Effect of non-optimal ambient temperature on preterm birth stratified by social positioning in Nepal: A space-time-stratified case-crossover study.

ABSTRACT

BACKGROUND: The ongoing increase in the mean global temperature due to human induced climate change, indicates that women and infants will have higher exposure to heat events leading to adverse outcomes. The study investigates the effect of non-optimal ambient temperature on the risk of preterm birth stratified by social position in Nepal. METHOD: This is a space-time-stratified case-crossover design, based on hospital-registered perinatal data between 2017 and 2021 (n = 47,807). A daily count of pregnant women residing in seven heat-prone districts was extracted together with their social status (ethnicity), obstetric complication and gestation of birth. The daily count of events was matched with the daily ambient temperature of their residence using the NOAA spatial temperature recording. Ambient temperature exposure was analysed using conditional Poisson regression and distributed lag non-linear models. FINDINGS: In the general population, with exposure to ambient temperature at the 75th centile (28 °C) the cumulative risk of preterm birth over 28 days was 1·29 times higher (RR, 1·29; 95% CI; 1·09, 1·54) than at median temperature (24.1 °C), and even higher among the socially disadvantaged population. Cumulative risk of preterm birth to cold ambient temperature at the 1st centile was high but not significant. Exposure to ambient temperature at the 90th centile (32·5 °C) had the highest cumulative risk of preterm birth for pregnant women from socially disadvantaged populations (RR 1·81; 95% CI; 1·28, 2·55). The delayed effect after exposure to temperatures above the 75th percentile was more prolonged in the disadvantaged than the advantaged social group. CONCLUSION: Although exposure to cold with certain effect on preterm births, heat (increase in ambient temperature) carries a risk of preterm birth in Nepal, and is more profound among socially disadvantaged populations.