Prenatal extreme heat exposure and neonatal mortality in Bangladesh, 2015 to 2025: a national birth-history study

Background: Heat in pregnancy is linked to preterm birth and stillbirth, but its link with neonatal death in South Asia is unclear. We estimated this association in Bangladesh and separated the role of prenatal heat from that of the season in which a child is born. Methods: We included 46,752 live births (February 2015 to May 2025) from the Bangladesh Multiple Indicator Cluster Survey 2025 birth history. Each birth was linked to daily maximum temperature from the ERA5 reanalysis at the location of its survey cluster. An extreme heat day was a day on which the maximum temperature reached the cluster's 95th percentile of daily maximum temperature for 2014 to 2024. Because gestational age was not recorded, we counted extreme heat days in estimated trimesters, dated back 280 days from birth. Weighted Poisson regression with fixed effects for district, division by calendar month of birth, and year of birth compared births within the same place, season and year. Extreme heat 28 to 55 days after birth, which cannot cause a neonatal death, served as a negative control exposure. Results: There were 1,039 neonatal deaths (22.7 per 1,000 live births, 95% CI 21.1 to 24.4). Mortality was highest for births in December to February (26.5 per 1,000, 23.0 to 29.9) and lowest in September to November (17.3, 14.5 to 20.0). Pregnancies had a mean of 13.5 extreme heat days (95% CI 13.4 to 13.6), rising from 4.0 for births in 2018 to 29.5 for births in 2024. The primary estimate was for the estimated third trimester. Per 5 extreme heat days, its adjusted rate ratio was 1.06 (0.98 to 1.15), a risk difference of 1.4 per 1,000 births (-0.6 to 3.3). The rate ratio was 1.04 (0.98 to 1.09) for the whole pregnancy and 0.92 (0.77 to 1.09) for the last 4 weeks of pregnancy. The negative control window showed no clear association (0.92, 0.79 to 1.08). Third trimester estimates were similar with MERRA-2 temperature (1.04, 0.97 to 1.12) and with a fixed 1991 to 2020 threshold (1.07, 1.00 to 1.14). Without control for calendar month of birth, each 1 degree C higher mean third trimester maximum temperature was associated with lower mortality (rate ratio 0.96, 0.93 to 0.99). Conclusions: In this national birth-history analysis, neonatal mortality varied by season of birth and was highest for births in December to February. When births were compared within the same district, season and year, estimated prenatal extreme heat was not clearly associated with neonatal death. The confidence intervals did not exclude modest increases in risk, and both gestational age and heat exposure were estimated with error. Studies of prenatal heat and newborn survival should control for season of birth, because season can hide or create an apparent effect of heat.