Lee J, Soh EY, Park SH, Kim JH, Park TJ (2014) TSH signaling overcomes B-RafV600E-induced senescence through regulation of DUSP6. Neoplasia 16(12):1107-20. Retrieved 19 Dec 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4309262

Koulori O, Moran C, Halsall D, Chatterjee K, and Gurnell M (2013) Pitfalls in the measurement and interpretation of thyroid function tests. Best Pract Res Clin Endocrinol Metab 27(6):745-762.  Retrieved 12 Jul 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3857600/

Krassas G, Karras SN, Pontikides N (2015) Thyroid diseases during pregnancy: a number of important issues. Hormones 14(1):59-69. Retrieved 29 Apr 2020 from https://www.hormones.gr/8550/article/article.html

Kratzsch J, Fiedler GM, Leichtle A, Brügel M, Buchbinder S, Otto L, Sabri O, Matthes G and Thiery J (2005) New reference intervals for thyrotropin and thyroid hormones based on National Academy of Clinical Biochemistry criteria and regular sonography of the thyroid. Clin. Chem. 51(8):1480-6   https://www.clinchem.org/content/51/8/1480.full

Kumar KC, Kant R, Ekka BR, Kumar A (2016) Comparative study of lipid profile, renale and muscle damage in subclinical and overt hypothyroidism. Intl J of Contemporary Medical Research 3(10):77-83. Retrieved 30 Nov 2019 from https://www.ijcmr.com/uploads/7/7/4/6/77464738/ijcmr_984_oct_16.pdf

Ladenson PW, Singer PA, Ain KB, Bagchi N, Bigos ST, Levy EG, Smith SA, Daniels GH, and Cohen HD (2000) American Thyroid Association Guidelines for Detection of Thyroid Dysfunction. Arch Intern Med 160(11):1573-5. Retrieved 9 July 2020 from https://pubmed.ncbi.nlm.nih.gov/10847249/ (abstract only)

Laferrère B, Abraham C, Russell CD, Bowers CY (2005) Growth Hormone Releasing Peptide -2 (GHRP-2), like ghrelin, increases food intake in healthy men. J Clin Endocrinol Metab 90(2):611–614. Retrieved 26 Nov 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2824650/#__ffn_sectitle

Lambert M, Zech F, De Nayer P, Jamez J, and Vandercam B (1990) Elevation of serum thyroxine-binding globulin (but not of cortisol-binding globulin and sex hormone-binding globulin) associated with the progression of human immunodeficiency virus infection. Am J Med. 89(6):748-51. https://www.ncbi.nlm.nih.gov/pubmed/2147539

Laurberg P, Knudsen N, Andersen S, Carlé A, Pedersen IB, Karmisholt J (2012) Thyroid function and obesity.  Eur Thyroid J 1(3):159-67.  Retrieved 10 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3821486/

Lee, SL (2014) Iodine deficiency. Retrieved 3 May 2015 from https://emedicine.medscape.com/article/122714-overview

LeGrys VA, Funk MJ, Lorenz CE, Giri A, Jackson RD, Manson JE, Schectman R, Edwards TL, Heiss G, Hartmann KE (2013) Subclinical hypothyroidism and risk for incident myocardial infarction among postmenopausal women. J Clin Endocrinol Metab 98(6):2308-17. Retrieved 20 Aug 2013 from https://www.ncbi.nlm.nih.gov/pubmed/23539723

Leproult R, Van Reeth O, Byrne MM, Sturis J and Van Cauter E (1997) Sleepiness, performance, and neuroendocrine function during sleep deprivation: effects of exposure to bright light or exercise. J Biol Rhythms 12: 245-258. Retrieved 28 Dec 2012 from https://jbr.sagepub.com/content/12/3/245 but now only this is available: https://www.ncbi.nlm.nih.gov/pubmed/9181436?dopt=Abstract (abstract)

LeFevre ML (2015) Screening for thyroid dysfunction: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med [Epub ahead of print]. Retrieved 3 May 2015 from https://annals.org/article.aspx?articleid=2208599

Liu Y-J, Qiang W, Liu X-J, Xu L, Hui G, Wu L-P, and Shi B (2011) Association of insulin-like growth factor-1 with thyroid nodules. Oncol Lett 2(6):1297-1301. Retrieved 18 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3406504/

Livingston, EH and RA McNutt (2011) The hazards of evidence-based medicine: assessing variations in care. JAMA 306(7):762-763. Retrieved 26 Jan 2013 from https://jama.jamanetwork.com/article.aspx?articleid=1104225#qundefined

Longhi S and Radetti G (2013) Thyroid function and obesity. J Clin Res Pediatr Endocrinol 5(Suppl 1):40-4. Retrieved 20 Feb 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3608008/

Love AG and Davenport CB (1920) Defects found in drafted men. Statistical information compiled from the draft records showing the physical condition of the men registed and examined in pursuance of the requirements of the Selective-Service Act. Washington DC:Government Printing Office.

McAninch EA, Jo S, Preite NZ, Farkas E, Mohácsik P, Fekete C, Egri P, Gereben B, Li Y, Deng Y, Elizabeth Patti M, Zevenbergen C, Peeters RP, Mash DC, and Bianco AC (2015) Prevalent polymorphism in thyroid hormone-activating enzyme leaves a genetic fingerprint that underlies associated clinical syndromes. J Clin Endocrinol Metab (Epub ahead of print). Retrieved 17 Jan 2015 from https://www.ncbi.nlm.nih.gov/pubmed/25569702 (abstract only)

Mairet-Coello G, Tury A, and DiCicco-Bloom (2009) IGF-1 promotes G1/S cell cycle progression through bidirectional regulation of cyclins and CDK inhibits via the PI3K/Akt pathway in developing rat cerebral cortex. J Neurosci 29(3): 775-788. Retrieved 4 Apr 2014 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3256126/pdf/nihms92857.pdf

Malinowski, H (1997) Bioavailability/bioequivalence studies in evaluation of new levothyroxine products.  FDA.  Retrieved 26 Jan 2013 from https://www.fda.gov/downloads/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/UCM186436.pdf

Manson JE (2015) Thyroid dysfunction: to screen or not to screen? Retrieved 3 May 2015 from https://www.medscape.com/viewarticle/843456?nlid=80283_1842&src=wnl_edit_medp_wir&spon=17?src=sttwit

Maraka S, Mwangi R, McCoy RG, Yao X, Sangaralingham LR, Ospina NMS, O’Keeffe DT, Espinosa De Ycaza AE, Rodriguez-Gutierrez R, Coddington,III CC, Stan MN, Brito JP, and Montori VM (2017) Thyroid hormone treatment among pregnant women with subclinical hypothyroidism: US national assessment. BMJ 356:i6865. Retrieved 7 Feb 2017 from https://www.bmj.com/content/bmj/356/bmj.i6865.full.pdf

Marino M and McCluskey RT (2000) Role of thyroglobulin endocytic pathways in the control of thyroid hormone release. Am J Physiol Cell Physiol 279(5):1295-306. Retrieved 11 Oct 2013 from https://ajpcell.physiology.org/content/279/5/C1295

Markman JD, Gewandter JS, Frazer ME (2020) Comparison of a pain tolerability question with the numeric rating scale for assessment of self-reported chronic pain. JAMA Netw Open 3(4):e203155. Retrieved 26 May 2020 from https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2764594

Markou K, Georgopoulos N, Kyriazopoulou V, and Vagenakis AG (2001) Iodine-induced hypothyroidism.  Thyroid 11(5):501-10.  Retrieved 12 Feb 2020 from https://www.ncbi.nlm.nih.gov/pubmed/11396709

Mebis l and van den Berghe G (2009) The hypothalamus-pituitary-thyroid axis in critical illness. Neth J Med 67(10):332-40. Retrieved 29 March 2013 from https://www.njmonline.nl/getpdf.php?id=10000518

Merck & Co. (2006a) The Merck manual of diagnosis and therapy. Whitehouse Station, NJ:Merck Research Laboratories:1200

Merck & Co. (2006b) The Merck manual of diagnosis and therapy. Whitehouse Station, NJ:Merck Research Laboratories:1194

Midgley JEM, Toft AD, Larisch R, Dietrich JW, and Hoermann (2019) Time for a reassessment of hypothyroidism.  BMC Endocrine Disorders 19 (Article Number 37).  Retrieved 7 Mar 2020 from https://bmcendocrdisord.biomedcentral.com/articles/10.1186/s12902-019-0365-4

Miller M (2009) What are the effects of statins on triglycerides and what are the results of major outcomes studies? Retrieved 26 May 2014 from https://www.medscape.org/viewarticle/589010

Minor RK, Chang JW, and de Cabo R (2009) Hungry for life: how the arcuate nucleus and neuropeptide Y may play a critical role in mediating the benefits of calorie restriction. Mol Cell Endocrinol 299(1):79-88. Retrieved 2 March 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2668104/

Monzani F, Di Bello V, Caraccio N, Bertini A, Giorgi D, Giusti C, and Ferrannini E (2001) Effect of levothyroxine in subclinical hypothyroidism: a double-blind, placebo-controlled study.  J Clin Endocrinol Metab 86(3):1110-5.  Retrieved 26 Apr 2020 from https://academic.oup.com/jcem/article/86/3/1110/2847567

Moon KM, Lee YJ, Choi SH, Lim S, Yang EJ, Lim J-Y, Paik N-J, Kim KW, Paik KS, Jang HC, Cho YC, and Young JP (2010) Subclinical hypothyroidism has little influences on muscle mass or strength in older people. J Korean Med Sci 25(8):1176-81. Retrieved 22 Jun 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2908787/

Morshed SA, Ma R, Latif R, and Davies TF (2013) How one TSH receptor antibody induces thyrocyte proliferation while another induces apoptosis. J Autoimmun 47:17-24. Epub 2013 Aug 17. Retrieved 14 Feb 2014 from https://www.ncbi.nlm.nih.gov/pubmed/23958398 (abstract only)

Murray CM, Egan SK, Kim H, Beru N and Bolger PM (2008) US Food and Drug Administration's Total Diet Study: Dietary intake of perchlorate and iodine.  J Expo Sci Environ Epidemiol 18:571-80. Retrieved 2 May 2013 from  https://www.nature.com/jes/journal/v18/n6/full/7500648a.html

Mutaku JF, Poma JF, Many M-C, Denef J-F, and van den Hove M-F (2002) Cell necrosis and apoptosis are differentially regulated during goitre development and iodine-induced involution. J Endocrinol 172:375-386. Retrieved 23 Jul 2013 from https://joe.endocrinology-journals.org/content/172/2/375.long

Mutlu S, Parlak A, Aydogan U, Aydogdu A, Soykut B, Akay C, Saglan K, and Taslipinar A (2013) The effects of levothyroxine replacement therapy on lipid profile and oxidative stress parameters in patients with subclinical hypothyroid. Arch Pharm Res 8 Aug 2013. Retrieved 10 Apr 2020 from https://www.ncbi.nlm.nih.gov/pubmed/23925559 (Abstract only)

Nagao T and Hirokawa M (2017) Diagnosis and treatment of macrocytic anemias in adults. J Gen Fam Med 18(5):200-4.  Retrieved 1 Oct 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5689413/

Nalin P (2005) What causes a fever? Retrieved 31 May 2014 from https://www.scientificamerican.com/article/what-causes-a-fever/

National Center for Biotechnology Information (2013) Thyroxine -- Compound Summary. Retrieved 31 March 2013 from https://pubchem.ncbi.nlm.nih.gov/summary/summary.cgi?cid=5819

National Institutes of Health (2011) Iodine. Retrieved 1 May 2013 from https://ods.od.nih.gov/factsheets/Iodine-HealthProfessional/

National Research Council (2007) Earth materials and health: research priorities for earth science and public health. Washington, DC:The National Academies Press. Retrieved from https://www.nap.edu/catalog.php?record_id=11809 (essentially not available)

Nazarpour S, Tehrani FR, Rahmati M, Minooee S, Simbar M, Noroozzadeh M, and Azizi F (2018) Validation of Billewicz scoring system for detection of overt hypothyroidism during pregnancy. Int J Endocrinol Metab Jul; 16(3): e64249. Accessed 4 Jan 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6174609/

Negro, R.,  G. Formoso, T. Mangieri, A. Pezzarossa, D. Dazzi and H. Hassan (2006) Levothyroxine treatment in euthyroid pregnant women with autoimmune thyroid disease: effects on obstetrical complications.  J Clin Endocrinol Metab. 2006 Jul;91(7):2587-91. Epub 2006 Apr 18 https://www.ncbi.nlm.nih.gov/pubmed/16621910

Ness-Abramof R, Nabriski DA, Shapiro MS, Tripto-Shkolnik l, Katz B, Weiss E, and Shenkman L (2009) Cardiac troponin T is not increased in patients with hypothyroidism. Intern Med J 39(2):117-20. Retrieved 8 May 2014 from https://www.ncbi.nlm.nih.gov/pubmed/9848718 (abstract only)

Nisen M (2013) How adding iodine to salt resulted in a decade's worth of IQ gains for the United States. Business Insider July 22. Retrieved 30 Jul 2013 from https://www.businessinsider.com/iodization-effect-on-iq-2013-7

Ock S, Ahn J, Lee SH, Kang H, Offermanns S, Ahn HY, Jo YS, Shong M, Cho BY, Jo D, Abel ED, Tee TJ, Park WJ, Lee IK, and Kim J (2013) IGF-1 receptor deficiency in thyrocytes impairs hormone secretion and completely inhibits TSH-stimulated goiter. FASEB J. Epub ahead of print. Retrieved 27 Nov 2013 from https://www.ncbi.nlm.nih.gov/pubmed/23982142 (abstract only)

Office of the Legislative Counsel for the Use of the U.S. House of Representatives (2010) Compilation of the patient protection and affordable care act.  483-484. Retrieved 12 Jan 2018 from https://www.hhs.gov/sites/default/files/ppacacon.pdf.

Okosieme O, Gilbert J, Abraham P, Boelaert K, Dayan C, Gurnell M, Leese G, McCabe C, Perros P, Smith V, Williams G, and Vanderpump M. Management of primary hypothyroidism: statement by the British Thyroid Association Executive Committee. Clinical Endocrinology (2015), 0, 1–10. Retrieved 21 Aug 2019 from https://www.british-thyroid-association.org/sandbox/bta2016/bta_statement_on_the_management_of_primary_hypothyroidism.pdf

Olesen R (1927) Iodization of public water supplies for prevention of goiter. Association of Schools of Public Health: Public Health Reports 42:1355-67. Retrieved 29 Sep 2013. (first page only)

O'Rourke M (2013) What's wrong with me? New Yorker, August 26, pp. 32-37.

PCORI. 2018 annual report: Patient-Centered Outcome Research Institute.  Retrieved 12 Feb 2020 from https://www.pcori.org/sites/default/files/PCORI-Annual-Report-2018.pdf

PR Newswire (2018) Viking Therapeutics Announces Positive Top-Line Results from Phase 2 Study of VK2809 in Patients with Non-Alcoholic Fatty Liver Disease (NAFLD) and Elevated LDL-Cholesterol.  Retrieved 19 Sep 2018 from https://www.prnewswire.com/news-releases/viking-therapeutics-announces-positive-top-line-results-from-phase-2-study-of-vk2809-in-patients-with-non-alcoholic-fatty-liver-disease-nafld-and-elevated-ldl-cholesterol-300714176.html

Paddock Laboratories (2013) Liothyronine sodium tablet. Retrieved 21 Jan 2013 from https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=cfa4a974-3d3a-4772-a2da-1b2383672e0a

Panicker V, Saravanan P, Vaidya V, Evans J, Hattersley AT, Frayling TM, and Dayan CM (2009) Common variation in the DIO2 gene predicts baseline psychological well-being and response to combination thyroxine plus triiodothyronine therapy in hypothyroid patients. J Clin Endocrinol Metab 94(5):1623-29. Retrieved 21 Sep 2015 from https://press.endocrine.org/doi/pdf/10.1210/jc.2008-1301

Pantalone KM and Nash C (2010). Approach to a low TSH level: patience is a virtue. Cleveland Clinic J of Med 77(11):803-11. Retrieved 10 Jul 2013 from https://www.ccjm.org/content/77/11/803.full.pdf+html  As of 9 Oct 2019, the abstract is available from this journal: https://www.ncbi.nlm.nih.gov/pubmed/21048053 and that page has a link to https://www.mdedge.com/ccjm/article/95238/endocrinology/approach-low-tsh-level-patience-virtue

Papadea C (1998) Generations of immunoassays. Laboratory Medicine 29(9):528-529. Retrieved 22 May 2020 from https://academic.oup.com/labmed/article/29/9/528/2504047

Papaleontiu M and Cappola AR (2016) Thyroid-stimulating hormones in the evaluation of subclinical hypothyroidism. JAMA 316(15):1592-3. Retrieved 6 Jan 2017 from https://jamanetwork.com/journals/jama/article-abstract/2569762

Paranjape SA, Chan O, Zhu W, Horblitt AM, McNay EC, Cresswell JA, Bogan JS, McCrimmon RJ, and Sherwin RS (2010) Influence of insulin in the ventromedial hypothalamus on pancreatic glucagon secretion in vivo. Diabetes 59(6):1521-7. Retrieved 2 Mar 2020 from https://diabetes.diabetesjournals.org/content/59/6/1521

Park JP, Yoon JW, Kwang II K, Kim SY et al. (2009) Subclinical hypothyroidism might increase the risk of atrial fibrillation in cardiac arterial bypass grafting. Ann Thorac Surg 87(6):1846-52. Retrieved 25 Mar 2021 from https://www.annalsthoracicsurgery.org/article/S0003-4975(09)00526-8/fulltext

Patashnik EM, Gerber AS, and Dowling CM (2017) Unhealthy politics: the battle over evidence-based medicine. Princeton University Press.

Patient Protection and Affordable Care Act; Coverage of Certain Preventive Services Under the Affordable Care Act, Fed. Reg. 41318 (July 14, 2015) (to be codified at 26 CFR pt. 54, 29 CFR pts. 2510 and 2590, and 45 CFR pt. 147).  Retrieved 14 Jan 2018 from https://www.gpo.gov/fdsys/pkg/FR-2015-07-14/pdf/2015-17076.pdf

Pavelka S (2004) Metabolism of bromide and its interference with the metabolism of iodine. Physiol Res 53 (Suppl. 1):S81-90. Retrieved 19 Aug 2013 from https://www.biomed.cas.cz/physiolres/pdf/53%20Suppl%201/53_S81.pdf

Pearce SHS, Brabant G, Duntas LH, Monzani F, Peeters RP, Razvi S, and Wemeau J-L (2013) 2013 ETA guideline: management of subclinical hypothyroidism. Eur Thyroid J 2(4):215-28. Retrieved 16 Feb 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3923601/

Peeters RP (2017) Subclinical hypothyroidism. N Engl J Med:2556-65. Retrieved 9 July 2019 from https://www.nejm.org/doi/full/10.1056/NEJMcp1611144

Pert, C. (1997) Molecules of emotion. NY:Scribner

Peterson SJ, Cappola AR, Castro MR, Dayan CM, Farwell AP, Hennessey JV, Kopp PA, Ross DS, Samuels MH, Sawka AM, Taylor PN, Jonklaas J, Bianco AC (2018) An online survey of hypothyroid patients demonstrates prominent dissatisfaction. Thyroid 28(6). Epub ahead of press. Retrieved 7 Jan 2020 from https://www.liebertpub.com/doi/10.1089/thy.2017.0681

Phillips, C. (2009) What is a QALY? Retrieved 2 Nov 2014 from https://www.medicine.ox.ac.uk/bandolier/painres/download/whatis/QALY.pdf

Pierce Protein Biology Products/Thermo Scientific (2014) Imject Freund's Adjuvants: Convenient ampules of complete (CFA) and incomplete (IFA) Freund's adjuvants. Retrieved from https://www.piercenet.com/product/imject-freunds-adjuvants

Pincott J (2020) The menopause connection. Sci American 322(5):37-41. 

Pingitore A, Galli E, Barison A, Iervasi A, Scarlattini M, Nucci D, L'abbate A, Mariotti R, Iervasi G (2008) Acute effects of triiodothyronine (T3) replacement therapy in patients with chronic heart failure and low-T3 syndrome: a randomized, placebo-controlled study. J Clin Endocrinol Metab. 93(4):1351-8. Retrieved 21 July 2019 from https://www.ncbi.nlm.nih.gov/pubmed/18171701

Pirahanchi Y and Jialal I (2019) Physiology, Thyroid Stimulating Hormone (TSH). Treasure Island (FL): StatPearls Publishing. Retrieved 1 Jan 2020 from https://www.ncbi.nlm.nih.gov/books/NBK499850/

Polak JF, Pencino MJ, Pencina KM, O'Donnell CJ, Wolf PA, and D'Agostino RB Sr. (2011) Carotid-Wall intima-media thickness and cardiovascular events. N Engl J Med 365:213-21. Retrieved 16 Jan 2020 from https://www.nejm.org/doi/full/10.1056/NEJMoa1012592

Portes, ES, JH Oliveira, P MacCagnan, and J Abucham (2000) Changes in serum thyroid hormones levels and their mechanisms during long-term growth (GH) replacement therapy in GH deficient children. Clin Endocrinol 53(2):183-9.  Retrieved 17 Jan 2013 from  https://www.ncbi.nlm.nih.gov/pubmed/10931099

Portillo-Sanchez P, Rodriguez-Gutierrez R, and Brito JP (2016) Subclinical hypothyroidism in elderly individuals -- overdiagnosis and overtreatment? a teachable moment. JAMA Int Med 176(12):1741. Retrieved 15 Mar 2017 from https://jamanetwork.com/journals/jamainternalmedicine/article-abstract/2560378

Posadas-Romero C, Jorge-Galarza E, Posadas-Sanchez R, Acunas-Valerio J, Juarez-Rojas JG, Kimura-Hayama E, Medina-Urrutia A, and Cardoso-Saldana GC (2014) Fatty liver largely explains associations of subclinical hypothyroidism with insulin resistance, metabolic syndrome, and subclinical coronary atherosclerosis.  Eur J Endocrinol 171(3):319-25. Retrieved 23 Feb 2018 from https://www.eje-online.org/content/171/3/319.long

Powers BW, Jain SH, Shrank WH. De-adopting Low-Value Care: Evidence, Eminence, and Economics. JAMA. Published online October 02, 2020. doi:10.1001/jama.2020.17534. Retrieved 11 Oct 2020 from https://jamanetwork.com/journals/jama/fullarticle/2771511

Problems of the equine thyroid (2005). Horse & Hound 13(59). Retrieved 23 Nov 2020 from https://www.horseandhound.co.uk/horse-care/vet-advice/problems-of-the-equine-thyroid-7136

Pugh, DE (2016) A New Regulatory Catch-22: Hypothyroidism and Depression. Retrieved 3 Jan 2017 from blog%2034%20hypothyroidism%20and%20depression.html

Punekar P, Sharma AK, Jain A (2018) A study of thyroid dysfunction in cirrhoses of liver and correlation with severity of liver disease. Indian J Endocrinol Metab 22(5):645-50. Retrieved 14 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6166553/

Qian M, Wang D, Watkins WE, Gebski V, Yan YQ, Li M, and Chen ZP (2005) The effects of iodine on intelligence in children: a meta-analysis of studies conducted in China. Asia Pac J Clin Nutr 14(1):32-42. Retrieved 13 Oct 2013 from https://apjcn.nhri.org.tw/server/APJCN/14/1/32.pdf

Qiu J, Bosch MA, Tobias SC, Grandy DK, Scanlan TS, Ronnekiev OK, and Kelly MJ (2003) Rapid signaling of estrogen in hypothalamic neurons involves a novel G-protein-couple estrogen receptor that activates protein kinase C. J Neurosci 23(29):9529-40. Retrieved 28 Dec 2013 from https://www.jneurosci.org/content/23/29/9529.full.pdf+html

Rao AS, Kremenevskaja, Resch J, and Brabant G (2005) Lithium stimulates proliferation in cultured thyrocytes by activating Wnt/ß-catenin signalling. Eur J of Endocrinol 153:929-38. Retrieved 14 Feb 2014 from https://www.eje-online.org/content/153/6/929.full.pdf

Raffensperger L (2013) How adding iodine to salt boosted Americans' IQ.  Discover Magazine, July 23, 2013. Retrieved 27 Sep 2013 from https://blogs.discovermagazine.com/crux/2013/07/23/how-adding-iodine-to-salt-boosted-americans-iq/#.Ukqz44bXQbK

Rastogi S (2021) Establishing equity in medical education -- supporting clinical trainees with disabilities. N Engl J Med 384(30):885-7. (accessed via subscription)

Razvi S, Shakoor A, Vanderpump M, Weaver JU, Pearce HS (2008) The influence of age on the relationship between subclinical hypothyroidism and ischemic heart disease: a metaanalysis. J Clin Metab 93(8):2998-3007. Retrieved 25 Jun 2020 from https://academic.oup.com/jcem/article/93/8/2998/2598481 

Razvi S, Weaver JU, Butler TJ, and Pearce SHS (2012) Levothyroxine treatment of subclinical hypothyroidism, fatal and nonfatal cardiovascular events, and mortality. Arch Intern Med: 172(10):811-7. Retrieved 3 May 2015 from https://archinte.jamanetwork.com/article.aspx?articleid=1149639

Razvi S, Waver JU, Vanderpump MP and Pearce SHS (2010) The incidence of ischemic heart disease and mortality in people with subclinical hypothyroidism: reanalysis of the Whickham survey consort.  J Clin Endocrinol Metab 95(4):1734-40 https://jcem.endojournals.org/content/95/4/1734.long

Razvi S, Peeters R, and Pearce SHS (2019) Letter: Thyroid hormone therapy for subclinical hypothyroidism. JAMA 321(8):804.

Reichert, LJM and de Rooy HAM (1989) Treatment of amiodarone induced hyperthyroidism with potassium perchlorate and methmazole during amiodarone treatment. BMJ 298:1547-8. Retrieved 28 Oct 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1836819/pdf/bmj00235-0019.pdf

Reinagel, M (2017) Can a gluten free diet help with thyroid disease? Scientific American, 7 Oct 2017. Retrieved 3 Sep 2019 from https://www.scientificamerican.com/article/can-a-gluten-free-diet-help-with-thyroid-disease/

Reyes Domingo, F, Avey MT, and Doull, M (2019) Screening for thyroid dysfunction and treatment of screen-detected thyroid dysfunction in asymptomatic, community-dwelling adults: a systematic review. Syst Rev 8, 260

Rizos CV, Elisaf MS, and Liberopoulos EN (2011) Effects of thyroid dysfunction on lipid profile. Open Cardiovasc Med J 5:76-84. Retrieved 8 May 2014 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3109527/pdf/TOCMJ-5-76.pdf

Rodondi N et al. (2010) Subclinical hypothyroidism and the risk of coronary heart disease and mortality. JAMA 304(12):1365-74. Retrieved 16 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3923470/

Roelfsema F  (2009) Diminished and irregular TSH secretion with delayed acrophase in patients with Cushing's syndrome. Eur J Endocrinol 161:695-703.  Retrieved 3 Jan 2013 from  https://eje-online.org/content/161/5/695.long

Roelfsema F and Veldhuis JD (2013) Thyrotropin secretion patterns in health and disease. Endocrine Reviews 34(5):619-57. Retrieved 16 May 2020 from https://academic.oup.com/edrv/article/34/5/619/2354647

Rotondi M, Cappelli C, Leporati P, Chytires S, Zerbini F, Fonte R, Magri F, Castellano M, Chiovato L (2010) A hypoechoic pattern of the thyroid at ultrasound does not indicate autoimmune thyroid diseases in paients with morbid obesity. Eur J Endocrinol 163(1):105-9. Retrieved 21 Feb 2020 from https://eje.bioscientifica.com/view/journals/eje/163/1/105.xml

Rubin R (2020) Confirmatory trial for drug to prevent preterm birth finds no benefit, so why is it still prescribed? JAMA 323(13):1229-32.  Retrieved 18 Apr 2020 from https://jamanetwork.com/journals/jama/article-abstract/2763422 (initial text only, but had access to publication through subscription)

Ruggeri, RM, E Vitarelli, G Barresi, F Trimarchi, S Benvenga, and M Trovato (2010) The tyrosine kinase receptor c-met, its cognate ligand HGF and the tyrosine kinase receptor trasducers STAT3, PI3K and RHO in thyroid nodules associated with Hashimoto's thyroiditis: an immunohistochemical characterization.  Retrieved 23 Feb 2013 from  https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3167304/

Ruiz-Nuñez B, Tarasse R, Vogelaar EF, Dijck-Brouwer, and Muskiet FAJ (2018) Higher prevalence of "Low T3 syndrome" in patients with chornic fatigue syndrome: a case-control study. Front Endocrinol Retrieved 31 July 2019 from https://www.frontiersin.org/articles/10.3389/fendo.2018.00097/full

Russell, W, RF Harrison, N Smith, K Darzy, S Shalet, AP Weetman, and RJ Ross (2008) Free Triiodothyronine has a distinct circadian rhythm that is delayed but parallels thyrotropin levels. J Clin Endocrinol Metab 93(6):2300-6.  Retrieved 18 Feb 2013 from https://jcem.endojournals.org/content/93/6/2300.full.pdf

Sackett DL, Rosenberg WMC, Gray JAM, Haynes RB, and Richardson WS (1996) Evidence-based medicine: what it is and what it isn't. BMJ 312(7023):71-2. Retrieved 5 Oct 2014 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2349778/pdf/bmj00524-0009.pdf

Sadek SH, Khalifa WA, and Azoz AM (2017) Pulmonary consequences of hypothyroidism. Ann Thorac Med 12(3):204-8. Retrieved 25 Jun 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5541969/

Saladin, K (2007) Anatomy & Physiology, 4th ed. NY:McGraw-Hill

Salamone JD, Correa M, Mingote S, and Weber SM (2003) Nucleus accumbens dopamine and the regulation of effort in food-seeking behavior: implications for studies of natural motivation, psychiatry, and drug abuse. J Pharmacol Exp Ther 305(1):1-8.  Retrieved 3 Nov 2014 from https://jpet.aspetjournals.org/content/305/1/1.full.pdf+html

Salamova A, Yuning M, Venier M, and Hites RA (2013) High levels of organophosphates in the Great Lakes atmosphere. Environ Sci Technol Lett. 1(1):8-14. Retrieved from https://pubs.acs.org/doi/abs/10.1021/ez400034n?source=cen

Saller B, Broda N, Heydarian R, Görges R, and Mann K (1998) Utility of third generation thyrotropin assays in thyroid function testing. Exp Clin Endocrinol Diabetes 106 Suppl 4:S29-33. Retrieved 21 May 2020 from https://pubmed.ncbi.nlm.nih.gov/9867193/

Samuels MH, Schuff KG, Carlson NE, Carello P, Janowsky JS (2007) Health status, mood, and cognition in experimentally induced subclinical hypothyroidism. J Clin Endocrinol Metab 92(7):2545-51. Retrieved 17 May 2020 from https://academic.oup.com/jcem/article/92/7/2545/2598343

Santi A, Duarte MMMF, Menezes CC, and Loro VL (2012) Association of lipids with oxidative stress biomarkers in subclinical hypothyroidism. Int J Endocrinol 856359.  Retrieved 3 Nov 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3517852/

Santini F, Galli G, Maffei M, Fierabracci P et al. (2010) Acute exogenous TSH administration stimulates leptin secretion in vivo. Eur J Endocrinol 163(1):63-7. Retrieved 30 Oct 2020 from https://eje.bioscientifica.com/view/journals/eje/163/1/63.xml

Santini, F, A Pinchera, G Ceccharini, M Castagna, V Rosellini, C Mammoli, L montanelli, V Zucchi, IJ Chopra, and L Chiovato (2001) Evidence for a role of the type III-iodothyronine deiodinase in the regulation of 3,5,3'-triiodothyronine cdontent in the human central nervous system. Eur J Endocrinol 144:577-583. Retrieved 13 April 2013 from https://eje-online.org/content/144/6/577.long

Santini F, Marzullo P, Rotondi M, Ceccarini G, Pagano L, Ippolito S, Chrovato L, And Biondi B (2014) Mechanisms in Endocrinology: The crosstalk between thyroid gland and adipose tissue: signal integration in health and disease. European J Endocrinol 171(4):R137. Retrieved 3 Oct 2020 from https://eje.bioscientifica.com/view/journals/eje/171/4/R137.xml

Sawhney, R.C., I Rastogi, GK Rastogi (1978) Effect of estrogens on thyroid function. I. Alteration in rhesus plasma thyrotropin and its kinetics. Endocrinology 102(4):1310-6 https://www.ncbi.nlm.nih.gov/pubmed/105876

Sawin CT, Castelli WP, Hershman JM, McNamara P, and Bacharach (1985) The aging thyroid. Thyroid deficiency in the Framingham Study. Arch Int Med 145(8):1386-8. Retrieved 19 May 2020 from https://www.ncbi.nlm.nih.gov/pubmed/4026469/ (abstract only)

Sawin, CT (2002) The heritage of Dr. Hakaru Hashimoto (1881-1934). Endocr J 49(4):399-403. Retrieved 24 Jan 2013 from https://www.jstage.jst.go.jp/article/endocrj/49/4/49_4_399/_pdf

Scanlon, M.F. and D.R. WEightman, D.J. Shale, B. Mora, M. Heath, M.H. Snow, M. Lewis, R. Hall (1979) Dopamine is a physiological regulator of thyrotropin (TSH) secretion in normal man. Clin Endocrinol 10(1): 7-15 https://www.ncbi.nlm.nih.gov/pubmed/436307 (abstract only)

Scheen, A.J. and O.M. Buxton, M. Jison, O. Van Reeth, R. Leproult, M. L'Hermite-Balériaux, and E. Van Cauter (1998) Effects of exercise on neuroendocrine secretions and glucose regulation at different times of day. Retrieved from Am J Physiol Endocrinol Metab 274(6):E1040-49. Retrieved 11 Oct 2013 from https://ajpendo.physiology.org/content/274/6/E1040.full

Scientific Committee on Food (2002) Opinion of the Scientific Committe on  Food on the tolerable upper intake of iodine. European Commission: Health & Consumer Protection Directorate-General.  Retrieved 5 Sep 2019 from https://ec.europa.eu/food/sites/food/files/safety/docs/sci-com_scf_out146_en.pdf

Senese R, Cioffi F, de Lange P, Leanza C, Iannucci LF, Silvestri E, Moreno M, Lombardi A, Goglia F, Lanni A (2017) Both 3,5-diiodo-L-thyronine and 3,5,3'-triiodo-L-thyronine prevent short-term hepatic lipid accumulation via distinct mechanism in rats being fed a high-fat diet. Front Physiol 8:706. Retrieved 20 Sep 2018 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5603695/pdf/fphys-08-00706.pdf

Serghiou S and Goodman SN (2019) Random-effects meta-analysis: summarizing evidence with caveats. JAMA 321(3):301-2. Retrieved 27 Jan 2019 from https://jamanetwork.com/journals/jama/article-abstract/2719518 (abstract only, had access to hard-copy full article)

Shan S, Aslam S, Saleem N, and Ritzvi NB (2017) Thyroid hormone deficiency and coronary artery disease. Anaesth Pain & Intensive Care 21(4):489-96. Retrieved 31 Jan 2020 from https://www.apicareonline.com/index.php/APIC/article/view/94/88

Shomon M (2005) Living well with hypothyroidism: what your doctor doesn't tell you... that you need to know. NY:Avon Books, Inc.

Shekhar S, Sinaii N, Irizarry-Caro JA, et al. (2020) Prevalence of hypothyroidism in patients with Erdheim-Chester Disease. JAMA Netw Open 3(10):e2019169. Retrieved 31 Oct 2020 from https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2772340#:~:text=Findings%20In%20this%20cross%2Dsectional,%2C%20respectively%2C%20in%20the%20community.

Shiel Jr. (2020) Medical definition of subclinical disease. Retrieved 7 May 2020 from https://www.medicinenet.com/script/main/art.asp?articlekey=5578

Sinha RA, Singh BK, and Yen PM (2018) Direct effects of thyroid hormones on hepatic lipid metabolism. Nat Rev Endocrinol 14(5):259-69. Retrieved 19 July 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6013028/

Sinha R, Bruinstroop E, Singh BK, and Yen PM (2019): Nonalcoholic fatty liver disease and hypercholesterolemia: roles of thyroid hormones, metabolites, and agonists. Thyroid 29(9):1173-91. Retrieved 3 Sep 2020 from https://pubmed.ncbi.nlm.nih.gov/31389309/

Skeaff SA, Lonsdale-Cooper E. (2013) Mandatory fortification of bread with iodised salt modestly improves iodine status in schoolchildren. Br J Nutr. 109(6):1109-13. Retrieved 5 Sep 2019 from https://www.ncbi.nlm.nih.gov/pubmed/22849786

Slawik M, Klawitter B, Meiser E, Zwermann O, Borm K, Peper M, Lubrich B, Hug MJ, Nauck M, Olschewski M, Beuschlein F, Reincke M (2007) Thyroid hormone replacement for central hypothyroidism: a randomized controlled trial comparing two doses of thyroxine (T4) with a cominbation of T4 and triiodothyronine. J Clin Endocrinol Metab 92(11):4115-22. Retrieved 8 Aug 2020 from https://academic.oup.com/jcem/article/92/11/4115/2598008

Smart, GA and SG Owen (1960) Thyroid antibodies and thyroiditis. Postgraduate Medical Journal: July 1960:442-6. Retrieved 24 Jan 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2481746/pdf/postmedj00476-0032.pdf

Society for Endocrinology (2019) Society for Endocrinology & British Thyroid Association issue statement against new treatment recommendation. Eurekalert! American Association for the Advancement of Science (AAAS). Retrieved 21 Aug 2019 from https://www.eurekalert.org/pub_releases/2019-05/sfe-sfe052919.php

Soldin, O. P., Chung, S. H., & Colie, C. (2013). The Use of TSH in Determining Thyroid Disease: How Does It Impact the Practice of Medicine in Pregnancy?. Journal of thyroid research2013, 148157. https://doi.org/10.1155/2013/148157. Retrieved 25 Nov 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3665256/

Sontag S (1978) Illness as metaphor. NY:Farrar, Straus & Giroux

Matthew Speer, J. Mac McCullough, Jonathan E. Fielding, Elinore Faustino, Steven M. Teutsch, “Excess Medical Care Spending: The Categories, Magnitude, and Opportunity Costs of Wasteful Spending in the United States”, American Journal of Public Health 110, no. 12 (December 1, 2020): pp. 1743-1748.  Retrieved 17 Feb 2021 from https://ajph.aphapublications.org/doi/10.2105/AJPH.2020.305865 (abstract only)

Spencer CA, Hollowell JG, Kazarosyan M, Braverman LE (2007) National Health and Nutrition Examination Survey II thyroid-stimulating hormones (TSH) - Thyroperoxidase antibody relationships demonstrate that TSH upper reference limits may be skewed by occult thyroid dysfunction. J Clin Endocrinol Metab 92(11):4236-40.  Retrieved 26 Oct 2019 from https://academic.oup.com/jcem/article/92/11/4236/2598356

Spencer CA (2017) Assay of hormones and related substances. Endotext [Internet]. Retrieved 1 Jan 2020 from https://www.ncbi.nlm.nih.gov/books/NBK279113/

Stagnaro-Green, A. (2008) Maternal thyroid disease and preterm delivery. J Clin Endocrinol Metab 94(1):21-5. Retrieved 18 Jan 2013 from https://jcem.endojournals.org/content/94/1/21.long

Stagnaro-Green, A., S.H. Roman, R.H. Cobin, E. el_Harazy, M. Alvarez-Marfany, T.F. Davies (1990) Detection of at-risk pregnancy by means of highly sensitive assays for thyroid autoantibodies.  JAMA Sep 19;264(11):1422-5 https://www.ncbi.nlm.nih.gov/pubmed/2118190 (abstract only)

Stoddard II, FR, AD Brooks, BA Eskin, and GJ Johannes (2008) Iodine alters gene expression in the MCF7 breast cancer cell line: evidence for an anti-estrogen effect of iodine. Int J Med Sci 5(4):189-96. Retrieved 13 Feb 2013 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2452979/

Stott DJ, Rodondi N, Kearney PM, Ford I, Westendorp RGJ, Mooijaart SP, Sattar N, Aubert CE, Aujesky D, Bauer DC, Baumgartner C, Blum MR, et al. (2017). N Engl J Med 376:2534-44. Retrieved 10 Nov 2019 from https://www.nejm.org/doi/full/10.1056/NEJMoa1603825

Strom S (2014) Coca-cola to remove an ingredient questioned by consumers. New York Times: May 5, 2014. Retrieved 6 May 2014 from https://www.nytimes.com/2014/05/06/business/coca-cola-to-remove-an-ingredient-questioned-by-consumers.html?_r=1

Sungro J et al. (2019) Type 2 deiodinase polymorphism causes ER stress and hypothyroidism in the brain. J Clin Inv 129(1):230-45. Retrieved 20 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6307951/

Surks MI, Hollowell JG (2007) Age-specific distribution of serum thyrotropin and antithyroid antibodies in the US population: implications for the prevalence of subclinical hypothyroidism. J Clin Endocrinol Metab 92(12):4575-82.  Retrieved 26 Oct 2019 from https://academic.oup.com/jcem/article/92/12/4575/2596923

Sutoo D and Akiyama K (2003) Regulation of brain function by exercise. Neurobiol Dis 13(1):1-14. Retrieved 2 Feb 2014 from https://www.ncbi.nlm.nih.gov/pubmed/12758062 (abstract only)

Tan ZS, Beiser A, Vasan RS, Au R, Auerbach S, Kiel DP, Wolf PA, Seshadri S (2008) Thyroid function and the risk of Alzheimer's disease: The Framingham Study. Arch Int Med (168)14:1514-20. Retrieved 16 Jun 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2694610/#:~:text=Whereas%20thyroid%20dysfunction%20has%20long,increased%20risk%20of%20irreversible%20dementia.

Taylor PN, Razvi S, Pearce SH, Dayan C (2013) A review of the clinical consequences of variation within the reference range. J Clin Endocrinol Metab 98:3562-71. Retrieved 3 Jan 2020 from https://academic.oup.com/jcem/article/98/9/3562/2833074

Taylor P and Bianco AC (2019) Urgent need for further research in subclinical hypothyroidism. Nat Rev Endocrinol 15:503-4. Retrieved from https://www.nature.com/articles/s41574-019-0239-x

Taylor PN, Iqbal A, Minassian C, Sayers A, Draman MS, Greenwood R, Hamilton W, Okosieme O, Panicker V, Thomas SL, and Dayan C (2014) Falling threshold for treatment of borderline elevated thyrotropin levels-balancing benefits and risks: evidence from a large community-based study. JAMA Intern Med 174(1):32-9. Retrieved 28 May 2019 from https://www.ncbi.nlm.nih.gov/pubmed/24100714

Taylor PN, Razvi S, Muller I, Wass J, Dayan CM, Chatterjee K, Boelaert K (2019) Liothyronine cost and prescriptions in England. The Lancet: Diabetes and Endocrinology 7(1):P11-12.  Retrieved 3 Mar 2020 from https://www.thelancet.com/journals/landia/article/PIIS2213-8587(18)30334-6/fulltext

Tazebay UH, Wapnir IL, Levy O, Dohan O, Zuckier LS, Zhao QH, Deng HF, Amenta PS, Fineberg S, Pestell RG, and Carrasco N (2000) The mammary gland iodide transporter is expressed during lactation and in breast cancer. Nat Med 6(8):871-8. Retrieved 12 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/10932223 (abstract only)

Thakur A (2019) A hospital-based study for clinico-investigative profile of newly diagnosed patients of hypothyroidism. Endocrinol Metab Syndr 8:304. Retrieved 15 Jul 2020 from https://www.longdom.org/open-access/a-hospitalbased-study-for-clinicoinvestigative-profile-of-newly-diagnosed-patients-of-hypothyroidism-44599.html#ai

Thayakaran R, Adderley NJ, Sainsbury C, Torlinskaya B, Boelaert K, Sumilo D, Price M, Thomas GN, Toulis KA, Nirantharakumar K (2019) Thyroid replacement therapy, thyroid stimulating hormone concentrations, and long term health outcomes in patients with hypothyroidism: longitudinal study. BMJ 366:I4892.  Retrieved 29 Oct 2019 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6719286/

Tortora G, Pepe S, Cirafici AM, Giardiello F, Porcellini A, Clair T, Colletta G, Cho-Chung YS, Bianco AR (1993) Thyroid-stimulating hormone-regulated growth and cell cycle distribution of thyroid cells involve Type I Isoenzyme of cyclic AMP-dependent protein kinase. Cell Growth Differ 4(5):359-65.  Retrieved 10 Oct 2013 from https://cgd.aacrjournals.org/cgi/reprint/4/5/359

Trost SU, Swanson E, Gloss B, Wang-Iverson DB, Zhang H, Volodarsky T, Grover GJ, Baxter JD, Chiellini G, Scanlan TS and Dillmann WH (2000) The thyroid hormone receptor-beta-selective agonist GC-1 differentially affects plasma lipids and cardiac activity. Endocrinol 141:3057-64.  Retrieved 28 Nov 2019 from https://academic.oup.com/endo/article/141/9/3057/2988394

Tsui S, Naik V, Hoa N, Hwang CJ, Afifyan NF, Hikim AS, Gianoukakis AG, Douglas RS, and Smith TJ (2008) Evidence for an association between TSH and IGF-1 receptors: A tale of two antigens implicated in Graves' Disease. J Immunol 181(6):4397-4405. Retrieved 29 Jan 2014 from https://www.jimmunol.org/content/181/6/4397.full.pdf

U.S. Centers for Medicare and Medicaid Services (2019a) Preventive care benefits for women. Retrieved 17 Dec 2019 from https://www.healthcare.gov/preventive-care-women/

U.S. Centers for Medicare and Medicaid Services (2019b) Preventive care benefits for adults. Retrieved 17 Dec 2019 from https://www.healthcare.gov/preventive-care-adults/

U.S. Environmental Protection Agency (2011) Fluoride risk assessment and relative source contribution. Retrieved 25 Aug 2013 from https://water.epa.gov/action/advisories/drinking/fluoride_index.cfm

U.S. Environmental Protection Agency (2013) Polybrominated Diphenyl Ethers (PDBEs) Action Plan Summary. Retrieved 7 Oct 2013 from https://www.epa.gov/oppt/existingchemicals/pubs/actionplans/pbde.html

U.S. Food and Drug Administration (2017) Total Diet Study Elements Results Summary Statistics Market Baskets 2006 through 2013.  Retrieved 12 Nov 2019 from https://www.fda.gov/media/77948/download

U.S. Preventive Services Task Force (2004) Screening for Thyroid Disease: Recommendation Statement.  Retrieved 8 Aug 2019 from https://annals.org/aim/fullarticle/717097/screening-thyroid-disease-recommendation-statement

U.S. Preventive Services Task Force. Final Recommendation Statement: Thyroid Disease: Screening. March 2015.  Retrieved 20 Oct 2016 from https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/thyroid-dysfunction-screening

U.S. Preventive Services Task Force. Final Recommendation Statement: Depression in Adults. January 2016. Retrieved 5 Oct 2016 from https://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/depression-in-adults-screening1

U.S. Preventive Services Task Force. December 2016. Sixth Annual Report to Congress on High-Priority Evidence Gaps for Clinical Preventive Services. Retrieved 3 January 2017 from https://www.uspreventiveservicestaskforce.org/Page/Name/sixth-annual-report-to-congress-on-high-priority-evidence-gaps-for-clinical-preventive-services.

U.S. Preventive Services Task Force. About the USPSTF. Retrieved 5 Sep 2014 from https://www.uspreventiveservicestaskforce.org/about.htm

U.S. Preventive Services Task Force (2016) Final recommendation statement. Breast cancer: screening. Retrieved 8 July 2019 from   https://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/breast-cancer-screening1

Udovcic M, Pena RH, Patham B, Tabatar L, and Abhishek A (2017) Hypothyroidism and the heart. Methodist DeBakey Cardiovasc J 13(2):55-9. Retrieved 6 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5512679/

Van den Berghe G,De Zegher F,Baxter RC,Veldhuis JD,Wouters P,Schetz M,Verwaest C,van der Vorst E,Lauwers P,Bouillon R, and Bowers CY (1998) Neuroendocrinology of prolonged critical illness: effects of exogenous thyrotropin-releasing hormone and its cominbation with growth hormone secretagogues. J Clin Endocrinol Metab 83(2):309-319. Retrieved 29 March 2013 from https://www.njmonline.nl/getpdf.php?id=10000518

Van den Berghe G (2014) Non-thyroidal illness in the ICU: a syndrome with several faces. Thyroid 24(10):1456-65. Retrieved 22 Mar 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4195234/

van de Graaf SAR, Ris-Stalpers C, Pauws, Mendive FM, Targovnik FM and de Vijlder JJM (2001) Up to date with human thyroglobulin. J of Endocrinol 170:307-21. Retrieved 16 Feb 2014 from https://joe.endocrinology-journals.org/content/170/2/307.full.pdf

Van Reeth, O., J Sturis, MM Byrne, JD Blackman, M L'Hermite-Baleriaux, R Leproult, C Oliner, S Refetoff, FW Tuerk and E Van Cauter (1994) Nocturnal exercise phase delays circadian rhythms of melatonin and thyrotropin secretion in normal men. Am J Physiol Endocrinol Metab 266:E964-E974.  Retrieved 29 Dec 2012 from https://ajpendo.physiology.org/content/266/6/E964.full.pdf+html

van Tienhoven-Wind LJN and Dullaart RPF (2015) Low-normal thyroid function and novel cardiometabolic biomarkers. Nutrients 7(2):1352-77. Retrieved 6 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4344592/

Van Nostrand D, Bloom G, and Wartofsky L (2010) Thyroid cancer: a guide for patients. pp 92-3. NH:Keystone Press. Excerpted by Kenneth D. Burman, MD with permission. Retrieved 4 May 2020 from https://www.thyca.org/pap-fol/more/hypothyroidism/

Vanderpump, MP, WM Tunbridge, JM French, D Appleton, D Bates, F Clark, J Grimley Evans, H Rodgers, F Tunbridge, ET Young (1996, abstract only) The development of ischemic heart disease in relation to autoimmune thyroid disease in a 20-year follow-up study of an English community.  Retrieved 30 Jan 2013 from https://www.ncbi.nlm.nih.gov/pubmed/8837320

Vanderpump, MPJ, WMG Tunbridge, JM French, D Appleton, D Bates, F Clark, J Grimley Evans, DM Hasan, H Rodgers, F Tunbridge and ET Young (1995) The incidence of thyroid disorders in the community: a twenty-year follow-up of the Whickham Survey. Clin Endocrinol 43: 55-68.  Retrieved 26 December 2012 from https://www.ieonline.com/cgi-bin/xFer/cg7t462r/quosa/3373850.pdf

Vandevijvere S, Mourri AB, Amsalkhir S, Avni F, Van Oyen H, and Moreno-Reyes R. (2012) Fortification of bread with iodized salt corrected iodine deficiency in school-aged children, but not in their mothers: a national cross-sectional survey in Belgium. Thyroid 22(10):1046-53. Retrieved 5 Sep 2019 from https://www.ncbi.nlm.nih.gov/pubmed/22947351

Varghese S, Montufar-Solis D, Vincent BH, and Klein JR (2008). Virus infection activates thyroid stimulating hormone synthesis in intestinal epithelial cells. J Cell Biochem 105(1):271-6. Retrieved 30 May 2014 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2552408/

Viartis (2013) Parkinson's Disease News. Retrieved 4 Dec 2013 from https://www.viartis.net/parkinsons.disease/news/101128.htm

Viking Therapeutics (2019) Selective thyroid receptor-β agonist.  Retrieved 28 Nov 2019 from https://www.vikingtherapeutics.com/pipeline/vk0214/

Villani J, Ngo-Metzger Q, Vincent IS, and Klabunde C (2018) Sources of funding for research in evidence reviews that inform recommendations of the US Preventive Services Task Force. JAMA 319(20):2132-3. Retrieved 21 Jun 2018 from https://jamanetwork.com/journals/jama/article-abstract/2681728 (first page only including table)

Villicev CM, Freitas FRS, Aoki MS, Taffarel C, Scanlan TS, Moriscot AS, Ribeiro MO, Bianco AC, and Gouveia CHA (2007) Thyroid hormone receptor beta-specific agonist GC-1 increases energy expenditure and prevents fat-mass accumulation in rats. J of Endocrinol 193(1):21-9. Retrieved 28 Nov 2019 from https://joe.bioscientifica.com/view/journals/joe/193/1/1930021.xml

Waggoner, M. (1986) North Carolina's junior senator commits suicide, police say.  Greenville, NC: Associated Press.  Retrieved 27 Dec 2012 from https://www.apnewsarchive.com/1986/North-Carolina-s-Junior-Senator-Commits-Suicide-Police-Say/id-3068cacfc00e4e139d79968a5c167bc7

Walsh JP, Bremner AP, Feddema P, Leedman PJ, Brown SJ, and O'Leary P (2010) Thyrotropin and thyroid antibodies as predictors of hypothyroidism: a 13-year, longitudinal study of a community-based cohort using current immunoassay techniques. J Clin Metab 95(3):1095-104. Retrieved 23 Dec 2019 from https://academic.oup.com/jcem/article/95/3/1095/2596745

Wasserstein RL and Lazar NA (2016) The ASA's statement on p-values: context, process, and purpose.  Am Stat 70(2):129-33. Retrieved 5 July 2018 from https://www.tandfonline.com/doi/full/10.1080/00031305.2016.1154108

Werneck de Castro JP, Fonseca TL, Ueta CB, McAninch EA, Abdalla S, Wittmann G, Lechan RM, Gereban B, Bianco AC (2015) Differences in hypothalamic type 2 https://academic.oup.com/jcem/article/95/3/1095/2596745 deiodinase ubiquitination explain localized sensitivity to thyroxine. J Clin Invest (Epub ahead of print). Retrieved 16 Jan 2015 at https://www.ncbi.nlm.nih.gov/pubmed/25555216 (abstract only)

Wilkinson J, Bass C, Diem S, Gravley A, Harvey L, Maciosek M, McKeon K, Milteer L, Owens J, Rothe P, Snellman L, Solberg L, Vincent P. Preventive services for adults. Updated Sdeptember 2013.  Retrieved 13 Oct 2014 from  https://www.icsi.org/_asset/gtjr9h/PrevServAdults.pdf

Witherspoon, LR (2005) Clinical utility of sensitive TSH measurements. Lab Med. 36(11):711-5.  Retrieved 28 Dec 2012 from https://www.medscape.com/viewarticle/515755

World Health Organization (2007) Salt as a vehicle for fortification: report of a WHO expert consultation. Retrieved 3 Oct 2013 from https://whqlibdoc.who.int/publications/2008/9789241596787_eng.pdf

Wu, P (2000) Practical pointers: thyroid disease and diabetes. Clinical Diabetes 18(1):38.  Retrieved 16 Jan 2013 from https://journal.diabetes.org/clinicaldiabetes/v18n12000/pg38.htm

Xu L, Ma H, Miao M, and Youming L (2012) Impact of subclinical hypothyroidism on the development of non-alcoholic fatty liver disease: A prospective case-control study. J Hepatol 57(5):1153-4. Retrieved 22 Feb 2018 from https://www.journal-of-hepatology.eu/article/S0168-8278(12)00448-5/fulltext

Yalow RS and Berson SA (1960) Immunoassay of endogenous plasma insulin in man. J Clin Invest 39:1157-75. Retrieved 5 Jan 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC441860/pdf/jcinvest00455-0122.pdf

Yan F, Wang Q, Lu M, Chen W, Song Y, Jing F, Guan Y, Wang L, Lin Y, Bo T, Zhang J, Wang T, Xin W, Yu C, Guan Q, Zhou X, Gao L, Xu C, and Zhao J (2014) Thyrotropin increases hepatic triglyceride content through upregulation of SREBP-1c activity. Hepatol 61(6):1358-64.  Retrieved 16 Jan 2019 from https://www.ncbi.nlm.nih.gov/pubmed/25016220

Yehuda-Shnaidman E, Kalderon B, Bar-Tana J (2005) Modulation of mitochondrial transition pore components by thyroid hormone. Endocrinology 146(5):2462-72.  Retrieved 6 Dec 2019 from https://academic.oup.com/endo/article/146/5/2462/2500461

Yoshimura M, Hershman JM, Pang XP, Berg L, Pekary AE (1993) Activation of the thyrotropin (TSH) receptor by human chorionic gonadotropin and luteinizing hormone in Chinese hamster ovary cells expressing functional human TSH receptors. J Clin Endocrinol Metab 77(4):1009-13. Retrieved 29 Jul 2013 from https://www.ncbi.nlm.nih.gov/pubmed/7691861 (abstract only)

Zeng Q, Cui YS, Zhang L, Fu G, Hou CC, Zhao L, Wang AG, Liu HL (2012) Studies of fluoride on the thyroid cell apoptosis and mechanism. Zhonghua Yu Fang Yi Xue Za Zhi 46(3): 233-6. Retrieved 21 Aug 2013 from https://www.ncbi.nlm.nih.gov/pubmed/22800594 (abstract only)

Zhao T, Chen B, Zhou Y, Wang X, Zhang Y, Wang H, and Shan Z (2017) Effect of levothyroxine on the progression of carotid intima-media thickness in subclinical hypothyroidism patients. BMJ Open 7(10):e16053.  Retrieved 21 April 2020 from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5665330/

Zhao W, Zhu H, Yu Z, Aoki K, Misumi J, Zhang X (1998) Long-term effects of various iodine and fluorine doses on the thyroid and fluorosis in mice. Endocri Regul 32(2):63-70. Retrieved 23 Aug 2013 from https://www.ncbi.nlm.nih.gov/pubmed/10330519 (abstract only)

Zhou J, Waskowicz LR, Lim A, Liao XH, Lian B, Masamune H, Refetoff S, Tran B, Koeberl DD, And Yen PM (2019) A liver-specific thyromimetic, VK2809, decreased hepatosteatosis in glycogen storage type 1a. Thyroid 29(8):1158-67.  Retrieved 28 Nov 2019 from https://www.ncbi.nlm.nih.gov/pubmed/31337282 (abstract only)

Zimmer KP, Scheumann GF, Brämswig J, Böcker W, Harms E, and Schmid KW (1997) Ultrastructural localization of IgG and TPO in autoimmune thyrocytes referring to the transcytosis of IgG and the antigen presentation of TPO. Histochem Cell Biol 107(2):115-20. Retrieved 11 Oct 2013 from https://www.ncbi.nlm.nih.gov/pubmed/9062796 (abstract only)

Zimmermann, MB, Y. Ito, SY Hess, K Fujieda and L Molinari (2005) High thyroid volume in children with excess dietary iodine intakes. Am J Clin Nutr 81:840-4. Retrieved 21 Jan 2013 from https://ajcn.nutrition.org/content/81/4/840.full.pdf+html

Zöphel, K., G. Wunderlich and J. Kotzerke (2006) Should we really determine a reference population for the definition of thyroid-stimulating hormone reference interval? Retrieved (actually, still available) 15 Oct 2019 from https://www.clinchem.org/content/52/2/329.full

Zuger A (2021) Tallying the waste of American healthcare: Worst case estimates suggest we might waste half the $3.6 trillion we spend on our health annually.  NEJM Journal Watch 41(2):28. Retrieved 17 Feb 2021 from https://www.jwatch.org/na52991/2021/01/05/tallying-waste-american-healthcare

Zulewski H, Müller B, Pascale E, Miserez AR, Staub J-J (1997) Estimation of tissue hypothyroidism with a new clinical score evaluation of patients with various grades of hypothyroidism and controls. J Clin Endocrinol Metab 82(3):771-6. Retrieved 16 Dec 2019 from https://academic.oup.com/jcem/article/82/3/771/2656260