Answers are grounded in this paper and cite their sources.
Subclinical thyroid dysfunction is a common laboratory diagnosis characterized by an abnormal serum TSH concentration with normal free T4 and T3 levels. Subclinical hypothyroidism is especially common, affecting about 4% of the general population in the United States, with much higher frequencies in older patients, whereas subclinical thyrotoxicosis affects less than 1% of the population (1, 2). Refinements in serum TSH measurements and ubiquitous laboratory testing during the past few decades have resulted in a sharp increase in the diagnosis of thyroid dysfunction, mirroring the tendency toward early diagnosis of many other conditions such as hyperlipidemia for which early treatment appears beneficial (3). Over the past 10 yr, there has been an explosion of published data regarding subclinical thyroid dysfunction and its clinical relevance, which has provoked the inevitable debates about screening and treatment. Many organizations have created guidelines, which are sometimes in conflict, in an attempt to help practicing clinicians determine what tests to obtain, when and in whom to get them, and when to treat patients when the tests are abnormal. Because of the astounding amount of time and effort directed toward this common condition, and because of the difficulties in interpreting data from many different sources, the American Association of Clinical Endocrinologists, The Endocrine Society, and the American Thyroid Association took the laudable step of creating a panel that included practicing thyroid specialists, cardiologists, statisticians, epidemiologists, and general physicians, some of whom had experience with similar analyses, to carefully evaluate the evidence and make recommendations based on the information available as of September 2002 (4). Nearly simultaneously, the U.S. Preventive Services Task Force (USPSTF) updated their recommendations regarding screening and treating thyroid dysfunction (5). Applying evidence-based methodologies, the society-sponsored panel evaluated the data and created a consensus document approved by the group, which represents a balance between the views of the epidemiologists and practicing clinicians on the panel. When evaluating their recommendations, it is important to remember several things about the panelists and their task: 1) many were not thyroidologists and were thus not biased by previous positions on the subject; 2) they were specifically asked to determine whether routine screening of asymptomatic patients was justified; and 3) they recognized that most patients with thyroid dysfunction are never seen by endocrinologists or thyroidologists. The USPSTF criteria were used to rate the strength of the evidence as it related specific outcomes to thyroid status or benefits of treatment. A positive recommendation was made if there was evidence that the condition is associated with adverse clinical outcomes that were reversed by treatment, the benefits of which outweighed the risks. When evidence was not available or was contradictory or insufficient, the members of the societies-sponsored panel relied on their experience, judgment, and interpretation of the literature in formulating recommendations for clinical practice. Differences of opinion were settled by a majority vote. The panelists concluded that the evidence supports treating patients with normal free T4 and T3 levels but with TSH concentrations below 0.1 mIU/liter or above 10 mIU/liter, whether or not the patient was symptomatic. This was based on “fair” evidence that pituitary TSH alterations of this magnitude are associated with clinical alterations that can be reversed by treatment. However, serum TSH levels among the asymptomatic population usually are not higher than 10 mIU/liter or lower than 0.1 mU/liter. The panel found good evidence that the progression to overt hypothyroidism is likely if there is any elevation in serum TSH (i.e. >4.5 mIU/liter). Moreover, it acknowledged that early levothyroxine therapy may prevent symptoms and signs in those who do progress to overt disease. Still, the panel’s conclusion was that the available data do not confirm clear-cut benefits for early therapy. This conclusion was reached on the basis of a risk-benefit analysis of treatment in which they found fair evidence that overtreatment with levothyroxine was common, perhaps more so than most of us would like to acknowledge, and might impact negatively on outcome. This was based on fair evidence that TSH levels below 0.1 mIU/liter cause clinically significant disease such as atrial fibrillation and osteoporosis and that normal TSH ranges change in different clinical situations (e.g. pregnancy) that often result in treatment errors. The importance of this concern in shaping the panel’s recommendations is confirmed by the commentary by Surks (6) in the current issue of JCEM. The panel thus concluded that routine screening was not justified for the general population. In other words, the evidence for therapeutic benefit and, therefore, screening did not clearly outweigh the potential risks of therapy, even for populations at high risk of having the disorder (e.g. older women) or those with special considerations (e.g. pregnant women). This leaves major unanswered questions concerning the potential consequences of delayed therapy for high risk patients, particularly the fetuses of pregnant women with abnormal TSH levels, women with osteoporosis, or older persons at increased risk for atrial fibrillation. In such individuals, the panel advocated aggressive case finding (defined as testing for a condition or disease other than the one for which a person is consulting a clinician), which in our opinion means clinical judgment based on a broad variety of circumstances, including the patient’s symptoms, demographics, potential for thyroid dysfunction, and general health. There are some inconsistencies in the report. For example, in sharp contrast to the recommendation not to screen, the panel recommended that if an elevated serum TSH level is (somehow) found in pregnant women or those of childbearing potential planning to become pregnant, they should be treated with levothyroxine to restore the TSH to normal. This recommendation was based on the possible association between high TSH and either increased fetal wastage or subsequent neuropsychological complications occurring in the offspring due to thyroid hormone insufficiency. The panelists acknowledged that although there are no published intervention trials that assess the benefits of thyroid hormone replacement in this special population, the potential benefit-risk ratio of treatment justifies its use. It underscores a point that the panel did indeed bring clinical judgment to bear on their conclusions. Yet it is also clear acknowledgment that without screening, subclinical hypothyroidism will be missed in some patients who will fare less well without levothyroxine therapy. It also acknowledges that one must accept the notion that it is more likely than not that levothyroxine treatment may be safely administered by a knowledgeable physician treating a compliant patient in such circumstances. A similar and nearly simultaneous analysis of the literature was performed by the USPSTF, a group charged with evaluating the usefulness of screening tests in the general population (5). Their analysis, conducted by nonendocrinologists, showed that the evidence is inconclusive for efficacy of treatment of patients with subclinical thyroid dysfunction found by screening. They found no formal treatment trials of subclinical hyperthyroidism with only several small, randomized trials of the treatment of subclinical hypothyroidism showing inconclusive results. They found the data on the adverse effects of levothyroxine therapy under these circumstances to be sparse and thus inconclusive. Otherwise their interpretation was similar to that of the societies-sponsored panel. In the current issue of the Journal of Clinical Endocrinology and Metabolism, Gharib et al. (7) present the view of the three societies that sponsored the original consensus panel. In this statement, they agree on some issues with the panel they sponsored, but disagree with their conclusions. They recommend 1) routine screening for subclinical thyroid disease in the general population; 2) routine screening for subclinical thyroid disease in women who are pregnant or planning pregnancy; and 3) routine treatment of patients with subclinical hypothyroidism with serum TSH levels of 4.5–10 mU/liter. The main thrust of their argument is that they believe that the panel’s negative recommendations are inappropriate because they are based primarily on a lack of evidence for benefit rather than evidence for a lack of benefit. This is not a new argument against evidence-based medicine. They feel there is fair evidence to favor therapy for patients with subclinical hypothyroidism and poor evidence of risk of therapy, thus leading to recommendations to screen for thyroid disease in patients at greatest risk of having an abnormality or those in whom delayed recognition of the disorder could potentially lead to clinically significant consequences. So, with the two analysis groups (the societies-sponsored panel and USPTF) coming to similar conclusions, why is there continued controversy? Why is there no consensus between the consensus sponsors and their panel? And, most importantly, what is the “take home” message for primary care providers and specialists who see patients at risk for an abnormal TSH level? The reason for the disagreement seems clear, using definition 2b (see above) for consensus: the recommendations are an amalgamation of the views of the members of the various groups and panels, each of which includes individuals with different experiences, roles, training background, and personal biases. The recommendations are logically based on the individual groups’ assessments of the data (i.e. fair vs. poor, no one seems to feel the evidence is “good”) and balancing those interpretations with clinical experience and judgment. However, the recommendations of the societies-sponsored consensus panel are anathema to the very organizations that represent clinical endocrinologists. This is a clear reflection of the “high bar” of evidence-based medicine Cochrane analysis of data (8, 9). When the data are not of high enough quality to clear that high bar, one must resort to clinical acumen and become what we all strove to be in medical school, “the cagey clinician” — identifying disease at an early stage and treating it to prevent poor outcomes, but balancing this desire with the dictum of primum non nocere, above all do no harm. The various recommendations and guidelines remind us that a consensus based on group solidarity and sentiment (definition 1, above) is possible between all of these groups only if the scientific data correlate with widely held clinical opinions. If there is disagreement with the estimates of treatment risk and thus the benefit of screening, and if a consensus is important to patient management, then the onus is on clinician-scientists to truly determine whether the risk of treatment is lower than the published data suggest. If there is concern regarding normal ranges and targets for therapy in pregnancy, then these reference ranges should be created based on carefully constructed reference ranges. It appears that these steps must be taken for all to agree that general population screening is mandated. For now, however, it remains our responsibility to be thoughtful clinicians, advocating for patients in the best manner to detect real disease early, to treat appropriately and safely, to choose to screen or not to screen all patients in our practices, and to treat and not to treat when it is in the best interest of our patients’ health. This is the clinician’s view. The old saw that absence of evidence is not evidence of absence supports the clinician’s view. Epidemiologists, on the other hand, have a more global view of the problem, citing wasted resources, flawed therapy, and poorly defined outcomes as a reason to withhold treatment. This dilemma is a clarion for further clinical research that fulfills the epidemiologists’ high standards: prospective long-term randomized studies aimed at evaluating the efficacy of treatment of subclinical thyroid dysfunction. Until this occurs, we will continue to vigorously debate this issue.
No takes yet. Share an insight, caveat, or question.
Ringel et al. (2005) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: