Skip to main content

Magnitude, demographics and dynamics of the effect of the first wave of the COVID-19 pandemic on all-cause mortality in 21 industrialized countries

Our take —

This study used data from 21 high income countries to estimate the excess deaths from mid-February 2020 to May 2020, compared to the estimated death rate during this period had the pandemic not occurred. Models estimated 206,000 excess deaths, equivalent to an 18% increase in mortality during this time period. Overall, England and Wales accounted for 28% of the excess deaths, followed by Italy (24%) and Spain (22%). The study was unable to determine whether these deaths were due specifically to COVID-19 or from other indirect causes or unrelated secular trends, but suggest a substantial impact of COVID-19 on populations throughout the world.

Study design

Ecological

Study population and setting

The study objective was to calculate the number of excess deaths among industrialized countries from mid-February to May 2020 due to the pandemic. The study used weekly death data as reported by the respective governments of 21 high income countries: Australia, Austria, Belgium, Bulgaria, Czechia, Denmark, England and Wales, Finland, France, Hungary, Italy, the Netherlands, New Zealand, Norway, Poland, Portugal, Scotland, Slovakia, Spain, Sweden, and Switzerland. It used a time-series through mid-February 2020 to estimate the model parameters for the next 15 weeks (ending in May). Excess deaths per 100,000 people and the relative increase were estimated using an ensemble of 16 Bayesian Poisson models and using an applied probabilistic average generated from these. The models adjusted for seasonality, temperature, public holidays, and used linear trend terms for secular changes in death. The different combinations of these parameters being included in the models led to 16 total models in the ensemble. The counterfactual prediction of no pandemic during the time period was generated using the same ensemble with data from January 2010 to February 2019 and assessed fit for 3 different subsets of this time period.

Summary of Main Findings

Deaths in all countries were at least at levels expected from the counterfactual model, but began to diverge by March 2020, with an estimated 206,000 (95% Credible Interval [CrI]: 178,100 – 231,000) additional deaths than expected equivalent to an 18% (95% CrI: 15- 21%) increase in mortality. Of these, 105,800 (95% CrI: 90,400 – 119,000) were in men and 100,000 (95% CrI: 82,000 – 117,500) were in women. England and Wales accounted for 28% of these deaths, followed by Italy with 24%, and Spain with 22% of the total excess deaths. The posterior probability, representing the likelihood of an increase in deaths, was more than 99% for men in Switzerland, and for all deaths in the Netherlands, France, Sweden, Belgium, Italy, Scotland, Spain, and English and Wales. For the timing of excess deaths, men in Italy were estimated to experience the earliest rise in mortality in the first week of March 2020. Deaths returned to expected levels in France and Spain in April, while mortality continued to remain elevated through May in England and Wales and Sweden. The total mortality was concentrated among people 65 years and older, who experienced 94% of all excess deaths; mortality was 40% greater than expected in Spain and England and Wales, and 30% higher in Belgium, Scotland, and Italy.

Study Strengths

This analysis used 16 models to create predictions and average them, which allowed it to remain robust even if some of the models were incorrectly specified. By using Bayesian analysis, they were able to integrate prior information into their model, which can improve model prediction accuracy. They also used a wide range of countries (21 in total), representing some climate and policy diversity, and still found increased rate of death consistently elevated.

Limitations

The major limitation is that this was an ecological study, and it could not discern which deaths were caused due to the pandemic, which may have been due to indirect causes of the pandemic such as delay in preventative care or disruption of chronic care, and which may have been due to other unrelated causes. The use of the counterfactual model to estimate the expected deaths attempted to distill the effect to the pandemic, but other temporal changes in excess deaths may still have played a role. Additionally, the model could not discern which underlying conditions or risk factors individuals may have had that may have impacted COVID-19 risk and/or the excess death rates in a given country.

Value added

This study examines a range of industrialized countries to estimate excess deaths of more than 4 million people, providing important information on the global impact of COVID-19.

This review was posted on: 7 April 2021