COVID-19 death toll gleaned from alternative data sources

In the study, an international team estimated the number of COVID-19 deaths in regions where a lack of testing capacity or robust civil registries may have led to underreporting. The researchers used burial reports, satellite imagery of cemeteries, and a social media survey of infection from the capital cities of Ethiopia, Yemen and Sudan respectively, to look for excess deaths not captured by the official data. 

“The idea came to us while providing modelling support to partners in low- and middle-income countries during the pandemic,” says Oliver Watson, a research fellow at the London School of Hygiene and Tropical Medicine, who led the study. “It was evident that the official COVID-19 death counts were failing to accurately reflect the intensity of the pandemic's impact within these countries. This led us to collaborate closely with local partners to identify data sources that could lend more precision in quantifying the excess mortality attributable to COVID-19.”

To do this, the team took advantage of an existing mathematical model that uses the infection fatality rate of COVID-19 to predict the seroprevalence of SARS-CoV-2 from the number of deaths. They used this to estimate the seroprevalence based on either the number of reported COVID-19 deaths or the estimated number of deaths from analysing the alternative data sources. When the two predictions were different, they checked which was closer to the actual seroprevalence rate reported in surveys in those locations around the same time. With these comparisons in hand, they could estimate the fraction of underreported COVID-19 deaths.

The analysis showed that most or all COVID-19 deaths were reported in Addis Ababa (67-100%), while in Aden and Khartoum the vast majority of deaths went unreported (<10% in both cases). For Watson, this isn’t simply about getting a more accurate picture of COVID-19. “It’s having an understanding of how many individuals lost their lives – historical memorialisation is a fundamental human right,” he says.

Watson says that a similar approach could be used to provide better information during an outbreak. “With the knowledge that these alternative data are reliable indicators, we could have used these data to estimate infection fatality rate in these settings, which would have disproved hypotheses that countries with limited resources were spared the worst of the pandemic,” he explains.

While the new study validates the use of these alternative data sources, Watson stresses that this shouldn’t replace conventional data collection. “The global community must continue to prioritize support and investment towards developing vital registration systems that capture all-cause mortality globally.”

Sign-up to receive our e-alert update every two weeks to keep up with everything new on the portal.