There is considerable debate as to the most appropriate metric for characterizing the mortality impacts of air pollution. Life expectancy has been advocated as an informative measure. Although the life‐table calculus is relatively straightforward, it becomes increasingly cumbersome when repeated over large numbers of geographic areas and for multiple causes of death. Two simplifying assumptions were evaluated: linearity of the relation between excess rate ratio and change in life expectancy, and additivity of cause‐specific life‐table calculations. We employed excess rate ratios linking PM2.5 and mortality from cerebrovascular disease, chronic obstructive pulmonary disease, ischemic heart disease, and lung cancer derived from a meta‐analysis of worldwide cohort studies. As a sensitivity analysis, we employed an integrated exposure response function based on the observed risk of PM2.5 over a wide range of concentrations from ambient exposure, indoor exposure, second‐hand smoke, and personal smoking. Impacts were estimated in relation to a change in PM2.5 from 19.5 μg/m3 estimated for Toronto to an estimated natural background concentration of 1.8 μg/m3. Estimated changes in life expectancy varied linearly with excess rate ratios, but at higher values the relationship was more accurately represented as a nonlinear function. Changes in life expectancy attributed to specific causes of death were additive with maximum error of 10%. Results were sensitive to assumptions about the air pollution concentration below which effects on mortality were not quantified. We have demonstrated valid approximations comprising expression of change in life expectancy as a function of excess mortality and summation across multiple causes of death.
