Associations between residential greenness and birth outcomes across Texas

Background: The amount of greenness around mothers' residences has been associated with positive birth outcomes; however, findings are inconclusive. Here we examine residential greenness and birth outcomes in a population-based birth cohort in Texas, a state with large regional variation in greenness levels, several distinct cities, and a diverse population.

Methods: We used Vital Statistics data to create a birth cohort (n=3,026,603) in Texas from 2000 to 2009. Greenness exposure measures were estimated from full residential addresses across nine months of pregnancy, and each trimester specifically, using the mean of corresponding MODIS satellite 16-day normalized difference vegetation index (NDVI) surfaces at a 250m resolution, which have not been previously used. Logistic and linear mixed models were used to determine associations with preterm birth, small for gestational age (SGA) and term birth weight, controlling for individual and neighborhood factors.

Results: Unadjusted results demonstrated consistent protective effects of residential greenness on adverse birth outcomes for all of Texas and the four largest cities (Houston, San Antonio, Dallas, and Austin). However, in fully adjusted models these effects almost completely disappeared. For example, mothers with the highest (>0.52) compared to the lowest (<0.37) NDVI quartiles had a 24.4g (95% CI: 22.7, 26.1) increase in term birth weight in unadjusted models, which was attenuated to 1.9g (95% CI: 0.1, 3.7) in fully adjusted models. Maternal and paternal race, ethnicity and education had the largest impact on reducing associations. Trimester-specific greenness exposures showed similar results to nine-month average exposures. Some evidence was seen for protective effects of greenness for Hispanics, mothers with low education and mothers living in low income neighborhoods.

Conclusions: In this large population-based study, across multiple urban areas in Texas and diverse populations, we did not observe consistent associations between residential greenness and birth outcomes.