ORIGINAL INVESTIGATION. Association Between Thyroid Dysfunction and Total Cholesterol Level in an Older Biracial Population

Transcription

1 ORIGINAL INVESTIGATION Association Between Thyroid Dysfunction and Total Cholesterol Level in an Older Biracial Population The Health, Aging and Body Composition Study Alka M. Kanaya, MD; Fran Harris, MS; Stefano Volpato, MD; Eliseo J. Pérez-Stable, MD; Tamara Harris, MD; Douglas C. Bauer, MD Background: Thyroid dysfunction increases with age. Less is known about the prevalence of thyroid disease in older black adults and whether an association between thyroid function and serum cholesterol level exists, as in older white adults. Methods: A cross-sectional study of 2799 well functioning white and black participants, aged 70 to 79 years, were recruited for a population-based study. Participants underwent thyrotropin, free thyroxine, and total cholesterol testing; a medical history; and physical measurements. Results: Among the entire cohort, 94% were euthyroid based on biochemical testing results. Approximately 10% were taking thyroid hormones. Subclinical hypothyroidism was the most prevalent disorder (3.1% of all participants not taking thyroid hormones), but black men and women had lower rates of this condition than white men and women. After excluding those taking thyroid or lipid medication and adjusting for potential confounders, an elevated thyrotropin level ( 5.5 miu/ml) was associated with a 9 mg/dl (0.23 mmol/l) higher cholesterol level, and a suppressed thyrotropin level ( 0.35 miu/ ml) was associated with a 19 mg/dl (0.49 mmol/l) lower cholesterol level. Conclusion: Healthy community-dwelling older black adults have a lower prevalence of thyroid dysfunction compared with older white adults, but the association between increased thyrotropin and increased cholesterol levels is similar in both races. Arch Intern Med. 2002;162: From the Division of General Internal Medicine, Department of Medicine (Drs Kanaya, Pérez-Stable, and Bauer), Prevention Sciences Group (Ms F. Harris and Dr Bauer), Center for Aging in Diverse Communities (Drs Kanaya and Pérez-Stable), Medical Effectiveness Research Center for Diverse Populations (Dr Pérez-Stable), University of California, San Francisco; and the National Institute on Aging, National Institutes of Health, Bethesda, Md (Drs Volpato and T. Harris). WITH THE advent of sensitive assays for thyrotropin, isolated elevation and suppression of thyrotropin levels have been increasingly recognized in older populations. Subclinical hypothyroidism, defined as an elevated thyrotropin level with a normal serum free thyroxine (FT 4 ) concentration, occurs in 5% to 10% of patients older than 65 years and is especially prevalent in older white women. 1-9 Less information exists regarding the prevalence of subclinical hyperthyroidism in older black adults; estimated prevalences range from 2% to 12%, but results from population-based studies often have included patients receiving thyroxine treatment, which may overestimate the true prevalence. 10 Many contend that identifying and treating older patients with subclinical hypothyroidism may retard deleterious effects on the cardiovascular system, mostly via beneficial changes on lipoprotein levels. 11 The relationship between overt hypothyroidism and lipid abnormalities has been recognized for decades. Frank hypothyroidism is associated with elevated total cholesterol and low-density lipoprotein levels. 12,13 Studies investigating the relationship between lipids and subclinical hypothyroidism have shown variable results, ranging from no correlation to deleterious effects on total cholesterol, lowdensity and high-density lipoprotein cholesterol, lipoprotein(a), and apolipoproteins B and A1. 1,17-20 A meta-analysis 21 found that subclinical hypothyroidism is 2 to 3 times more frequent in persons with elevated cholesterol levels and that total cholesterol levels are slightly elevated in patientswithsubclinicalhypothyroidism. Similarly, intervention studies of thyroxine treatment for subclinical hypothyroidism have foundmixedresultsonlipidprofile, although the meta-analysis of 13 treatment studies found that thyroid substitution therapy resulted in an overall decrease in cholesterol of 15 mg/dl (0.4 mmol/l). 773

2 PARTICIPANTS AND METHODS STUDY DESIGN AND SETTING Participants in this cross-sectional study were enrolled in the Health, Aging and Body Composition study, an ongoing prospective cohort study of 3075 communitydwelling men and women, aged 70 to 79 years, enrolled or recruited at 2 clinical centers, in Memphis, Tenn, and Pittsburgh, Pa. To be eligible for this study, participants had to report no difficulty walking 0.4 km (one quarter mile) or up a flight of stairs. For our analyses, we used data gathered from 2799 participants who had thyrotropin test specimens, medical history, medication inventory, and physical measurements obtained at the second annual visit. All Health, Aging and Body Composition study participants were interviewed and examined at one of the clinical centers during the first and second annual visits between April 28, 1997, and June 14, MEASUREMENTS Questionnaire variables gathered from the first annual visit included self-identified racial group, level of education completed, and estimated family income. Physical activity was assessed at this visit by self-report of walking and exercise, assigning kilocalories per week to categories based on the Harvard Alumni Health Study variables. 22 In addition, participants reported smoking history (never, former, or current smoker) and alcohol use (number of drinks per week). Detailed data about physician-diagnosed medical conditions were collected at the first visit as well. Participants were asked specifically about previous diagnosis of Graves disease, hyperthyroidism, hypothyroidism, or an overactive or underactive thyroid gland. Participants were classified as having diabetes if they reported physician-diagnosed diabetes, were taking medication to control their diabetes, had a fasting serum glucose level greater than 125 mg/dl ( 6.9 mmol/l), or had a 2-hour oral glucose tolerance test result 200 mg/dl or greater than ( 11.1 mmol/l) at the baseline or second annual visit. At the second visit, their weight was measured using standardized protocols, 23 and body mass index (BMI) was calculated using the height measurement from the first visit. Trained interviewers confirmed medication used in the preceding 2 weeks by examination of pill bottles at the second visit as well. Fasting serum specimens were collected from each participant at the second annual visit. We measured thyrotropin levels by immunoassay (ACS; Chiron Diagnostics Corp, Emeryville, Calif) in 2799 participants. The normal range of thyrotropin by this assay is 0.35 to 5.5 miu/ml. The lower limit of detection of thyrotropin was 0.03 miu/ml. The coefficient of variation for thyrotropin was 4.1% at a level of miu/ml and was 3.6% at a level of 1.26 miu/ml. We routinely measured FT 4 by competitive immunoassay (ACS; Chiron Diagnostics Corp) on all of the participants with a thyrotropin value of 0.1 miu/ml or less or greater than 7.0 miu/ml. The normal range of FT 4 for this assay is 0.8 to 1.8 ng/dl (10-23 pmol/l). We determined a fasting total cholesterol level (VITROS; Johnson & Johnson, Rochester, NY ) for all participants at this visit as well. Little is known about the prevalence of thyroid dysfunction in older black adults and whether a similar association between thyroid function and cholesterol level exists among black adults, as in white adults. A crosssectional study 7 of middle-aged and older volunteers screened at health fairs found that white participants had a higher prevalence of abnormally suppressed or increased thyrotropin levels compared with black adults. We are unaware of any study that has examined the relationship between thyrotropin and cholesterol levels among older black adults. To examine the prevalence of self-reported thyroid disease and biochemically measured thyroid dysfunction and the relationship between thyroid disorders and total cholesterol levels by race, we used data from the Health, Aging and Body Composition study, a study of well functioning white and black men and women aged 70 to 79 years. We hypothesized that the black participants in this cohort might have a lower prevalence of measured thyroid dysfunction than the white participants, and that a significant relationship between thyrotropin and cholesterol levels would exist for both racial groups. RESULTS Among the entire cohort, 2799 (91%) had laboratory tests performed at the second annual visit. Sixty percent of them were white, and 51% were women. The mean age was similar among all subgroups (Table 1). men and women were more educated and reported higher family income than black men and women, respectively. Men of both races had similar mean BMIs, while black women had the highest BMIs of all. Overall, white men had the highest activity levels, and black women were the least active. Women of both races reported more known diagnoses of thyroid disease and increased use of thyroid hormones compared with men. Compared with black women, white women reported a significantly greater prevalence of known hypothyroidism (16.5% vs 6.2%, P=.001), while black women reported a higher prevalence of hyperthyroidism (9.7% vs 6.0%, P=.01). However, among the entire study population, only 2 black women were currently taking antithyroid medications. Significantly more white participants were taking lipid-lowering medications in both sex groups, and more white women were taking oral estrogen than black women. Mean values and ranges for thyrotropin, FT 4, and cholesterol for all participants are reported in Table 2. On average, FT 4 and cholesterol means were similar for both races in each sex group, but women had higher mean total cholesterol levels than men. The mean thyrotropin values were significantly higher for white men and women compared with black men and women. 774

3 DEFINITION OF THYROID DISEASE AND DYSFUNCTION Biochemically documented hypothyroidism was defined as a thyrotropin level of 7.0 miu/ml or higher and an FT 4 value of less than 0.8 ng/dl ( 10 pmol/l), while subclinical hypothyroidism was defined as a thyrotropin value of 7.0 miu/ml or higher with a normal FT 4 level. Biochemically documented hyperthyroidism was defined as a thyrotropin level of 0.1 miu/ml or less with an FT 4 value of greater than 1.8 ng/dl ( 23 pmol/l), and subclinical hyperthyroidism was defined as a suppressed thyrotropin level ( 0.1 miu/ml) with a normal FT 4. All participants with a thyrotropin value of greater than 0.1 miu/ml or less than 7.0 miu/ml were considered to be euthyroid. To distinguish between participants who were currently being treated with thyroid hormone therapy or antithyroid medications from those with laboratory evidence of a thyroid disorder, we termed the former group as those with thyroid disease and the latter group as those with thyroid dysfunction. Suboptimal thyroid treatment was evaluated for those with existing thyroid disease (hypothyroidism) who were taking thyroid hormone therapy. Participants who were not euthyroid while taking thyroid hormones were included in the analysis of suboptimal thyroid treatment. STATISTICAL ANALYSIS Means and SDs of the baseline characteristics were determined. Bivariate associations were examined by 2 test, Fisher exact test, or t test, when appropriate. Differences were considered significant at P.05. In analyzing the biochemically measured thyroid status of participants in each sex and race group, we stratified our analyses by use of thyroid hormone. Among the cohort taking thyroid hormones, we performed a stepwise logistic regression analysis to assess characteristics associated with suboptimal treatment with thyroid hormones. In addition, we evaluated the relationship of mean cholesterol levels adjusted for age, BMI, physical activity, current smoking, alcohol use, oral estrogen use, and diabetes with levels of thyroid dysfunction among those who were not taking thyroid medications or lipid-lowering drugs. We calculated a P for trend for this relationship for each sex and race subgroup. The relationship between thyrotropin and total cholesterol levels was analyzed using multivariate linear regression models, adjusting for potential confounders. Confounders were determined based on biologic plausibility (age and BMI) or strong association with serum cholesterol levels (alcohol use, oral estrogen use, and diabetes) or thyrotropin level (current smoking). Thyrotropin level was analyzed as a categorical, continuous, and log-transformed variable. All statistical analyses were performed using commercially available software (SAS, version 6.12; SAS Institute, Inc, Cary, NC). Because there were small numbers of persons with thyroid dysfunction in each sex and race subgroup and because a similar trend was seen between cholesterol level and thyroid function regardless of sex or race, we performed multivariate analyses of the entire group and adjusted for age, sex, race, BMI, current smoking, alcohol use, oral estrogen use, and diabetes. We excluded persons who were taking lipid-lowering medications, thyroid hormones, or antithyroid medications in these analyses. We performed similar analyses stratified by sex as well. Most (94%) of the cohort were euthyroid based on biochemical testing results (Table 3). The overall frequency of overt hypothyroidism and hyperthyroidism in the whole cohort was low ( 1% for each condition). Subclinical hypothyroidism was the most prevalent thyroid disorder in the entire cohort (3.9%), with white women having a significantly higher rate compared with the other 3 subgroups. Among the 271 participants taking thyroid hormones (10% of the entire cohort), most (83%) were euthyroid. women taking thyroid hormones appeared to have a lower percentage (76%) with euthyroid status compared with the other 3 sex and race subgroups (84% for the 3 subgroups combined), but this difference was not statistically significant (P=.34). Again, subclinical hypothyroidism was the most prevalent thyroid disorder overall (11.1%). We determined predictors of suboptimal treatment with thyroid hormones by stepwise logistic regression analysis. Low family income was the strongest predictor for having an abnormal thyrotropin level (P=.003), and those with the highest reported physical activity were least likely to be euthyroid (P=.03). Neither sex nor race was associated with suboptimal treatment of hypothyroidism. In analyses based on those who were not taking thyroid hormones or antithyroid medications, again subclinical hypothyroidism was the most prevalent thyroid abnormality, with white women having a significantly higher prevalence than black women (4.6% vs 1.7%, P=.005). There was a trend toward a higher prevalence of subclinical hypothyroidism in white men compared with black women (P=.06) and an equal prevalence among black men and black women. Table 4 shows adjusted mean cholesterol values by thyroid status for each sex and race strata of participants, excluding those who were taking lipid-lowering medications, thyroid hormones, or antithyroid medications. Participants with overt hypothyroidism had higher mean cholesterol values than those who were euthyroid among all subgroups. For all subgroups, there was a significant trend seen with decreasing cholesterol values for distinct levels of thyroid function, from frank biochemically documented hypothyroidism to hyperthyroidism. The only subgroup that had significantly elevated cholesterol levels associated with subclinical hypothyroidism, compared with the euthyroid state, was black women (P.001). The unadjusted cholesterol values by thyroid status for all sex and race groups were similar to these adjusted values. We performed multivariate linear regression analysis to determine whether an association existed between thyrotropin and cholesterol levels after adjusting for age, sex, race, BMI, current smoking, alcohol use, oral estrogen use, and diabetes. Similar results were obtained 775

4 Table 1. Characteristics of the Participants at the First and Second Annual Visits* Men Women Characteristic (n = 888) (n = 478) P Value (n = 797) (n = 636) P Value Mean age ± SD, y 74.9 ± ± ± ± Education, y (39.8) 350 (73.4) 411 (51.6) 460 (72.8) (60.2) 127 (26.6) 385 (48.4) 172 (27.2).001 Family income (yearly) $ (1.3) 84 (19.2) 47 (6.9) 174 (31.7) $ $ (68.9) 314 (71.9) (74.7) 362 (65.9).001 $ (29.8) 39 (8.9) 125 (18.4) 13 (2.4) Body mass index, kg/m ± ± ± ± Physical activity in prior week, kcal/wk (34.3) 262 (54.8) 418 (52.4) 439 (69.0) (29.4) 132 (27.6) (31.5) 139 (21.9) (36.3) 84 (17.6) 128 (16.1) 58 (9.1) Smoking Never 262 (29.5) 150 (31.4) 467 (58.6) 359 (56.5) Former 583 (65.7) 231 (48.3) (34.1) 202 (31.8).01 Current 43 (4.8) 97 (20.3) 58 (7.3) 75 (11.8) Alcohol, drinks/wk (55.0) 359 (75.6) 559 (70.2) 573 (90.2) (16.5) 49 (10.3) (12.8) 35 (5.5) (28.4) 67 (14.1) 135 (17.0) 27 (4.3) Self-reported medical problems Hypothyroidism 48 (5.6) 8 (1.7) (16.5) 37 (6.2).001 Hyperthyroidism 19 (2.2) 15 (3.2) (6.0) 58 (9.7).01 Diabetes 196 (22.1) 150 (31.4) (12.9) 175 (27.5).001 Current medication use at second annual visit Thyroid hormone 54 (6.1) 13 (2.7) (18.2) 59 (9.3).001 Lipid-lowering 206 (23.2) 67 (14.0) (23.2) 90 (14.2).001 Oral estrogen NA NA 261 (32.8) 86 (13.5).001 *Data are given as number (percentage) unless otherwise indicated. Percentages were calculated by dividing the number of participants with the characteristic by the total number who responded to the question, multiplied by 100. Some percentages do not sum to 100 because of rounding. The number of participants varies because of missing data for some characteristics. NA indicates not applicable. Table 2. Laboratory Values of Participants at Second Annual Visit* Men Women (n = 888) (n = 478) P Value (n = 797) (n = 636) Thyrotropin, miu/ml 2.96 ± ± ± ± Thyrotropin, range No. of participants with FT 4 testing FT 4, ng/dl 1.00 ± ± ± ± FT 4, range Cholesterol ± SD, mg/dl ± ± ± ± Cholesterol, range *Data are given as mean ± SD unless otherwise indicated. FT 4 indicates free thyroxine. To convert FT 4 from nanograms per deciliter to picomoles per liter, multiply by To convert cholesterol from milligrams per deciliter to millimoles per liter, multiply by P Value using thyrotropin level as a continuous, categorical, or log-transformed explanatory variable. We examined the results with thyrotropin levels categorized into high, normal, or low, using 2 definitions for euthyroid state ( 0.1 to 7.0 miu/ml and 0.35 to 5.5 miu/ml). For the entire cohort, a high thyrotropin level ( 5.5 miu/ml) was associated with a 9 mg/dl (0.23 mmol/l) higher total cholesterol, while a low thyrotropin level ( 0.35 miu/ml) was associated with a 19 mg/dl (0.49 mmol/l) lower total cholesterol (Table 5). Separate models were constructed for each sex, which showed similar associations between continuous and log-transformed thyrotropin levels, examined relative to cholesterol levels, while the relationship between categorical thyrotropin levels and cholesterol levels trended in the same direction, but did not reach statistical significance for many of the categories. Separate 776

5 Table 3. Prevalence of Thyroid Abnormalities by Race and Sex* Men Women Thyroid Status No. not taking thyroid drugs (n = 2526) Euthyroid 797 (95.6) 451 (97.0) 608 (93.3) 555 (96.5) Hypothyroid 5 (0.6) 2 (0.4) 7 (1.1) 4 (0.7) Subclinical hypothyroid 28 (3.4) 10 (2.2) 30 (4.6) 10 (1.7) Hyperthyroid 1 (0.1) 0 3 (0.5) 1 (0.2) Subclinical hyperthyroid 3 (0.4) 2 (0.4) 4 (0.6) 5 (0.9) No. taking thyroid hormones (n = 271) Euthyroid 46 (85.2) 12 (92.3) 121 (83.5) 45 (76.3) Hypothyroid (0.7) 4 (6.8) Subclinical hypothyroid 5 (9.3) 0 18 (12.4) 7 (11.9) Hyperthyroid (1.7) Subclinical hyperthyroid 3 (5.6) 1 (7.7) 5 (3.5) 2 (3.4) No. in total sample (N = 2797) Euthyroid 843 (94.9) 463 (96.9) 729 (91.5) 600 (94.6) Hypothyroid 5 (0.6) 2 (0.4) 8 (1.0) 8 (1.3) Subclinical hypothyroid 33 (3.7) 10 (2.1) 48 (6.0) 17 (2.7) Hyperthyroid 1 (0.1) 0 3 (0.4) 2 (0.3) Subclinical hyperthyroid 6 (0.7) 3 (0.6) 9 (1.1) 7 (1.1) *Data are given as number (percentage) unless otherwise indicated. Some percentages do not sum to 100 because of rounding. women vs black women, P =.01. women vs black women, P =.005. Excludes 2 participants who were taking antithyroid medications. women vs black women, P =.02. women vs black women, P =.003. Table 4. Cholesterol Levels by Thyroid Function for Those Not Taking Thyroid Hormones* Thyroid Status Men (n = 618) Men (n = 372) Women (n = 493) Women (n = 463) Hypothyroid (n = 13) ± ± ± ± 22.8 Subclinical hypothyroid (n = 62) ± ± ± ± 15.0 Euthyroid (n = 1858) ± ± ± ± 1.9 Subclinical hyperthyroid (n = 8) ± ± 36.6 NA ± 19.8 Hyperthyroid (n = 5) ± 34.8 NA ± ± 39.6 P for trend *Data are given as adjusted mean ± SE, unless otherwise indicated. NA indicates not applicable. To convert cholesterol from milligrams per deciliter to millimoles per liter, multiply by Table 5. Multivariate Results for the Association Between Thyrotropin and Cholesterol Levels* Entire Cohort Women Men Thyrotropin Level (95% CI) P Value Level (95% CI) P Value Level (95% CI) P Value Continuous 1.2 (0.7 to 1.7) (0.8 to 2.2) (0.2 to 1.5).01 Log 6.8 (4.5 to 9.1) (5.0 to 11.5) (1.7 to 8.2).003 Euthyroid, 0.1 to 7.0 miu/ml High 19.2 (10.3 to 28.2) (11.1 to 36.5) (1.9 to 27.2).02 Low 26.5 ( 45.6 to 7.3) ( 54.5 to 4.9) ( 51.5 to 10.6).20 Euthyroid, 0.35 to 5.5 miu/ml High 9.0 (3.4 to 15.5) (5.1 to 24.0) ( 6.1 to 12.1).51 Low 18.5 ( 32.7 to 4.3) ( 34.0 to 3.6) ( 44.7 to 0.6).04 *Adjusted for age, sex, race, body mass index, current smoking, alcohol use, oral estrogen use, and diabetes. CI indicates confidence interval. analyses of participants taking thyroid hormones but not lipid-lowering medications found significantly higher cholesterol levels for those with high thyrotropin values (15 mg/dl [0.39 mmol/l] higher total cholesterol if the thyrotropin level was 5.5 miu/ml) and a trend to lower cholesterol levels with low thyrotropin values, but these did not reach statistical significance. Among the 271 participants taking thyroid hormones, only 60 (22%) were taking lipid-lowering medications. Similarly, among the 96 participants who had 777

6 subclinical or overt hypothyroid dysfunction, 15 (16%) were taking lipid-lowering medications. The thyroid dysfunction group was not more likely to be treated with lipid-lowering agents than was the thyroid disease group (P=.17). COMMENT In this study of healthy community-dwelling older adults, the overall prevalence of any unrecognized thyroid dysfunction was lower than previously reported, and the use of thyroid supplementation was more common. Subclinical hypothyroidism was the most prevalent disorder, but black men and women had lower rates of this condition than white men and women. The lower prevalence of thyroid dysfunction may be partially explained by the selection of healthy participants who were able to walk 0.4 km (one quarter mile) and climb stairs at baseline as part of the Health, Aging and Body Composition study. We also confirmed an association between thyrotropin and cholesterol levels and extended this association to older black adults. For men and women in both racial groups, an elevated thyrotropin level was associated with a 9 mg/dl (0.23 mmol/l) higher total cholesterol, and a suppressed thyrotropin level was associated with a 19 mg/dl (0.49 mmol/l) lower total cholesterol. The prevalence of thyroid dysfunction in older populations has varied by ethnicity. There are limited published data reporting prevalence rates of subclinical thyroid disease in nonwhite populations. Lindeman et al 24 found that non-hispanic white women have higher rates of subclinical hypothyroidism compared with Hispanic women. However, there was insufficient power to examine the relationship between thyroid status and lipid abnormalities among the different sex and ethnicity strata in this study. The only prior publication reporting the prevalence of biochemically measured hypothyroidism or hyperthyroidism in blacks is a cross-sectional analysis of community-dwelling volunteers, 55 and older, at community health fairs. 7 The authors reported a lower prevalence of hypothyroidism and hyperthyroidism in blacks vs whites, and both ethnicities had higher prevalences of the 2 disorders in women. Our results support this finding, although we found much lower prevalences of overt thyroid disease in all 4 sex and ethnicity strata overall. Likewise, older ambulatory Chinese patients in Hong Kong had low frequencies ( 2%) of abnormal thyrotropin values, with Chinese women having more thyrotropin abnormalities compared with men. 25 Among the participants who were taking thyroid hormones, most were euthyroid based on biochemical testing results. Neither sex nor race was associated with being overtreated or undertreated with thyroid hormones among this subgroup. Those with lower family income were less likely to be euthyroid, possibly because of poorer access to health care, expense of successive laboratory testing, and prescription drug costs. Those with higher levels of physical activity were also less likely to be euthyroid, an observation that is unexplained. In our evaluation of thyrotropin abnormalities and cholesterol levels, we found that a similar association exists between these 2 biochemical test results for both sexes and ethnicities. However, despite finding elevated cholesterol levels associated with high thyrotropin levels, we did not detect overtreatment of hypercholesterolemia in patients with thyroid dysfunction. Similar proportions of older adults with existing treated hypothyroidism and those with thyroid dysfunction detected on laboratory screening were taking cholesterollowering medications. Although screening older adults with increased cholesterol levels for thyroid dysfunction seems prudent, our data do not show overuse of lipidlowering drugs for those with thyroid dysfunction. We were limited to using questionnaire results and physical measurements obtained for participants at the first annual visit and using thyrotropin, FT 4, and cholesterol test results from the second annual visit for our analyses. It is unlikely that there was a significant change in most self-reported variables, such as income, physical activity, smoking, or alcohol use, during one year. With regard to the biochemical tests, serum lipid levels have high day-to-day variability, and a more accurate association between thyroid function and cholesterol level may be seen with multiple fasting levels of the test specimens drawn simultaneously. However, this additional variability would only obscure the association that we observed. We also had limited power for some associations, such as race-specific relationships between thyrotropin and cholesterol levels. When a participant s thyrotropin level was 0.1 miu/ml or less or greater than 7.0 miu/ml, FT 4 test results were obtained. Other cut points might have given slightly different results. By using our conservative cut points for ordering FT 4 testing, we were unable to determine a biochemically defined thyroid status for 41 persons with thyrotropin levels of less than 0.35 miu/ml but greater than 0.1 miu/ml and for 106 participants with thyrotropin levels of greater than 5.5 miu/ml but less than 7.0 miu/ml. Because the overall prevalences of frank hypothyroidism and hyperthyroidism were so low in persons with more strikingly abnormal thyrotropin values, we doubt that we have misclassified overt thyroid disease. However, by using these conservative variables, we may have failed to recognize more subclinical thyroid dysfunction and may have underestimated the true prevalence of subclinical hypothyroidism and hyperthyroidism. We conclude that, in individuals who were not taking thyroid hormones, overt thyroid dysfunction was less prevalent than subclinical hypothyroidism in an ambulatory biracial older population. women have equally low prevalences of subclinical thyroid disease compared with white and black men. There is an association between thyrotropin and cholesterol levels, with an elevated thyrotropin being associated with a 9 mg/dl (0.23 mmol/l) increase in cholesterol and a suppressed thyrotropin being associated with a 19 mg/dl (0.49 mmol/l) decrease in total cholesterol. Future studies should examine the association between thyrotropin levels and specific lipoprotein fractions and investigate whether thyroid autoantibodies explain the differences in thyroid dysfunction prevalences between racial groups. 778

7 Accepted for publication August 7, The Health, Aging and Body Composition study was funded by grants N01-AG , N01-AG , and N01-AG from the National Institute on Aging, National Institutes of Health, Bethesda, Md. This work was also supported by pilot investigator grant to the Center for Aging in Diverse Communities, University of California, San Francisco, funded by the National Institute on Aging, National Institute of Nursing Research, and Office of Research on Minority Health, National Institutes of Health. It was also supported by faculty development in general internal medicine grant 1D08PE from the US Department of Health and Human Services, Washington, DC (Dr Kanaya). We acknowledge and thank the reviewers of the Health, Aging and Body Composition study publications committee for their valuable contributions to manuscript revision. Corresponding author: Alka M. Kanaya, MD, Division of General Internal Medicine, Department of Medicine, University of California, San Francisco, 1701 Divisadero St, Suite 554, San Francisco, CA ( REFERENCES 1. Canaris GJ, Manowitz NR, Mayor G, Ridgway EC. The Colorado Thyroid Disease Prevalence Study. Arch Intern Med. 2000;160: Nystrom E, Bengtsson C, Lindquist O, Noppa H, Lindstedt G, Lundberg PA. Thyroid disease and high concentration of serum thyrotropin in a population sample of women: a 4-year follow-up. Acta Med Scand. 1981;210: Falkenberg M, Kagedal B, Norr A. Screening of an elderly female population for hypo- and hyperthyroidism by use of a thyroid hormone panel. Acta Med Scand. 1983;214: Sawin CT, Bigos ST, Land S, Bacharach P. The aging thyroid: relationship between elevated serum thyrotropin level and thyroid antibodies in elderly patients. Am J Med. 1985;79: Rosenthal MJ, Hunt WC, Garry PT, Goodwin JS. Thyroid failure in the elderly: microsomal antibodies as discriminant for therapy. JAMA. 1987;258: Robuschi G, Safran M, Braverman LE, Gnudi A, Roti E. Hypothyroidism in the elderly. Endocr Rev. 1987;8: Bagchi N, Brown TR, Parish RF. Thyroid dysfunction in adults over age 55 years: a study in an urban US community. Arch Intern Med. 1990;150: Sawin CT, Chopra D, Azizi F, Mannix JE, Bacharach P. The aging thyroid: increased prevalence of elevated serum thyrotropin levels in the elderly. JAMA. 1979;242: Samuels MH. Subclinical hypothyroid disease in the elderly. Thyroid. 1998;8: Marquess E, Haden ST, Utiger RD. Subclinical thyrotoxicosis. Endocrinol Metab Clin North Am. 1998;27: Arem R, Escalante D. Subclinical hypothyroidism: epidemiology, diagnosis, and significance. Adv Intern Med. 1996;41: O Brien T, Dinneen SF, O Brien PC, Palumbo PJ. Hyperlipidemia in patients with primary and secondary hypothyroidism. Mayo Clin Proc. 1993;68: Stone NJ. Secondary causes of hyperlipidemia. Med Clin North Am. 1994;78: Althaus BU, Staub JJ, Ryff-De Leche A, Oberhansli A, Stahelin HB. LDL/HDLchanges in subclinical hypothyroidism: possible risk factors for coronary heart disease. Clin Endocrinol (Oxf). 1988;28: Parle JV, Frandlyn JA, Cross KW, Jones SR, Sheppard MC. Circulating lipids and minor abnormalities of thyroid function. Clin Endocrinol (Oxf). 1992;37: Bogner U, Arntz HR, Peters H, Schleusener H. Subclinical hypothyroidism and hyperlipoproteinaemia: indiscriminate L-thyroxine treatment not justified. Acta Endocrinol (Copenh). 1993;128: Bauer DC, Ettinger B, Browner WS. Thyroid function and serum lipids in older women: a population-based study. Am J Med. 1998;104: Ozata M, Yildirimkaya M, Yilmaz K, et al. The effects of thyroid status on serum apolipoprotein A-I containing lipoprotein particles. Horm Metab Res. 1998;30: Elder J, McLelland A, O Reilly DS, Packard CJ, Series JJ, Shepherd J. The relationship between serum cholesterol and serum thyrotropin, thyroxine and triiodothyronine concentrations in suspected hypothyroidism. Ann Clin Biochem. 1990;27: Kung AW, Pang RW, Lauder I, Lam KS, Janus ED. Changes in serum lipoprotein(a) and lipids during treatment of hyperthyroidism. Clin Chem. 1995;41: Tanis BC, Westendorp GJ, Smelt HM. Effect of thyroid substitution on hypercholesterolaemia in patients with subclinical hypothyroidism: a reanalysis of intervention studies. Clin Endocrinol (Oxf). 1996;44: Paffenbarger RS Jr, Blair SN, Lee IM, Hyde RT. Measurement of physical activity to assess health effects in free-living populations. Med Sci Sports Exerc. 1993; 25: Lohman TG, Going SB. Multicomponent models in body composition research: opportunities and pitfalls. Basic Life Sci. 1993;60: Lindeman RD, Schade DS, LaRue A, et al. Subclinical hypothyroidism in a biethnic, urban community [see comments]. J Am Geriatr Soc. 1999;47: Kung AW, Janus ED. Thyroid dysfunction in ambulatory elderly Chinese subjects in an area of borderline iodine intake. Thyroid. 1996;6: