The article published by Kowall and colleagues in PLoS ONE on the 03.08.2021 reported no excess mortality for Germany in 2020 as compared to the four preceding years. The main finding that ”In Germany, cumulative mortality by the end of 2020 was relatively 2.4% lower than in the average of the preceding four years”, which was also highlighted by the press release of the University Duisburg-Essen in October 2021 [1], was widely cited by the German media [e.g., 2-5], largely due to its contradictory nature. However, various studies that were based either on preliminary estimates or on the final data for 2020, clearly suggest that there was indeed an increase in mortality in Germany [6-10].
The published article suffers from methodological shortcomings. The major problem in any research dealing with the analysis of excess mortality is the estimation of the baseline mortality. Indeed, there is no consensus how one should compute the expected deaths. Different models are shown to produce “very similar patterns of weekly excess deaths but disagree substantially on the level of excess“ [11]. Nevertheless, the approach chosen by Kowall and colleagues does not conform to the ‘state-of-the-art’ methodology for mortality analyses in general. For example, in the estimates of age-specific mortality rates for 2020, the authors used as the denominator the population as on 31.12.2019, which leads to the numerator-denominator bias [12]. Another issue is focusing on estimating the annual overall excess mortality instead of providing a thorough analysis of the weekly mortality trends. Given the current pandemic situation that does not affect mortality at a constant rate throughout the year, the analysis of data for the whole year is insufficient. It is difficult to disentangle the effects of the COVID-19 pandemic with other effects on mortality. For example, in Germany in the beginning of the year and before the outbreak of the pandemic, there was a relatively mild flu wave. The mild mortality during the first weeks of 2020 in the older age groups partially balanced the excess in the higher age groups later in the year [13]. On the other side, the second wave of the pandemic only peaked shortly before the end of the year and extended well into 2021. Therefore, in the analysis of the annual data, various effects are mixed with each other and the already known effects on deaths are only partially considered. This should be kept in mind when making the conclusions about the excess mortality estimated for the whole year.
The work of Kowall and colleagues lacks any explanation on how exactly the week 53 in year 2020 was handled in the estimates of expected number of deaths. This is an important piece of information when a conclusion about the annual mortality estimates is made. If the deaths in week 53 were excluded from the analysis, it would lead to the underestimation of mortality for the whole year. According to the official statistics published by the German Federal Statistical Office (Destatis), in Germany about 25.5 thsd. deaths were recorded in week 53 [14]. Four days of this week with high mortality have been within the calendar year 2020.
To adjust their estimates for the demographic development between 2016 and 2020, the authors often refer in their work to the indicator of life expectancy. It is, however, questionable how exactly the extrapolated number of deaths (based on the exponential model) and the estimated “relative mortality change” account for the trends in life expectancy. These estimates rather illustrate the changes in the age-specific mortality trends between the two periods. The seasonal adjustments were not taken into account in the extrapolation of the mortality rates (e.g., for the strong flu waves in 2017 and 2018 as in the case of Germany), which are important to address, when data used as the basis for extrapolation regard such an unconventional short period of time for mortality extrapolations.
The work does not suffer from the methodological flaws only but is also based on aggregated and preliminary data: the death counts and population figures were extracted from the Eurostat database in the beginning of February 2021 - that gives a reason to assume that these counts were still provisional at that time. A lack of sex-specific mortality analysis and the use of broadly defined age groups also limit the analysis and might bias the results. The 20-year age groups, for example, do not allow considering relevant shifts within these age groups.
The results provided by the authors clearly disagree with the results published elsewhere. For instance, Destatis has reported around 46 thsd. more deaths in 2020 as compared to the previous year, which corresponds to a 5% increase in the absolute death counts [10]. According to the conservative estimates reported in the same press release, if the time trend towards increasing life expectancy and the previously foreseeable shifts in the age structure of the population were considered, a 1-2% increase in number of deaths would still have been expected in 2020 [10]. An increase of about 20 thsd. deaths between 2019 and 2020 can be attributed to changes in the age structure of the population [15]. The estimates reported by Islam and colleagues [6] which account for temporal trends and seasonal and natural variability suggest that in 2020 in Germany there were about 26 thsd. excess deaths as compared to the expectation based on mortality in 2016-2019, corresponding to the results published by Destatis. A finding of excess mortality was also confirmed by both the crude and age-standardized deaths rates [6, 15]. The study of Morfeld and colleagues [16], who looked at the data only for the first ten months of 2020 and did not yet account for the effect of the second pandemic wave at the end of the year, already reported a slight increase in mortality overall (Standardized Mortality Ratio (SMR) of 1.01) and for men in particular (SMR of 1.03) as compared to the average of 2016-2019 [16]. The most recent work of the OECD also reported after adjusting for population change a 7% increase in the total number of deaths in 2020 as compared to the average of 2015-2019 in Germany [17]. All these findings contradict to the results by Kowall et al.
Previous research that used estimates of weekly death counts showed that despite the increasing trend in life expectancy in Germany during the last decades, in year 2020 it declined by 0.38 years for men and 0.23 years for women as compared to the previous year [8]. The official age-specific death rates published by the German Federal Statistical Office [15] also show an increase in mortality between 2019 and 2020, particularly for the very old ages. Finally, in October 2021 the Federal Institute for Population Research (BiB) published life expectancy estimates based on the official and complete data for 2019-2020. The decline constituted 0.3 years for men and 0.1 years for women [9].
To sum up, the main conclusion of the paper on Germany is not supported by evidence presented elsewhere. The contradictory result by Kowall et al. may be caused by methodological flaws and incompleteness of data used. Pushing such results based on provisional data with the press release in October 2021 [1] is not a good scientific practice as it has been misleading the public. More detailed and final data as well as the contradictory findings based on these data were already available at that time but were not considered in the release.

This comment is supported by:
Felix zur Nieden (Federal Statistical Office of Germany)
Olga Grigoriev (Federal Statistical Office of Germany)
Sebastian Klüsener (Federal Institute for Population Research, Germany)
Pavel Grigoriev (Federal Institute for Population Research, Germany)
Vladimir M. Shkolnikov (Max Planck Institute for Demographic Research, Germany)
Jonas Schöley (Max Planck Institute for Demographic Research, Germany)


References:
1. Pressemitteilung der Universität Duisburg Essen vom 21.10.2021: https://www.uni-due.de/20...
2. Welt (23.10.2021). Keine Übersterblichkeit in Deutschland. https://www.welt.de/wisse...
3. Rheinische Post (25.10.2021). https://rp-online.de/nrw/....
4. Neue Osnabrücker Zeitung (26.10.2021). Untersterblichkeit statt mehr Tote im Corona-Horrorjahr 2020? https://www.noz.de/deutsc....
5. Aerzteblatt (2021). Analyse: 2020 keine oder nur geringe Übersterblichkeit in Deutschland. https://www.aerzteblatt.d...
6. Islam N. et al. (2021). Excess deaths associated with covid-19 pandemic in 2020: age and sex disaggregated time series analysis in 29 high income countries. BMJ 2021; 373: n1137
7. Islam N. et al. (2021). Effects of covid-19 pandemic on life expectancy and premature mortality in 2020: time series analysis in 37 countries. BMJ 2021; 375:e066768.
8. Aburto J.M. et al. (2021). Quantifying impacts of the COVID-19 pandemic through life-expectancy losses: a population-level study of 29 countries. International Journal of Epidemiology, 1–12.
9. BiB (2021). COVID-19: Lebenserwartung in Deutschland im internationalen Vergleich nur gering gesunken. Pressemitteilung, 12.10.2021.
10. Destatis (2021). Mehr Sterbefälle, weniger Geburten und Eheschließungen im Jahr 2020. Pressemitteilung Nr. 200 vom 26. April 2021. Available at: https://www.destatis.de/D...
11. Schöley J. (2021). Robustness and bias of European excess death estimates in 2020 under varying model specifications. Available at: https://doi.org/10.1101/2...
12. Kramer M.S. (1988). Rates. In: Clinical Epidemiology and Biostatistics. Springer, Berlin, Heidelberg. https://doi.org/10.1007/9...
13. De Nicola G., Kauermann G., Höhle M. (2021). On assessing excess mortality in Germany during COVID-19 pandemic. Available at: https://arxiv.org/pdf/210...
14. Destatis (2021). Sterbefälle - Fallzahlen nach Tagen, Wochen, Monaten, Altersgruppen, Geschlecht und Bundesländern für Deutschland 2016 – 2021. Available at: https://www.destatis.de/D...
15. Destatis (2021). Lebenserwartung in Deutschland nahezu unverändert. Pressemitteilung Nr. 331 vom 9. Juli 2021. Available at: https://www.destatis.de/D...
16. Morfeld P. et al. (2021). COVID-19: Wie änderte sich die Sterblichkeit? – Mortalität von Frauen und Männern in Deutschland und seinen Bundesländern bis Oktober 2020. Dtsch Med Wochenschr146: 129–131.
17. OECD (2021). Health at a Glance 2021: OECD Indicators.