Disparities in the Diagnosis and Pharmacologic Treatment of High Serum Cholesterol by Race and Ethnicity: Data From the Third National Health and Nutrition Examination Survey

Background  Serum cholesterol is one of the most important modifiable risk factors for coronary artery disease. There are conflicting data on racial and ethnic variation in the treatment of high cholesterol.

Methods  We analyzed data from the Third National Health and Nutrition Examination Survey, a nationally representative cross-sectional survey conducted between 1988 and 1994. Participants included 7679 white, 4467 African American, and 4113 Mexican American adults older than 25 years who completed the household adult questionnaire. The adjusted odds of serum cholesterol screening and of taking a prescription medication to lower serum cholesterol among African Americans and Mexican Americans were compared with those of whites, controlling for differences in age, sex, income, educational level, insurance status, comorbid illness, and having a regular source of health care.

Results  African Americans and Mexican Americans were significantly less likely than whites to report ever having had their blood cholesterol checked (odds ratio, 0.7 for both; P<.001). Among individuals with high cholesterol who were told to take a medication, African Americans (P<.001) and Mexican Americans (P = .05) were less likely than whites to be taking a cholesterol-lowering agent (odds ratios, 0.3 and 0.5, respectively). Individuals who reported being told they had high cholesterol had significantly higher serum cholesterol measurements (from the laboratory examination) than those who reported being told their cholesterol was not high (234 vs 198 mg/dL [6.05 vs 5.12 mmol/L]; P<.001).

Conclusions  African Americans and Mexican Americans were less likely to report serum cholesterol screening than whites. Even when identified as having high cholesterol that required medication, African Americans and Mexican Americans were less likely than whites to be taking cholesterol-lowering agents.

RECENT DECLINES in death rates from coronary heart disease (CHD) have been more significant in whites than in minority populations in the United States.^1-3 Many studies^4-8 have shown that Latinos and African Americans receive less aggressive treatment for CHD than do whites. However, there are conflicting data on racial and ethnic variation in the primary prevention of CHD. Elevated serum cholesterol is one of the most important modifiable risk factors for CHD,^9-11 and treating hypercholesterolemia lowers the risk of developing disease.^12-16 Given the strength of this evidence, the National Cholesterol Education Program recommends measuring the serum cholesterol in all adults older than 20 years at least once every 5 years.⁹

Previous studies have documented low rates of treatment for elevated cholesterol levels^17-22 and variation in management by physician^23-28 and patient^29-37 characteristics. In the San Antonio Heart Study, Mexican Americans were less likely than whites to be aware of and to be undergoing treatment for high cholesterol.^38,39 In the 1988-1990 Behavioral Risk Factor Surveillance System (BRFSS), African Americans and Latinos were less likely than whites to report cholesterol screening.^40,41 Paradoxically, in the same study, African Americans and Latinos were more likely than whites to report taking a prescription medication for high cholesterol.⁴¹ In the Cardiovascular Health Study,⁴² among a cohort 65 years or older, only 20% of eligible individuals were treated for high cholesterol and no treatment difference was noted between whites and African Americans. These reports conflict with data from the Atherosclerosis Risk in Communities study⁴³ and the Heart and Estrogen/Progestin Replacement Study,³⁴ in which African Americans were less likely than whites to receive lipid-lowering medication.

The last national study⁴¹ examining differences in treatment for hypercholesterolemia by race and ethnicity used the 1988-1990 BRFSS data. This study may not fully reflect the change in treatment patterns after the approval of the 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (the "statins") in 1987.¹⁷ The use of data from the Third National Health and Nutrition Examination Survey (NHANES III), collected between 1988 and 1994, permits further examination of time trends in treatment as the statins became widely used. To accurately assess treatment for high blood cholesterol, detailed medication information obtained by review of actual medication bottles collected during NHANES III will be analyzed. With these nationally representative data, the purpose of our study is to examine the effect of race and ethnicity on the receipt of cholesterol screening and on the treatment of high serum cholesterol.

NHANES III was conducted by the National Center for Health Statistics at 89 survey locations between 1988 and 1994.^44,45 The survey is a cross-sectional nationally representative sample of the US civilian noninstitutionalized population. NHANES III used a stratified multistage probability cluster design with oversampling of Mexican Americans, African Americans, and persons older than 60 years.⁴⁶ The survey consists of multiple components, including a household interview, a physical examination, and laboratory tests.

A total of 16 884 people older than 25 years completed the household adult questionnaire. Given guidelines recommending a serum cholesterol measurement every 5 years beginning at the age of 20 years,¹ we limited our sample to those 25 years or older. Respondents were asked to identify themselves as white, black, Mexican American, or other. We excluded those individuals (n = 625) identified as other race, including those of Hispanic origin who did not identify themselves as Mexican American.

Respondents were asked a series of questions regarding serum cholesterol screening and treatment in the household adult questionnaire. Of the subjects older than 25 years, 96% (n = 15 686) answered these questions and compose the sample for this study. All respondents were asked: "Have you ever had your cholesterol checked?" If respondents answered yes, they were asked: "Have you ever been told by a doctor or other health professional that your blood cholesterol level was high?" If they reported being told their blood cholesterol level was high, they were asked: "Because of your high blood cholesterol, have you ever been told by a doctor or other health professional to take prescribed medication?"

If a respondent reported taking any prescription medicine, the interviewer asked to see the medication container to record the name of the product. If the container was not available, the interviewer probed for the medication name. Medication data were available for 14 655 individuals. Lipid-lowering agents identified included cholestyramine, clofibrate, cholestipol, gemfibrozil, lovastatin, niacin, and probucol. If a person reported taking 1 or more of these medications, the person was considered to be taking a lipid-lowering agent. Laboratory data were obtained from respondents at the mobile examination center and were available for 10 820 participants. Coronary heart disease risk factors, including hypertension, diabetes mellitus, coronary artery disease, and smoking, were based on self-report. Participants were considered smokers if they reported smoking at least 100 cigarettes during their lifetime and were current smokers.

Data were weighted to account for the unequal probability of selection that resulted from the survey cluster design, nonresponse, and oversampling of certain target populations.^46,47 Statistical analysis was performed using computer software (Stata, version 6.0)⁴⁸ to take into account the complex sampling design. Sampling weights were used to calculate population estimates, and sampling strata and primary sampling units were accounted for to estimate variances and to test for significant differences. All results are presented as unweighted counts and weighted percentages and odds ratios.

Bivariate analyses were performed to assess the significance of associations between race and ethnicity and receipt of cholesterol screening, being told your serum cholesterol level was high, being told to take a cholesterol-lowering medication, and taking a cholesterol-lowering medication. Multivariate analyses were performed to examine the association of race and ethnicity on the receipt of cholesterol screening and on taking a cholesterol-lowering medication, accounting for the independent effects of sociodemographic factors (age, sex, income, and educational level), health insurance coverage, comorbid disease (diabetes mellitus, hypertension, CHD, and tobacco use), and having a regular source of care. Multicollinearity was assessed using correlation coefficients between the independent variables. None of the correlation coefficients between independent variables were greater than 0.6.

Of all adults older than 25 years, 63% reported that they had their cholesterol level checked at least once (Table 1). Mexican Americans and African Americans were much less likely to report serum cholesterol screening compared with whites (37% and 50%, respectively, vs 66%; P<.001). The unadjusted and adjusted odds ratios and predicted percentages for receipt of cholesterol screening are displayed in Table 2. The difference in screening by race and ethnicity persists in a multivariate analysis controlling for the independent effect of sociodemographic factors, type of health insurance, comorbid illness, and having a regular source of care. Those who were poor, less educated, uninsured, or smokers were also less likely to report the receipt of cholesterol screening.

Overall, 35% of those who had their cholesterol level checked were told the value was high (Table 1). Of those with high cholesterol, 22% were told to take medication. Of those told to take medication for high cholesterol, 46% were documented to be taking a cholesterol-lowering agent during the month before the survey. Most individuals were taking either lovastatin or gemfibrozil (data not shown). Of those who were told to take prescription medication, Mexican Americans and African Americans were less likely than whites to be taking medication to control high blood cholesterol (28% and 29%, respectively, vs 49%; P<.001). Table 3 displays the unadjusted and adjusted odds ratios and predicted percentages for taking a cholesterol-lowering medication among those who were told to take medication. The difference in taking a cholesterol-lowering medication by race and ethnicity persists after adjusting for other sociodemographic factors, type of health insurance, comorbid illness, and having a regular source of care. As expected, other factors that were associated with taking a cholesterol-lowering agent included a previous history of coronary artery disease, having diabetes mellitus, and reporting a regular source of care.

To assess the validity of the self-report of high cholesterol, we compared mean cholesterol levels for those who reported being told they had high cholesterol with those who were told they did not have high cholesterol. The mean total serum cholesterol level of those who reported being told their cholesterol level was high was 234 mg/dL (6.05 mmol/L) compared with 198 mg/dL (5.12 mmol/L) for those who were told their cholesterol level was not high (P<.001). The mean cholesterol level did not significantly (P = .13) differ by race or ethnicity among those who were told their serum cholesterol was high (data not shown). In addition, 99% (n = 298) of the individuals who were taking a cholesterol-lowering medication reported being told that their cholesterol was high and that they should be taking a cholesterol-lowering medication.

African Americans and Mexican Americans were less likely than whites to report ever having been screened for high cholesterol, and of those who were told to take a prescription medication, were less likely to be taking a cholesterol-lowering agent. Our results add important information to the growing literature on racial and ethnic variation in the treatment of CHD. Hypercholesterolemia is an important risk factor in the development of CHD, and recent national data⁴⁹ suggest that the proportion of individuals with an elevated cholesterol level is similar across different racial and ethnic groups. Although the death rate from CHD is lower among Latinos, declines in mortality have occurred to a smaller degree among Latinos than whites.³ Among African Americans, CHD mortality is higher and declines have also been less than in the white population.² The difference in primary prevention we describe could contribute to this variation in disease rates.

The strength of our study is the use of data designed to provide accurate estimates about African Americans and Mexican Americans. Collected through 1994, data from NHANES III represent the most current nationally representative data on the treatment of high blood cholesterol after the introduction and widespread use of the statins. In addition, estimates about pharmacologic treatment were made from direct review of medication bottles.

Cholesterol screening rates have steadily increased during the past 20 years. In 1983, only 35% of adults reported that they had their cholesterol level checked, compared with 65% in 1990.^20,50 Consistent with these national surveys, we found that 63% of adults older than 25 years reported having undergone a serum cholesterol test. Our results of the racial and ethnic variation in screening rates are consistent with data from the 1988-1990 BRFSS.⁴¹ However, we report a lower rate of cholesterol screening for Mexican Americans than found for Hispanic individuals in the BRFSS (42% vs 58%). The higher screening rate for Hispanic individuals in the BRFSS is not adjusted to account for insurance status or income. In our study, income and insurance status were strong correlates for cholesterol screening. It is possible that differences in income and insurance status could explain the observed variation in screening rates between the 2 studies.

In our analyses, differences in screening for high cholesterol by race and ethnicity persist after accounting for the independent effects of income, educational level, health insurance status, comorbid disease, and having a regular source of care. These results suggest that lower rates of cholesterol screening in African Americans and Mexican Americans are not solely a result of problems with health care access. Potential explanations for the observed differences may include unmeasured socioeconomic or clinical factors, patient preferences, or physician characteristics, such as knowledge of screening recommendations or bias.

Although the first National Cholesterol Education Program guidelines were released in 1988⁵¹ and the statins were approved for use in 1987,¹⁷ NHANES III data were collected before the publication of the most definitive primary and secondary prevention trials.^52-55 Recent studies^56-58 have demonstrated a modest increase in the use of cholesterol-lowering agents since the publication of these landmark studies. In NHANES III, we estimated that 2.5% of US adults older than 25 years were undergoing treatment with a lipid-lowering agent (data not shown), consistent with other national surveys.²⁰ The period for NHANES III may account for the low rate of screening and treatment for hypercholesterolemia overall, but would not explain the variation in treatment by race and ethnicity that we have demonstrated.

In the only previous analysis⁴¹ of nationally representative data, African Americans and Hispanic individuals were more likely to report that they were taking a lipid-lowering agent than were whites. These data were based on self-report and not on review of medication bottles. Using data from a medication bottle review, Mexican Americans and African Americans in NHANES III were less likely to actually be taking a cholesterol-lowering agent among those who were told to take medication. It is reassuring that we did not find any racial or ethnic variation in the report of physician recommendation for nonpharmacologic interventions to lower serum cholesterol (data not shown) or to take a medication for high cholesterol.

Our results regarding lower treatment rates of high blood cholesterol among African Americans are consistent with 2 previous studies. In the Atherosclerosis Risk in Communities study,⁴³ Nieto and colleagues observed a cohort of individuals in 4 US communities and assessed cholesterol treatment through medication bottle review. They found that of those with high cholesterol, 28% of whites and 20% of African Americans were undergoing treatment.⁴³ In the Heart and Estrogen/Progestin Replacement Study,⁵⁹ a large secondary prevention trial of estrogen replacement therapy, a high percentage of women were taking cholesterol-lowering agents. However, even among clinical trial participants, white women were still more likely to be treated for hypercholesterolemia compared with African American women (71% vs 52%).³⁴

We found that less than half of the individuals who reported being told by a physician to take a lipid-lowering agent were actually taking medication to treat high cholesterol, suggesting either a low likelihood of starting a recommended therapy or, as documented by previous studies,^60,61 a high discontinuation rate. The observed racial and ethnic differences in pharmacologic treatment could reflect concerns about cost or differences in coverage for prescription medications⁶² that were not measured in NHANES III. Patient- or physician-level health beliefs, social norms, and differential knowledge regarding the benefits of treatment of high cholesterol may also play a role. Differences in treatment rates could also reflect lack of patient trust or ineffective communication between the physician and patient regarding treatment.

Our results also highlight the importance of access to preventive services for the primary prevention of CHD and are consistent with previous reports^63,64 on limited access to preventive care among the uninsured. In this study, having a regular source of care was one of the strongest predictors for reporting cholesterol screening and for taking a cholesterol-lowering medication. The uninsured were significantly less likely to report ever having their cholesterol level checked. Although we did find fewer uninsured individuals taking a cholesterol-lowering medication, the limited sample size did not provide enough power to detect a statistically significant difference in prescription drug use by insurance categories.

The main limitation of this study is the self-report of cholesterol screening. We performed several validity checks of the self-reported screening data. Of those taking a lipid-lowering agent, 99% reported that they had their cholesterol checked and were told that they required medication. In addition, individuals who reported being told their cholesterol was elevated had a significantly higher measured serum cholesterol from the laboratory examination compared with those who were told that their cholesterol was not high. Previous studies^65-67 suggest that individuals underreport cholesterol screening, although there is no evidence that African Americans or Mexican Americans are more likely to underreport screening than whites. To diminish the racial and ethnic differences we describe, the rate of underreporting would have to be significantly different among these populations.

This description of racial and ethnic variation in the primary prevention of CHD adds important information to the growing literature on disparities in health care in the United States. African Americans and Mexican Americans are less likely than whites to report cholesterol screening and, if indicated, are less likely to be taking a cholesterol-lowering medication. Further research is needed to understand the interaction of patient, physician, and organizational factors that contribute to this variation to address this disparity in the primary prevention of CHD. To understand differences in the treatment of high cholesterol, identifying patient preferences and barriers to adherence for lipid-lowering agents, such as trust in physician recommendations and knowledge about the benefits of treatment, will be essential. In addition, larger studies of secondary prevention are needed to understand if these racial disparities exist in higher-risk populations.

Accepted for publication August 29, 2001.

This study was supported in part by grant U54RR14616 from the Research Centers in Minority Institutions, National Center for Research Resources, National Institutes of Health, Bethesda, Md.

Corresponding author and reprints: Karin Nelson, MD, MSHS, VA Puget Sound Health Care System, 1660 S Columbian Way, Mail Stop S-111-GIMC, Seattle, WA 98108-1597 (e-mail: karin.nelson@med.va.gov).