A unique aspect of the EHDIC-SWB is the similarity of demographic and health-related characteristics between African American and non-Hispanic whites, which is not the case in NHANES or any nationally representative sample. By design EHDIC-SWB and NHANES are quite different; however, it is instructive to specify the ways in which the samples differ. displays the demographic characteristics of the EHDIC-SWB and NHANES samples. As expected the African Americans and non-Hispanic whites in EHDIC-SWB were younger and substantially more likely to have never been married than African Americans and non-Hispanic whites in NHANES, respectively. Although the proportion of African American males was similar in both samples, there was a much smaller proportion of non-Hispanic whites in EHDIC-SWB relative to the whites in NHANES. With regard to SES indicators, African Americans and non-Hispanic whites in EHDIC-SWB had, on average, lower incomes than their respective counterparts in NHANES. However, African Americans in EHDIC-SWB were more likely to be a high school graduate/GED than those in NHANES; whereas the opposite is true when comparing the non-Hispanic whites in EHDIC-SWB to those in NHANES.
Distribution of Demographic Variables of the EHDIC-SWB and NHANES 1999–2004 Participants by Racea
presents age-adjusted proportions for the health-related characteristics of the EHDIC-SWB and NHANES samples. In general African Americans and non- Hispanic whites in EHDIC-SWB had poorer health status than African Americans and non-Hispanic whites in NHANES, with greater likelihood of self-reporting fair/poor health, being a drinker, hypertensive, and a higher mean systolic and diastolic BP. Also, African Americans and non-Hispanic whites in EHDIC-SWB were less likely to report having health insurance. African Americans in EHDIC-SWB were less likely to be physically inactive, obese, and diagnosed with diabetes compared to African Americans in NHANES. Non-Hispanic whites in EHDIC-SWB and NHANES were similar with regard to being physically inactive and obese. Non-Hispanic whites were more likely to be diagnosed with diabetes than whites in the national sample.
Age-Adjusted Distribution of Health Related Characteristics of EHDIC-SWB and NHANES 1999–2004 Participants by Racea
In we examined the association between race and hypertension. Model 1 tested for race differences in age-adjusted hypertension. The analysis found that African Americans exhibited greater odds of being hypertensive (odds ratio (OR = 1.48, 95 % confidence interval (CI): 1.16 – 1.89) relative to non-Hispanic whites. After adjusting for demographic variables in Model 2, African Americans still displayed greater odds of being hypertensive (OR=1.45, 95% CI:1.12–1.87) compared to non- Hispanic whites. However, the race odds ratio decreased. Thus, race differences in demographic factors partially accounted for race differences in hypertension.
Odds Ratios and 95 Percent Confidence Intervals for the Association between Race and Hypertension Prevalence in the EHDIC-SWB Study
In Model 3, we added health insurance status to the analysis, African Americans continued to display greater odds of being hypertensive (OR=1.43, 95% CI:1.11–1.85) than non-Hispanic white adults. This relationship remained virtually unchanged, but reduced slightly when accounting for health-related characteristics in Model 4 and both the number of assets and having a regular doctor in Model 5.
In , we estimated Model 1 through Model 4 in NHANES to determine the degree to which findings from EHDIC-SWB differed from a national sample previously used to test race disparities in hypertension that does not account for the race/SES confounding or the different environments that non-Hispanic whites and African Americans live. In Model 1, the age-adjusted model, African Americans displayed a greater odds of being hypertensive (OR = 2.25, 95 % CI: 1.95 – 2.59) relative to non- Hispanic whites. Adjusting for demographic factors in Model 2 reduced the race disparity in hypertension (OR = 2.07, 95 % CI: 1.79 – 2.04) Also, accounting for health insurance status in Model 3 and health-related characteristics in Model 4 further reduced the race effect, but, did not eliminate the disparity. Although the patterns for the relationship between race and hypertension in this national sample were consistent with those observed in EHDIC- SWB, the magnitude of the association was much smaller in EHDIC- SWB, with an average of 31% difference in OR between similar models using the different data sets. Compared to the national sample, the OR in the EHDIC- SWB sample was 34% lower in the age-adjusted model, 29% lower when adjusting for demographic variables, 31% lower after accounting for health insurance status, and 29% lower when adjusting for health-related characteristics.
Association between Race and Hypertension by Dataset*