|Home | About | Journals | Submit | Contact Us | Français|
Minorities have a higher prevalence of hypertension, a major risk factor for cardiovascular disease, which contributes to racial/ethnic disparities in morbidity and mortality in the US. Many modifiable health behaviors have been associated with improved blood pressure control, but it is unclear how racial/ethnic differences in these behaviors are related to the observed disparities in blood pressure control.
Cross-sectional analyses were conducted among 21,489 US adults aged >20 years participating in the National Health and Nutrition Examination Survey from 2001–2006. Secondary analyses were conducted among those with a self-reported diagnosis of hypertension. Blood pressure control was defined as systolic values <140 mmHg and diastolic values <90 mmHg (or <130 mmHg and <80 mmHg among diabetics, respectively).
In primary analyses, Non-Hispanic Blacks had 90% higher odds of poorly controlled blood pressure compared to non-Hispanic Whites after adjustment for sociodemographic and clinical characteristics (p <0.001). In secondary analyses among hypertensive subjects, non-Hispanic Blacks and Mexican-Americans had 40% higher odds of uncontrolled blood pressure compared to non-Hispanic Whites after adjustment for sociodemographic and clinical characteristics (p <0.001). For both analyses, the racial/ethnic differences in blood pressure control persisted even after further adjustment for modifiable health behaviors, which included medication adherence in secondary analyses (p <0.001 for both analyses).
Although population-level adoption of healthy behaviors may contribute to reduction of the societal burden of cardiovascular disease in general, these findings suggest that racial/ethnic differences in some health behaviors do not explain the disparities in hypertension prevalence and control.
African-Americans and US-born Mexican-Americans, have higher risks for all-cause mortality and cardiovascular disease (CVD) morbidity and mortality compared to Whites.1–6 A large proportion of excess morbidity and mortality among minorities may be attributable to hypertension.7–9 As a result, identifying modifiable factors that contribute to racial and ethnic differences in rates of hypertension and blood pressure (BP) control is an essential step toward eliminating disparities in CVD outcomes.
Current literature suggests that some modifiable health behaviors, including moderate alcohol consumption, decreased sodium intake, smoking cessation, increased physical activity, and increased intake of fruits, vegetables, and whole grains may reduce incidence of hypertension and improve BP control among those with and without a prior diagnosis of hypertension.10–24 To date, however, data regarding racial and ethnic differences in the distributions of modifiable health behaviors is varied. Prior studies have shown African-Americans and Mexican-Americans as having a higher prevalence of some risk factors (e.g., low rates of physical activity) but similar or lower prevalence of others (e.g., smoking and alcohol intake) when compared to Whites.25–30
Despite numerous studies of potential pharmacological and socioeconomic mediators for racial differences in hypertension prevalence and severity, there has been little examination of the possible contribution of racial and ethnic differences in health behaviors to disparities in hypertension prevalence.2, 31–33 Therefore, we examined a nationally representative sample of adults participating in the 2001–2006 National Health and Nutrition Examination Survey (NHANES) to assess the relative contribution of racial and ethnic differences in modifiable health behaviors to disparities in hypertension prevalence and control.
NHANES is a complex, stratified multistage probability sample of the non-institutionalized US population with over-sampling for certain groups including African-Americans and Mexican-Americans. Since 1999, NHANES has been conducted continuously with data released every two years. We considered 2001 to 2006 as one time period and generated 6-year sample weights, as recommended in the online survey documentation, for all analyses.34 We limited our analyses to 2001–2006 data because the 2007–2008 survey cycle no longer contains data on individual physical activities needed to calculate total physical activity (MET-hours per week) in the same way as prior cycles and the NHANES 1999–2000 nutrient intakes were calculated using an older version of the USDA Survey Nutrient database compared to the other cycles.35 Detailed descriptions of the history, design, informed consent, data collection, and available data for NHANES are available on the National Center for Health Statistics website.36 Our protocol was approved by the Partners Institutional Review Board.
Using data from the NHANES 2001–2006, the eligible sample consisted of non-Hispanic White, non-Hispanic Black, and Mexican-American adult participants aged >20 years who were not pregnant, completed the survey questionnaire, had BP measured during the mobile exam center (MEC) evaluation, and reliably completed the dietary recall for the nutrients of interest as defined by NHANES online documentation. Participants reporting race and ethnicity of “Other Hispanic” and “Other” were excluded due to relatively small sample sizes. Secondary analyses were conducted among participants with a prior diagnosis of hypertension.
The primary outcome variable was BP control defined, as a mean systolic BP <140 mmHg (<130 mmHg in diabetic patients) and/or mean diastolic BP <90 mmHg (<80 mmHg in diabetic patients). Mean blood pressure was calculated from four separate measurements obtained by certified examiners during the MEC evaluation according to the study protocol.37 The primary independent variable was self-reported race/ethnicity, coded in NHANES as non-Hispanic White, non-Hispanic Black, or Mexican-American.
Covariates included participants’ self-reported age (continuous) and sex, financial status using the poverty income ratio (PIR), low educational level, and insurance status. Other potential confounders included impaired functional status, self-reported general health status, body mass index, and diabetes.
We included variables related to modifiable health behaviors based on prior literature implicating their role in the incidence and control of high BP.11, 13–15, 20, 22, 38, 39 These included: self-reported current smoking status (yes/no), total MET-hours per week calculated from all self-reported individual leisure time physical activities, total energy intake, total daily nutrient intakes of sodium, alcohol, and fiber based on participants’ 24-hour dietary recall. Fiber was used as a proxy for fruit, vegetable, and whole grain intake based on literature suggesting that fiber content mediates their antihypertensive effects.38, 39
Patients with previously diagnosed hypertension may change their health behaviors as a consequence of physician counseling, and thus these factors could have different associations with BP control in this group compared to the general population.40, 41 Therefore, we conducted secondary analyses restricting the sample to subjects with known hypertension. We defined hypertensive participants as those with a positive response to the NHANES survey question “Have you ever been told by a doctor or other health professional that you had…high blood pressure?” Among hypertensives, those responding “no” to “because of your high blood pressure/hypertension, have you ever been told to take prescribed medicine?” were classified as “medication ineligible”. Among medication eligible hypertensives, those who also endorsed current use of these medications by answering “yes” to “Are you now taking prescribed medicine” were classified as “medication adherent”, with the others classified as “non-adherent”.
Bivariable analyses comparing proportions of subjects meeting recommendations for lifestyle behaviors and BP control were conducted by race/ethnicity, age-standardized based on the 2000 US Census Population. Multivariable regression modeling was used to determine the relationship of race with BP control. The base model controlled for age, sex, education, income, insurance status, general health status, functional status, diabetes, and body mass index (Model 1). In the next model, all modifiable health behaviors (smoking status, physical activity, sodium, fiber, and alcohol) were added to the base model (Model 2). Interaction terms between race and each behavior were added separately to Model 2 to evaluate whether associations differed by race/ethnicity. In secondary analyses, these analyses were repeated among known hypertensive participants and, in addition, medication adherence was added to generate a final adjusted model (Model 3). For Model 2 (or Model 3 for hypertensive subpopulation), a greater than 10% change in the estimated odds ratio for race/ethnicity was considered a significant difference compared to their respective Model 1. To account for the complex sampling design of NHANES, we analyzed all data with SAS® version 9.2 (SAS Institute) and SUDAAN version 10.1 (RTI International).
The final sample for analysis included 21,489 subjects who represent about 37.7 million Americans. Of these, 83.2% were Non-Hispanic Whites, 10.2 % were Non-Hispanic Blacks, and 6.6% were Mexican-Americans. The age adjusted proportion of subjects with the demographic, sociodemographic, or clinical characteristics within each racial/ethnic group is shown in Table 1. Compared to non-Hispanic Whites, a higher proportion of minorities were impoverished, had less than twelve years of education, and were uninsured. Minorities also were more likely to have diabetes and high BMI.
The age-adjusted proportions of subjects with modifiable health behaviors within each racial/ethnic group are shown in Table 1. Compared to non-Hispanic Whites, minorities reported less physical activity, with almost 27% of Non-Hispanic Blacks and about one-third of Mexican-Americans reporting no leisure time physical activity. For sodium intake, more minorities consumed <2000 mg/day when compared to non-Hispanic Whites, who had the largest proportion of subjects consuming >4500 mg/day. Fiber intake was generally low across all racial and ethnic groups; however, a higher proportion of Mexican-Americans achieved the recommended levels of >25 g/day. A higher proportion of African-Americans and Mexican-Americans reported low total energy intake of less than 1400 kcal/day. Rates of smoking were similar across all groups, but a higher proportion of non-Hispanic Whites consumed more than the equivalent of two drinks per day.
In age-adjusted analyses, Non-Hispanic Blacks had a much higher prevalence (27.4%) of uncontrolled BP when compared to other racial/ethnic groups (Non-Hispanic White 17.0%, Mexican-Americans 20.2%). Data from the multivariate regression models is shown in Table 2. In the base model, non-Hispanic Blacks had 90% higher odds of poorly controlled BP when compared to non-Hispanic Whites after adjustment for age, gender, and sociodemographic and clinical characteristics (p <0.001). This difference remained unchanged after addition of self-reported physical activity, smoking status, and dietary intakes of nutrients of interest (p <0.001). Interaction terms between race and each health behavior were added separately to Model 2 for each behavior and were not significant (data not shown).
In the final model, age, race, gender, diabetes, and BMI are all strong independent predictors of BP control (please see Table S1 of the Online Supplement http://hyper.ahajournals.org). Among health-related behaviors, smoking and sodium intake were not significant predictors of BP control. Although there appeared to be trend toward poorer BP control with increasing total daily caloric intake, this was not statistically significant. However, the highest level of physical activity (>50 MET-hours/week) had a 30% lower odds of poorly controlled BP compared to all other levels of activity (p=.02). In addition, those who reported consuming >28 g/day of alcohol had 85% higher odds of poor BP control compared with lower levels of alcohol consumption (p=.006); those consuming less than five grams of fiber per day had 74% higher odds of poor BP control compared with those meeting or exceeding the recommended 25 grams/day (p=.06).
The sub-sample of respondents who reported being told by a physician they had high BP included 6043 subjects who represent about 9.2 million Americans, of whom 82.2% were Non-Hispanic Whites, 13.7% were Non-Hispanic Blacks, and 4.1% were Mexican-Americans. Across all racial groups, the sample of known hypertensive respondents had higher proportions of subjects reporting fair/poor general health status, at least one functional limitation, diabetes, smoking, and obesity than the total sample (please see Table S2 of the Online Supplement http://hyper.ahajournals.org). The racial/ethnic disparity in the prevalence of uncontrolled BP persisted; 36.8% non-Hispanic Black and 37.6% Mexican-American hypertensives had poor control compared to 27.7% non-Hispanic Whites.
In multivariate regression models, known hypertensive non-Hispanic Blacks had 40% higher odds of uncontrolled BP after adjusting for sociodemographic and clinical characteristics when compared to non-Hispanic Whites in the base model (p=.02) (Table 3). Mexican-Americans had a 41% higher adjusted odds compared to Whites (p=.06) of borderline statistical significance. Adjustment for self-reported physical activity, smoking status, dietary intakes of specific nutrients, and medication adherence did not substantially change the estimates in either group.
In the final model including health behaviors and adherence to antihypertensive medications, only age, race/ethnicity, diabetes, physical activity and medication adherence were strong independent predictors of BP control (please see Table S3 of the Online Supplement http://hyper.ahajournals.org). There also appeared to be a non-significant trend toward poorer BP control with higher levels total daily caloric intake and lower levels of fiber intake. Participants reporting non-adherence to anti-hypertensives and diabetes had nearly three times higher odds of uncontrolled BP. Similar to the overall population, the threshold for a statistically significant effect of physical activity occurred at the highest level of activity (>50 MET-hours/week), with the most active known hypertensive patients having 43% lower odds of poorly controlled BP after adjustment for medication adherence.
In our examination of a nationally-representative cohort representing 37.7 million adult individuals in the United States, we found that racial disparities in the prevalence of uncontrolled BP among Non-Hispanic Blacks when compared to Whites persist even after adjustment for demographic, socioeconomic status, clinical characteristics, and modifiable health behaviors in both the overall sample. Among a subsample with self-reported prior diagnoses of hypertension, the disparity in blood pressure control persisted for both Non-Hispanic Blacks and Mexican-Americans after adjustment for all covariates of interest. Further, among known hypertensive subjects, medication adherence was strongly associated with BP control, but did not explain the racial/ethnic disparity in the prevalence of uncontrolled BP. Last, after multivariate adjustment, only participants with self-reported levels of physical activity above 50 MET-hours/week had significantly lower odds of uncontrolled BP. To our knowledge, this is the first study to document the threshold of leisure-time physical activity associated with BP control among a nationally-representative sample utilizing a measure that incorporates duration as well as intensity and frequency of activity.
Differences in hypertension prevalence, treatment, and outcomes between minorities and non-Hispanic Whites have long been noted in the U.S. In a recent study, only 50% of hypertensive participants in NHANES were noted to have controlled hypertension, and Mexican-Americans were found to have higher prevalence of uncontrolled BP compared to non-Hispanic Whites.43 Our findings are consistent with prior literature, showing that age, race/ethnicity, and diabetes were associated with uncontrolled BP overall and among the subsample of known hypertensives.1, 31, 44, 45 We also noted that the minority groups did report lower sodium and alcohol intake as well as lower total energy intake than Whites; however, this did not impact the racial/ethnic disparity in BP control in multivariate regression modeling.
Furthermore, increasing levels of physical activity were associated with improved BP control overall and among hypertensives in NHANES, with extremely high levels of exercise (>50 MET-hrs/week) reaching statistical significance. Although a prior study examining NHANES III showed that greater than five episodes per week of moderate to vigorous activity was associated with decreased hypertension prevalence, frequency of activity was not assessed at that time.31 To put our findings in perspective, the current recommendation for exercise is equivalent to 7.5–12.5 MET-hours per week.46 White men in the highest quartile of leisure activity in the Atherosclerosis Risk in Communities (ARIC) study had 34% lower odds of developing hypertension over six years compared to the least active in multivariate models.10 Further, data from the Nurses’ Health Study, a sample of predominately White professional women, suggest that only the highest levels of activity were associated with decreased risk of incident hypertension.13
None of the self-reported health behaviors, including physical activity, mitigated the disparity in BP control in our study. This is consistent with findings in the ARIC study that although White men who reported high physical activity levels had lower odds of developing hypertension over time, baseline activity was not associated with incident hypertension in white women or blacks, even after controlling for other health behaviors.10 In the Coronary Artery Risk Development in (Young) Adults Study (CARDIA), accounting for lifestyle behaviors narrowed, but did not eliminate, the racial disparity in BP incidence that appeared after seven years of follow-up.47 However, this study did not include sodium intake in the analyses, and utilized a sample of adults who were relatively young (ages 18–30) at study entry. Thus, it is plausible that differences in dietary sodium intake (particularly if higher among Whites as was the case in the current study) may attenuate their findings. Alternatively, the influence of lifestyle behaviors on BP disparities may not persist as these individuals continue to age.
This study is subject to some limitations. First, NHANES is a cross-sectional study and thus temporal associations cannot be established. Second, the nutrient intakes were taken from a single 24-hour dietary recall for each subject. Since dietary recalls only cover one 24-hour period, it may not represent an individual’s usual intake if this period is not a typical diet day for the individual. However, the mean of the population’s distribution of usual intake can be estimated from a sample of individuals’ 24-hour recalls without sophisticated statistical adjustment.48 Third, in addition to nutrient intakes, many of the other behavioral and sociodemographic variables were from subject self-report, and therefore misclassification and reporting bias is a concern. Age, sex, obesity, and socioeconomic status have been associated with under-reporting dietary intakes and other self-reported health behaviors; however, it is unclear how this may relate to racial differences in self-reported nutrient intakes.49–52 For example, if racial/ethnic minority groups are more likely to under-report their caloric intake compared to non-Hispanic whites, and higher caloric intake is inversely related to blood pressure control, our findings of racial/ethnic disparities in control may underestimate the true magnitude of the racial/ethnic differences. Alternatively, if minorities are less likely to under-report caloric intake compared to non-Hispanic whites, our findings may overestimate the gap between minorities and non-Hispanic whites in rates of blood pressure control. Our findings should be interpreted in light of these limitations.
Last, there may be other variables that were not measured and/or included in our study that may impact the association of race with uncontrolled BP; i.e., occupational-related physical activity53 and/or exposure to psychosocial stressors such as discrimination. A recent study at a single academic medical center reported that Blacks reported more discrimination and concern about antihypertensive medications than Whites; after accounting for these factors, race was no longer a significant predictor of BP control.54 We also could not evaluate differences in health care provider behaviors such as adherence to prescribing guidelines or intensification of antihypertensive medications which may have a role in addressing racial/ethnic health disparities.55 In addition, racial/ethnic differences in genetic susceptibility, gene-environment interactions, and regional/geographic variations also may be more important contributors to the racial disparity in hypertension as well as other metabolic states (i.e., diabetes, obesity) that are associated with poor BP control.11, 56
In this nationally representative cohort of adults, we found that racial/ethnic disparities in BP control were not fully explained by racial/ethnic differences in potentially modifiable health behaviors, including self-reported adherence to anti-hypertensive medications among participants with previously diagnosed hypertension. Our findings suggest that although population-wide adoption of healthy behaviors may be critical to reduce the societal burden of hypertension and its sequelae, interventions targeting racial/ethnic differences in these behaviors may not mitigate disparities in rates of uncontrolled BP. Future research should further assess the potential relationships of gene-environment interactions, job-related stress, racism, and other psychosocial factors to racial/ethnic disparities in the prevalence and outcomes of hypertension in the United States.
Drs. Redmond and Hicks conceived and designed the study. Dr. Redmond acquired, analyzed, and interpreted the data and drafted the article. Drs. Hicks and Baer contributed to the interpretation of the data and revised the manuscript critically for important intellectual content. All authors approved the final submitted manuscript. The authors would like to thank Shimon Shaykevich, MS and Lenny Lopez, MD, MDiv, MPH for their advice on statistical programming.
Dr. Redmond and this study were supported by grant number T32HP10251 from the Health Resources and Services Administration of the Department of Health and Human Services. The study’s contents are solely the responsibility of the authors and do not necessarily represent the official views of the Department of Health and Human Services.
Dr. Hicks was supported by the Health Disparities program of Harvard Catalyst | The Harvard Clinical and Translational Science Center (NIH Grant #1 UL1 RR 025758-01 and financial contributions from participating institutions).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Dr. Hicks is a scientific advisor to the Health Management Corporation and a member of the Board of Directors to Health Resources in Action. The authors have no other disclosures or conflicts of interest to report.