Studies have shown that temperature could modify the effect of ambient fine particles on mortality risk. In assessing air pollution effects, temperature is usually considered as a confounder. However, ambient temperature can alter people's physiological response to air pollution and might "modify" the impact of air pollution on health outcomes. This study investigated the interaction between daily PM2.5 and daily mean temperature in Beijing, China, using data for the period 2005-2009. Bivariate PM2.5-temperature response surfaces and temperature-stratified generalized additive model (GAM) were applied to study the effect of PM2.5 on cardiovascular, respiratory mortality, and total non-accidental mortality across different temperature levels. We found that low temperature could significantly enhance the effect of PM2.5 on cardiovascular mortality. For an increase of 10 μg/m(3) in PM2.5 concentration in the lowest temperature range (-9.7∼2.6 °C), the relative risk (RR) of cardiovascular mortality increased 1.27 % (95 % CI 0.38∼2.17 %), which was higher than that of the whole temperature range (0.59 %, 95 % CI 0.22-1.16 %). The largest effect of PM2.5 on respiratory mortality appeared in the high temperature range. For an increase of 10 μg/m(3) in PM2.5 concentration, RR of respiratory mortality increased 1.70 % (95 % CI 0.92∼3.33 %) in the highest level (23.50∼31.80 °C). For the total non-accidental mortality, significant associations appeared only in low temperature levels (-9.7∼2.6 °C): for an increase of 10 μg/m(3) in current day PM2.5 concentration, RR increased 1.27 % (95 % CI 0.46∼2.00 %) in the lowest temperature level. No lag effect was observed. The results suggest that in air pollution mortality time series studies, the possibility of an interaction between air pollution and temperature should be considered.
Keywords: Interaction; Mortality; Particulate matter; Temperature.