BACKGROUND: We assessed the relationship between neighborhood socioeconomic status (NSES) and incident coronary heart disease (CHD) among women, adjusting for individual sociodemographic characteristics, baseline health status and health behaviors. Data came from the Women’s Health Initiative Clinical Trial (WHI CT), a longitudinal study of women ages 50 to 79 baseline, enrolled between 1993 and 1998 and followed until at least March 2005. Participants were recruited at 40 clinical centers and 36 satellite locations. The sample (n=68,132) was 81.7% nonHispanic white, 10.3% nonHispanic black, 4.2% Hispanic, and 3.8% other; 60.9% were married at baseline, 94.3% had at least a high school education, and 66% had household incomes between $20,000 and $75,000 (categories ranged from <$10,000 to $150,000).
METHODS: Using 2-level hierarchical Cox proportional hazard regression models (e.g. shared frailty models), we analyzed the WHI CT data merged with tract-level Census data on neighborhood sociodemographic characteristics. We examined 3 outcomes: time until first CHD event (myocardial infarction (MI), revascularization, and hospitalized angina), time until CHD death or first MI, and time until CHD death. The NSES index included 6 educational and economic measures at the Census tract-level.
RESULTS: After controlling for a number of key individual-level sociodemographic characteristics including age, race, education, income, martial status, region, family history of MI, and study arm, we found that women residing in lower NSES census tracts experienced higher risk for each of the outcomes and shorter time to first CHD event or death by CHD. The relationship between NSES and incident CHD was mediated by baseline health status (BMI, waist hip ratio, self-reported history of diabetes, hyperlipidemic medication use and/or self reported high cholesterol, hypertension) and health behaviors (smoking packyears, alcohol use, hormone use). After controlling for these baseline measures in addition to sociodemographic variables, the effect of NSES
decreased but remained statistically significant for all outcomes except time to CHD death. To illustrate the effect size of NSES on our outcomes, we compared the hazard ratios for our outcomes for a woman living in Anacostia and northwest DC (neighborhoods which represent the bottom and top quartiles of NSES in the US), controlling for both demographics and baseline health measures. Compared to the same woman living in northwest DC, one living in Anacostia has a 1.20 times greater risk for first CHD event (CI:1.01,1.42) and 1.28 times greater risk for CHD death or MI (CI:1.02,1.62). There is no significant difference between the risk of these two women for CHD death (HR 1.28, CI:0.86,1.92). Propensity analyses confirmed that these effect sizes were
not sensitive to selection on observed characteristics.
CONCLUSION: Living in a lower NSES neighborhood was independently associated with greater CHD risk, above and beyond individual- level baseline characteristics. Our findings suggest that the observed effects operate in part through health behaviors and diseases which increase cardiovascular risk. This study is part of a larger effort aimed at assessing whether and how neighborhood characteristics affect women’s health and understanding whether
changing neighborhood features could improve health and reduce health disparities.