The WHO estimates of excess mortality associated with the COVID-19 pandemic

Abstract

The World Health Organization has a mandate to compile and disseminate statistics on mortality, and we have been tracking the progression of the COVID-19 pandemic since the beginning of 2020¹. Reported statistics on COVID-19 mortality are problematic for many countries owing to variations in testing access, differential diagnostic capacity and inconsistent certification of COVID-19 as cause of death. Beyond what is directly attributable to it, the pandemic has caused extensive collateral damage that has led to losses of lives and livelihoods. Here we report a comprehensive and consistent measurement of the impact of the COVID-19 pandemic by estimating excess deaths, by month, for 2020 and 2021. We predict the pandemic period all-cause deaths in locations lacking complete reported data using an overdispersed Poisson count framework that applies Bayesian inference techniques to quantify uncertainty. We estimate 14.83 million excess deaths globally, 2.74 times more deaths than the 5.42 million reported as due to COVID-19 for the period. There are wide variations in the excess death estimates across the six World Health Organization regions. We describe the data and methods used to generate these estimates and highlight the need for better reporting where gaps persist. We discuss various summary measures, and the hazards of ranking countries' epidemic responses.

Fig. 1. Global excess and reported COVID-19 deaths and death rates per 100,000 population.

a, Cumulative global excess death estimates and the cumulative reported COVID-19 deaths by month from January 2020 to December 2021. b, Global excess death rates per 100,000 population and the reported COVID-19 death rates per 100,000 population, also by month, from January 2020 to December 2021. On both plots, the central lines of the excess mortality series show the mean estimates and the shaded regions indicate the 95% uncertainty intervals. Source data

Fig. 2. Global and WHO region P-scores (excess deaths relative to expected deaths).

Monthly estimates of P-scores, expressed as a percentage, aggregated globally and for the six WHO regions for the period January 2020 to December 2021. All plots show the mean estimates and the 95% uncertainty intervals. Source data

Fig. 3. The 25 countries with the highest total estimated excess deaths between January 2020 and December 2021.

The red dots show the total reported COVID-19 death numbers. The purple dots show the mean total estimated excess death numbers with the width of the bars showing the 95% uncertainty intervals. Source data

Fig. 4. Mapping estimated P-scores (excess deaths relative to expected deaths).

The map shows the geographic distribution of the mean P-scores for years 2020 and 2021 across all 194 WHO member states. The darker the colour the higher the estimated mean P-score. The patterns indicate the quality of the all-cause mortality data that were available for each respective country with the solid pattern showing full or partial data, dots for mixed data and diagonal lines for no data. Source data

Fig. 5. The 25 countries with the highest mean P-scores (excess deaths relative to expected deaths).

The plot shows the 25 countries with the highest mean P-scores for years 2020 and 2021 after ranking all WHO member states with populations greater than 200,000 by mean P-score from highest to lowest. The mean for each country is shown using the central grey dots and the widths of the bars show the 95% uncertainty intervals. Source data

Fig. 6. Mapping the ratio of total excess deaths to total reported COVID-19 deaths.

The map shows the geographic distribution of the mean ratio of the total excess deaths to total reported COVID-19 deaths for years 2020 and 2021 across all 194 WHO member states. The darker the colour the higher the estimated mean ratio. The patterns indicate the quality of the all-cause mortality data that were available for each respective country with the solid pattern showing full or partial data, dots for mixed data and diagonal lines for no data. Source data

Extended Data Fig. 1. Mapping the availability of all-cause mortality data.

The countries in dark blue have all 24 months of data available for January 2020 to December 2021 whereas those in purple have monthly data available but for less than 24 months. For the countries in green we only have either subnational or annual data for some or all of the period and for those in yellow do not have any representative all-cause mortality data available for the pandemic period.

Extended Data Fig. 2. Overview of modelling strategy to produce excess mortality estimates for all countries.

The flowchart outlines how the excess mortality associated with the COVID-19 pandemic is estimated. By definition, this is the difference between the all-cause mortality (ACM) during the pandemic period and that which was expected had the pandemic not occurred (light-green). Starting with the historic monthly ACM data (light-blue) we apply the Negative Binomial Spline model to generate the expected deaths for each country. The data that are available inform the modelling strategy employed to calculate the pandemic period ACM (light-pink). For the places with full data, the reported ACM are taken as is. For those countries with only subnational ACM data, a Multinomial Subnational model is employed to derive national level ACM based on the historic fractions of deaths observed in subnational regions. The overdispersed Poisson Covariate model is fitted to countries with monthly pandemic data, and this model is used to estimate pandemic period ACM for countries without any reported ACM. The covariate model is also used to infer the within-year monthly pandemic ACM in countries with only annual data, via the Multinomial Covariate model.

Extended Data Fig. 3. Excess mortality densities and ranking probabilities for select countries.

Left: Ridgeplots representing the uncertainty in the cumulative excess monthly mortality rate over January 2020–December 2021 for six European countries. Right: Bivariate plots of pairs of excess rates (lower triangular), 1-dimensional summaries for individual countries (diagonal), and probabilities that the excess for the country labelled on the left exceeds the rate for the country labelled at the top (upper triangular). These probabilities are the fraction of points that lie below the red line in the corresponding bivariate plot.

Extended Data Fig. 4. Excess rates over time and ranking probabilities for select countries.

: Left: excess monthly mortality rate (per 100,000 of population), with 95% uncertainty intervals, by month, for six European countries. The panels to the right display the ranking probabilities for each of the individual countries. The probabilities sum to one for each country within each month, i.e., a country has to be in one of the six ranking positions. Also for a fixed month and ranking position across all countries the probabilities sum to one because one of the countries has to be in rank 1, 2, etc, in each month.