Fig. 1: Patterns of COVID-19-associated mortality across settings.

a, Countries with age-specific death data (beige) and locations with seroprevalence data (dots). b, Estimated median and 95% credible intervals of the proportion of the population that has died in each age group, relative to the proportion of population that has died among 55–59-year-old individuals in that country (black dots and lines), plotted on a log-linear scale. Coloured dots represent the country- and age-specific relative risks (RR) of COVID-19-associated death in the population relative to the risks of 55–59-year-old individuals observed from reported death data, accounting for population age distributions (Supplementary Methods 1) (n = 538,477 reported deaths). All data points are plotted at the midpoint of the reported age group. The grey shaded areas highlight the relative risks of death by age for age groups of 65 years and older, which were excluded from model fitting and black asterisks represent estimates inferred solely from England data, which were derived independent of deaths that occurred in nursing homes. c, Comparing the reconstructed incidence of deaths with reported data for age groups aged 60 or 65 years and older for a subset of countries from which deaths that occurred in nursing homes could be excluded. Black dots and lines indicate the estimated median and 95% credible intervals derived from only the England data (blue bar); orange bars show the reported incidence of deaths of individuals aged 60 years and older in 13 additional countries that did not occur in nursing homes. Countries labelled with an asterisk indicate those in which the number of deaths were reconstructed for those aged 65 years and older, to align with the reported age groups for each country.