Multi-pollutant urban study on acute respiratory hospitalization and mortality attributable to ambient air pollution in Canada for 2001-2012

Humans inhale a mixture of various air pollutants, and there is ambiguity whether these pollutants act independently or in an additive, synergistic, or antagonistic manner. We aimed to examine differences in adverse health effects of short-term exposure to air pollution through single-versus multi-pollutant models. We collected daily ambient air pollutants (O-3, NO2 and PM2.5) concentrations from 24 Canadian urban centers from 2001 to 2012. We examined between-pollutant correlations and estimated their associations with respiratory hospitalization and mortality. We applied a generalized quasi-Poisson model to each city adjusting for one and/or two pollutants. Then we employed a Bayesian hierarchical model to pool the city-specific estimates for national estimates. We also examined 0- to 6-day air pollutant lagged effects by season. Overall, we found more frequently significant effects from single-pollutant models (vs. multi-pollutant models), for hospitalization (vs. mortality), for O-3 (vs. NO2, PM2.5), for warm (vs. cold) season, and for lags <= 2 days (vs. 3-6 days). The most consistent effects appeared on respiratory hospitalization for 1-day lagged O-3 and PM2.5, but not NO2. This study found little additive or antagonistic risk of respiratory health outcomes from multi-pollutant models due to low-to-medium correlations among the specified air pollutants.