Ambient temperature and stillbirth: Risks associated with chronic extreme temperature and acute temperature change.

ABSTRACT

BACKGROUND: Ambient temperature events are increasing in frequency and intensity. Our prior work in a U.S. nationwide study suggests a strong association between both chronic and acute temperature extremes and stillbirth risk. OBJECTIVE: We attempted to replicate our prior study by assessing stillbirth risk associated with average whole-pregnancy temperatures and acute ambient temperature changes in a low-risk U.S. METHODS: Singleton deliveries in the NICHD Consecutive Pregnancies Study (Utah, 2002-2010; n = 112,005) were identified using electronic medical records. Ambient temperature was derived from the Weather Research and Forecasting model. Binary logistic regression determined the adjusted odds ratio (aOR) and 95% confidence interval (95% CI) for stillbirth associated with whole-pregnancy exposure to extreme cold (<10th percentile) and hot (>90th percentile) versus moderate (10th-90th percentiles) average temperature, adjusting for maternal demographics, season of conception, hypertensive disorders of pregnancy, and gestational diabetes. In a case-crossover analysis, we estimated the stillbirth aOR and 95% CI for each 1° Celsius increase during the week prior to delivery using conditional logistic regression. In both models, we adjusted for relative humidity, ozone, and fine particulates. RESULTS: We observed 500 stillbirth cases among 498 mothers. Compared to moderate temperatures, whole-pregnancy exposure to extreme cold (aOR 4.42, 95% CI3.43, 5.69) and hot (aOR 5.06, 95% CI 3.34, 7.68) temperatures were associated with stillbirth risk. Case-crossover models observed a 7% increased odds (95% CI 1.04, 1.10) associated with each 1° Celsius increase during the week prior to delivery. DISCUSSION: Both chronic and acute ambient temperature were associated with odds of stillbirth in this low-risk population, similar to our prior nationwide findings. Future increases in temperature extremes are likely and the observed risk in a low-risk population suggests this association merits attention.