Africa’s COVID-19 statistics point to bias in excess mortality modeling

Despite the high level of informality, the mortality rate from COVID-19 in Africa has been paradoxically low compared to the countries of the Global North. Examining studies in which low rates are attributed to poor statistical reporting, Kate Meagher argues that mortality overestimations tell us more about the assumptions of the modellers than about the realities of COVID-19 in Africa.

Africa was expected to be hit hard by the COVID-19 pandemic due to extreme socioeconomic vulnerability. The informal sector accounts for 72% of African urban employment, exacerbated by weak health systems and the lowest level of social protection in the world. International organizations have warned that Africa’s huge informal workforce will face a tough choice between disease or hunger. UNECA predicted up to 3.3 million deaths in Africa by the end of 2020, while Melinda Gates spoke of “corpses in the streets‘. However, statistics show that Africa had the lowest death rate from COVID-19. WHO data show that the death rate from COVID-19 in Africa is about 13% of that in Europe and North America. Africa has not only been hit less by the pandemic than developed countries, but, as I have shown in recent article that mortality rates from COVID-19 are inversely related to levels of informal employment, both interregional and among African subregions.

What explains this paradoxical relationship between informal employment and COVID-19 deaths? Is it a product’bad numbers‘ in Africa and other regions with large informal economies? Or is this further evidence of a poor understanding of how the pandemic intersects with the economic, social and political realities of Africa? A closer look raises questions about the trend towards revisiting COVID-19 death rates in Africa, using statistical modeling to validate rather than test global public health narratives.

Paradoxes and statistical tricks

Data from the International Labor Organization on informality show that informality in Africa is about 50% higher than in Latin America and more than four times higher than in developed countries. However, 18 months after the start of the pandemic, globally recognized Worldometer data show that the death rate from COVID-19 in Africa is 18% of the level in Latin America and 20% of the level in developed countries. In Africa, too, where the level of informality varies by region, West Africa more than twice as informal as South Africahowever, the death rate from COVID-19 in West Africa was less than one tenth victims in South Africa.

Africa’s conflicting mortality data has been challenged by a plethora of excess mortality models arguing that the continent’s low mortality rates drastically underestimate the true impact of the pandemic.

Africa’s conflicting mortality data has been challenged by a plethora of excess mortality models arguing that the continent’s low mortality rates drastically underestimate the true impact of the pandemic. Estimates of excess mortality obtained Economist show that sub-Saharan Africa has the highest excess mortality rate of any region. A a study by Wang et al. published in The Lancet, the gap between estimated and reported COVID-19 deaths is higher in South Asia and Sub-Saharan Africa than in other regions.

Although excess mortality models are seen as low quality corrective data for Africa, a closer look raises important questions about their reliability. Two statistical issues are particularly problematic: the emphasis on excess mortality relationship not death rates, as well as the tendency to extrapolate data from the Global North to model “real” results in Africa. Famous reports of excess mortality show a tendency to sensational mortality figures in Africa, highlighting the excess mortality. relationship is the ratio of excess mortality estimates to reported deaths. A table from the Economist report shows that Africa has an excess mortality rate of +700%, the highest of any region, while Wang et al. Sub-Saharan Africa has been repeatedly noted to have one of the highest gaps between estimated surplus and reported deaths. However, among these dramatic claims, both studies note that sub-Saharan Africa, with the exception of South Africa, had particularly low mortality. rates. High excess death relationship are simply an artifact of the high overestimation of mortality compared to the low reported mortality. Even these liberal estimates still leave most African countries with some of the lowest estimated COVID-19 death rates in the world.

Bad numbers or bad modeling?

The second problem is related to the tendency of global studies of excess mortality to use data from developed countries to estimate excess mortality in Africa, despite apparent differences in conditions. Although known excess mortality models account for demographic differences, they universalize environmental and biophysical factors on the assumption that they will operate in Senegal as they do in Spain. The Economist model was trained on data from rich and middle-income countries, and Wang et al. also acknowledge that they are using data from developed countries to estimate missing data for Africa, ignoring large differences in the context of transmission and underlying health conditions. The Economist also notes that his model overestimates mortality when access to health services is limited, as is the case in much of Africa.

A growing body of evidence indicates that dramatic global estimates of excess mortality in Africa are implausible and do not correlate with known events and realities in the region.

Excess Mortality Study in the African Region by Kabore etc. shows that the desire for universalization inherent in global modeling runs counter to the evidence that the trajectory of COVID-19 in Africa has been different from other regions. A number of factors have limited susceptibility to the virus in Africa, including the world’s youngest population profile, lower urbanization, more outdoor food, pre-existing immunity, and a timely public health response. Taking into account local data limitations, Kabore etc. caution against the tendency to fill data gaps with globally modeled estimates that ignore important local specific factors. The detailed methodology highlights the importance of using available local data to fill gaps and ensure reliable estimates. The authors demonstrate that estimates based on African conditions are possible and preferable to building models by extrapolating data from other regions.

The tyranny of implausible estimates

A growing body of evidence indicates that dramatic global estimates of excess mortality in Africa improbable And does not correlate with known events and realities in the region. Known models of excess mortality exhibit a propensity for extravagant inclusion errors, inappropriate estimation procedures, and apparently baggy confidence intervals. The Economist model allows for inflated mortality estimates, sometimes exceeding the population of an entire country. Global models estimate COVID-19 mortality rates in sub-Saharan Africa at 14 times official numbers. Wang et al. suggest that roughly 480,000 deaths in West Africa went unnoticed during the pandemic—nearly 50 times more than from Ebola. Over a million people are estimated to have died in East Africa during the same period, but the statistical machine recorded only 29,000 deaths. Given the absence of apocalyptic news coverage outside of South Africa, it is frankly unlikely that such high death rates would escape the attention of the local and international community. More realistic estimates by Cabore et al. the excess death rate is just under three times Africa’s official figures, while real-time Worldometer estimates are 1.5 times the official figures.

The Economist acknowledges that their own estimates of excess mortality are “extremely rough”, but claims of significantly higher mortality in Africa are accepted without rigorous scrutiny. Instead of provoking a deeper exploration of the paradox of Africa’s low mortality rate under conditions of high informality, statistical modeling is simply being used to “normalize” the anomalous outcomes of the pandemic in Africa. Yet excess mortality patterns provide circumstantial evidence that COVID-19 had the same impact in Africa as it did elsewhere, obscuring the important question of why Africa’s death rate was so low amid rampant informality. This calls for the decolonization of statistical modeling as a key part of the project. decolonization of global health.

The content created on this blog is for informational purposes only. This article represents the views and opinions of the authors, but does not reflect the views and opinions of the Impact of Social Science blog (blog) or the London School of Economics and Political Science. Please see our comment policy if you have any concerns about posting a comment below.

Image credit: Evin van Berghijk – Quantum via Unsplash.com.