Why we need to count the coronavirus pandemic's invisible deaths

Eighteen months into the Covid-19 pandemic, we don’t know exactly how many people have died. A true global tally is not just overdue. It’s vital. 
 

The task is a Herculean one, given that so much of the world struggles to record basic information about deaths even in normal times. But without a better grasp of the toll, we will strain to assess the precise, real-world impact of vaccines and other preventative measures, or to understand just how lethal variants have been. Absent efforts to fill the gaps even in developed nations, the wealthy will be counted, but too many of the rest will not, making it easy for economic recovery to overlook the underserved. 

Deaths are a sure indicator of the damage wrought. The trouble is just how bad we are at counting them.

The World Health Organization puts the pandemic’s toll at more than 195 million infections and over 4 million deaths. Both anecdotal evidence and academic estimates suggest reality is far worse. That’s not necessarily because of deliberate misreporting, though there have certainly been instances of politically motivated lowballing. More commonly, it’s a combination of insufficient testing and patchy data collection, families’ reluctance to register coronavirus deaths to avoid funeral restrictions and stigma, and a narrow view of what counts as a Covid-19 fatality amid complications or confusion from underlying ailments. Probable coronavirus deaths that aren’t confirmed by a diagnostic test can also go unrecorded, despite global guidelines to the contrary.

Considering excess mortality — a measure that includes all fatalities above what would be expected in normal conditions, based on historic trends — gets around some of those issues and is arguably a better metric for the overall loss, capturing deaths from mental health strain, drug abuse, or illnesses left undiagnosed while hospitals were overwhelmed. It just isn’t necessarily any easier to track.

The developing world has gotten better at registering births, partly because of incentives or the need for families to provide legal identity to access schooling or other services. But still too many deaths are either not reported at all, or are recorded inadequately. Civil registration systems reflect the capacity of the states supporting them. Often they are slow and fragmentary, even for deaths in hospital. One Bangladesh study done outside the capital, Dhaka, found that out of all deaths in the households surveyed, only 17% were registered and most of those were men. Only 5% of female deaths made it into the record.

When it comes to Covid-19, India has provided one of the most dramatic examples of undercounting. The government’s death toll is over 420,000, suggesting a lower case fatality rate and fewer deaths per capita than the U.S. or Latin America’s hotspots. In fact, seroprevalence surveys that estimate the number of infections, plus data-based evaluations from journalists and researchers, suggest numbers several times higher. So does anecdotal evidence, including reports during the most recent wave of corpses dumped in the Ganges, firewood shortages for cremations, and, in one instance, of furnaces burning the dead so continuously that metal parts began to melt. 

Unsurprisingly, under-reporting is most acute where access to health care is poor. Murad Banaji, a mathematician at Middlesex University London who has tracked India’s pandemic, has highlighteddiscrepancies in data between states that point to this weak surveillance. Bhramar Mukherjee, a professor at the School of Public Health at the University of Michigan, warns of inequities skewing the findings even in the data that has been available, much of it coming from cities and private hospitals accessed by the middle classes. She estimates that the Covid-19 toll alone could be as high as 2.5 million, five times the official number. 

Overall excess mortality, the wider metric of all deaths above the trend, comes in somewhere between 3.4 million and 4.9 million, according to a report by the Washington-based Center for Global Development that leans on a major household survey, civil registration, and applying international fatality rates to Indian infection data.

New Delhi, wary of opportunities to call into question its handling of the pandemic, has said reports on higher death counts are “conjectures and speculation.” The trouble is, gaps in knowledge this wide not only limit the official understanding of who suffered and how — slum dwellers or villagers, as opposed to city folk — but also have direct consequences for pandemic management. Had numbers been more realistic and uncertainties acknowledged after India’s first wave of infections, there might have been fewer government victory laps and less complacency. Ordinary people may just have been more cautious, too.

India’s government isn’t alone. Russia initially had significant gaps between excess deaths and official figures, though that has since narrowed. Mexico acknowledged in March that Covid-related fatalities were far higher than the official count; excess deaths since early 2020 are now running at close to a half-million, nearly 50% more than the country would normally expect and twice confirmed Covid deaths.

Fixing this on a global scale isn’t easy. But there are steps we can take.

One, put the focus on excess deaths. That bypasses the complexities of definition in the throes of a pandemic and has the advantage of taking a holistic approach to considering the human cost to include, say, malnutrition during a collapsing economy. As Dale Fisher, an infectious diseases physician at Singapore’s National University Hospital, put it to me, even if there was a consistent methodology to count fatalities from Covid-19 alone, the number might not be as valuable.

Then, focus on getting emerging economies — and even the less affluent corners of the developed world — to the point where more deaths are counted. In the long term, that means supporting administrative capacity and encouraging community and faith leaders to promote the importance of timely registrations, especially in an epidemic. Short-term, it’s building workarounds that should include targeted surveys but might even involve, as Mukherjee suggests, a range of data points, from (in India) inactive biometric identification cards to suspended bank accounts, plus extrapolation from better-covered but demographically similar regions.

Every life deserves to be counted, and that’s a powerful moral argument. But there’s an economic and social one, too. Without data, we can’t even begin to set right the pandemic’s wrongs — much less tackle the next one.