Screening for Thyroid Disease: An Update

1.0 Introduction

In 1990, the American College of Physicians published guidelines [1] and a background paper [2] supporting the appropriate use of thyroid function tests in symptomatic women older than 50 years of age who sought care in the primary care setting. The guidelines recommended against screening for clinically inapparent “subclinical” or “mild” thyroid failure because the evidence that earlier detection and treatment improves health outcomes in these patients was insufficient.

1.1 Since 1990, new information about the long-term risks of untreated subclinical thyroid dysfunction has led some experts to support screening in the general population. The sensitive thyroid-stimulating hormone (TSH) test has been proposed as a good screening tool because it can identify persons who have abnormal thyroid function before other clinical or laboratory abnormalities occur. Screening is done with a sensitive TSH test followed by a test of free thyroxine if the result of the sensitive TSH test is abnormal. If the level of TSH is undetectable and the free thyroxine level is normal, a free triiodothyronine test is also done.

1.2 As shown in Table 1, this approach to screening identifies four groups of patients with abnormal thyroid function. Patients with overt hyperthyroidism have an undetectable TSH level and a high free thyroxine or free triiodothyronine level. Patients with overt hypothyroidism have an elevated TSH level and a low free thyroxine level. Patients with subclinical hyperthyroidism have an undetectable TSH level and normal free thyroxine and free triiodothyronine levels. Finally, patients with subclinical hypothyroidism have an elevated TSH level and a normal free thyroxine level.

1.3 Controversy centers on whether early treatment or close follow-up is warranted in apparently healthy persons in whom the only indication of a thyroid disorder is an abnormal test result. Screening is warranted if the evidence meets three criteria. First, an abnormal test result should be associated with future complications that affect health status and quality of life. Second, follow-up and early treatment should improve outcomes by reducing the risk for these complications. Third, these benefits must clearly outweigh the adverse effects and long-term risks of treatment.

1.4 In this paper, we consider whether primary care physicians should screen for thyroid function in patients who are seen in general medical practice, who have no specific indication for thyroid testing, and who come to the physician for other reasons. For each condition in Table 1, we appraised the evidence supporting early treatment and estimated the number of persons who could benefit from screening. We focus on whether screening should be aimed at detection of subclinical thyroid dysfunction and on whether persons who have mildly abnormal TSH levels can benefit.

2.0 Methods

2.1 We reviewed studies of screening for thyroid dysfunction and studies of the complications, course, or treatment of subclinical hyperthyroidism and subclinical hypothyroidism. We defined screening as “the application of a test to detect a potential disease or condition in a person who has no known signs or symptoms of that condition at the time the test is done” [3]. Studies of screening can be classified according to the setting in which the decision to screen takes place [3]. In office-based screening, or case finding, testing for thyroid dysfunction is performed in patients who come to their physicians for unrelated reasons. Studies of case-finding programs provide the most realistic estimates of the effects and costs of screening in clinic or office practice [2]. Population-based studies of screening use special methods to recruit, contact, and follow patients in the context of an epidemiologic research effort. Such studies show the extent of unsuspected thyroid disease in a population sample in a particular geographic area but do not reflect the yield or costs of screening in practice. We used population-based studies of screening as a benchmark against which the yield and benefits of more practical office-based screening programs can be measured.

2.2 Data Sources

We reviewed the original literature from 1989 to 1996 to identify articles that were not included in the original College background paper [2]. In a MEDLINE search, the Medical Subject Headings thyroid function tests and thyroid diseases were combined with the term mass screening and the text words screening or case finding. We conducted a separate MEDLINE search (1970 to 1996) for controlled studies of the effect of thyroid-directed treatments on potential complications of subclinical thyroid disease (symptoms, hyperlipidemia, coronary heart disease, atrial fibrillation, osteoporosis, and bone fracture). Periodic hand searching of endocrinologic and major medical journals, review of the reference lists of retrieved articles, and retrieval of articles reviewed in the first background paper [2] supplemented the MEDLINE searches.

2.3 Study Selection

We selected studies in which a test of serum TSH or thyroxine was performed in the general adult population, in a demographic segment of the adult population, or among patients seen in the general office setting. All controlled studies of treatment in patients with subclinical hypothyroidism or subclinical hyperthyroidism were also included regardless of whether they were performed in patients identified by screening. Studies of screening for congenital or familial thyroid disorders and studies of screening in inpatients; institutionalized patients; and patients known to have diabetes, depression, obesity, or other conditions were excluded.

2.4 Thirty-three screening studies were included. The MEDLINE search identified 56 articles published since 1989, 18 of which were included in the review. Of the 38 excluded studies, 9 concerned inpatients or institutionalized patients and 23 did not report the age and sex of the participants or the results of thyroid function tests; in the latter group, most studies concerned screening for thyroid cancer or familial thyroid disorders. Twenty-three controlled studies of treatment for subclinical thyroid dysfunction were also included in the review.

2.5 Data Extraction

For each condition listed in Table 1, we sought to answer three questions.

1. Prevalence. How many persons have the condition?

2. Burden of illness. What are the risks for and severity of each potential complication?

3. Effectiveness. Is there evidence that early treatment reduces the burden of illness?

2.6 We pooled information from population-based and office-based studies of screening to estimate the prevalence of each type of thyroid dysfunction (Table 1). To assess the burden of illness, we pooled longitudinal studies of patients with subclinical hypothyroidism or subclinical hyperthyroidism to estimate how often patients develop the following complications: 1) symptoms, 2) progression of subclinical to overt dysfunction, 3) elevated cholesterol in hypothyroidism, 4) atrial fibrillation in hyperthyroidism, and 5) osteoporosis in hyperthyroidism. We used a variance-weighted equal effects model [4] implemented in FAST*PRO software for meta-analysis [5] to combine information from observational studies.

2.7 To assess effectiveness, we reviewed randomized trials and other comparison studies that contained information about the effect of treatment on each complication. For each controlled trial, we recorded the difference between the probability of a response in the treatment and control groups for each complication studied. The inverse of this difference is the number needed to treat (NNT) to benefit one patient [6]. For each complication, we used estimates of the prevalence and effectiveness of treatment to calculate the NNT and the number needed to screen to benefit one patient.

3.0 Data Synthesis

3.1 Benefits of Detecting Overt Hyperthyroidism and Overt Hypothyroidism

Screening can identify patients who have overt but clinically unrecognized thyroid disease [7-11]. Older persons, in particular, may present with subtle, nonspecific signs and symptoms of thyroid dysfunction. Hyperthyroidism can cause weight loss, apathy, tremor, heat intolerance, and muscle weakness. Hypothyroidism can cause muscle cramps, dry skin, intolerance to cold, constipation, poor energy levels, fatigue, and mental slowness.

3.2 The cause of these symptoms may be over-looked until an abnormal result on a screening test leads a physician to ask about them. For example, suppose a 70-year-old woman with varicose veins sees her primary physician for a routine check-up. Because she takes a diuretic for pedal edema, the physician orders a serum potassium test, and the laboratory adds a screening sensitive TSH test to this request. Because the TSH level is elevated, the physician calls the patient, learns that she has noticed slight fatigue and dry skin, and, after finding that she also has a low free thyroxine level, decides to prescribe L-thyroxine. Even though this patient has symptomatic, overt hypothyroidism, it is fair to say that her condition was detected by screening.

3.3 How Many Persons Benefit from Detection of Overt Thyroid Dysfunction?

The prevalence of overt thyroid dysfunction depends on the age and sex of the population. Most recent studies concern older persons [11-22] (Table 2 and Table 5). Of women 70 to 80 years of age, 2% had overt hypothyroidism; this rate is significantly higher than that in all other age groups [8, 10-1214, 16-19, 23-31] (Figure 1). Among all women older than 60 years of age, office-based screening detected 14 cases of overt hypothyroidism or hyperthyroidism per 1000 women, or 1 case per 71 women (95% CI, 59 to 111 women) screened (Figure 1, Figure 2, and Figure 3).

3.4 The prevalence of overt thyroid dysfunction was lower in younger women and in men. Among women 40 to 60 years of age, the prevalence of overt hypothyroidism was 5 per 1000, and the prevalence of overt hyperthyroidism was 4.5 per 1000. In men older than 60 years of age, the prevalence of unsuspected hypothyroidism was 8 per 1000 (CI, 5 to 13.5 per 1000), but the prevalence of hyperthyroidism was only 1.3 per 1000 (CI, 0 to 1.6 per 1000). Unsuspected overt thyroid dysfunction is rare in women younger than 40 years of age and men younger than 60 years of age [20].

3.5 Overt hypothyroidism is seldom diagnosed in persons who have a mildly elevated TSH level (6 to 9 mU/L) [11, 17]. Similarly, almost all patients found to have overt hyperthyroidism have an undetectable rather than a mildly decreased TSH level [2, 9, 16, 32]. Therefore, overt thyroid dysfunction can be identified by performing a follow-up free thyroxine test only when the TSH level is undetectable or markedly elevated (≥ 10 mU/L).

3.6 What Is the Burden of Illness and What Is the Effectiveness of Treatment?

Relief of symptoms in patients with symptomatic but unsuspected overt thyroid disease is the most established benefit of screening. Screening in persons who receive regular office-based care has identified patients with overt hypothyroidism or hyperthyroidism that was not diagnosed in the usual course of care [8-11]. Such patients almost always have symptoms that are relieved by treatment [8, 23].

3.7 Would a careful history and physical examination in each patient followed by selective use of thyroid function tests be as effective as screening for detecting symptomatic overt thyroid dysfunction? A nonselective screening test, broadly applied, was more effective than clinical judgment in two studies that addressed this question. In one of these studies [33], screening of 1152 Swedish women detected 3 cases of overt hypothyroidism and 2 cases of overt hyperthyroidism. On the basis of history and examination, clinicians would have tested for thyroid disease in 286 women but would have failed to diagnose the condition in all 3 women who had overt hypothyroidism and 1 of the 2 who had overt hyperthyroidism. In the other study, screening in 2000 consecutive patients seen in a primary office setting detected 19 cases of overt hypothyroidism or hyperthyroidism [9]. On the basis of their own clinical assessments, clinicians ordered thyroid function tests in 35 of the 2000 patients, none of whom had thyroid disease.

3.8 Although some clinicians may have better results than these, data show that selective testing based on signs and symptoms is neither as practical nor as effective as a screening test followed by a history and physical examination in selected patients. The high yield of screening in studies of patients who regularly seek primary care suggests that selective testing based on clinical judgment is usually not effective in actual practice. For these reasons, we recommend a sensitive TSH test in women older than 50 years of age to detect overt but unsuspected hypothyroidism and hyperthyroidism. Data to guide the frequency of repeated screening are sparse, but longitudinal studies indicate that a patient who has a TSH level within the normal range is unlikely to develop overt hypothyroidism within 5 years after testing.

3.9 Screening for overt thyroid dysfunction will also detect persons who have subclinical thyroid dysfunction. Should these persons be treated, or is observation appropriate? In the following sections, we examine whether persons found by screening to have subclinical thyroid dysfunction benefit from treatment or follow-up.

4.0 Benefits of Detecting Subclinical Hyperthyroidism

4.1 When defined as an undetectable TSH level in a person with a normal free thyroxine level, the prevalence of subclinical hyperthyroidism is about 1% (CI, 0.4% to 1.7%) in men older than 60 years of age and 1.5% (CI, 0.8% to 2.5%) in women older than 60 years of age. In follow-up studies that used second-generation sensitive TSH assays, only 59% of these patients (range, 14% to 87%) had an undetectable TSH level on repeated testing [11, 16]. The newer, third-generation sensitive TSH assays may be more specific, but no studies of their use as a screening tool in the general population have been reported to date.

4.2 Patients with clinical findings of thyroid disease, such as a functioning solitary nodule or a pronounced multinodular goiter, have a high risk for progression [34-37], but 50% to 92% of patients in whom thyroid disease is diagnosed by screening had no signs or symptoms of thyroid disease on further evaluation [11, 16, 21, 27]. In studies of screening, the yearly rate of progression to overt hyperthyroidism is 0% in men and 1.5% in women [11, 16, 26].

4.3 What Is the Burden of Illness and What Is the Effectiveness of Treatment?

Prevention of atrial fibrillation and osteoporosis are the main potential benefits of treating subclinical hyperthyroidism. No randomized trials of early treatment for subclinical hyperthyroidism to prevent these complications have been done, nor were we able to locate studies of functional status or quality of life in patients with subclinical hyperthyroidism identified by screening. Patients found by screening to have an undetectable TSH level have not been shown to have more symptoms than age- and sex-matched controls.

4.4 Prevention of Atrial Fibrillation. Persons older than 60 years of age who have an undetectable TSH level are at increased risk for atrial fibrillation. Sawin and colleagues [38] studied 2007 persons in the Framingham study population who were 60 years of age or older and did not have atrial fibrillation. Overall, 61 patients (3%) had a low TSH level (mean, 0.1 mU/L) and a thyroxine level within the normal range; 36 were receiving L-thyroxine, and 25 (1.8% of the total study population) were not receiving L-thyroxine. In the latter group, the risk for atrial fibrillation over 10 years was 32% (CI, 14% to 71%) compared with 8% for patients who had a normal TSH level. Assuming that treatment would reduce the risk for atrial fibrillation to that of the general population, the NNT to prevent one case of atrial fibrillation over 10 years would be 4.2.

4.5 Prevention of Osteoporosis. Overt hyperthyroidism is associated with accelerated bone loss [39]. Information about the risk for osteoporosis in subclinical hyperthyroidism is scant. Two cross-sectional studies evaluated bone density in patients with subclinical hyperthyroidism associated with multinodular goiter, a small subset of patients who would be detected by screening. The patients in these studies had statistically and clinically significant decreases in bone mineral density at the femoral neck and at the radius compared with age-matched controls [40, 41].

4.6 In a nested case–control study from the Study of Osteoporotic Fractures cohort, women with a low TSH level had an increased risk for hip or vertebral fracture over 4 years (odds ratio for each type of fracture, 2.5 [CI, 1.1 to 8.0]) [42]. In many of these women, however, the low serum TSH level was associated with a history of hyperthyroidism or use of L-thyroxine. The risk for fracture in women with a low serum TSH level but no history of thyroid disease has not been reported.

4.7 Summary

Currently, the potential benefits of treating subclinical hyperthyroidism are theoretical. A suppressed TSH level increases the long-term risk for atrial fibrillation, but two thirds of patients with this finding will remain in normal sinus rhythm over 10 years. Because no randomized trials of antithyroid treatment in asymptomatic persons have been done, the effect of early treatment in preventing atrial fibrillation or bone fracture in the future is not known. Given the sparse data that are available, it is not clear that monitoring these persons with thyroid function tests would 1) prevent the spontaneous development of over hyperthyroidism or atrial fibrillation or 2) reduce the time spent in these states. Additional research is needed to determine whether markers of disease progression exist that could identify persons at high risk for these complications.

5.0 Benefits of Detecting Subclinical Hypothyroidism

5.1 Subclinical hypothyroidism, also called mild thyroid failure, is diagnosed when a patient has a TSH level less than 6 mU/L and a thyroxine level within the laboratory's normal range. In these patients, the serum thyroxine level is reduced from the physiologically normal level for that patient but does not fall below the lower limit of the laboratory's reference range for the thyroxine assay.

5.2 Subclinical hypothyroidism is the most common condition found by screening with thyroid function tests. Five percent to 10% of adult women have an elevated TSH level. Screening and treatment for all U.S. women older than 50 years of age with subclinical hypothyroidism would generate 4 000 000 new lifetime prescriptions for L-thyroxine in the first year and an additional 600 000 to 1 000 000 prescriptions every 5 years.

5.3 Recently, Danese and colleagues [43] presented the rationale for screening and proposed a specific strategy for selecting patients who might benefit from treatment. They concluded that screening was cost-effective and that 82% of women found to have subclinical hypothyroidism should receive L-thyroxine. Specifically, they recommended screening women older than 35 years of age and treating all patients with subclinical hypothyroidism who also have symptoms, serum antithyroid antibodies, or an elevated serum total cholesterol level (≥ 6.2 mmol/L).

5.4 In interpreting these conclusions, it is important to note that the investigators assumed that early treatment improved health outcomes and then proceeded to examine the likely consequences and costs of a screening program. Other researchers [37, 44-47] have questioned these assumptions, citing flaws in the literature on which they are based. In the next section, we examine the basis for assumptions about the ability of early detection, follow-up, and treatment to improve health outcomes in persons with subclinical hypothyroidism.

5.5 What Is the Burden of Disease and What Is the Effect of Treatment?

The potential complications of subclinical hypothyroidism are symptoms, hyperlipidemia, and progression to overt hypothyroidism. The chance that a patient will have complications depends on the patient's age and TSH level and the results of tests for serum lipids and antithyroid antibodies. A young, asymptomatic patient with a mildly elevated TSH level (6 to 9 mU/L) has the lowest risk for complications. At the other extreme is an older patient with a markedly elevated TSH level (≥ 10 mU/L), a high titer of antithyroid antibodies, and hyperlipidemia. In such a patient, the burden of illness and the risks for complications are greater and treatment with L-thyroxine is more likely to be of benefit.

5.6 Most patients in whom subclinical hypothyroidism is diagnosed have a mildly elevated TSH level (6 to 9 mU/L) and therefore have a low risk for complications. The Whickham survey, a population-based study of 2738 persons in an English town, provides realistic estimates of the distribution of TSH results in patients found by screening to have subclinical hypothyroidism [23]. In that survey, 8% of women 35 years of age or older had subclinical hypothyroidism, defined as a TSH level of 6 mU/L or less and a normal free thyroxine level. Of these women, 80% had a mildly elevated TSH level and 20% had a markedly elevated TSH level. The group with the lowest risk was women younger than 54 years of age who had a mildly elevated TSH level. This group comprised 31% of all women in whom subclinical hypothyroidism was diagnosed. The results of screening in a general office population are similar [9].

5.7 Most studies of complications focus on older women with markedly elevated TSH levels-the highest-risk group-but most persons in whom subclinical hypothyroidism would be diagnosed in a screening program are younger, have a mildly elevated TSH level, and have a lower risk. Little is known about the benefits of treating subclinical hypothyroidism in these low risk groups. In examining the published data, we sought direct evidence that early detection improves outcomes in persons who have mildly elevated TSH levels and in other low-risk groups.

5.8 Relief of Symptoms. Most patients found by screening to have subclinical hypothyroidism have at least one symptom that could be related to this diagnosis [11, 13]. Symptoms include muscle cramps, dry skin, intolerance to cold, constipation, poor energy levels, fatigue, and mental slowness.

5.9 Data from randomized trials of the effects of L-thyroxine treatment in subclinical hypothyroidism are shown in Table 3. Only three small trials have been done, and their results conflict. The first trial, by Cooper and colleagues [48], was in patients who had undergone thyroid ablation for hyperthyroidism 20 years earlier. Patients had a mean TSH level of 10.9 mU/L, three times the upper limit of the reference range for the assay used in the study. Symptoms were rated on the Cooper questionnaire, a 24-point scale that records how six symptoms of hypothyroidism change over time. After 1 year, patients receiving L-thyroxine improved by 2.1 points and patients receiving placebo deteriorated by 1.2 points (P = 0.037). The difference (3.3 points) is roughly equivalent to complete relief of one symptom and partial relief of a second symptom per patient. Eight (47%) of 17 treated patients reported reduced or milder symptoms, 4 felt worse, and 5 reported no change. In the placebo group, 3 (19%) of 16 patients felt better, 6 felt worse, and 7 reported no change. The difference between the proportion of patients who felt better in each group is 0.28 (CI, −0.09 to 0.65); the NNT to benefit 1 patient is 3.5. However, this result indicates the substantial uncertainty over the degree of benefit.

5.10 The study by Cooper and colleagues [48] supports treatment in symptomatic patients with a history of treated Graves disease, especially if the TSH level is greater than 10 mU/L. However, it has little relevance to screening because the natural history of treated Graves disease differs from the natural history of spontaneous hypothyroidism in the general population. Most persons found to have subclinical hypothyroidism have no history of hyperthyroidism, and 80% have a TSH level less than 10 mU/L.

5.11 The second trial, a small, short-term cross-over study of L-thyroxine therapy by Nystrom and associates [49], was in women identified by screening in the general population (Table 3). The 20 participants were women older than 50 years of age who had an initial TSH level between 4 and 15 mU/L. The strength of this study is that the patients are representative of patients with subclinical hypothyroidism identified in a typical screening program. Unfortunately, the design and analysis of the study limit its value in assessing the effectiveness of treatment. After 6 months of treatment, the mean symptom score improved by 1.81 units, equivalent to the complete relief of one symptom per patient. As judged by subjective improvement and cognitive measures, 4 (24%) of the 19 patients improved while receiving L-thyroxine and 2 (12%) felt worse with treatment.

5.12 In the third trial, by Jaeschke and colleagues [50], 37 patients with subclinical hypothyroidism were recruited from the outpatient clinics of a community hospital and randomly assigned to receive L-thyroxine or placebo (Table 3). Patients who received placebo did as well as or better than those who received L-thyroxine. After 6 months, 8 patients in the L-thyroxine group improved, 3 became worse, and 5 remained the same according to the Cooper questionnaire. In the placebo group, 11 patients improved, 1 became worse, and 4 remained the same. After 11 months, patients who received L-thyroxine had a small but statistically significant improvement in short-term memory, but treatment had no significant effects on symptom scores or disease-related quality of life [50].

5.13 These three small, short-term trials leave important questions unanswered. To date, only 90 patients have been studied, and no patient has been followed for longer than 1 year. Only 16 men and a similar number of women younger than 50 years of age were included. Enrollment was too small to determine whether treatment relieves symptoms in these groups.

5.14 The trials also did not examine whether patients with mildly elevated TSH levels were as likely to improve as those with markedly elevated TSH levels. The conflicting results raise two questions about patients with mildly elevated TSH levels that should be addressed in future studies. First, how often are symptoms causally related to a thyroid disorder? In unselected populations, persons with mild thyroid failure have not been shown to have more symptoms or more neuropsychiatric abnormalities than euthyroid persons. In two studies of screening, for example, most older women found to have subclinical hypothyroidism had nonspecific symptoms or signs, such as dry skin, fatigue, depression, muscle cramps, decreased memory, or mental slowness. However, these symptoms were no more frequent in patients with subclinical hypothyroidism than in age-matched controls [27, 49].

5.15 Complicating the question of causality, mild elevations of the TSH level were transient in some patients. In the third trial, 8 of 19 patients in the placebo group had at least one TSH value that had reverted to normal by 11 months. This occurred even though all patients had two consecutive elevated TSH levels before study entry. Reversion also occurred in 4 of 17 patients in the second trial, and 6 patients in the first trial were excluded because the TSH level reverted to normal. If early intervention is to succeed, patients who have transient abnormalities must be distinguished from those who are likely to have persistent and progressive disease.

5.16 How do subclinical hypothyroidism and L-thyroxine treatment affect overall health and quality of life [51]? In the third randomized trial [50], treatment did not improve health-related quality of life as measured with a standardized questionnaire, the Sickness Impact Profile. In that study, the mean Sickness Impact Profile score in patients with subclinical hypothyroidism was initially 3.1 out of 10. On this scale, a score of 3.0 is usually interpreted as the border between no disability and mild disability. In another study [52], a random sample of healthy older adults had a similar mean Sickness Impact Profile of 3.4.

5.17 On the basis of the three randomized trials, we conclude that among patients with mild thyroid failure who have one or more symptoms suggestive of hypothyroidism, 0 to 28 patients per 100 treated may benefit. The conflicting results of the trials make the effectiveness of treatment for subclinical hypothyroidism uncertain and suggest that the results from any one study should not be overgeneralized. To strengthen confidence in the utility of treatment, future trials should focus on patients who are found by screening and have markers of disease, such as antithyroid antibodies, that may increase the likelihood that symptoms are causally related to mild thyroid dysfunction.

5.18 Prevention of Overt Hypothyroidism. Over time, asymptomatic patients with subclinical hypothyroidism may progress to overt hypothyroidism. Progression is diagnosed when a patient with subclinical hypothyroidism develops a low free thyroxine level (Table 1). The presence of thyroid auto-antibodies in serum is a strong risk factor for progression. Older age and a higher TSH level also increase the chance that a patient with subclinical hypothyroidism will progress.

5.19 Treatment of asymptomatic persons with L-thyroxine is intended to prevent symptoms that might accompany progression to over hypothyroidism. Because this potential benefit has not been studied in randomized trials, it is necessary to estimate it on the basis of data from observational studies. The benefit of early treatment depends on the likelihood that a patient will progress and the risk for morbidity associated with progression.

5.20 How often does subclinical hypothyroidism progress to overt hypothyroidism? The Whickham survey provides the best data on the chance that a person with subclinical hypothyroidism found by screening will develop overt hypothyroidism [23, 53]. The Whickham investigators defined a TSH level of 6 mU/L or more as elevated. Two thirds of women with an elevated TSH level had antithyroid antibodies in serum. During 20 years of follow-up, 55% of women with a TSH level of 6 mU/L or more and a positive antibody test result developed overt hypothyroidism [54]. About 25% of these women had an initial TSH level of 10 mU/L or more; for these women, the risk for overt hypothyroidism over 20 years was close to 90%, or about 0.11 per year.

5.21 Younger women and women with a mildly elevated TSH level (6 to 9 mU/L) had a lower risk for progression. A logistic regression Equation developed by the Whickham investigators can be used to estimate risk in these groups [54]. For a 50-year-old woman with a positive test result for thyroid antibodies, the 20-year risk for developing overt hypothyroidism was 0.57 if the TSH level was 6 mU/L and 0.72 if the TSH level was 9 mU/L. For a 35-year-old woman with a positive antibody test result, the 20-year risk for developing overt hypothyroidism was 0.47 if the TSH level was 6 mU/L and 0.67 if the TSH level was 9 mU/L.

5.22 The risk for progression was not evenly distributed throughout the follow-up period. By 5 years, 8% of patients with an elevated TSH level developed overt hypothyroidism and 92% remained well [53]. Almost all patients who progressed within 5 years had a TSH level of 10 mU/L and high titers of circulating antibodies. Of 57 women who had an initial TSH level between 6 and 10 mU/L, none progressed within 2 years and 3 (5.2%) progressed within 5 years. In patients with a mildly elevated TSH level, the risk for progression after 5 years was not distinguishable from that of euthyroid patients.

5.23 We used data from the Whickham study to estimate the effect of early treatment in patients with mildly elevated TSH levels (6 to 9 mU/L). If 1000 women 35 years of age and older are screened, 80 will receive a diagnosis of subclinical hypothyroidism; 43 of these women have a mildly elevated TSH level and positive results on antithyroid antibody testing.

5.24 If these 43 women receive L-thyroxine, how many could benefit and when will they benefit? In their decision analysis, Danese and colleagues [43] assumed a constant progression rate of 0.054 per year. In this case, by 5 years, early treatment would prevent overt hypothyroidism in 10 women, but 33 asymptomatic women will have taken medication for 5 years even though they would not have progressed in that time (NNT = 4.3). This approach overestimates the early benefits of treatment because when the TSH level is only mildly elevated, the risk for progression is lower in the first few years after diagnosis. When the interim results of the Whickham study are used, overt hypothyroidism would be prevented in only 3 women by 5 years, but 40 will have taken medication for 5 years without a clear benefit (NNT = 14.3).

5.25 By 20 years, overt hypothyroidism would be prevented in 29 (67%) of the 43 women, but 14 otherwise healthy women will have taken medication for 20 years. Overall, for every 1000 women screened, 860 person-years of medication would be dispensed; 505 of these person-years would be prescribed to women before the onset of over hypothyroidism, when benefit has not yet been derived.

5.26 In asymptomatic patients who advance to overt hypothyroidism, how much morbidity can early treatment prevent? As noted above, the NNT to prevent one asymptomatic patient from progressing to overt hypothyroidism is 4.3 to 14.3; this implies that 77% to 93% of these patients will take L-thyroxine for 5 years without clear benefit. Whether an NNT in this range is desirable depends on whether spontaneous progression from subclinical to overt hypothyroidism is associated with significant, reversible morbidity and on the morbidity caused by the treatment itself.

5.27 The case for treatment to prevent progression of mild thyroid failure would be strengthened by data showing that this progression is associated with a substantial burden of illness that could be prevented by earlier treatment. Unfortunately, no studies have measured or even reported on the severity of symptoms or degree of disability in newly hypothyroid patients, the length of time spent in that state, or the adverse effects on health outcomes of wider use of L-thyroxine in apparently healthy persons.

5.28 Could patients found by screening to have an elevated thyroid-stimulating hormone level benefit from careful follow-up? Targeted follow-up of patients with subclinical hypothyroidism could be beneficial if it enables the physician to diagnose and treat complications earlier than would occur in the usual course of care. Close follow-up (consisting of a history, physical examination, and thyroid function testing) is more likely to be beneficial in patients who have a markedly elevated TSH level (≥ 10 mU/L). Among these patients, the high incidence of overt hypothyroidism in the first few years after detection is high. Monitoring within this period might detect progression early, allowing treatment to begin before the disease progresses further.

5.29 The same approach is less likely to be useful for patients who have a mildly elevated TSH level. In this group, the chance of an increase in TSH level or development of overt hypothyroidism is low, especially in the first 2 years after diagnosis. No data have been published on the effect of careful follow-up on health outcomes in patients with subclinical hypothyroidism.

5.30 Reduction of Hyperlipidemia and Risk for Vascular Complications

In patients with subclinical hypothyroidism and an elevated total cholesterol level, L-thyroxine treatment may reduce serum cholesterol and thereby decrease the incidence of coronary artery disease, stroke, and peripheral vascular disease. L-thyroxine therapy is most likely to reduce lipid levels in patients who have an elevated total cholesterol level (≥ 6.2 mmol/L) [55-57] and a markedly elevated TSH level (≥ 10 mU/L) [58-61]. One in four patients with subclinical hypothyroidism has a total cholesterol level more than 6.2 mmol/L, and one in five patients with subclinical hypothyroidism has a markedly elevated TSH level; therefore, treatment with L-thyroxine could reduce the cholesterol level in about 1 in 20 patients (5%) with subclinical hypothyroidism.

5.31 How much reduction in cholesterol can be expected in these patients? In their decision analysis, Danese and colleagues assumed that in patients with a total cholesterol level greater than 6.2 mmol/L, treatment with L-thyroxine reduces the total cholesterol level by 17% [43]. This assumption overestimates the effectiveness of treatment. In a recent meta-analysis of 13 studies of patients with subclinical hypothyroidism, the pooled mean reduction in total cholesterol level after L-thyroxine therapy was 0.4 mmol/L (CI, 0.2 to 0.6 mmol/L), representing a 6% decrease from a pretreatment mean of 6.7 mmol/L [57]. The effect was slightly higher (0.6 mmol/L, an 8% decrease) in the subset of studies in which the mean pretreatment cholesterol level was 6.5 mmol/L. Of the 13 studies, 6 reported a decrease in total cholesterol level of 5% or less; 6 reported a decrease between 6% and 12%; and 1, a study of seven patients treated with doses of L-thyroxine sufficient to suppress TSH to subnormal levels, reported a 17% decrease [58].

5.32 There is considerable uncertainty about any estimates because 12 of the 13 studies in the meta-analysis discussed above concerned patients referred to an endocrinologist for elevated lipid levels, a history of thyroid disease, or elevated TSH levels. Therapy may be less effective in lowering cholesterol levels in patients identified by screening than in more selected populations. In the 1 study of patients identified by screening, L-thyroxine therapy reduced cholesterol levels by only 2.5% [49] compared with 6% to 8% in all studies combined.

5.33 The amount of vascular disease that could be prevented by screening women older than 50 years of age is unknown. The following considerations suggest that the potential benefit is small. First, many clinicians already screen for hypercholesterolemia and test thyroid function in patients with elevated total cholesterol levels. This approach may be just as effective as general screening for thyroid dysfunction in identifying candidates for L-thyroxine therapy.

5.34 Second, on the basis of the 1990 U.S. female population [62], if women older than 50 years of age were screened, 35% of the candidates for cholesterol reduction would be older than 75 years of age. So far, there is no direct evidence that treating hypercholesterolemia prevents cardiovascular disease in this older cohort. The total cholesterol level alone is a weak marker of risk in the elderly [63], probably because chronic illness masks the underlying relation between total cholesterol level and cardiovascular mortality [64]. Put differently, frail elderly patients who have illnesses that limit their life expectancy are less likely to benefit from reduction of cholesterol levels. High-density lipoprotein cholesterol is a stronger predictor of cardiovascular mortality than total cholesterol in elderly persons, but L-thyroxine therapy has not consistently increased levels of high-density lipoprotein cholesterol in patients with subclinical hypothyroidism [50, 58, 59, 64-66].

5.35 Putting these concerns aside, how much would a reduction in total cholesterol level of 0.4 to 0.6 mmol/L (6% to 8%) decrease the incidence of cardiovascular disease? In a 60-year-old woman with a pretreatment total cholesterol level of 6.5 mmol/L and no other risk factors, a reduction of 0.6 mmol/L (8%) would decrease the risk for developing ischemic heart disease from 10% to 9%, and a reduction of 1.1 mmol/L (17%) would decrease the risk to 8% over 10 years [67]. For each year of treatment, the NNT to prevent 1 case of ischemic heart disease would be 1000 for a reduction of 0.6 mmol/L and 476 for a reduction of 1.1 mmol/L.

5.36 In summary, L-thyroxine treatment may reduce the serum cholesterol level by 8% in selected patients who have both a TSH level of 10 mU/L and an elevated total cholesterol level (≥ 6.2 mmol/L). However, there is still no direct evidence that L-thyroxine treatment reduces total cholesterol levels by this amount in patients identified in a screening program.

5.37 Potential Risks of Treating Subclinical Hypothyroidism

Advocates of screening have an obligation to determine whether screening will have unexpected adverse effects. Adverse effects of replacement doses of L-thyroxine include nervousness, palpitations, atrial fibrillation, and exacerbation of angina pectoris. Adverse effects were noted in two of the randomized trials discussed earlier (Table 3). In one of the trials, 2 of 20 patients (10%) receiving L-thyroxine quit the protocol because of nervousness and palpitations [49]. In the most recent trial, 2 of the 18 patients (11%) assigned to L-thyroxine therapy withdrew because of complications: One had an increase in angina, and 1 had new-onset atrial fibrillation [50]. Patients who experience adverse effects may require medical attention or cessation of therapy, both of which reduce the effectiveness and increase the costs of screening.

5.38 Overtreatment with L-thyroxine, indicated by an undetectable TSH level, is another potential risk. Fifteen percent to 50% of patients receiving L-thyroxine for primary hypothyroidism are unintentionally maintained on doses sufficient to cause the TSH level to be undetectable [38, 68, 69]. Data from the Framingham cohort suggest that one excess case of atrial fibrillation may occur for every 114 patients treated with doses of L-thyroxine sufficient to suppress TSH [38]. A well-done meta-analysis [70] found that suppressive doses of L-thyroxine caused small (5% to 9%) but statistically significant decreases in bone density of the spine, femoral neck, distal radius, and trochanter. The authors concluded that overtreatment with L-thyroxine probably contributes to the development of osteoporosis in postmenopausal women. Long-term excessive doses of L-thyroxine may also cause left ventricular hypertrophy [71] and other findings associated with untreated hyperthyroidism, but these complications have not been shown to be clinically significant.

5.39 Summary

Table 4 summarizes our findings on the effects of screening for subclinical hypothyroidism, and it highlights gaps in information that underlie disagreement about the value of screening. Most estimates are based on observational studies in selected groups of patients rather than in patients identified by screening.

5.40 Early treatment is most likely to be effective in patients who have a markedly elevated TSH level (≥ 10 mU/L). On the basis of results of randomized trials, 1 in 8.3 of these women (1 in 519 screened) may benefit from L-thyroxine therapy given to relieve symptoms. By 5 years, treatment would prevent overt hypothyroidism in 1 of 2 patients, or 1 in 112 patients screened. Treatment may reduce the cholesterol level by an average of 8% in patients who have a TSH level of 10 mU/L or more and an elevated serum cholesterol level. One in every 250 patients screened may benefit. If therapy were effective in every patient (NNT = 1), 1 case of coronary artery disease could be prevented for every 95 to 200 patients treated by 5 years.

5.41 In patients who have mildly elevated TSH levels, randomized trials have not specifically proven that treatment relieves symptoms, and therapy does not seem to decrease elevated serum lipid levels substantially. Preventing overt hypothyroidism is a potential benefit, but if treatment is given for this purpose, many patients who would never develop symptoms or would not develop them for years will take L-thyroxine before there is a clear reason for them to do so. After 5 years, 4.3 to 14.3 patients would have to be treated and 100 to 333 would have to be screened for 1 patient to benefit. The actual degree of benefit will remain uncertain until we know how long overt hypothyroidism typically remains undiagnosed and how much morbidity is associated with its onset.

6.0 Conclusions and Recommendations

6.1 Primary care office-based screening with a sensitive TSH test in women older than 50 years of age may be indicated.

6.1.1 A free thyroxine test should be done in women older than 50 years of age when the TSH level is undetectable or is 10 mU/L or greater. The rationale for this recommendation is that in this age group, 1 in 71 women has unsuspected but symptomatic overt hypothyroidism or overt hyperthyroidism that will respond to treatment.

6.1.2 Screening in women younger than 50 years of age and in men is not warranted because the prevalence of unsuspected overt thyroid dysfunction is low. The evidence for this recommendation would be level III in the U.S. Preventive Task Force system [46] or level C in the system used by Hayward and colleagues [72, 73], reflecting the lack of direct evidence from randomized trials about the outcomes of screening.

6.2 Treatment in patients found by screening to have subclinical hyperthyroidism has not been studied. This condition is associated with an increased risk for atrial fibrillation and osteoporosis. However, the role of early treatment is not clear because most persons with this condition remain healthy.

6.2.1 Patients who are found to have relatively specific symptoms and signs (such as goiter, nodule, eye findings of Graves disease, or tremor) should be referred to an endocrinologist for consideration of treatment, but the management of patients with none of these findings is not clear.

6.2.2 Information about factors that could help predict which patients are likely to develop complications is needed to create a practical strategy for managing this condition.

6.3 The available evidence is not sufficient to recommend for or against treatment for subclinical hypothyroidism. Results from randomized trials of treatment to relieve symptoms are mixed.

6.3.1 There is little doubt that treatment can benefit a subset of patients, often seen in the endocrine referral setting, who have a long history of thyroid disease or more marked abnormalities and symptoms than do patients who are identified by screening. However, it is a mistake to overgeneralize these results to the much broader spectrum of patients who are found by screening in the general population.

6.3.2 Larger, well-designed randomized trials are needed to determine whether treatment would be effective in office-based screening of otherwise healthy primary care patients.

6.3.3 Women older than 50 years of age with markedly elevated TSH levels (≥ 10 mU/L) have the highest risk for complications. Evidence is not sufficient to recommend or discourage treatment of these patients, and either treatment or observation is a reasonable approach. One option is to treat patients who have symptoms that may be caused by hypothyroidism with the understanding that a follow-up evaluation may be needed to discontinue therapy in patients who do not improve. Physicians who believe that substantial morbidity accompanies progression to overt hypothyroidism may wish to treat asymptomatic patients as well.

6.3.4 Other groups, including younger women, men, and patients with a mildly elevated TSH level (6 to 9 mU/L), have a lower risk for complications. In these groups, no strong evidence shows that treatment is effective or ineffective in relieving symptoms. The question can only be resolved by conducting a large controlled trial of treatment in persons identified by screening as having mild thyroid failure. Their quality of life, response to treatment, and rates of adverse effects must be known before wider use of L-thyroxine can be recommended in these patients.

6.3.5 Patients who have a mildly elevated TSH level present a challenge to the physician. Concerns about the medicolegal or ethical consequences of failing to act on abnormal test results on the one hand and the uncertain benefit of targeted follow-up or treatment on the other might lead some physicians to not screen at all.

6.3.6 There is no clear answer to this dilemma, but clinicians might consider a trial of therapy in selected, more symptomatic patients and a specified plan for follow-up in the others. Follow-up could be done by taking a history, performing a physical examination, and repeating thyroid function tests at regular intervals. How often this evaluation should be done has not been established. Follow-up every year or every 2 years is probably not necessary because otherwise healthy persons who have a mildly elevated TSH level seldom progress within 2 years to the point of being identified by screening. On the basis of this observation, follow-up every 2 to 5 years may be adequate.

6.4 For screening to be most cost-effective, the laboratory, clinic, or office should be equipped to identify candidates who meet age and sex criteria for screening and to obtain appropriate follow-up tests on the same serum sample without requiring a second phlebotomy [2]. In particular, the office or clinic must follow patients who have overt but clinically unrecognized disease because these patients derive the most definite benefit from treatment. These patients have an elevated TSH level and a low free thyroxine level or an undetectable TSH level and an elevated level of thyroxine or triiodothyronine. A formal system to identify these patients and arrange for follow-up evaluation should be part of the screening program.

6.5 Future prospective, concurrent comparison studies should examine the effects of screening on health outcomes. These studies should be large enough to determine the likelihood that specific subgroups of patients will benefit from treatment in relation to their age, sex, clinical findings, quality of life, and TSH level.

Dr. Redfern: PACE Center, 5050 NE Hoyt #540, Portland, OR 97213.