ABSTRACT

The authors studied the relative predictive powers of several bioclimatic indices as predictors of population mortality during heat waves. Daily mean and maximum values of air temperature, Humidex, apparent, and physiological equivalent temperatures (PETs) were examined. The numbers of daily deaths and daily meteorological data in Rostov-on-Don (southern Russia) were used. The study period spanned April–September between 1999 and 2011. The eight selected bioclimatic indices were used to identify heat waves and calculate the expected increases in mortality during such events from Poisson generalized linear model of daily death counts. All of the bioclimatic indices considered were positively and significantly associated with mortality during heat waves. The best predictor was chosen from a set of similar models by maximization of relative mortality risk estimates. Having compared the relative increases and their significance levels in several cause- and age-specific mortality rates, the authors concluded that PET was the most powerful predictor.