The effects of summer ambient temperature on total mortality in Serbia.

In the context of recent climate change, temperature-attributable mortality has become an important public health threat worldwide. A large number of studies in Europe have identified a relationship between temperature and mortality, while only a limited number of scholars provided evidence for Serbia. In order to provide more evidence for better management of health resources at the regional and local level, this study aims to assess the impact of summer temperature on the population in Serbia, using daily average temperature (T) and mortality (CDR (crude death rate) per 100,000). The analysis was done for five areas (Belgrade, Novi Sad, Niš, Loznica, and Vranje), covering the summer (June-August) period of 2001-2015. In order to quantify the T-related CDR, a generalized additive model (GAM) assuming a quasi-Poisson distribution with log as the link function was used. Five regression models were constructed, for each area, revealing a statistically significant positive relationship between T and CDR in four areas. The effect of T on CDR was defined as the relative risk (RR), which was obtained as the exponential regression coefficient of the models. RR indicates that a 1 °C increase in T at lag0 was associated with an increase in CDR of 1.7% for Belgrade, Novi Sad, and Niš and 2% for Loznica. The model for Vranje did not quantify a statistically significant increase in CDR due to T (RR=1.006, 95% CI 0.991-1.020). Similar results were confirmed for gender, with a slightly higher risk for women. Analysis across lag structure showed different exposure, but the highest effect of T mainly occurs over the short term and persists for 3 days.