Countrywide analysis of heat- and cold-related mortality trends in the Czech Republic: growing inequalities under recent climate warming

Abstract

Background: Only little is known about trends in temperature-mortality associations among the most vulnerable subgroups, especially in the areas of central and eastern Europe, which are considered major climatic hotspots in terms of heatwave exposure. Thus, we aimed to assess trends in temperature-related mortality in the Czech Republic by sex, age and cause of death, and to quantify the temporal evolution of possible inequalities.

Methods: We collected daily time series of all-cause (1987-2019) and cause-specific (1994-2019) mortality by sex and age category, and population-weighted daily mean 2-metre temperatures for each region of the Czech Republic. We applied a quasi-Poisson regression model to estimate the trends in region-specific temperature-mortality associations, with distributed lag non-linear models and multivariate random-effects meta-analysis to derive average associations across the country. We then calculated mortality attributable to non-optimal temperatures and implemented the indicator of sex- and age-dependent inequalities.

Results: We observed a similar risk of mortality due to cold temperatures for men and women. Conversely, for warm temperatures, a higher risk was observed for women. Results by age showed a clear pattern of increasing risk due to non-optimum temperatures with increasing age category. The relative risk (RR) related to cold was considerably attenuated in most of the studied subgroups during the study period, whereas an increase in the RR associated with heat was seen in the overall population, in women, in the age category 90+ years and with respect to respiratory causes. Moreover, underlying sex- and age-dependent inequalities experienced substantial growth.

Conclusions: Our findings suggest ongoing adaptation to cold temperatures. Mal/adaptation to hot temperatures occurred unequally among population subgroups and resulted in growing inequalities between the sexes and among age categories.

Keywords: Heat; adaptation; climate change; cold; inequalities; mortality; temperature.

Figure 1.

Distribution of daily mean temperatures between (a) October and April and (b) May and September in the Czech Republic between 1987 and 1999 and between 2007 and 2019

Figure 2.

Temperature–mortality associations from the model for the whole study period stratified by (a) sex, (b) age category and (c) cause of death. Dashed vertical lines denote the 1st and 99th percentiles of the study period-specific temperature distribution. Solid lines denote subgroup-specific minimum mortality temperatures. RR,  relative risk

Figure 3.

Relative risk of death at the 1st (cold) and 99th (heat) temperature percentiles of the whole study period from the model with subsets of 13-year moving periods. Mann–Kendall’s tau value and P-value from the Nonparametric Block Bootstrapped Mann–Kendall Trend Test. RR, relative risk; P01, temperature percentile 1; P99, temperature percentile 99

Figure 4.

Evolution of inequality index at the 1st (cold) and 99th (heat) temperature percentiles during the study period. Mann–Kendall’s tau value and P-value from the Nonparametric Block Bootstrapped Mann–Kendall Trend Test. P01, temperature percentile 1; P99, temperature percentile 99

Figure 5.

Spatial variation in relative risks and the inequality index at the 1st (cold) and 99th (heat) temperature percentiles by sex and their temporal trends during the study period. Mann–Kendall’s tau value and P-value from the Nonparametric Block Bootstrapped Mann–Kendall Trend Test. Hatched regions are not significant. RR, relative risk; TP01, temperature percentile 1; TP99, temperature percentile 99

Figure 6.

Spatial variation in relative risks and the inequality index at the 1st (cold) and 99th (heat) temperature percentiles by age category and temporal trends during the study period. Mann–Kendall’s tau value and P-value from the Nonparametric Block Bootstrapped Mann–Kendall Trend Test. Hatched regions are not significant. RR, relative risk; TP01, temperature percentile 1; TP99, temperature percentile 99