Update of the Exposure Response Function for the Evaluation of the Impact of PM2.5 on Natural Mortality

Abstract

Background/Aim: Epidemiologic evidence provides a reliable basis for air pollution impact assessment summarized in the form of exposure-response functions (ERFs) that quantify the increase in the risks of mortality in relations to exposure to different levels of air pollution. A WHO Regional Office for Europe document (HRAPIE, 2013) provided rationale and indications in order to perform health impact assessment of air pollution (in the context of cost-benefit analysis of policies at European level) based on linear or log-linear functions relating particulate matter of a diameter less than 2.5 µm (PM2.5) to natural mortality and other specific outcomes. In the HRAPIE report, the meta-analysis by Hoek (2013) was based on 11 studies and provided the ERF for the PM2.5-all-cause (or natural-cause) mortality association (6% excess mortality, 95% Confidence Interval, CI = 4-8.3%, per 10 µg/m3 PM2.5). Methods: We reviewed 19 studies available until December 2016 on the relationship between long-term exposure to PM2.5 and natural mortality (10 studies included in Hoek et al, the extension of the Canada National Cohort, and 8 new studies). We performed meta-analysis using random effects methods of DerSimonian and Laird (1986). Results: The new synthesis indicates a summary effect estimate considerably higher (10%, 95% CI = 6-14% per 10 µg/m3 PM2.5 over the range of 5-30 µg/m3) than the value previously estimated. The resulting ERF tends to be non-linear with higher coefficients in the region 5-20 µg/m3 and lower values above 20 µg/m3. There are few observations above an annual average of 20 µg/m3, as there are few epidemiological studies on mortality risk from long-term exposure to PM2.5 in the world’s most polluted regions. Conclusions: Given the importance of health impact estimation for normative and policy evaluation activities, including Sustainable Development Goals reporting, the development of stable exposure-response functions is key and natural mortality represents a reliable epidemiologic measure. Scientific literature is progressing and adaptation and improvement of the exposure-response functions over time is also needed.