Prevention potential of daily deaths in Switzerland between 2001 and 2010 under a realistic air pollution abatement scenario

Abstract

Background: Obtaining risk estimates of the association between daily mortality and PM10 on time-series data is highly informative for local policy makers. Translating this information into the health prevention potential that could have been achieved in the past can help policy-makers better understand the impications of the times-series findings. Aims: We estimated the mortality prevention potential that would have resulted from lower daily PM10 levels between 2001 to 2010 in Switzerland for a realistic air pollution abatement scenario, which assumed a constant reduction of daily PM10-levels by 20%. Methods: The prevention potential of deaths over the study period was obtained by subtracting from the daily predicted cases given by a Poisson regression model the hypothesized new cases that would have occurred under this same predictive model if the daily values of PM10 were according to the hypothesized abatement scenario. Poisson regression models were fitted with spline functions of calendar time, polynomial functions of meteorological variables and influenza frequency, and indicator variables for different days of the week and holidays. To evaluate potential latency in effects, PM10 levels in the regression model were introduced with both a 7 and a 28-day lag. Results: Considering a 7-day lag effect of PM10, we estimated that 1493 (95% CI = 735 to 2251) deaths, 0.34% of all deaths, could have been avoided in Switzerland if daily PM10 levels had been 20% lower between 2001 and 2010. When the lag effect was extended to 28 days, the corresponding preventable number were 2857 (95% CI = 1674 to 4040), corresponding to 0.66% of all deaths. Conclusion: Even small abatement strategies can result in improved public health. The prevention potential was three times larger with a 28-day lag than with a 7-day lag suggesting a potential latency of the PM10 effects over several weeks.