Excess mortality and vaccination

Philip Schellekens  |  
How front-runners in the vaccination space compare across countries on excess mortality

The relationship between excess mortality and vaccination can be explored at different levels. Micro-level evidence suggests a strongly beneficial effect of vaccination on the risk of developing severe sickness or death following infection by the coronavirus. This post examines the relationship between excess mortality and vaccination at the macro level and explores the performance of front-runners across countries. It first takes a look at 3-month rolling averages for excess mortality and vaccination and then examines the cumulative numbers for the entire period of 2021 and 2022. It concludes with a discussion of the various confounding factors at play.

Contents

Excess mortality and vaccination: the three-month rolling view

The visualization below illustrates the evolving relationship across countries between excess mortality and vaccination. We consider the period of early 2021 (when the global vaccination campaign started to take off) until the present day. 

Let’s unpack the different elements of the chart before we discuss the results:

  • Horizontal axis: the total vaccine coverage ratio, expressed as a rolling average over the last 90 days and defined as the sum of adjusted primary doses and booster doses per 100 people. Primary doses are adjusted by converting doses of 1- and 3-dose vaccines into 2-dose equivalents (we scale them respectively by x2 and x2/3 so that full vaccination corresponds to 2 shots; more on that here). 
  • Vertical axis: the excess death rate. We take the daily 90-day rolling average of excess deaths per 100,000 people, where excess deaths are the mid-point values of the excess death model by The Economist (more here). 
  • Bubbles: population size of countries (measured by the medium variant of the WPP’s projection for 2021). he two largest bubbles are China and India, respectively in light green and turquoise. 
  • Colors: the World Bank income classification, which divides the world into four income groups: high-income (HIC), upper-middle-income (UMIC), lower-middle-income (LMIC) and low-income (LIC) countries. 
  • Regression line: a linear weighted regression between excess mortality and vaccination, weighted by population size.

The chart illustrates the following aspects about the relationship between excess mortality and vaccination. To zoom in on them, we are also adding two snapshots below of what the chart looked like mid-2021 (when 90-day excess deaths reached a global peak) and what it looks like today based on the most recent information. 

  • High excess deaths in developing countries.  Some of the highest and more persistent bounces in excess deaths are observed in developing countries (which correspond to the observations not colored in red, i.e. all countries that are not high income). Most significant among them all was the spike in India May 2021 (see the large turquoise bubble). Considering the full horizon of the pandemic, the group of lower-middle-income countries have suffered the highest cumulative excess death rates in the world whereas those for other income groups have been surprisingly similar. More details here and here
  • Large dispersion in excess deaths initially, much lower later. Unconditionally, dispersion seems to be highest for lower- and middle-income countries. Conditional on the level of vaccination, we see a huge drop in dispersion between 2021 and 2022. Country outcomes varied a lot more during the first year of the vaccination campaign. 
  • Vaccination front-runners do generally a lot better than the rest. The relationship between excess mortality and vaccination tilts strongly downwards as time progresses and front-runner get better-vaccinated. The population-weighted regression line is strongly influenced by population outliers such as China, which did well on vaccination (even though as discussed below there are confounding factors). 

Excess mortality and vaccination: the cumulative view

The previous analysis was based on rolling averages for the excess death rate and the vaccine coverage ratio over a period of three months. That allowed us to detect any changes in the cross-country relationship between excess mortality and vaccination between 2021 and 2022. 

In what follows we take the analysis forward and come up with an aggregate view that is based on the full 2021 and 2022 period. In other words, we relate the cumulative build-up in excess deaths with the cumulative progress in vaccination. On the y axis, we have excess deaths cumulated since the start of 2021 up through the date indicated and per 100,000 people. On the x axis, we have the cumulative mean of the total vaccine coverage ratio. The full picture is summarized by the dynamic visualization below. 

As we can see, the relationship over the entire period is a negative. The slope flattened somewhat during the course of 2022. Note also that the flattening is less dramatic than in the three-monthly averages as the cumulative numbers incorporate more information from the past. This is visible, for example, in the total vaccine coverage ratio which has a much-more compressed scale reflecting the fact in the earlier part of 2021 vaccination levels were still very low.

We can also decompact the above visualizations by World Bank income group. This is what the chart below does.

Interestingly, the negative relationship pops up for high income countries and upper-middle income countries. However, due to what happened in India May 2021, the relationship in lower-middle income countries is positive. These results seem the support the suggestion that the cross-country relationship starts to tilt beyond a minimum threshold level of vaccination.

Confounding factors

The cross-country comparisons suggest that vaccine coverage above a certain threshold is associated with lower excess mortality. We see that low vaccination rates are associated with much higher excess death rates, particularly during 2021. We also note that front-running countries in the vaccination space typically perform best on excess mortality. 

As this post has shown, the cross-country relationship between excess mortality and vaccination flattened considerably during 2022, even though the relationship remained negative over the full 2021-2022 period as the cumulative numbers show. 

This reflects the fact that more countries got better vaccinated, which produces a flatter and more elongated regression line. But it also has to do with natural immunity. As time passed and the pandemic matured, a large share of the global population got infected which bolstered immunity levels not just through vaccination but also natural infection. 

Several other confounding factors affect this cross-country relationship. These include (1) the significant evolution in the pathogenicity of the prevalent variant in circulation, with delta (which was particularly dominant during 2021) more lethal than omicron (which dominated especially during 2022); (2) the variation across countries and over time in the quality and strictness of containment and suppression policies; (3) a country’s age and health profile, and (4) the general quality and resilience of country health systems. 

The simple bivariate cross-country correlations between excess mortality and vaccination should not be over-interpreted as many other factors are at play. Yet, as tentative as they are, the results do complement the finding from micro studies that show the beneficial effects of vaccination on avoiding severe pandemic outcomes (see for example herehere and here).

Note: Thanks to Tom Frieden for helpful suggestions on this post.

Disclaimer: Posts by the Center for Global Development reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions. Likewise, views expressed do not necessarily reflect those of the United Nations, the United Nations Development Programme, its programmes/projects or governments.  The designations employed do not imply the expression of any opinion whatsoever concerning the legal status of any country, territory or area, or its frontiers or boundaries.  

Share
Visuals