An exposure-response meta-analysis of ambient PM2.5 during pregnancy and preeclampsia.

ABSTRACT

Relationships between PM2.5 exposure and preeclampsia have been the focus of four recent systematic reviews and meta-analyses. We expand on knowledge gaps in these reviews by characterizing the shape of the exposure-outcome relationship, and by assessing the heterogeneity in these associations by study characteristics. Studies of PM2.5 and preeclampsia were identified from reviews, and confounder-adjusted estimates were extracted. Estimates were derived using a random-effects model. Potential non-linearity was evaluated using a one-stage dose-response meta-analysis. Contrary to previous meta-analyses reporting stronger relationships, the overall adjusted relative risk (RR) for a 10 µg/m3 average increase in PM2.5 during pregnancy and preeclampsia was modest and not statistically significant (RR 1.07, 95% CI 0.99-1.15). This was mainly attributable to inclusion/exclusion decisions for studies made during this review. In addition, there was no evidence of non-linearity, and no important sub-group differences by characteristics such as region, exposure assessment, participant exclusions, and early versus late-onset preeclampsia. Overall, our analysis suggests a modest relationship between ambient PM2.5 and preeclampsia. We provide details on inclusion and exclusion decisions that were lacking in previous studies, and report novel investigations of non-linearity and heterogeneity.