Prenatal exposure to ambient temperature and preterm birth: A historical cohort

Background Accumulating evidence suggests links between ambient temperature and preterm birth. We aimed to explore susceptible exposure weeks and groups concerning temperature and preterm birth in an innovative methodological approach. Methods We conducted a historical cohort study of 131 599 singleton live births in Southern Israel in 2005-19. Weekly mean temperatures were assessed based on residential address and a spatiotemporal model. We fitted Cox models with time-dependent covariates and distributed lag non-linear models, adapting them for the challenges of examining prenatal exposures. We further extended the models with time-dependent coefficients to assess variations by preterm birth categories. Finally, we estimated associations of cumulative exposures by using predicted survival curves contrasting realistic exposure trajectories by month of the last menstrual period (LMP). Results Exposures to high temperatures in gestation Weeks 29-37 were associated with increased preterm birth risk. The highest hazard ratio was 1.04 [95% confidence interval (CI): 1.01-1.07] in Week 37 for the 99th percentile (31°C) compared with the minimum-risk temperature (16°C). There was a strong seasonal pattern in the estimated risk, with pregnancies with LMP in autumn having a higher risk. The average estimated risk by LMP month varied between 6.5% and 7.6% for pregnancies with LMP in March and October, respectively, corresponding to a relative risk of 1.17 (95% CI: 1.07-1.27). Conclusion The final weeks of the third trimester are the key window for heat exposure in Southern Israel, creating a distinguished estimated risk by LMP month, with the highest risk for pregnancies conceived in the autumn.