Is greener better? Associations between greenness and birth outcomes in both urban and non-urban settings

Background Beneficial effects of greenness on birth outcomes have been reported, but few studies have investigated the associations in both urban and non-urban settings. We aimed to evaluate and compare linear and nonlinear associations between greenness and birth outcomes in urban and non-urban settings. Methods From October 2015 to December 2018, participants were recruited into the Maoming Birth Cohort Study. A total of 11 258 live birth records were obtained. Greenness exposure was assessed using the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI). Linear regression and nonlinear restricted cubic spline models were implemented to investigate the associations between greenness and birthweight, birth length, gestational age, preterm birth, low birthweight, small for gestational age and the potential for effect variation under urban or non-urban settings, after adjusting for covariates. Results A 0.1-unit increase in NDVI-500m was significantly associated with an increase of 35.4 g in birthweight [95% confidence interval (CI): 13.2, 57.7], 0.15 cm in birth length (95% CI: 0.03, 0.26), 0.88 days in gestational age (95% CI: 0.05, 1.71) and lower odds of low birthweight [odds ratio (OR) = 0.69, 95% CI: 0.56, 0.85] and preterm birth (OR = 0.70, 95% CI: 0.58, 0.85). No association with head circumference was observed. For all outcomes, no significant linear associations were observed among non-urban dwellers. Inversed 'U-shaped' associations between greenness exposure and birth outcomes were observed in the total study population. Conclusions Greenness exposure was associated with increased gestational age, birthweight and birth length in urban dwellers. Nonlinear associations assessed by restricted cubic splines suggested that health benefits could be larger when increasing greenness levels from low to medium compared with increasing greenness from medium to high levels. Further studies adopting nonlinear methods are warranted to verify our findings.