Counting the Dead: What is COVID’s True Effect on Africa?
In our last post, we discussed the ways in which low COVID-19 testing levels could misleadingly indicate that a country is doing well in the fight against coronavirus. Here, we continue the discussion of the implications of limited health information systems on the COVID-19 response, with a focus on the inability to count deaths in Africa.
A 2007 Lancet special series focused on the importance of well-functioning civil registration and vital statistics systems and described the considerable gaps in birth notification, death registration and medical cause of death declaration in low-income countries — especially Africa. These articles point to the importance of birth and death registration systems in effectively managing health and development programs and policy planning. WHO, the World Bank, and the Global Financing Facility have each covered the importance of vital statistics systems with regards to human rights and responding to health crises.
Mortality data at the national and sub-national level allow policy makers and public health officials to understand major contributors to death and allocate research and programmatic effort accordingly. In an emerging outbreak situation, monitoring mortality can be an early indicator that an otherwise undetected disease might be spreading. By comparing current monthly trends in death to historical levels, one can estimate “excess mortality” which might not be expected if not for a novel cause (i.e. an emerging infectious disease, severe weather event, or war). In the case of COVID-19, estimates of excess death have been instrumental in contextualizing and quantifying the true death burden of the disease. This application, however, has been restricted to mostly high-income settings where mortality registration is both frequent enough and of high enough quality to detect trends.
One of the first investigations into excess mortality was in northern Italy. In the earliest and hardest hit high-income countries, our understanding of the severity of COVID-19 in the region comes not from their official confirmed counts of coronavirus deaths, but rather analysis of historically “normal” levels of mortality.
At the peak of the outbreak, researchers looking into excess mortality were able to estimate that true COVID-19 mortality was likely one-third higher than the official death toll. This has been important to understanding the true fatality ratio of COVID-19 as well as the potential public health impact of the pandemic. Videos from Lombardy showed scenes of overrun hospitals and morgues in one of the world’s highest functioning health systems. Afterward, these images would unfortunately become symbolic of a truly rampant COVID-19 epidemic, setting the standard for the need for political action.
So, what then of the observed COVID spread in sub-Saharan Africa? As of the end of June, case fatality ratios ranged from a high in Chad of 8.5% to a low in Rwanda of 0.2% — half as much as Iceland which has one of the best COVID-19 responses in the world. Further, five countries have still not reported a single COVID-19 death. In total, there are approximately 5,500 confirmed deaths and over 290,000 cases for a case fatality ratio of 1.9%. [1] This is significantly lower than the average of high-income countries, which hovers around 7%, while some of the hardest hit countries in western Europe have rates in the teens.
[1] Here we differentiate case fatality from infection fatality, as the former considers only confirmed data, while the latter is a modeled estimate accounting for underreported diagnosis.
Over the course of the pandemic we have regularly seen articles describing how Africa has dodged the worst of the pandemic and even how the disease has been suppressed. There are certainly valid reasons why we might expect COVID to be less severe in Africa. The median age on the continent is less than 20 years old and absolute rates of diabetes and hypertension (all risk factors for COVID severity) are lower than in many high-income countries where the epidemic has been most severe. By standardizing estimates of the case fatality ratio to African nations according to age, we control for a younger age structure in Africa. In doing so we estimate COVID-19 might be half as lethal at the population-level compared to high-income settings — 3.5% vs 7.1%. Still, there are many potential countervailing factors and unknowns.
For example, HIV and tuberculosis might increase risk of death due to COVID by over 2 times — with both conditions being highly prevalent throughout Africa. Furthermore, it is unknown what extent to which common co-morbidities like malaria and malnutrition might contribute to COVID susceptibility and severity, or what role a weaker health system might play on survival. Additionally, per capita testing levels are hundreds of times lower than many other countries and are not expected to increase significantly in the near- to mid-term.
This suggests that an age-adjusted case fatality ratio of 3.5% is likely overly optimistic relative to high-income settings. The observed case fatality ratio in Africa is nearly 75% less than what has been seen in high-income settings. This is a far greater difference in lethality than what can be explained by age alone and there are many other factors which suggest an even poorer fatality ratio.
While there are many potential explanations for the very low mortality seen in Africa, one challenge we highlight here is limited availability of mortality data and death registration. Mortality data availability for Africa at the global level is extremely sparse. One major source, the WHO Mortality Database is very limited in both low-income country coverage and recency. Currently, only three countries in sub-Saharan Africa are represented in the database, with most recent data sourced from 2015. Another global source, the Institute for Health Metrics and Evaluation’s Global Burden of Disease, has estimates of mortality up to 2017; however these are highly reliant on modeled rather than empirical estimates in low-income settings. In each case, the data are neither frequent enough, recent enough, widespread enough, or complete enough to understand rapid fluctuations in deaths.
The poor availability of mortality data at a global level is linked to systems limitations at the country-level. For national death registration systems, only one country in sub-Saharan Africa (South Africa) has a sufficient, high quality death registration system. Fewer than 10% of deaths on the continent may be recorded, with even fewer having a cause of death attributed to them. Anecdotally, there are reports of health districts without an official recorded death in years. This ultimately means that the inability to detect unexpected upticks in mortality makes it extremely difficult to understand to what extent, if any, COVID spread might be controlled.
Of the aforementioned excess mortality analyses, only two have been demonstrated in middle- or low-income countries, where the uncounted COVID mortality is extremely high. In Istanbul, COVID mortality may be twice what has been recorded, while in Jakarta COVID-related deaths could be as much as 10-times the official number. Interestingly, these data were not drawn from proper death registries but rather by looking at trends of funerals and burials. While these estimates are troubling, they bring to light the true health burden of the COVID pandemic in those countries and a call to action for policy makers and health agencies.
The exact extent to which COVID-19 has spread and caused death in Africa remains an open question. We have pointed to the need for the continent to accelerate testing capabilities and uptake. But here we also have shown that the inability to simply count the number of people dying each month — from any cause — has also limited our understanding of the severity of disease on the continent.
While we recognize that rapidly standing up high quality death registration systems in 48 low-resourced countries is unlikely, rapidly deploying simple data collection tools to communities to tally deaths, while integrating and digitizing existing records of burials could be feasible and enlightening. In our view, it is obvious that the impact of the epidemic is understated — both in terms of cases and deaths. But appropriate policy action can only be enabled with quality information, which in this instance involves the most basic effort to count the dead.
Thank you for reading, we welcome your comments below.
About Cooper/Smith
When implemented correctly, data collection and analysis ensures that programs succeed and achieve actionable results. In international development, that means concrete improvements for those who need it most. At Cooper/Smith, we use hard data to increase the effectiveness and efficiency of development programs worldwide.
Write to us at contact@coopersmith.org or visit www.coopersmith.org to learn more about our data-driven approach to health and development.
