Background/Aim: Exposure to ambient air pollution has been associated with elevated incidence of ischemic stroke, but less is known about how green space may affect post-stroke prognosis. We investigated the impact of Normalized Difference Vegetation Index (NDVI) on all-cause readmissions following ischemic stroke among older adults in the United States.
Methods: Participants were Medicare beneficiaries ≥65 years of age hospitalized with a principal discharge diagnosis of ischemic stroke from 2008 to 2013. Logistic regression was used to evaluate associations between green space at the residential ZIP code level, determined as a 365-day average of NDVI prior to hospital admission, and readmission within 30 days post-discharge. Models were adjusted for individual and area level demographic and lifestyle characteristics, PM2.5in the prior year, and National Climatic Data Center (NCDC) region (Ohio Valley, Upper Midwest, Northeast Northwest, South, Southeast, Southwest, West, Northern Rockies and Plains). Sensitivity analyses included random effects meta-analysis combining estimates from models for each NCDC region and restricted cubic splines to examine region-specific exposure-response relationships.
Results: There were 1,248,426 participants (565,367 men (45%), mean ± standard deviation age 79±8) and 174,336 (14%) were readmitted within 30 days. An interquartile range difference in NDVI (0.18) was associated with lower odds of readmission (OR=0.98, 95% Confidence Interval (CI) 0.97, 0.99). The pooled estimate combining data from all regions was OR= 0.99 (95%CI: 0.97, 1.007) with substantial heterogeneity observed across the regions (I2=53%, p=0.03). Evaluation of restricted cubic splines suggested a steeper slope at levels of exposure below 0.4 in several of the regions including the Northeast, Ohio Valley and West and wide confidence intervals at higher levels.
Conclusions: Our findings suggest that NDVI may be a predictor of prognosis following stroke in some regions of the United States and that these associations may also vary by other weather and land-use characteristics.