Effect Modification by Influenza on the Relationship between Fine Particles, Temperature and Cardiovascular Mortality

Background: Cardiovascular mortality (CVD) has been associated with periods of cold temperature, air pollution and influenza epidemics. However, the manner in which the three exposures interact on the outcome of CVD has yet to be determined. We investigated effect modification by seasonal influenza estimated by influenza-like illness (ILI), on the relationship between cold temperatures, particulate matter 2.5 (PM2.5) on daily cardiovascular death. The data available for this study includes 67 United States MSAs throughout the country from the fall of 2003 through the end of 2006. Methods: To estimate short-term effects of PM2.5 and coldest quintile temperature, we used a Poisson time-series regression model. The city-specific beta estimates from the regression were then applied to second stage meta-analysis. We were able to control for day-of-week, same-day and delayed temperature, seasonal trends in the first stage model. In the second stage we included the flu variable and other predictors such as average annual temperature, population density, and population over age 65. Results: The meta-analysis of 67 cities shows the relationship between CVD and PM2.5 exposure is not significantly modified by ILI estimates. The three day lagged cold temperature and ILI estimates show a greater association with increased CVD, combined relative risk and 95% confidence interval 1.028 (1.018-1.038). Conclusions: These findings suggest that the association between PM2.5 effects on CVD may not be strong enough to look effect modification by seasonal influenza. In contrast, ILI may modify the relationship of three day lagged of the coldest quintile of temperature effects on CVD.