Relationship between Excess Mortality Derived from Death Counts and Age-Standardised Mortality Rates: The Case of Germany During the COVID-19 Pandemic

Excess mortality can be estimated from crude death counts or from age-standardised mortality rates (ASMRs). Although both approaches are widely used, their analytical relationship and the effects of methodological choices such as standard population, age-group definition, and baseline selection are not always explicitly considered. The analytical relationship between relative excess mortality based on death counts and ASMRs was derived and both approaches were applied to mortality in Germany during the COVID19 pandemic 2020-2023. Expected mortality was estimated by linear extrapolation of pre-pandemic trends. ASMR calculations were repeated using different standard populations, age-group definitions, open-ended age groups, and reference periods. Death-count- and ASMR-based P-scores were analytically equivalent when the age distribution of the standard population corresponded to that of the population under study. Empirically, both approaches yielded closely corresponding estimates for Germany. The effect of the standard population depended on the age distribution of excess mortality and was particularly pronounced in 2020, when excess mortality was concentrated at older ages. The definition of the open-ended age group had an additional effect: aggregating ages 80+ altered the pre-pandemic ASMR trend, and consequently changed the estimated excess mortality. Variation of the pre-pandemic reference period substantially affected P-scores, but death-count- and ASMR-based estimates changed in a closely corresponding manner. Relative excess mortality derived from death counts and ASMRs is closely related, and differences between the approaches can be explained by their underlying weighting and baseline assumptions. For trend-based excess mortality estimation, the standard population, age-group definition, and baseline specification are therefore integral methodological choices. Death counts provide a simple and transparent reference measure for analyses within a single population, whereas age-standardised estimates require careful consideration of these additional assumptions.