Estimating all-cause excess mortality during COVID-19 pandemic in Serbia, 2020-2022

Early reports indicated that Serbia was among the best-performing European countries in dealing with COVID-19 health issues based on relatively small COVID-19 mortality. This success can be partly attributed to the government's rapid response and implementation of restrictive measures to curb the spread of the virus. It can also be noted that the high level of solidarity among citizens contributed to the effective containment of the pandemic, with many adhering to experts' prescribed measures and recommendations. Serbia has become an example of good practice in the fight against COVID-19, resulting in positive assessments by international organizations and experts. However, Mortality from the virus alone is insufficient to describe the pandemic's health effects, unlike excess mortality from all causes. This paper aims to estimate excess mortality in Serbia during 2020-2022 and to compare estimated mortality with the reported COVID-19 deaths. Excess was calculated using a negative binomial regression with historical 2015-2019 data. Estimation provides a P-score as a percentage difference between the reported and expected number of deaths. Mortality excess in Serbia was 15,437 in 2020 and 35,836 in 2021, with 224 and 524 per 100,000 population rates and P-scores of 15% and 36%, respectively. Three prominent waves of excess were observed: the winter of 2020 and the spring and last quarter of 2021. The highest monthly excess was noticed in December 2020, with a rate of 113 per 100,000 and a P-score of 84%. The ratios of reported COVID-19 deaths to calculated excess mortality were 20% in 2020 and 27% in 2021. The excess mortality dramatically increases with age. Serbia faced high levels of mortality excess in 2020 and 2021, particularly among older people. During 2022, there was a decrease in mortality trends compared to pandemic years. That year 109,203 died with a rate of 1,639 per 100,000. The excess in 2022 was 7733 (116 per 100,000) with a P-score of 8%.