Deaths in Germany in 2022 - Evaluation with updated population data

Guess the mess

This week, Destatis carried out the annual update of the population figures, which enables a more precise evaluation. Compared with the previous results based on updated population figures, there are only minor differences. In some respects, 2022 has set negative records.

There is no generally applicable method for determining excess mortality. The method used here calculates separately in 14 age cohorts on a weekly basis of the "Destatis-Sonderauswertung-Sterbefälle". From the comparison period 2016-2020, trend and seasonality of mortality risks were determined for each cohort in a linear time series model (TSLM) and a forecast ("Forecast") of future cohort mortality risks was generated. Known cohort population changes at each year-end were distributed linearly over the 52 weeks of each year, providing an estimate of the population picture for each week. An expected value of the number of deaths for each week and each cohort is obtained by multiplying the calculated population by the predicted risk of death. The advantages of this approach are, first, the inclusion of all relevant parameters and, second, the possibility of a week-specific and age-specific assessment. All key figures of interest, e.g. mean age at death, life expectancy, risk ratios, absolute and percentage excess mortality can be calculated from this.

According to the now final population figures, 2022 recorded a cumulative excess mortality of 70,800 (7.1%) in this model, after 35,600 (3.6%) in 2021. Very similar results (66,000 in 2022 and 34,000 in 2021 ) were recently obtained by Kuhbandner & Reitzner in a methodologically elaborate study. The mortality risks were taken from the official period mortality tables, in contrast to the method described above. The estimation of risk trends was also realized differently and calculated separately by gender for each month. If almost the same result is obtained in different ways, this indicates a high degree of validity.

Fig. 1 shows that excess mortality occurred mainly in the second half of the year, which was also observed in 2020 and 2021 and represents a deviation from the normal seasonal pattern. In the historical time series, the highest mortality waves are found at the beginning of the year with a focus around the 10th calendar week.

Fig. 1: Weekly total deaths (orange) and expected values since 2021 (blue)

Also noticeable are repeated increases in the number of deaths in the 2nd half of the year, which some media have incorrectly declared to be heat waves. From previous years, one can see how heat waves typically emerge (as short, steep increases). However, the 2022 waves show a breadth atypical of heat waves and also occurred in weeks with moderate temperatures. Also, the summer of 2022 was not excessively hot. This does not mean there are no temperature influences in these signal patterns, but if there are, they do not explain these patterns.

A robust parameter for assessing the health status of an overall population is life expectancy. It can be calculated from the age-specific mortality risks for any desired time period. The calculation is mathematically challenging, especially if the mortality risks are only available in cohorts (and not in age cohorts by year). Excess mortality as an absolute or percentage number is the result of an "observed versus expected" method, and expected values assume data from the past and assumptions about how they will evolve. Life expectancy is different. Because it does not require information from the past, it is in this sense an incontestable quantity and a touchstone against which every mortality assessment must be measured. Admittedly, comparisons of life expectancies from different periods are used to assess the development.

Over time, a curve mirroring the mortality figures becomes visible (Fig. 2). In phases with excess mortality, life expectancy decreases and vice versa.

Fig. 2: Life expectancy trend since 2020 (orange) with forecast (blue) with scatter band (one standard deviation). The forecast was generated with an ARIMA model (Auto Regressive Integrated Moving Average) from the reference period 2016-2020.

Annual values of life expectancy increased over decades, but the increase slowed over time (Fig. 3). Until 2020, the values lined up along the trend line like a pearl necklace, with some outliers always followed by a countermovement in the year after.

Fig. 3: Long-term development of life expectancy with trend line (blue) from the years 2000-2020 (blue)

The record value was reached in 2019, and so the slight decline in 2020 was almost inevitable in accordance with the rule established above. Completely out of the ordinary, however, are the crashes in 2021 and 2022. Two subsequent years below trend, and the second even lower than the first, must be classified as a historical anomaly.

The updated population figures show stronger immigration than in previous years. It was strongest in the 0-30 cohort, with about 3% compared to the figures obtained by updating the end-2021 population. All age cohorts were significantly over-mortalized in 2022, but persons below the age of about 35 had a noticeably higher risk ratio than all other cohorts in 2022 (Fig. 4).

Fig. 4: Risk ratios of the year 2022 compared to the expected value by age with 95% confidence intervals

Looking back at the three pandemic years 2020-2022, a serious contradiction to the mainstream Corona narrative becomes apparent. While 2020 was exactly on trend in terms of life expectancy, it plummeted in 2021 and sagged further in 2022. This is inconsistent with the view that the C19 death counts collected by PCR testing according to WHO protocol had any relevance to the overall death incidence. Over the course of these three years and the various C19 variants, case fatality had not increased but at best decreased. Thus, life expectancy should have declined most sharply in 2020 rather than later. In summary, the figures suggest a massive deterioration in general health, especially in 2022, but not in 2020. This general finding is flanked by numerous anomalies in the analysis of temporal and age-related references, which has already been elaborated in the earlier heatmap article.

In 2023, the mortality wave from the end of 2022 initially had an impact. In a countermovement, life expectancy rose above the expected values at the beginning of the year. This was followed by a decline due to a wave of influenza. Since then, the trends in death rates and life expectancy have meandered along the expected values. The wave motion described is similar to the trajectories in 2020 and 2021.

In this article, I would like to limit myself to provide clear results. You, dear reader, can draw your own conclusions from this.