Assessing excess mortality and heat-attributable risk during the summer of 2022 in Catalonia, Spain: a Bayesian spatiotemporal analysis

Abstract

In 2022, excess mortality in Spain was the third highest on record, surpassed only by 2020 and 2015. However, of that excess only 23% has been directly attributed to extreme heat. The problem with this figure is that the estimate is based on models that could present biases and limitations. We proposed to estimate the excess mortality attributable to extreme temperatures in Catalonia during the summer of 2022, and assessing how the risk of death from these extreme heat temperatures may be modified by other factors, particularly socioeconomic variables. We employed a longitudinal ecological design covering the period from 2015 to 2022, using data at the health area level. We used generalized linear mixed models for all ages and for those aged 65 and older. These models corrected for biases by using small-scale geographic units and explicitly took spatial variability into account. According to our results, during the summer months of 2022, 49.41% of excess mortality was attributable to extreme heat. Not only did heatwaves increase the risk of death, but so too did maximum temperature extremes. Effect modifiers found to increase the risk of dying on days with extreme heat were namely: being 65 years or older, high relative humidity, extreme minimum temperature, and low income. Our results suggest the following methodological considerations: (i) minimize the effects of exposure misclassification by using smaller geographic units than those typically used in other studies; (ii) explicitly take spatial variability into account by using, for example, a hierarchical Bayesian spatiotemporal models; and (iii) control for spatial and temporal dependències

This work was partially financed by AGAUR, the Department of Climate Action, Food and Rural Agenda and by the Department of Research and Universities, both from the Government of Catalonia (Generalitat de Catalunya) (grants number 2023 CLIMA 00037 and 2021 SGR 01197). Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature