• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Rugge B, Balshem H, Sehgal R, et al. Screening and Treatment of Subclinical Hypothyroidism or Hyperthyroidism [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 Oct. (Comparative Effectiveness Reviews, No. 24.)

Results

Systematic Reviews

Our search identified 86 unique reviews whose abstracts were reviewed for inclusion (Figure 2). After review, 18 were pulled for full-text review. Of those 18 reviews, four were not systematic reviews;56-59 six did not address the key questions of this review;12,43-45,60,61 two fell outside the timeframe for included reviews;62,63 and three were excluded for reasons discussed below. The studies included in the three remaining reviews are compared in Table 6.1,21,64

Figure 2 is a flow diagram illustrating the process for inclusion and exclusion of systematic reviews.  The process began with identification of 86 abstracts from a search of relevant bibliographic databases. Of these, 68 articles were excluded at the abstract stage. Eighteen full-text articles (since 2002) were reviewed for relevance to the key questions. Of these 18 articles, 15 were excluded from further review: four were not systematic reviews, six were not relevant to the key questions of this review, and five were related to the topic but were either from the wrong timeframe or of poor quality.  Of the three remaining reviews, one (Helfand 2004) was relevant to Key Question 1, Screening; one (Helfand 2004) was relevant to Key Question 2, Harms of Screening; three (Villar 2007, Helfand 2004, Surks 2004) were related to Key Question 3, Efficacy of Treatment; and one to Key Question 4, Harms of Treatment (Helfand 2004).

Figure 2

Systematic review literature flow diagram.

Table 6. Systematic review study concordance.

Table 6

Systematic review study concordance.

Three of these reviews examined the effect of treating subclinical hypothyroidism on lipids.65-67 All of these demonstrated an improvement in lipids with treatment, but had serious flaws—they did not distinguish between studies of otherwise healthy patients from those who acquired hypothyroidism after treatment for hyperthyroidism, or they included poorly done observational studies.

A recent, good-quality Cochrane review provided the best information about several relevant outcomes, including cardiovascular mortality and morbidity, symptom improvement, and health-related quality of life.64 That review also examined all-cause mortality, lipid levels, systolic and diastolic heart function, and adverse effects of levothyroxine. The review concluded that while there might be some short-term improvement in lipid profiles and left ventricular function with treatment, trials of thyroid hormone therapy for subclinical hypothyroidism were not suited to assess survival or cardiovascular morbidity. These trials did not demonstrate that treatment improved health-related quality of life or symptoms.64

The earlier reviews1,21 addressed screening and treatment of both subclinical hypothyroidism and hyperthyroidism. As seen in Table 6, there was considerable overlap in the trials reviewed by these earlier reviews and the Cochrane review. A total of 14 trials were included in one or more of these reviews. The conclusions of the two earlier reviews largely agreed with those of the more recent Cochrane review. Because of the convergence of these disparate reviews on a finding of uncertainty, we considered the evidence from these three reviews to adequately reflect the state of the evidence through 2007 (the most recent search date for the Cochrane review). This report, therefore, only addressed the question of whether there is any new evidence from more recent studies.

Individual Studies

The current review includes eight trials, six of which were not examined in previous systematic reviews. Four were studies of treatment for subclinical hypothyroidism and two were studies of treatment for subclinical hyperthyroidism. The flow of studies from initial identification of titles and abstracts to final inclusion or exclusion using prespecified criteria (Appendix C) is diagrammed in Figure 3. Our original search found a total of 948 individual articles published from 2002 to May 2010. After selection of articles for review, we found that no foreign language articles had been included. To ensure that we had not erred in our selection of articles we rereviewed the abstracts of foreign language articles from the original search and expanded our search to include CINAHL and the WHO Global Health Library, databases with broader indexing of foreign language journals. After the abstracts were reviewed, 297 articles from our original search were found to meet criteria for full article review. The supplemental review resulted in identifying 12 articles for full-text review. Of the 12 foreign-language articles reviewed, none met criteria for inclusion in this review. The flow of studies from this supplementary search is diagrammed in Figure 4. Excluded studies are described in Appendix D. No controlled trials of the benefits or harms of screening (Key Questions 1 and 2) were identified, but two observational studies relevant to the harms of screening were identified. Six trials assessed one or more relevant outcomes of treatment for subclinical hypothyroidism (Key Question 3), but none of them assessed harms systematically.

Figure 3 is a flow diagram illustrating the process for inclusion and exclusion of articles of individual studies.  The process began with identification of 948 abstracts from a search of relevant bibliographic databases. An additional 42 articles were identified through a search of the gray literature, and 108 from references gleaned from retrieved articles. Of these, 651 articles were excluded at the abstract stage. Two hundred ninety-seven full-text articles were reviewed for relevance to the key questions. Of these, 287 were excluded from further review: 48 were used as background (3 regarding complications of subclinical thyroid disease, 6 regarding the effects of elevated TSH, 11 were reviews); 78 were excluded because they were not relevant to the population of interest; 52 had no data by design; 16 had no data for other reasons; 98 had no relevance to the key questions; 41 were related to the topic but did not meet other inclusion criteria of study design or quality.  Of the 10 remaining reviews, no studies were found for Key Question 1 (Screening); 2 studies were found for Key Question 2 (Harms of Screening); 8 studies were found for Key Question 3 (Efficacy of Treatment), 6 studies of subclinical hypothyroidism and 2 studies of subclinical hyperthyroidism; and no studies were found for Key Question 4 (Harms of Treatment).

Figure 3

Individual study literature flow diagram.

Figure 4 is a flow diagram illustrating the process for inclusion and exclusion of foreign language articles.  The process began with identification of 173 abstracts from a search of relevant bibliographic databases. Of these, 161 articles were excluded at the abstract stage. Twelve full-text articles were reviewed for relevance to the key questions. Of these twelve articles all twelve were excluded from further review: two dealt with the wrong population; one was not relevant to the key questions; one had no data by design; and eight did not meet inclusion criteria for study quality. No studies were found for Key Question 1 (Screening); no studies were found for Key Question 2 (Harms of Screening); no studies were found for Key Question 3 (Efficacy of Treatment), and no studies were found for Key Question 4 (Harms of Treatment).

Figure 4

Foreign language literature flow diagram.

Several aspects of these trials make their applicability to a U.S. primary care population problematic. None of the studies was conducted in the United States. Most of the studies recruited subjects from specialty clinics, rather than from the primary care setting. Additionally, the studies were relatively short, the longest lasting 12 months. A longer duration would be needed to compare early treatment with annual retesting, the main alternative in practice.

A meta-analysis was not performed due to the methodological and clinical diversity among the included studies. The TSH value that was used to diagnose subclinical thyroid dysfunction varied among the studies. Different dosages of medications were used in the different trials.

Key Question 1. Does screening for subclinical thyroid dysfunction reduce morbidity or mortality?

In 2004, Helfand1 did not find any studies that evaluated the impact of screening for thyroid disease on morbidity or mortality. We also identified no controlled trials of screening for subclinical hypothyroidism or subclinical hyperthyroidism in the general population.

Key Question 2. What are the harms of screening for subclinical thyroid dysfunction?

Information about the harms of screening is still sparse. We identified no RCTs or controlled observational studies that evaluated harms associated with screening for subclinical thyroid dysfunction. Overdiagnosis is a potential harm because TSH levels return to the reference range in a substantial proportion of individuals.

Potential harms of screening for subclinical hypothyroidism include false positive test results, anxiety related to test results, and overdiagnosis. There is little direct evidence about these potential adverse effects of screening.

In the context of screening for subclinical hypothyroidism, overdiagnosis can be defined as diagnosing subclinical hypothyroidism in a patient who cannot benefit from the diagnosis. Overdiagnosis is relevant to screening for subclinical hypothyroidism because many individuals who have a mildly elevated TSH level and a normal free T4 never develop complications and, in some, the TSH reverts spontaneously to a value below the upper reference limit.

Spontaneous reversion of mildly elevated TSH levels was observed in the Whickham study, other natural history studies, and in trials of levothyroxine therapy. Recent studies suggest that it is more frequent than previously thought and that it can occur after a long period of persistent elevation (Table 7). An observational, population-based study followed elderly patients living in Leiden, Netherlands, screening for thyroid disease at age 85 and then rechecking thyroid function at age 88.87 Twenty-one subjects were initially found to have subclinical hypothyroidism (all were 85 years old). Three years later, none of the subjects had progressed to overt hypothyroidism, eight continued to have subclinical hypothyroidism, two had developed subclinical hyperthyroidism, and 11 subjects now had normal thyroid function tests. In the same study, 12 subjects were initially found to have subclinical hyperthyroidism. Three years later, one had progressed to overt hyperthyroidism while five continued to have subclinical hyperthyroidism, five individuals had normal thyroid function, and one subject had developed subclinical hypothyroidism.

Table 7. Screening evidence table.

Table 7

Screening evidence table.

In a natural history study, 107 patients age 55 to 83 with newly diagnosed and untreated subclinical hypothyroidism referred from general practice and other specialty clinics to an endocrinology clinic were followed for a mean period of 32 months.88 All patients had two measurements of TSH above 5.0 mIU/L and free T4 in the normal range of 0.75-2.0 ng/dL one to three months prior to entering the study. During the 32 month followup period a total of 40 patients (37.4%) reverted back to normal TSH levels without treatment. While this is only a natural history study, it suggests that overdiagnosis may occur unless treatment is withheld until repeat testing confirms persistent and progressive elevations.

Key Question 3. Does treatment of patients with subclinical hypothyroidism or subclinical hyperthyroidism detected by screening affect outcomes?

Subclinical Hypothyroidism

An earlier systematic review conducted for the USPSTF appraised eight trials of treatment for subclinical hypothyroidism.1 Three of the studies11,72,75 included individuals with known thyroid disease, making them less relevant to a population identified by screening. Only one of these studies was rated as a good-quality study.11 Cooper found that patients with a history of Graves disease had a modest improvement in hypothyroid symptoms following one year of therapy with levothyroxine, but no improvement in lipid levels. Another three studies included individuals without a history of thyroid disease,68,70,76 thus are likely more relevant to primary care. However, only one was considered of fair quality70 (the others were rated poor68,76) and that study only found improvement in short-term memory, and failed to find improvement in the Sickness Impact Profile (SIP). The final two studies included individuals who were euthyroid but either had a TSH in the 4 to 4.5 mIU/L range71 or had at least three symptoms of hypothyroidism.74 One study was rated as poor (Michalopoulou, 1998).71 The fair-quality study, Pollock (2001),74 paradoxically, found a reduced SF-36 vitality score in euthyroid patients treated with levothyroxine.

A 2007 Cochrane review64 provided the most recent, good-quality review and meta-analysis of levothyroxine treatment for subclinical hypothyroidism. The review included 12 studies, five of which were not included in the reviews by Surks21 or Helfand1 (seeTable 6). None of the trials assessed cardiovascular mortality or morbidity, and seven evaluated symptoms, mood, and quality of life, and found no statistically significant improvement. The authors also noted methodological deficiencies in previous meta-analyses of the effect of levothyroxine on cholesterol levels and other lipids. In their systematic review, six of the trials measured plasma cholesterol or LDL cholesterol. They noted baseline differences in cholesterol levels in several of the trials, and conducted two analyses finding: (1) the post-treatment difference of the means favored treatment, but (2) analysis of changes from baseline favored placebo.

Our review includes six trials that were not included in the 2004 USPSTF review,80-82,84-86 four of which were not included in the Cochrane review.81,84-86 (See Table 6 and Tables 8-15.) These studies were not included because they were published after the literature searches were performed for the earlier reviews and/or due to differences in inclusion/exclusion criteria. The largest enrolled 120 individuals, and four enrolled fewer than 70 individuals. The followup period ranged from 5 to 12 months. Five of the trials did not report their method for allocation concealment;80-82,85,86 four did not report their method of randomization;80-82,86 and three reported the number of patients enrolled at baseline but not at study conclusion, and did not indicate that intention-to-treat analysis was followed.81,85,86 These studies were rated fair quality.

Table 8. Treatment evidence table (subclinical hypothyroidism).

Table 8

Treatment evidence table (subclinical hypothyroidism).

Table 9. Quality of life (subclinical hypothyroidism).

Table 9

Quality of life (subclinical hypothyroidism).

Table 10. Weight (kg)/body mass index (BMI) (kg/m2) (subclinical hypothyroidism).

Table 10

Weight (kg)/body mass index (BMI) (kg/m2) (subclinical hypothyroidism).

Table 11. Blood pressure (mmHg) (subclinical hypothyroidism).

Table 11

Blood pressure (mmHg) (subclinical hypothyroidism).

Table 12. Total cholesterol (mg/dL) (subclinical hypothyroidism).

Table 12

Total cholesterol (mg/dL) (subclinical hypothyroidism).

Table 13. Low-density lipoprotein (LDL) cholesterol (mg/dL) (subclinical hypothyroidism).

Table 13

Low-density lipoprotein (LDL) cholesterol (mg/dL) (subclinical hypothyroidism).

Table 14. High-density lipoprotein (HDL) cholesterol (mg/dL) (subclinical hypothyroidism).

Table 14

High-density lipoprotein (HDL) cholesterol (mg/dL) (subclinical hypothyroidism).

Table 15. Triglycerides (mg/dL) (subclinical hypothyroidism).

Table 15

Triglycerides (mg/dL) (subclinical hypothyroidism).

Cardiac Events

None of the studies evaluated the effect on cardiovascular morbidity or mortality from treatment of subclinical hypothyroidism with levothyroxine.

Quality of Life

Two studies, with a total of 169 subjects, evaluated the effect of treatment on measures of quality of life. One good-quality, 12-week crossover study (Razvi) enrolled 100 participants from 322 patients from urban, general practices in the United Kingdom identified as eligible from thyroid function tests from a laboratory database.84 Of the 322 patients originally identified as eligible, 63 percent had been tested either for symptoms attributable to hypothyroidism (n=179) or because of familial history of thyroid disease (n=24). Fifty of the enrollees were given 100 mcg of levothyroxine without dose titration and 50 were given placebo for 12 weeks. Patients then switched treatment arms without a wash-out period. (The half-life of thyroxine is estimated to be 5 to 9 days.) While the absence of a washout period could introduce bias, it would bias the study towards the null, that is, decrease the likelihood of demonstrating a treatment effect. The primary endpoints of this study were improved brachial artery flow-mediated dilatation (as a marker of vascular endothelial function) and total cholesterol. Secondary endpoints assessed were changes in weight (as measured by BMI) and patient-reported outcomes including perceived health status, hypothyroidism-specific quality of life, and hypothyroid symptoms (as assessed by questionnaires). This trial found no improvement in overall quality of life measures; health status, as measured by the SF-36v2; or treatment satisfaction.

The other trial (Jorde, N=69) recruited subjects who had a TSH level between 3.5 and 10 mU/L from a population-based sample in Norway. Asymptomatic subjects who had no history of thyroid disease were invited to participate in a 12-month trial of levothyroxine treatment. After 12 months, there was no significant improvement in cognitive or emotional function or in hypothyroid symptoms.82

Weight/Body Mass Index (BMI)

Four studies with a total of 305 subjects looked at either weight or BMI. Study periods ranged from 586 to 12 months.81 Three were rated fair quality;80,81,86 one was rated good quality.84 None found any significant change in either weight or BMI after 5 months,86 24 weeks,84 6 months,80 or 1 year.81

Blood Pressure

Two studies, with a total of 195 subjects looked at blood pressure. One five-month study (n=95) was rated fair quality86 the other (n=100), a 12-week cross-over study, was rated good.84 Neither found significant change in blood pressure.

Lipids

Four studies with a total of 379 subjects evaluated the impact of treatment on lipids. Two found improvements to lipids; two found no improvement.

The good-quality randomized, controlled 12-week, cross-over trial showed modest improvement in total cholesterol and LDL.84 The study also reported results for changes in LDL, HDL, and triglycerides. Razvi found a significant decrease in total cholesterol of 5.8 percent (from 235.5 mg/dL to 220.0 mg/dL; p = <0.001) and a significant 5.6 percent decline in LDL (from 139.0 mg/dL to 131.2 mg/dL; p=<0.05).

We also found one fair-quality RCT that demonstrated improvement in total cholesterol and LDL from treatment for subclinical hypothyroidism.85 While this study described itself as a double-blind randomized, placebo-controlled trial, blinding was not described nor was allocation concealment. In addition, the number of enrolled patients was described at baseline, but not at the end of the study, and there was no mention of intention to treat analysis. This study of 120 patients from an endocrinology outpatient clinic in Kuwait found a significant decline in LDL of 12.4 percent (from 127.38 mg/dL to 111.55 mg/dL; p<0.01) and a significant 6.1 percent decline in total cholesterol (194.9 mg/dL to 183.0 mg/dL; p< 0.0001).

Iqbal 2006,81 a fair quality study with 64 subjects did a post hoc analysis and found a significant decline in total cholesterol and LDL in the subgroup of 23 patients with serum TSH at the end of the 12-month study within the range 0.2 – 2.0 mIU/L but no difference in lipids when considering the full study sample. This study reported the number of patients enrolled at baseline but not at study conclusion and did not indicate that intention-to-treat analysis was followed. Nagasaki 2008,86 a fair-quality study of 95 subjects found no improvement in any lipids. Like Iqbal, Nagasaki reported the number of patients enrolled at baseline but not at study conclusion and did not indicate that intention-to-treat analysis was followed.

Subclinical Hyperthyroidism

No controlled trials for the treatment of subclinical hyperthyroidism were found in the Helfand 2004 systematic review for the USPSTF.1 We identified two poor-quality controlled trials that assessed the effect of treatment of subclinical hyperthyroidism.77,83 (See Tables 16 through 20) Buscemi (2007)83 (N=14) was designed to assess the effects of treatment of subclinical hyperthyroidism on the heart as measured by blood pressure, basal heart rate, 24-hour heart rate, and atrial and ventricular premature beats; on bone turnover; and on bone density as measured by heel ultrasonometry (Stiffness Index). That study found a small, but significant weight gain (BMI 27.3 +/- 1.3 kg/m2 to 27.8 +/- 1.4 kg/m2 in treated patients vs. 27.9 +/- 1.2 kg/m2 to 28.1 +/- 1.0 kg/m2 in untreated controls; p < 0.05) after 12 months of treatment with methimazole. However, this study was weakened by failure to conceal allocation (alternating assignment) and lack of blinding (patients were given the option of changing their randomly assigned group; blinding of assessors was not described). In Yonem (2002)77 (N=20), 10 patients with subclinical hyperthyroidism randomized to treatment with propylthiouracil (nine patients) or radioiodine (one patient) were compared with 10 patients randomized to no treatment and found no change in bone mineral density or lipids, a small, but significant decrease in mean daytime systolic blood pressure (from 115.00 +/- 2.78 mmHg to 112.42 +/- 2.66 mmHg in treated patients vs. from 114.50 +/- 3.21 to 113.70 +/- 2.62 mmHg in untreated controls; p < 0.05), and inconsistent findings with regard to patient-reported outcomes. Of the 10 patients randomized to treatment, nine received propylthiouracil and one received radioactive iodine, and the analysis did not distinguish between treatments. In addition, patients randomized to control received no treatment and so the study was not adequately blinded.

Table 16. Treatment evidence table (subclinical hyperthyroidism).

Table 16

Treatment evidence table (subclinical hyperthyroidism).

Table 17. Body mass index (BMI) (kg/m2) (subclinical hyperthyroidism).

Table 17

Body mass index (BMI) (kg/m2) (subclinical hyperthyroidism).

Table 18. Blood pressure (mmHg) (subclinical hyperthyroidism).

Table 18

Blood pressure (mmHg) (subclinical hyperthyroidism).

Table 19. Bone mineral density (g/cm2) (subclinical hyperthyroidism).

Table 19

Bone mineral density (g/cm2) (subclinical hyperthyroidism).

Table 20. Lipids (mg/dL) (subclinical hyperthyroidism).

Table 20

Lipids (mg/dL) (subclinical hyperthyroidism).

Key Question 4. What are the harms of treatment of subclinical hypothyroidism and subclinical hyperthyroidism?

Until August, 2000, levothyroxine sodium was an unapproved marketed drug. In August 1997, the U.S. Food and Drug Administration declared levothyroxine sodium tablets a “new drug,” requiring manufacturers to submit a new drug approval to continue manufacturing it. The first product was approved in August, 2000. The product label mentions adverse effects on bone mineral density and the cardiovascular system, such as provocation of angina and arrhythmias and increased cardiac wall thickness. The U.S. Food and Drug Administration medical reviewer cited the following evidence regarding the safety of levothyroxine:

  • Elderly patients ≥ 60 years, with TSH suppressed to ≤0.1 mIU/L due to either subclinical hyperthyroidism or overtreatment with levothyroxine had approximately a 3-fold increased incidence of atrial fibrillation over a 10-year period compared with those with normal TSH levels89
  • Two systematic reviews found that long-term suppression therapy with levothyroxine led to decreases in bone mass in post-menopausal women. No adverse effect was found in men.62,63

The approval was conducted without requiring manufacturers to conduct studies to estimate the actual risks of short-term or long-term adverse effects. Consequently, scant information regarding the adverse effects of thyroid replacement is available. The 2004 Helfand review1 and the Cochrane review both mentioned that information is lacking about the frequency and severity of side effects when levothyroxine is used to treat subclinical hypothyroidism.

Three of the studies reviewed by Helfand (2004) provided some information on harms.71,72,75 In one study,11 out of 33 individuals, four treated with levothyroxine (and six with placebo) “felt worse.” Another study reported that out of 37 total individuals, one developed atrial fibrillation, and one developed angina.70 A third study found anxiety scores to be higher in the levothyroxine treated group.76 In another study of 20 individuals, two in the treatment arm dropped out, one for “nervousness” and another for “a sense of tachycardia.”68 The final study reported that within the treatment group a reduction in SF-36 vitality scores was found.74

None of newer studies of treatment for either subclinical hypothyroidism or subclinical hyperthyroidism systematically evaluated harms. An assessment of harms was likely not a part of any of the studies’ protocols, nor does it appear that study participants were provided with a list of potential harms and asked to identify those that they experienced. Of the six studies of treatment of subclinical hypothyroidism included in the current review, only one reported on harms, stating that none of the patients reported side effects that would have required withdrawal or dose reduction.86 One84 of the other five reported that one participant withdrew from the study after reporting side effects from 12 days of placebo treatment; a second82 reported that one subject in the placebo group dropped out after six months because of serious disease unrelated to thyroid function.

The long-term adverse effects of levothyroxine therapy may depend on careful clinical and laboratory monitoring and adjustment of dosage accordingly. The previous systematic review demonstrated that overtreatment with levothyroxine leading to undetectable TSH levels is common in practice.2 We did not identify more recent data to estimate the frequency of overtreatment in current practice.

For subclinical hyperthyroidism, one of the two studies of treatment reported that they “did not find any increase in confusion, myopathy, atrial fibrillation, and deep tendon reflexes incidence” in either the treatment or observation group before or after treatment/observation.77 The second study83 did not discuss harms.

While none of the six included subclinical hypothyroidism studies discussed unnecessary treatment as a harm, three of the studies reported a decline in the TSH value in the placebo subjects.82,85,86 This suggests that at least some individuals who are classified with subclinical hypothyroidism will spontaneously improve, and therefore, unnecessary treatment can occur if a strategy of early treatment is undertaken.

Cover of Screening and Treatment of Subclinical Hypothyroidism or Hyperthyroidism
Screening and Treatment of Subclinical Hypothyroidism or Hyperthyroidism [Internet].
Comparative Effectiveness Reviews, No. 24.
Rugge B, Balshem H, Sehgal R, et al.

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...