Estimating Multi-Annual Excess Deaths: A Formula and an Application to the Chronic Us Mortality Disadvantage, 1956-2023

Patrick Heuveline, University of California, Los Angeles
Nanum Jeon, University of California, CA

With low data requirements, relatively easy to estimate and readily interpretable, the number of excess deaths is arguably the most salient indicator of cross-country mortality differences. While sensitivity analyses have brought attention to the selection of counterfactual rates, we discuss here the other component of excess deaths estimation, population exposure. Incorrectly assuming exposure to be independent of mortality rates yields biases often small enough for annual numbers of excess deaths. This conventional assumption can be more problematic for longer reference periods and conceals the non-additivity of numbers of excess deaths across periods. We provide a formula to calculate multi-annual numbers of excess deaths from conventional annual estimates. We then use the chronic US mortality disadvantage relative to Canada and the 5 largest European countries to illustrate the dynamics of multi-annual excess mortality and its cumulative impact on population size. We show for instance that depending on the reference period, the multi-annual estimate of excess deaths is alternatively lower and higher than the simple sum of the annual estimates for the intervening years. Substantively, we show that excess mortality since 1960 has reduced the 2023 end-of-year US population by nearly 10 million, of which slightly over half are of working ages.

Keywords: Data and Methods, Mortality and Longevity, Civil Registration and Vital Statistics

See extended abstract.