Is ambient temperature associated with risk of infant mortality? A multi-city study in Korea.

BACKGROUND: Although numerous studies have shown increased risk of mortality from elevated temperatures for adults, limited studies have examined temperature's effect on mortality for infants. Our study investigated the city-specific and overall effects of ambient temperature on infant mortality in seven major cities in Korea, 2004-2007. METHODS: Birth cohort using a linked birth and death records included 777,570 births with 557 all-cause deaths. We estimated city-specific hazard ratios for each city using an extended Cox proportional hazards model with time-dependent covariates. Then we combined city-specific hazard ratios to generate overall hazard ratio across the seven cities using a Bayesian hierarchical model. Stratified analyses were conducted by cause of death (total and SIDS), exposure period (whole gestation, each trimester, lifetime, 1 month before death, and 2 weeks before death), sex, and maternal characteristics. RESULTS: Overall across the cities, we found significantly positive associations between ambient temperature during 1 month before death or 2 weeks before death and infant mortality from total or SIDS. The overall hazard ratio of infant mortality from total deaths and SIDS for a 1°C increase during 1 month before death was 1.52 (95% CI, 1.46-1.57) and 1.50 (95% CI, 1.35-1.66), respectively. We also found suggestive evidence that some factors such as mother's age may modify the association. CONCLUSIONS: Our findings have implications for establishment of policy to reduce the risk of infant mortality from high ambient temperature under climate change.