Changes in residential greenness between pregnancies and birth outcomes: longitudinal evidence from Michigan births 1990--2012

Background Residential exposure to greenness is associated with better birth outcomes, but it remains unknown whether this is explained by maternal characteristics associated with both place of residence and birth outcomes. We examined whether changes in residential greenness are associated with preterm birth (PTB) and birthweight. Methods We examined cross-sectional associations between maternal exposure to residential greenness [normalized difference vegetation index (NDVI)] and PTB (<37 weeks of gestation) and birthweight in grams, using all births in Michigan (1990–2012) linked by mother (n = 1 730 424). We used maternal fixed effects analysis to estimate associations within mothers across multiple pregnancies and associations for mothers who did not move, but for whom greenness changed between pregnancies, to mimic an intervention. Results Each 0.1-unit change in NDVI was associated with 0.98 [95% confidence interval (CI): 0.97, 0.99] times lower odds of PTB and a 9.0 (95% CI: 8.1, 9.9)-gram increase in birthweight after adjusting for individual and neighbourhood covariates. When we controlled for time-invariant maternal unmeasured confounders; these associations were close to null [odds ratio (OR): 1.00 (95% CI: 0.98, 1.01); β: -0.3 (95% CI: -2.0, 3.6)]. We did not find a relationship between greenness and birth outcomes among women who did not move between pregnancies, but for whom greenness changed within their residential location (as in an intervention). Conclusions Residential greenness does not predict birth outcomes, after controlling for time-invariant maternal characteristics, using longitudinal evidence. Future research should explore residential selection factors, spatial and individual heterogeneity and experimental study designs.