Increased Risk of Infant Mortality Associated with Pre and Postnatal Exposures to PM2.5 in South Korea: A Propensity Score-Matched Analysis

摘要

Background Evidence on the effect of pre and postnatal exposures to PM10 and PM2.5 on infant mortality is still scarce. We aimed to investigate exposures in six exposure periods (EPs) to explore the associations with infant mortality and their changes over time during pregnancy. Methods We used a 2010-15 retrospective birth cohort data obtained from Statistics Korea. We examined 5 cause-specific infant mortalities based on the ICD-10: sudden infant death syndrome (R95), circulatory (100-99), respiratory (J00-99), perinatal conditions (P00-96), and congenital malformations (Q00-99). Out of 2,525,109 infants 1,530 deaths (0.06%) were identified. With propensity score matching stratified into quintile a 1:2 match was conducted, which was controlled for sex, gestational age, and city size (metropolitan/others). We calculated mean PM2.5 concentrations, using ground monitoring data and Community Multiscale Air Quality model for the following six EPs: each trimester (EP 1-3), gestational period (EP 4), postnatal period (EP 5), and whole period (EP 6; from conception to death). Models were adjusted for infant's address, maternal education level, maternal age, birth weight, and season of birth. A conditional logistic model was applied to take matching and stratification into account. Results Adjusted conditional logistic regression resulted in statistically significant associations between PM2.5 and infant death for all 6 EPs. A 1μg/m3 elevation in PM2.5 concentrations was associated with 2% (95% CI, 1%-4%), 2% (1%-4%), 3% (2%-5%), 6% (4%-9%), 3% (2%-5%), and 15% (12%-19%) increase in infant mortality for ETs 1 to 6, respectively. Conclusion the study findings suggest that pre and postnatal exposures to PM2.5 could be a risk factor for infant mortality and that the association depends on the length of ET: The longer, the higher. Since the 5 causes were combined, further studies on each cause in relation with infant mortality are necessary as more data become available.