Causal inference in assessing the impact of air pollution on mortality

Abstract

Background: Numerous studies have investigated the association between particles less than 2.5 μm in aerodynamic diameter (PM2.5) and mortality. However, these have been associational studies that did not use formal causal modeling approaches. We considered the impact of high daily exposure level of PM2.5 on mortality in the metropolitan area of Seoul (Republic of Korea), during the year 2003-2012. Particularly, focusing on the historical health impacts of daily air pollution levels under pre-fixed thresholds on health using different causal approaches.

Methods: We applied propensity score matching methods (nearest neighbor matching and Caliper matching) and inverse probability weighting method to estimate the relative risk (RR) during the study period. We matched each exposed day, namely each day with an exposure higher than pre-fixed threshold exposure, with a day with similar background characteristics but an exposure lower than pre-fixed threshold exposure, using the propensity score. The pre-fixed threshold exposure level was based on the daily high PM2.5 (past, and present) in Korea Enforcement Decree of the Framework Act on Environmental Policy (50, 35 μg/m3), and the WHO daily high PM2.5 exposure level (25 μg/m3). Then, we estimated the impact by comparing the risk between the matched days. The risks were compared within different methods and by the changes in pre-fixed threshold in the same method.

Results: For days exceeding the limit of 50 ㎍/m3 for PM2.5 average had 0.982 times more risk of mortality than the days not exceeding 50 ㎍/m3, and days exceeding 35 ㎍/m3 PM2.5 average had 1.015 times more risk of mortality than the days below 35 ㎍/m3 PM2.5, for 1:1 nearest neighbor matching. The relative risk above pre-fixed threshold 35 ㎍/m3 was higher than the days above 50 ㎍/m3, in the nearest matching method.

Conclusions: We conclude that having strengthened the policy (changing daily bad indication for PM2.5 from above 35 to above 50 ㎍/m3), could have avoided higher mortality risks in Seoul during the study period.

Keywords: particles less than 2.5 μm in aerodynamic diameter (PM2.5), mortality, causal association, propensity score, Seoul, Republic of Korea