Assessing the effects of compound events of temperature and air pollution on weekly mortality in Spain using random forests

Abstract

Numerous studies have shown the link between extreme weather events and mortality. Specifically, the impacts of extreme temperatures on death rates have been extensively evaluated and documented. Likewise, exposure to air pollution is well known to adversely affect health, with extreme pollution episodes also being related to elevated mortality rates. Hence, it is reasonable to expect that the combination of these phenomena could result in elevated mortality episodes. In this study, we demonstrate that the temporal variability of mortality rates across several Spanish provinces can be effectively explained by a multivariate model that incorporates both meteorological factors and air quality. While a Random Forest analysis shows that temperature is the primary factor in most provinces, the inclusion of pollutant concentration significantly enhances the model’s predictive accuracy. Moreover, a seasonal analysis of extreme events reveals a strong relationship between high mortality episodes and the occurrence of compound events. These events encompass different variables depending on the season. During summer (June–August), situations characterized by extreme temperatures combined with elevated ozone levels result in marked mortality peaks within the same week. In winter (December–February), conditions involving very low temperatures along with high nitrogen dioxide concentrations are associated with extreme mortality rates (above the 90th percentile) during the following week in half of the cases considered. These results suggest that early warning systems should include not only the individual variables but also their combination.