Levothyroxine has been among the top three most-prescribed medications in the United States for over a decade, with annual prescription volume exceeding 100 million fills. A substantial fraction of these prescriptions are written for patients whose biochemical thyroid status falls within or just outside the population reference range for thyroid-stimulating hormone, whose symptoms are nonspecific and arguably could be explained by a dozen other conditions, and for whom the clinical trial evidence supporting thyroid hormone replacement is weak. The pattern has been documented repeatedly in the endocrine and primary care literature for at least twenty years. Practice has barely changed.
The clinical category at the center of this pattern is subclinical hypothyroidism—elevated TSH with normal free T4, generally with no specific clinical features. The category is itself contested. The TSH reference range has shifted over the past two decades as population data has accumulated, with ongoing debate about whether the upper limit of normal should be 4.5, 4.0, 2.5, or somewhere else depending on age, pregnancy status, and various other factors. A patient whose TSH is 5.5 may be hypothyroid by one set of reference standards and normal by another, and the clinical decision about treatment depends partly on which standards the prescribing clinician learned during training.
The clinical trial evidence on subclinical hypothyroidism treatment, when it has been generated, has been largely null. The TRUST trial, published in NEJM in 2017, randomized older adults with subclinical hypothyroidism to levothyroxine or placebo and found no difference in symptom scores after one year. The trial’s findings were robust within their population. They were also nearly invisible to the prescribing patterns that continued essentially unchanged. Subsequent meta-analyses have generally confirmed that thyroid hormone replacement in subclinical hypothyroidism produces no measurable improvement in symptoms, quality of life, or cardiovascular outcomes for most patient populations.
Why the prescribing pattern persists despite the evidence is a question with several layered answers. The diagnosis of subclinical hypothyroidism is, in primary care, often a diagnostic placeholder for symptom complexes that the encounter format does not adequately address—fatigue, weight gain, low mood, sleep disruption. Prescribing levothyroxine produces a measurable lab response over weeks (TSH normalizes), patient-reported symptom improvement that is often substantial regardless of whether it reflects pharmacological effect, and an ongoing therapeutic relationship that the patient and clinician both find satisfying. The clinical evidence that the improvement is non-specific is widely available; the evidence does not, in practice, govern the prescribing decision because the prescribing decision is responding to a different pressure than the evidence is meant to address.
There is a parallel question about thyroid hormone replacement in patients whose TSH is normal but who present with hypothyroid-spectrum symptoms. The clinical category sometimes called ’tissue hypothyroidism’ or, in alternative medicine framings, as ‘Wilson’s syndrome’ has no widely accepted clinical or biochemical definition but has nonetheless produced a substantial body of off-label prescribing—often involving combination T4/T3 therapy, sometimes natural desiccated thyroid extract, occasionally even more elaborate protocols. The clinical evidence for these interventions is, in most cases, considerably weaker than the evidence for treating subclinical hypothyroidism. The patient demand persists nonetheless.
What complicates the clinical analysis is that a small subset of patients does, genuinely, appear to derive benefit from thyroid hormone replacement at biochemical thresholds that the trial evidence does not support. Whether this reflects subgroup effects that the trials have been underpowered to detect, placebo response that is robust enough to be clinically meaningful regardless of mechanism, or clinical anecdote that has not been systematically validated is genuinely difficult to determine. The endocrinologists who treat these patients tend to credit the clinical experience. The trialists who designed the studies tend to credit the trial findings. Both positions have evidence behind them; the synthesis has not been achieved.
The combination T4/T3 question deserves separate discussion because it has produced a peculiar regulatory pattern. American Thyroid Association guidelines have generally not recommended combination T4/T3 therapy outside of specific clinical situations, citing the absence of trial evidence for benefit over T4 monotherapy. A subset of clinicians, however, has continued to prescribe combination therapy on the basis of clinical experience and patient preference. The mainstream prescription products for combination therapy are limited; many patients receive combination therapy through compounded preparations from compounding pharmacies, which has the consequence of moving a substantial volume of thyroid hormone prescribing outside the conventional manufacturer supply chain.
Natural desiccated thyroid extract, derived from porcine thyroid tissue, occupies an even more unusual regulatory position. The product is regulated as a drug, with several FDA-approved formulations available, but it is also widely compounded. Patient preference for natural desiccated thyroid over synthetic levothyroxine is, in some patient populations, intense and not easily addressed by clinicians who consider it clinically equivalent or inferior. The market for natural desiccated thyroid has grown over the past decade alongside broader trends in patient preference for ‘natural’ formulations, and the compounded versions of the product have been a particular focus of growth despite the availability of FDA-approved alternatives.
There is a substantial gender dimension to the thyroid prescribing pattern that the field has engaged unevenly. Women receive thyroid hormone prescriptions at rates roughly four times higher than men, partly reflecting genuinely higher rates of thyroid pathology in women, partly reflecting the population’s higher engagement with primary care, and partly reflecting that the symptom complexes that drive thyroid hormone prescribing—fatigue, weight changes, mood symptoms—are more commonly attributed to thyroid pathology in female patients than in male patients. JAMA Internal Medicine analysis has documented these gender patterns and noted the contribution of clinical training and cultural assumption to the prescribing decisions. The implication that some fraction of female-coded thyroid prescribing reflects symptom dismissal in other clinical domains is uncomfortable but probably accurate.
The downstream consequences of widespread levothyroxine prescribing are mostly modest but not zero. Patients on chronic thyroid hormone replacement require periodic monitoring, dose adjustments, and lifelong medication management for what may not have been a clinical problem requiring treatment. The medication itself is inexpensive at generic prices, but the surrounding clinical infrastructure—office visits, lab work, occasional specialist consultation—generates substantial system-level cost over time. The patients receiving unnecessary thyroid hormone replacement are, in addition, exposed to the small but real risks of overtreatment, including atrial fibrillation, accelerated bone loss, and cardiovascular effects that accumulate with prolonged exposure to supraphysiologic thyroid hormone levels.
What the thyroid prescribing pattern has resisted, despite repeated efforts at clinical guideline communication, is the reduction in low-value prescribing that the evidence would support. The American Society for Clinical Pathology’s Choosing Wisely list has, for over a decade, recommended against routine treatment of mild subclinical hypothyroidism. The recommendation has had measurable effect on prescribing patterns but not nearly the effect that its clinical justification would predict. The recommendation has been incorporated into specialty society guidelines, into electronic health record decision support, and into medical education curricula, with prescribing patterns continuing largely as they were.
What the thyroid case demonstrates, perhaps more clearly than any other example in primary care medicine, is the limit of evidence-based medicine when the evidence runs against patterns of clinical practice that have institutional momentum. The evidence is clear, the guidelines are aligned, the educational infrastructure has communicated the recommendations, and the prescribing pattern continues. The continuation reflects, in part, that the patients want the prescriptions, that the clinicians find the prescribing satisfying, that the alternatives—careful explanation of normal lab values to dissatisfied patients, multidisciplinary evaluation of symptom complexes that don’t fit specific diagnostic categories, primary care visit formats that allow for genuine therapeutic engagement with non-specific complaints—are more demanding than the existing prescribing model. The thyroid hormone prescription is, in many cases, what the system has available when it has nothing else to offer. Whether the field can produce something else to offer, and whether the prescribing pattern would change if it could, is a question the next generation of primary care reform will have to engage. The current generation has not.
