According to official figures, 5.7 million people worldwide will have died from Covid-19 by February 2022. In reality, there are probably many more, because not all countries report their death rates in the same way or with the same reliability. Depending on the model used, the worldwide excess mortality from Covid-19 may be as high as 12.7-24 million.
Calculating global covid mortality is not so easy: the available information is not immediately comparable. For example, at the start of the pandemic, the Netherlands only counted patients who had died in hospital after a positive test, while the highest covid deaths occurred in nursing and care homes.
The excess mortality can provide a better picture. To do this, the total mortality during the pandemic is compared with the average mortality over the previous 5 years. The disadvantage of this is that the change in population composition must also be taken into account. For example, many European countries have a large number of people over 80, which means that the national death rate was already rising. The figures for excess mortality also ‘include’ the smaller number of flu deaths, plus the deaths indirectly caused by the Covid pandemic, such as those caused by delayed care.
Unfortunately, there are also many countries that do not keep demographic data such as death rates at all. For those cases, researchers can fall back on models: on the basis of indicators such as population density, the number of care providers and the level of prosperity, it is possible to model how high the mortality is in those countries, and thus what the global mortality rate from Covid-19 is.
The model of the British The Economist used machine learning and identified about 100 indicators associated with excess mortality in countries for which data is available. The model shows that the official statistics are an underestimation of the actual mortality – only the degree of that underestimation varies. In wealthy countries, the excess mortality is about a third higher than the reported numbers, but in poor countries the model estimates as many as 20 times as many deaths from Covid-19, albeit with considerable uncertainty.
The US Institute for Health Metrics and Evaluation (IHME) estimates are significantly lower. Information about how that model works has not yet been published.
Meanwhile, researchers are trying other ways to count the number of deaths, such as by counting graves on satellite images, by getting death rates from a local newspaper or through telephone surveys. But as long as basic mortality figures are not or hardly collected, we will probably never know the real number of deaths from Covid-19, despite all the modeling exercises.