Analysis of weekly mortality of seniors: a comparison between Slovenia and the member states of the EU27

Abstract

We analysed data on the weekly mortality of seniors aged 65 and older in the member states of the EU27. We aimed to model normal mortality in the time before covid-19 and use this model to analyse the deviations of true values from model predictions in the time after the emergence of covid-19. We used linear regression with autocorrelated errors to model normal mortality, which shows seasonality with increased mortality during the winter months. In the time before covid-19 deviations above three sigma occurred regularly, but deviations above five sigma were rare. In contrast, we detected deviations above five sigma in all but three member states after the emergence of covid-19. The member state with the most extreme deviation was Spain (32 sigma), the member state with the longest continuous deviation above five sigma was Czechia (25 weeks), and the member state with the highest excess mortality was Slovakia (12.8 deaths per 1000 seniors). We attempted to explain differences in extreme deviations with multivariate linear regression analysis. Extreme deviations were lower in member states with higher health care expenditure in euros per inhabitant.