NEW YORK — The early death toll from the coronavirus pandemic was far higher than official U.S. statistics, a new study highlights the dramatic disparity in uncounted deaths.
Approximately 840,000 deaths due to COVID-19 were reported on death certificates in 2020 and 2021. But a group of researchers, using some form of artificial intelligence, estimate that as many as 155,000 additional unrecognized deaths may have occurred outside hospitals over the same period. This means that about 16% of deaths from COVID-19 were not counted that year.
Coronavirus skeptics are starting to win in court. As a result, access to vaccines may be reduced.
The overall findings, published Wednesday in the journal Science Advances, were close to estimates from other studies of pandemic deaths at the time. But the authors of the new study set out to pinpoint which deaths are likely to be missing from the official tally.
A: Undiagnosed deaths are likely Hispanics and other people of color who died during the first months of the pandemic and who were in certain states in the South and Southwest, such as Alabama, Oklahoma, and South Carolina.
Six years after the coronavirus spread in the United States, barriers remain for many of the same people, said Stephen Wolf, a researcher at Virginia Commonwealth University who was not involved in the study.
“People on the margins continue to die at disproportionate rates because of lack of access to health care,” he said in an email.
Access to care wasn’t the only issue
Hospitalized patients were regularly tested for COVID-19, but many of those who became ill and died outside the hospital were not tested. In many cases, that’s because home tests weren’t readily available early in the pandemic, said one of the study authors, Elizabeth Wrigleyfield of the University of Minnesota.
Global health experts discuss lessons from coronavirus and warnings for the future
In some parts of the country, death investigations are handled by elected coroners who do not necessarily have the specialized training that coroners do. Some studies suggest that partisan views may influence whether sick people and their families seek coronavirus tests and whether coroners perform post-mortem coronavirus tests. In fact, some coroners said they were pressured by families not to list COVID-19 as a cause of death.
“One of the main reasons we haven’t been able to get an accurate count of deaths, especially outside of metropolitan areas, is our outdated death surveillance system,” said lead author Andrew Stokes of Boston University.
Death toll skyrockets due to coronavirus politics
More than 1.2 million people have died from COVID-19 since the pandemic began in early 2020, according to data from the Centers for Disease Control and Prevention. More than two-thirds of the reported deaths occurred in 2020 and 2021.
The numbers have long been debated due to false claims on social media that the death toll from the coronavirus has been inflated. Adding to the disgust, in August 2020, President Donald Trump retweeted a post that was later deleted by Twitter, claiming that only 6% of reported deaths were actually caused by COVID-19.
To be sure, there have been deaths from other types of pandemics. For example, uninfected people died from other conditions because they could not receive treatment in hospitals full of coronavirus patients. People addicted to drugs were socially isolated and lost access to treatment, resulting in their deaths from overdoses. Other studies that have estimated the actual number of deaths from the pandemic have taken those deaths into account.
But Stokes and his collaborators wanted to focus on the deaths of people infected with the coronavirus. They used machine learning to sift through death certificates of infected patients who died in hospitals, and used the patterns observed in those records to evaluate death certificates of people who died outside of hospitals and whose deaths were thought to be caused by things like pneumonia or diabetes.
Although scientists’ understanding of the pros and cons of research that relies on machine learning is still evolving, Wolfe said the team’s use of machine learning is “interesting.”
— Mike Stobbe
