Coronavirus death toll may be undercounted by thousands in Pennsylvania and New Jersey

The coronavirus may have killed up to 9% more Pennsylvanians than the reported death toll for the month of April, an Inquirer data analysis suggests.

The state’s official number of COVID-19 deaths for the month is 3,200, but the number of deaths from all causes was 3,500 above the five-year average for April. That means as many as 300 additional deaths could have been due to coronavirus.

In New Jersey, which has been more up to date in recording death certificates, the possible undercount is even larger. As of May 9, some 9,100 deaths had been attributed to COVID-19, while the number of deaths from any cause was more than 12,000 above the state’s five-year average in the same time period. If those deaths were due to coronavirus, that would increase the virus’ death tally by a dramatic 32%.

As Yale University researchers have found elsewhere in the country, much of the likely undercount in each state can be explained by an unprecedented spike in deaths attributed to pneumonia and influenza.

Generally, deaths attributed to the flu are not confirmed with a laboratory test. And deaths due to pneumonia, whether blamed on the flu or not, are usually in steep decline by spring, said Esther Chernak, an associate professor at Drexel University’s Dornsife School of Public Health. Not so this year.

She and other infectious-disease experts say there is little explanation for the grim nationwide surge in flu and pneumonia deaths during March and April beside the obvious: COVID-19. It appears that doctors, pathologists, and funeral directors, unable to test for the new virus, mislabeled many early cases of the disease, which as of Tuesday has killed more than 5,100 Pennsylvanians.

“There’s no question that we’ve missed a number of cases,” Chernak said.

Precisely how many Americans have succumbed or likely will die from the virus has become an issue not only for medicine but also for politics. Fox News commentators and other conservatives have argued that the official tallies overstate the deaths and that some fatalities were mislabeled as COVID to exaggerate the crisis.

President Donald Trump, who has argued vigorously that his policies have kept deaths far lower than forecast, skeptically took note of New York City’s revision in April that increased its death tally upward by 3,000. On Friday, Trump said he accepted the current death toll, but said that the figures could be “lower than” the official count, which is drawing close to 100,000.

The Yale analysis found no evidence for an overcount, said Daniel Weinberger, an associate professor of epidemiology.

"The supposition that COVID deaths are being overestimated doesn’t seem to hold water," he said.

Some of the additional deaths may be due to other causes, such as heart attacks or strokes that proved fatal because patients were reluctant to go to coronavirus-slammed hospitals, Chernak said.

On the other hand, some heart-related deaths might be the result of the virus, as it is known to cause dangerous clots and other circulatory problems, Chernak said. With the shortage of testing early in the pandemic, a precise accounting may remain elusive.

But the unusual increase in deaths attributed to pneumonia and flu is a strong indicator of “missed” COVID deaths, Chernak and Weinberger agreed.

For Pennsylvania and New Jersey, The Inquirer counted the number of deaths due to pneumonia, flu, and COVID for the period from March 22 through May 9. That number was then compared with what pneumonia and flu deaths likely would have been during that period, based on historical trends. In fact, deaths from pneumonia and flu this year were actually running 15% under the typical numbers, until the virus hit.

Pennsylvania saw 141 additional such deaths beyond the expected number, while New Jersey recorded an extra 750. As COVID diagnoses apparently improved, The Inquirer’s analysis showed, pneumonia and flu cases fell back to the predicted figure.

Nationally, The Inquirer has found 8,500 likely COVID-19 deaths that were attributed to pneumonia or flu through May 2. Many states are still reporting their fatality data to the CDC, which is why the U.S. analysis ends a week earlier than the calculation for Pennsylvania and New Jersey.

At Yale, Weinberger and colleagues tackled the same question. The team’s methodology was more complex than The Inquirer’s, and only goes through April 25, but their findings are similar. Through April 25 the Yale team found 5,760 likely misdiagnosed deaths in the United States, 470 in New Jersey, and 120 in Pennsylvania.

As virus deaths began to mount nationwide, researchers have repeatedly found a likely undercount in COVID fatalities, both for the country and for individual states.

In their latest death-toll analysis, released May 15, Weinberger and his Yale team found as of April 25 an additional 30,000 deaths nationwide over and above regular death rates after accounting for the 50,000 virus deaths. That suggests an undercount by that point of 60%.

However, until recently, Pennsylvania lacked death data reliable enough to figure out what could be happening in the state. The problem was due to lags in the compilation of death data, an issue the state addressed in March when it instructed funeral directors and others to submit death records electronically, rather than on paper.

Over the last week, Pennsylvania has updated its death figures, especially for recent months. But it still cannot offer county-level data that would show potential undercounts for hard-hit Philadelphia and its collar counties.

As some of the missed COVID-19 deaths may be hidden in other categories besides pneumonia, researchers also are looking at deaths by all causes.

The U.S. Centers for Disease Control and Prevention has been tracking how much overall deaths in every state have differed from historical trends.

That historical baseline number in each state is based on the five-year average number of deaths, plus or minus a statistical margin of error.

In New Jersey, the CDC has calculated the number of deaths this year to be at least 11,500 above what would have been expected through May 9 (The Inquirer’s analysis found the difference was more than 12,000).

Either way, New Jersey’s deaths at that point were well beyond the 9,100 officially attributed to COVID.

An accounting of the year to date is not yet possible in Pennsylvania, as data are incomplete prior to mid-March, when some funeral directors were still submitting death certificates on paper.

As more data come in, the state continues to update its statistics.

Sara Simon of Spotlight Pa contributed to this article.