Quantifying the health burden of COVID-19 using individual estimates of years of life lost based on population-wide administrative level data

Background The COVID-19 pandemic caused substantial health losses but not much is known about how these are distributed across the population. We aimed to estimate the distribution of years of life lost (YLL) due to COVID-19 and investigate its variation across the Dutch population, taking into account pre-existing differences in health. Methods: We used linked administrative data covering the entire 50+ Dutch population over 2012-2018 (n=6102334) to estimate counterfactual individual-level life expectancy for those who died from COVID-19 in 2020 and 2021. We estimated survival models and used Cox-LASSO and Cox-Elastic Net to perform variable selection among the large set of potential predictors in our data. Using individual level life-expectancy predictions, we generated the distribution of YLL due to COVID-19 for the entire 50+ population by age and income. Results: On average, we estimate that individuals who died of COVID-19 had a counterfactual life expectancy about 28% lower than that of the rest of the population. Within this average there was substantial heterogeneity, with 20% of all individuals who died of COVID-19 having an estimated life expectancy exceeding that of the age-specific population average. Both the richest and poorest COVID-19 decedents lost the same average number of YLL, which were similarly dispersed. Conclusions: Accounting for pre-existing health problems is crucial when estimating YLL due to COVID-19. While average life expectancy among COVID-19 decedents was substantially lower than for the rest of the population, the popular notion that only the frail died from COVID-19 is not true.