How the World Missed More Than Half of All Covid-19 Deaths

By Umair Irfan, Vox

08 May 21

A new IHME analysis finds 6.9 million deaths worldwide, and suggests countries have been undercounting since the beginning of the pandemic.

he world may have undercounted Covid-19 deaths by a staggering margin, according to an analysis released Thursday by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington School of Medicine. The actual count may actually be 6.9 million deaths, more than double official tolls.

The United States alone is estimated to have had 905,000 Covid-19 fatalities, vastly more than the 579,000 deaths officially reported, and more than any other country. The calculation is based on modeling of excess mortality that has occurred during the pandemic.

The drastic difference highlights how difficult it is to keep track of even basic metrics like deaths when a deadly disease is raging. The higher toll also means the ripples of the pandemic have spread wider than realized, particularly for health workers on the front lines who have repeatedly faced the onslaught with limited medical resources and personal protection. And the undercounts have important consequences for how countries allocate resources, anticipate future hot spots, and address health inequities.

Researchers who weren�t involved with the analysis say it confirms what many already presumed: that official death counts were far, far off.

�Big picture, it�s not really surprising,� said Jennifer Nuzzo, an epidemiologist and a senior scholar at the Johns Hopkins Center for Health Security. �We�ve long suspected that the tolls of Covid are undercounted for a number of reasons, but probably a big part is having capacity to diagnose infections and count them.�

Now, with the number of reported cases around the world reaching new highs, the findings should serve as a stark reminder that disease surveillance and tracking remain dangerously inadequate, and that the world may have already overlooked some of the greatest tragedies of the pandemic. Preventing deaths going forward demands a coordinated international effort to contain Covid-19, vaccinate as many people as possible, and monitor the spread of the virus, led by countries with the most resources helping those with the fewest.

Otherwise, an even greater toll may lie ahead.

Almost every part of the world is underreporting Covid-19 deaths

To come up with the new estimate of 6.9 million total Covid-19 deaths so far around the world, the IHME team constructed a model that incorporated observations about the pandemic. They also constructed a baseline estimate of how many deaths there would have been in a world without Covid-19. The team drew on weekly and monthly death records from 56 countries and 198 sub-national locations � city, state, and provincial records � from places like the US and Brazil.

Researchers also drew on previously published death estimates. They then subtracted the anticipated deaths from the actual number of deaths to find the excess mortality stemming specifically from the disease.

Excess mortality is mostly due to deaths directly from Covid-19, but it also includes deaths indirectly caused by the pandemic like people unable or unwilling to receive medical care, a decline in vaccination rates for other diseases, an increase in drug use, and a rise in depression. So researchers tried to correct for these factors to get their Covid-19 death estimate.

It�s a well-worn approach in public health circles and has been used to calculate other health indicators like the global burden of disease.

The model showed that, around the world, more than half of Covid-19-related deaths are not labeled in the official tallies. And the actual number could still be higher.

According to Christopher Murray, the director of IHME, while just about every part of the world missed cases of Covid-19, some countries missed more than others.

�In many parts of the world � sub-Saharan Africa, India, Latin America, differences by state in Brazil and Mexico � you can account for much of the under-reporting because of lower testing rates,� Murray said during a press conference. �But there is this phenomenon � Egypt stands out, as do a number of different countries in Eastern Europe and Central Asia � where these excess mortality rate numbers suggest dramatically larger epidemics than have been reported that cannot be accounted for through testing.�

Egypt has officially reported just over 13,000 Covid-19 deaths, but IHME found its estimated death toll was more than 170,000. It�s not clear why the discrepancy is so large, but it shows Covid-19 epidemics in different countries can be far worse than the death reports reveal.

�We are absolutely, absolutely undercounting deaths,� said Ruth Etzioni, a professor and biostatistician at the Fred Hutchinson Cancer Research Center who was not involved in the study.

IHME�s Covid-19 models missed the mark before, but researchers say they�ve improved

Scientists have also been critical of IHME�s past modeling work during the Covid-19 pandemic.

IHME�s forecasts last spring were criticized for projecting many fewer deaths than actually occurred. In March 2020, the organization projected fewer than 161,000 deaths total in the US. Then in April 2020, the group revised their death toll projections through August to be 60,415, with an uncertainty range between between 31,221 and 126,703 deaths. The projections were out of step with other epidemiological models, which were anticipating far more casualties from Covid-19.

The Trump White House, however, was eager to use the rosy IHME projections as the basis for planning for the pandemic and lifting public health restrictions, as well as a political tool to downplay the severity of Covid-19. �I was furious with [IHME], and I�m still kind of getting over it,� Etzioni said. �In the beginning, it was unacceptably un-rigorous.�

By the end of August 2020, more than 180,000 Americans had died of the disease.

�So far as I can tell, IHME has substantially improved their modeling from the early days of the pandemic,� said Alexey J. Merz, a professor of biochemistry at the University of Washington, in an email. �My major criticisms pertain to those early efforts, and IHME�s ongoing failure to address what went wrong, or to assess the (in my opinion, considerable) damage arising from those flawed estimates.�

Asked about IHME�s track record, Murray explained how his team�s Covid-19 forecasting improved and even outperformed other models. �For example, if you go back to August last year, we were forecasting the winter surge, and nobody else thought there was going to be a winter surge in the United States,� he said. �We spend a lot of time on our model trying to look at what are the long-term drivers so we have been able to pick up these long-term trends quite a bit sooner than others.�

Why the US official count is so low compared to the new analysis

It makes sense that countries with less robust health care systems and fewer resources would struggle to keep track of how many people are dying of Covid-19. But the US, a wealthy country that has a national Covid-19 death reporting system, also missed almost 40 percent of Covid-19 deaths, according to the IHME model.

That�s because while death can seem like a pretty obvious health indicator, the causes of death can be mercurial.

The problems start with the death certificate. Ivor Douglas, chief of the Pulmonary Sciences and Critical Medicine division at the Denver Health Medical Center, explained that death certificates emphasize the primary cause of death, which is the most immediate condition leading to the fatality. Death certificates also have space for secondary and indirect causes.

As the Covid-19 pandemic has revealed, the disease can manifest in myriad ways and leave lasting damage, even in people who had a mild illness.

So a Covid-19 death certificate could list something like a blood clot in the lungs as the primary cause of death, with Covid-19 as a secondary or indirect cause. Whether that specific death is then coded as a Covid-19 fatality could differ depending on the state. That local-level reporting has sometimes become politicized and led to discrepancies in death tolls.

And when Covid-19 first arrived in the US, many health workers didn�t realize what they were dealing with and thus didn�t include it in their paperwork. �I think the preponderance of missed cases were early on in the pandemic,� Douglas said. �Often, certainly early in the pandemic, there was the primary diagnosis without Covid-19 attribution.�

The missing Covid-19 deaths are also another manifestation of the inequities in US society. �If you�re poor, don�t have access [to health care], and die at home, you�re much less likely to have an attribution of Covid pneumonia as a cause of your death than �oh, you�re a sad old person with diabetes� and that was the cause of death,� Douglas said.

That means that the groups that are being most severely harmed by Covid-19 may also be underrepresented in the official numbers. That makes it harder to properly allocate resources like tests, vaccines, and treatment to the most vulnerable people, forcing them to bear an even greater health burden.

�There�s real policy implications, it has political implications, and social justice implications, in my mind,� Douglas said. On the other hand, accurate monitoring could help mitigate the harms of the Covid-19 pandemic, helping health officials figure out not just where to deploy vaccines and treatments, but other factors driving transmission, like crowded living conditions. Intervening before infections begin to spread is what will yield the greatest dividends in containing the disease. �You cannot simply vaccinate your way out of this problem,� Douglas said.

Finding the true toll of Covid-19 is more urgent than ever

Regardless of how high the actual number of deaths is, the devastation of Covid-19 is clear. �Even the reported numbers are so utterly staggering that I�m not even sure doubling it should make us even more horrified,� Nuzzo said.

Still, the fact that Covid-19 deaths appear so vastly underreported should be a warning that the virus can still take millions more lives, and why containing Covid-19 is imperative for every country in the world. �We should feel more personally threatened by these numbers. And we should recognize it as a societal threat,� Etzioni said.

The devastating Covid-19 outbreak in India is all the more urgent now that multiple variants of Covid-19 that are more transmissible and better able to evade immunity are spreading around the world. As the virus continues to spread, the likelihood of even more dangerous variants arising will grow.

What�s more, the countries that have been reporting lower deaths so far deserve more attention. �Many of us contend that sub-Saharan Africa has been extensively devastated by the pandemic but because of lack of testing medical reporting, it appears as if there has been a relatively minor event there,� Douglas said.

As for countries that have so far been genuinely spared from Covid-19, they must remain vigilant and take active measures to keep the disease at bay. �It may be that they haven�t yet been hit or it could be that we don�t fully understand how they�ve been hit, but I want to put to bed this idea that any country has simply escaped the worst of it,� Nuzzo said. �The countries that have done the best are ones that have been very, very aggressive in responding to it.�