Abstract:
Introduction: Epidemiologic studies have shown an association between fine particulate matter (PM2.5.) air pollution and mortality, but few large cohort studies have been performed. It is also unclear what populations are at greatest risk. We hypothesized that those with diabetes mellitus (DM) or chronic obstructive pulmonary disease (COPD) would be more susceptible to death due to long term exposure to PM2.5.
Methods: We studied 90,063 U.S. Veterans seen at seven different Veterans Affairs (VA) hospitals (White River Junction, Little Rock, Birmingham, Richmond, Seattle, San Francisco, and West Los Angeles) during 1995 and 1996. Subject characteristics and the outcome of death were obtained from sources which included VA hospital data, Medicare data, and mailed health survey information. Subjects were assigned fine particulate matter exposures using data from the population-oriented EPA monitor closest to their residential zip code centroid within 30 miles. We estimated the hazard ratio (HR) of all-cause mortality associated with chronic exposure to fine particulate matter, using several different Cox proportional hazards models. Adjustments were made for age, race, hypertension (HTN), high cholesterol, DM, smoking, education, income, and treating VA hospital. Interaction terms for DM and chronic obstructive pulmonary disease (COPD) were separately analyzed in all models.
Results: Hazard ratios f or all-cause mortality associated with PM2.5 were less than or equal to 1.0 and not significant in any of the main Cox proportional hazard models that we performed. After entering interaction terms for DM or COPD into each main model, PM2.5 point estimates varied by model type, but tended to be lower for veterans with DM and higher for veterans with COPD. Adjusting for VA hospital site and restricting to subjects within 10 miles of an EPA monitor impacted PM2.5 point estimates and confidence intervals.
Discussion: We were not able to replicate the findings of most prior cohort studies with regard to fine particulate exposure and mortality, though our findings are consistent with a prior study of air pollution and all-cause mortality in a cohort of hypertensive VA veterans. The presence of residual confounding by measured or unmeasured factors may explain these results. Selection bias was also of considerable concern. More research is needed to more definitively address the question of effect modification by DM or COPD.