Long-term Effect of Air Pollution on Hospital Admissions among Medicare Participants Using a Doubly Robust Additive Hazards Model (DRAHM)

Abstract

TPS 684: Long-term health effects of air pollutants 1, Exhibition Hall, Ground floor, August 26, 2019, 3:00 PM - 4:30 PM Background/Aims: Our goal is to look at the relationship between long-term exposure to air pollutants and hospital admissions for cardiovascular and respiratory outcomes among Medicare participants in the United States on an additive scale. The hazard difference obtained from this analysis is of great utility in risk assessment studies as well as in looking at effect modification. Methods: We used a doubly robust additive hazard model (DRAHM) to assess the effect of long-term exposure to PM2.5 on hospital admissions for stroke, myocardial infarction (MI), lung cancer, heart failure, pneumonia, and chronic obstructive pulmonary disease (COPD) among Medicare fee-for-service participants from 2000 to 2013. This model is unbiased if either the inverse probability weight (IPW) model for exposure or the outcome regression model are correctly specified. It also does not require a proportional hazards assumption. PM2.5 levels were obtained from a previously validated high-resolution prediction models which utilized machine learning algorithms. These predictions were averaged spatio-temporally to obtain annual exposure on a zip code level. Admission information was assessed using discharge diagnosis codes from the Medpar database. Covariates included demographic and socioeconomic variables. Effect measure modification was assessed for sex, age, race, and socioeconomic variables. We then repeated the analyses among observations that were less than the federal standard of 12 mcg/m3 for PM2.5. Results: Preliminary results from the New England region show an increase in the hazard of admission 0.0025% (-0.0031%, 0.0081%) for MI, 0.00081% (-0.0037%, 0.0053%) for stroke, 0.0090% (0.0058%,0.012%) for lung cancer, and 0.070% (0.062%, 0.078%) for heart failure for each mcg/m3 increase in PM2.5 levels. Further results are pending. Conclusions: Long-term exposure to air pollution increases the hazard of admission for cardiovascular and respiratory outcomes on an additive scale.