A Direct Estimate of the Effect of PM2.5 on Life Expectancy using Doubly Robust Quantile Regression

A Direct Estimate of the Effect of PM2.5 on Life Expectancy using Doubly Robust Quantile Regression Introduction: To date, studies of the impact of air pollution on survival have estimate hazard ratios, not loss of life expectancy; have estimated the mean effect of exposure and not how it changes the distribution of life expectancy, and not used causal modeling. Methods: We introduce a doubly robust quantile survival analysis with inverse probability weights to estimate the causal effects of PM2.5 in New England on life expectancy in all Medicare participants (3.3 million participants over age 65) followed from 2000-2013. Particle exposure was estimated at zip code of residence using a validated model based on satellite remote sensing (cross validated R2=0.88). Results: PM2.5 exposure reduced the median life expectancy by 0.25 years per 1µg/m3 (95% CI 0.21, 0.29 years). The effects were not equally distributed across the percentiles of life expectancy, with large reductions in life expectancy at the lower...