Short-term effects of air temperature on mortality and effect modification by air pollution in three cities of Bavaria, Germany: a time-series analysis

Background: Air temperature has been shown to be associated with mortality; however, only very few studies have been conducted in Germany. This study examined the association between daily air temperature and cause-specific mortality in Bavaria, Southern Germany. Moreover, we investigated effect modification by age and ambient air pollution.

Methods: We obtained data from Munich, Nuremberg as well as Augsburg, Germany, for the period 1990 to 2006. Data included daily cause-specific death counts, mean daily meteorology and air pollution concentrations (particulate matter with a diameter<10 μm [PM10] and maximum 8-h ozone). We used Poisson regression models combined with distributed lag non-linear models adjusting for long-term trend, calendar effects, and meteorological factors. Air pollutant concentrations were categorized into three levels, and an interaction term was included to quantify potential effect modification of the air temperature effects.

Results: The temperature-mortality relationships were non-linear for all cause-specific mortality categories showing U- or J-shaped curves. An increase from the 90th (20.0 °C) to the 99th percentile (24.8 °C) of 2-day average temperature led to an increase in non-accidental mortality by 11.4% (95% CI: 7.6%-15.3%), whereas a decrease from the 10th (-1.0 °C) to the 1st percentile (-7.5 °C) in the 15-day average temperature resulted in an increase of 6.2% (95% CI: 1.8%-10.8%). The very old were found to be most susceptible to heat effects. Results also suggested some effect modification by ozone, but not for PM10.

Conclusions: Results indicate that both very low and very high air temperature increase cause-specific mortality in Bavaria. Results also pointed to the importance of considering effect modification by age and ozone in assessing temperature effects on mortality.