Official U.S. coronavirus death toll may be missing many thousands

A confluence of testing shortages, inconsistent death reporting practices and a high concentration of nursing home deaths means the overall fatality number is likely higher.
Image: Cordarial O. Holloway, Robert L. Albritten, Eddie Keith, Ronald Costello
A funeral at Cedar Hill Cemetery in Dawson, Ga., on April 18, 2020.Brynn Anderson / AP

Breaking News Emails

Get breaking news alerts and special reports. The news and stories that matter, delivered weekday mornings.
SUBSCRIBE
By Josh Lederman

WASHINGTON — As President Donald Trump and his allies suggest that the number of deaths from COVID-19 may be overinflated, epidemiologists are warning that the U.S. death toll is actually far higher than official statistics show, almost surely missing many thousands of virus victims.

On paper, more than 72,000 people have died from the coronavirus, according to an NBC News tally from each state and U.S. territory. But that number is likely an incomplete picture because of a perfect storm of testing shortages, inconsistent death reporting practices across the country and a high concentration of nursing home deaths.

Dr. Matt Heinz, a hospitalist who’s been treating COVID-19 patients in Arizona, said he’s seen cases, particularly early in the pandemic, in which patients still waiting for coronavirus testing died of unknown respiratory illness that in retrospect looked very much like the coronavirus.

“Is that COVID? Is that not? In some cases, we can’t even go back and test for that, because it's been weeks and weeks,” he said.

A stroke landed retired musician Charlie Terhune in a rehabilitation facility in Islip, New York, late last year, and weeks later, his family feared he would contract the coronavirus if he stayed there — a fear that later seemed to prove all too real. Although he’d made great progress recovering from his stroke symptoms, Terhune’s breathing deteriorated around the same time that he was discharged.

When the paramedics came to his home, they told his wife, Suzanne, that if he went to the hospital for testing, he’d likely never come home. A retired nurse, Suzanne knew she could care for him with the right equipment at home, so he was never tested before he died April 5.

“I spoke with the doctor every day from the time he came home until the time that he died, and there was no question in her mind" that he had coronavirus, Suzanne said. “The hospice nurses who came in, no question. The ambulance, no question. That’s what he had.”

Yet, the death certificate listed “acute respiratory failure” as Terhune’s cause of death. And so as America counts its dead, he became one of an unknown number of likely coronavirus fatalities to go uncounted.

Researchers at the Yale School of Public Health found that in hard-hit areas of the country, the death toll may be off by 50 percent or more.

“Those numbers are only as good as the death certificates that they're based on. And that depends upon the people who are filling them out,” said Dr. James Gill, Connecticut’s chief medical examiner, and vice president of the National Association of Medical Examiners.

Last month, he reviewed a death certificate for a woman who’d had a fever and shortness of breath but was never tested for the coronavirus. The cause of death was respiratory failure, but the state sent an investigator to the nursing home to take a sample post-mortem — and it was positive for the coronavirus. The certificate was changed.

“Without us doing that further investigation, that death would have been missed,” Gill said.

To get a more accurate picture, epidemiologists use a concept called “excess deaths.” They start with the total number of reported deaths, then subtract the number of deaths that are normally expected for that period, based on the past several years. The gap between the two is known as “excess deaths” — and may be a clue as to what’s really happening.

“There's not been a hurricane, there's not been a nuclear bomb or anything else. Then these excess deaths, in the setting of low testing rates, are a very good way for us to estimate the impact of this coronavirus,” said Dr. Camara Phyllis Jones, a Harvard University fellow and a past president of the American Public Health Association.

Data from the Centers for Disease Control and Prevention show there were thousands more deaths in March and April than normal for those months — even when known coronavirus deaths are factored in.

In New Jersey, there were about 5,200 excess deaths in the month ending April 11, according to a New York Times analysis, but less than 2,200 reported coronavirus deaths, leaving an unexplained gap of 3,000.

Not all of those excess deaths are necessarily from the coronavirus itself. Some could be other consequences of the new normal, such heart attacks or other diseases going untreated because people don’t want to risk going to the hospital in the middle of the pandemic. Higher rates of suicide and domestic violence could also explain some of the deaths.

In Washington, the numbers have also become highly politicized -– and a grim yardstick against which Trump may be measured as he runs for re-election.

In February, before the U.S. had any known coronavirus deaths, Trump predicted the number of cases would drop to zero. His later estimates of how many would die have changed repeatedly as the public health models have shifted up, down and back up again.

Trump has defended the comparatively high mortality rate by arguing that other countries are missing deaths that the U.S. is capturing because of what he describes as better reporting. He raised doubt when New York state announced that another 3,700 residents who died without being tested were now presumed to have died from the coronavirus.

“I see, this morning, where New York added 3,000 deaths because they died,” Trump said in mid-April. “And they're now saying, rather than, ‘It was a heart attack,’ they're saying, ‘It was a heart attack caused by this.’ So they're adding.”

Only 1 in 3 Americans trust the death toll is accurate, an Axios-Ipsos poll this week found. Not surprisingly, the split was partisan: Republicans were more likely to believe the death toll is overinflated, while Democrats are more likely to believe it’s underinflated.

One reason the U.S. may be undercounting its coronavirus dead is the high level of deaths occurring in nursing homes, where the outbreak has run rampant. In nursing homes and adult care facilities, residents are often cremated with no autopsy or death investigation.

In some parts of the U.S., only people who test positive for the coronavirus before they die get counted in the official tally. The CDC says medical examiners and coroners should use their “professional judgment” about whether to test post-mortem.

Another factor is the hodgepodge of systems used for death investigations across more than 2,000 jurisdictions in the U.S. Some areas have medical examiners, while others have coroners or justices of the peace. Some report electronically, others like Connecticut still have paper systems. It can take weeks for deaths to get counted.

It’s been four weeks since Terhune died with all the signs and symptoms of the coronavirus, but no place on the official tally of its victims. His wife said that while the official designation wouldn’t have had any effect on his outcome, it does matter for society at large.

“People who go by the number of deaths that's broadcast, I know that it's far more,” Suzanne said. “It has to be far more.”