New concentration-response functions for seven morbidity endpoints associated with short-term PM_2.5 exposure and...

Morbidity burdens from ambient air pollution are associated with market and non-market costs and are therefore important for policymaking. The estimation of morbidity burdens is based on concentration-response functions (CRFs). Most existing CRFs for short-term exposures to PM_2.5 assume a fixed risk estimate as a log-linear function over an extrapolated exposure range, based on evidence primarily from Europe and North America.

We revisit these CRFs by performing a systematic review for seven morbidity endpoints previously assessed by the World Health Organization, including data from all available regions. These endpoints include all cardiovascular hospital admission, all respiratory hospital admission, asthma hospital admission and emergency room visit, along with the outcomes that stem from morbidity, such as lost work days, respiratory restricted activity days, and child bronchitis symptom days.

We estimate CRFs for each endpoint, using both a log-linear model and a nonlinear model that includes additional parameters to better fit evidence from high-exposure regions. We quantify uncertainties associated with these CRFs through randomization and Monte Carlo simulations.

The CRFs in this study show reduced model uncertainty compared with previous CRFs in all endpoints. The nonlinear CRFs produce more than doubled global estimates on average, depending on the endpoint. Overall, we assess that our CRFs can be used to provide policy analysis of air pollution impacts at the global scale. It is however important to note that improvement of CRFs requires observations over a wide range of conditions, and current available literature is still limited.

The higher estimates produced by the nonlinear CRFs indicates the possibility of a large underestimation in current assessments of the morbidity impacts attributable to air pollution. Further studies should be pursued to better constrain the CRFs studied here, and to better characterize the causal relationship between exposures to PM_2.5 and morbidity outcomes.