Improving consistency in estimating future health burdens from environmental risk factors: Case study for ambient air pollution

Future changes in exposure to risk factors should impact mortality rates and population. However, studies commonly use mortality rates and population projections developed exogenously to the health impact assessment model used to quantify future health burdens attributable to environmental risks that are therefore invariant to projected exposure levels. This impacts the robustness of many future health burden estimates for environmental risk factors. This work describes an alternative methodology that more consistently represents the interaction between risk factor exposure, population and mortality rates, using ambient particulate air pollution (PM2.5) as a case study. A demographic model is described that estimates future population based on projected births, mortality and migration. Mortality rates are disaggregated between the fraction due to PM2.5 exposure and other factors for a historic year, and projected independently. Accounting for feedbacks between future risk factor exposure and population and mortality rates can greatly affect estimated future attributable health burdens. The demographic model estimates much larger PM2.5-attributable health burdens with constant 2019 PM2.5 (-10.8 million deaths in 2050) compared to a model using exogenous population and mortality rate projections (-7.3 million), largely due to differences in mortality rate projection methods. Demographic model-projected PM2.5-attributable mortality can accumulate substantially over time. For example, -71 million more people are estimated to be alive in 2050 when WHO guidelines (5 mu g m(-3)) are achieved compared to constant 2019 PM2.5 concentrations. Accounting for feedbacks is more important in applications with relatively high future PM2.5 concentrations, and relatively large changes in non-PM2.5 mortality rates.