The Predictability of Heat-Related Mortality in Prague, Czech Republic, during Summer 2015: A Comparison of Various Thermal Measures

Development and innovation of measures protecting populations against meteorological and climatic hazards are among the priorities of biometeorological research. Understanding the meteorological factors that are most closely associated with health risks can help refine and improve these measures. For example, heat warning systems should be used to alert the public of elevated health concerns on days with the highest risk. We compared selected thermal measures in their ability to predict days with elevated mortality due to heat in Prague, Czech Republic during the extraordinarily hot summer of 2015. Relatively novel thermal measures - Universal Thermal Climate Index and Extreme Heat Factor (EHF) - were compared with the more traditional ones (Apparent Temperature, Wet-Bulb Globe Temperature (WBGT), Physiologically Equivalent Temperature) to evaluate their suitability for heat warning systems. After adjusting mortality for long term trend and seasonality, relationships between thermal measures and all-cause mortality deviations were estimated by Generalized Additive Models for summer months during 1994–2014. Resulting models were applied to predict mortality deviations in summer 2015 based on observed meteorology, and the mortality estimates by individual models were compared.

All models showed a clear and strong association between thermal conditions and mortality deviations. Although the differences between models were generally small, WBGT appeared to be the best predictor of heat-related mortality in Prague according to the model diagnostics. When the predictions of prolonged periods with positive mortality deviations (heat waves) in summer 2015 were compared, all measures showed comparable estimates of their beginning and duration. However, we observed considerable differences in the magnitude of estimated cumulative mortality deviations during these periods. While model based on EHF in general overestimated the impact of heat waves compared to observations, other indices underestimated the heat effects. Results of this study suggest that current heat warnings in the Czech Republic based on the arbitrary air temperature threshold may have been missing a substantial part of heat-related deaths and improvements in the warning system should be considered.