In this issue, Helfand and Redfern [2] have made an effort to develop guidelines to assist clinicians when they encounter patients with subclinical thyroid disease. Their recommendations for checking a serum TSH level in patients who come to physicians for other reasons are consistent with those of the American Thyroid Association [3] and the American College of Pathology [4]. Given the high prevalence of thyroid disease in women older than 50 years of age, case finding in this population may be justifiable. More problematic for Helfand and Redfern is what to do when subclinical thyroid disease is found. Unfortunately, their literature review is incomplete; thus, they did not incorporate pertinent clinical information that could have been brought to bear on the subject. I review their recommendations for subclinical hyperthyroidism and subclinical hypothyroidism and attempt to add a clinical perspective to their views. In each section, I address three questions: 1) Do patients have symptoms attributable to their disease? 2) Is there evidence of organ system involvement and potential morbidity? 3) Is treatment beneficial and worth the cost and risk?

Subclinical Hyperthyroidism

Helfand and Redfern conclude that the role of early diagnosis and treatment is unclear. I agree that the clinical impact of subclinical hyperthyroidism is still not fully known, but many important clinical studies have not been taken into consideration in their analysis.

Do Patients with Subclinical Hyperthyroidism Have Symptoms?

Helfand and Redfern state that "patients found by screening to have an undetectable serum TSH level have not been shown to have more symptoms than age- and sex-matched controls." Although this is technically true in screened populations, some patients with subclinical hyperthyroidism do have symptoms. Stott and colleagues [5] observed that elderly persons with subclinical hyperthyroidism had mild symptoms on the Wayne index (a clinical index of hyperthyroidism) that differed significantly from those of age-matched euthyroid controls. Furthermore, elderly hyperthyroid patients, even those with relatively severe thyrotoxicosis, typically have few symptoms; the most extreme example of this phenomenon is apathetic thyrotoxicosis. Therefore, the absence of symptoms is not really relevant to the discussion.

Is There Evidence of Organ System Involvement and Potential Morbidity?

Helfand and Redfern state that "no randomized trials of early treatment for subclinical hyperthyroidism to prevent [osteoporosis] have been done." However, Mudde and coworkers [6] prospectively followed 16 postmenopausal women with subclinical hyperthyroidism. Eight were randomly assigned to receive methimazole to normalize thyroid function by normalizing TSH levels, and 8 did not receive therapy. After 2 years, distal forearm bone density had increased slightly in the treatment group and had declined by 5% in the control group, a statistically significant difference. A recently published, nonrandomized prospective study reported similar findings in the spine and hip [7]: Over 2 years, 16 women with subclinical hyperthyroidism who received radioiodine therapy had an increase in bone density of approximately 2% at both sites, whereas 12 women who did not receive therapy had a 5% loss in the spine and a 2% loss at the hip.

Helfand and Redfern cite epidemiologic data from the Framingham study associating subclinical hyperthyroidism and atrial fibrillation [8]. However, they do not refer to several studies showing higher heart rate, increased frequency of atrial premature beats, and greater left ventricular mass in patients with thyroxine-induced subclinical hyperthyroidism [9] that was reversible with ß-adrenergic blocking agents [10]. Although a recent study did not substantiate these findings, except for increased left ventricular mass [11], these data provide some support for the potentially deleterious cardiac effects of subclinical hyperthyroidism. There is no reason to believe that the cardiac effects of endogenous thyroid hormone would differ from those of exogenous thyroid hormone; indeed, most subclinically hyperthyroid patients who developed atrial fibrillation in the Framingham study cohort were receiving thyroxine.

Although the potential benefits of therapy can be debated, I submit that the discovery of a low serum TSH level requires an evaluation that Helfand and Redfern seem to ignore. The possible causes of a low TSH level include early Graves disease, a solitary autonomous nodule, a multinodular goiter, various forms of self-limited thyrotoxicosis (postpartum, silent, and subacute thyroiditis and iodine-induced hyperthyroidism), and central hypothyroidism. Furthermore, some patients with apparent subclinical hyperthyroidism have elevated serum free triiodothyronine levels and normal levels of circulating free thyroxine and total triiodothyronine [12], placing them in the category of overt hyperthyroidism. Don't Helfand and Redfern believe that these conditions warrant diagnosis, even if treatment is not given? They also state that "it is not clear that monitoring these patients with thyroid function tests would prevent the spontaneous development of overt hyperthyroidism." However, hyperthyroidism is generally insidious in onset, and although some patients' test results revert to normal with follow-up, it is my belief and my clinical experience that morbidity is decreased when progression to overt disease is detected at an earlier stage.

Is Treatment Beneficial and Worth the Cost and Risk?

Although no randomized clinical trials of treatment for subclinical hyperthyroidism have been done, apart from the small study by Mudde and coworkers [6], expert opinion does exist. A case for the treatment of subclinical hyperthyroidism has been made for elderly women with osteoporosis [13], and in older persons who have known coronary artery disease or who are at high risk for this condition, treatment is justifiable [14]. I agree that more data are needed, but I also believe that in these high-risk groups, the benefits outweigh the potential risks and costs. Is it worth trying to prevent an embolic stroke or hip fracture?

Subclinical Hypothyroidism

Helfand and Redfern review the extensive literature on the prevalence of subclinical hypothyroidism and the rate of conversion to overt hypothyroidism, which is highly but not exclusively dependent on the presence of circulating antithyroid antibodies. They then examine the results of three small, prospective controlled trials of thyroxine therapy, as well as other data on the effects of subclinical hypothyroidism on serum lipids. After noting the potential deleterious effects of excessive thyroxine, they conclude that the "evidence is not sufficient to recommend for or against treatment for subclinical hypothyroidism." Again, for unclear reasons, they ignored pertinent data that should help in clinical decision making in the absence of large controlled trials.

Do Patients with Subclinical Hypothyroidism Have Symptoms?

Although Helfand and Redfern acknowledge that patients found by screening to have subclinical hypothyroidism have at least one symptom that could potentially be caused by hypothyroidism, they stress that older patients have symptoms no more frequently than age-matched controls. They do not cite the study by Staub and associates [15] that clearly shows significant changes in a clinical symptom index (the Billewicz scale) in subclinically hypothyroid women (mean age, 50 years) compared with age-matched controls. In a more recent report, this group again noted significant differences in symptoms between subclinically hypothyroid patients and controls and correlated the symptom score with serum free thyroxine and TSH levels [16].

Is There Evidence of Organ System Involvement and Potential Morbidity?

Helfand and Redfern summarize the literature on abnormal lipid values and the risk for cardiovascular disease in subclinical hypothyroidism. However, they do not comment on studies of subclinical hypothyroidism and lipoprotein(a) levels, an independent risk factor for atherosclerosis. One report found no differences in lipoprotein(a) levels in patients with subclinical hypothyroidism compared with controls [17], but two others found higher serum levels in the former group [18, 19]. In one study, the lipoprotein(a) level returned to normal with thyroxine therapy [19]. Helfand and Redfern also do not cite a study showing that smokers with subclinical hypothyroidism (who are already at higher risk for cardiac events) have significantly more deranged lipid patterns than similarly affected nonsmokers do [20]; this suggests that smokers with subclinical hypothyroidism may benefit more from thyroxine therapy. Finally, a recent coronary angiographic study found greater progression of lesions in hypothyroid patients with TSH levels in the range seen in subclinical hypothyroidism compared with patients whose TSH levels were assiduously maintained in the normal range [21].

Helfand and Redfern also fail to mention other potential complications of subclinical hypothyroidism. For example, a group of patients with a mean serum TSH level of 8.8 mU/L had significant alterations in memory and mood, including anxiety, depression, and somaticism, all of which returned to normal with thyroxine therapy [22]. In another study [23], approximately 30% of patients with subclinical hypothyroidism were found to have elevated intraocular pressure with reversal after institution of thyroxine therapy. Of less certain clinical significance are electrophysiologic changes in peripheral nerve function in subclinical hypothyroidism [24], abnormalities in skeletal muscle energetics in patients with subclinical hypothyroidism and musculoskeletal symptoms [25], and significant decreases in intestinal transit time if hypothyroid patients stop thyroxine therapy for 1 week when thyroid function is in the range seen in subclinical hypothyroidism [26]. Other possible effects on reproductive function and depression have been reviewed [27].

Is Treatment Beneficial and Worth the Cost and Risk?

Helfand and Redfern state that hundreds of patients would have to be treated for one to benefit. I contend that they have not assessed all of the data currently available and that evidence is sufficient to warrant treatment with thyroid hormone in most patients with subclinical hypothyroidism. I believe that the potential benefits outweigh the minimal risks of thyroxine therapy in most patients, even when they have no obvious symptoms, no dyslipidemia, and negativity for circulating antithyroid antibodies. My opinion is generally shared by others in the field [13, 27] and the American Thyroid Association [28]. The presence of a serum TSH level of 10 mU/L or more or positivity for antithyroid antibodies, abnormal serum lipid levels, a history of smoking, or symptoms when the TSH level is less than 10 mU/L makes the decision to begin therapy more compelling.

The correct approach to subclinical thyroid disease is far from settled. However, a probabilistic "evidence-based" strategy may not assist in the management of an individual patient who is sitting in an individual physician's office and has a TSH level that is undetectable or slightly above normal. Not all relevant studies were reviewed by Helfand and Redfern, and small studies that failed to show positive results may have been underpowered. Although the call for better, larger, randomized, placebo-controlled trials is intellectually satisfying, it only begs the question because it is unlikely that such trials will be initiated. Therefore, in the absence of authoritative answers, physicians must do what they have always done: sift through all of the available evidence, consider expert opinion, discuss the potential risks and benefits with the patient, and try to make an informed decision. For subclinical hyperthyroidism, further diagnostic testing is warranted; therapy with ß-adrenergic blocking agents, antithyroid drugs, or radioiodine may be indicated in patients with known or suspected heart disease or osteoporosis. For subclinical hypothyroidism, I believe that the evidence is adequate to justify therapy in most patients.