Impact of neighborhood socioeconomic status, income segregation, and greenness on blood biomarkers of inflammation

Abstract

Background:

Neighborhood deprivation is linked with inflammation, which may explain poorer health across populations. Behavioral risk factors are assumed to largely mediate these relationships, but few studies have examined this. We examined three neighborhood contextual factors that could exert direct effects on inflammation: (1) neighborhood socioeconomic status, (2) an index of concentration at extremes (that measures segregation), and (3) surrounding vegetation (greenness).

Methods:

Using blood samples and addresses collected from prospective cohorts of 7,930 male (1990–1994) and 16,183 female (1986–1990) health professionals with at least one inflammatory marker, we prospectively linked neighborhood contextual factors to inflammatory biomarkers (adiponectin, C-reactive protein, interleukin-6, soluble tumor necrosis factor receptor-2). Log-transformed, z-scaled component measures were used to calculate an inflammation score. Neighborhood socioeconomic status and index of concentration of extremes were obtained from the 1990 decennial census and linked to participant addresses. Surrounding greenness was assessed from satellite data and focal statistics were applied to generate exposures within 270 m and 1230 m of the participants’ address. We fit multiple linear regression models adjusting for demographic, clinical, and behavioral risk factors.

Results:

Higher neighborhood socioeconomic status was associated with lower inflammation score in women (β for interquartile range increase = –27.7%, 95% CI: –34.9%, –19.8%) and men (β = –21.2%, 95% CI: –31.0%, –10.1%). Similarly, participants in neighborhoods with higher concentrations of high-income households were associated with lower inflammation score in women (β = –27.8%, 95% CI: –35.8%, –18.7%) and men (β = −16.4%, 95% CI: –29.7%, –0.56%). Surrounding greenness within 270 m of each participant’s address was associated with lower inflammation score in women (β = −18.9%, 95% CI: –28.9%, –7.4%) but not men. Results were robust to sensitivity analyses to assess unmeasured confounding and selection bias.

Discussion:

Our findings support the hypothesis that adverse neighborhood environments may contribute to inflammation through pathways independent of behavioral risk factors, including psychosocial stress and toxic environments.