By Kathleen McGrory

In places with large coronavirus outbreaks, researchers have recently found thousands of unexpected deaths beyond those captured in the official tally of COVID-19 fatalities.

But what about Florida?

To answer that question, we teamed up with a health researcher from the University of South Florida to analyze all deaths in the state, not just those certified as COVID-19.

They found that during the 5-week period ending April 25, there had been hundreds of unexpected deaths from illness and disease across Florida, more than can be explained by the coronavirus death count.

The analysis suggests the epidemic’s true toll may be between 17 percent and 58 percent higher than published death figures. Health experts say that likely includes some people who died of coronavirus but were never diagnosed as well as others who might have lived had the pandemic not kept them from getting care.

“There are a lot of challenges to identifying COVID deaths,” said Troy Quast, the researcher who conducted the analysis.

At the same time, the analysis also showed a drop in deaths from external causes such as homicides and motor-vehicle accidents — presumably because people stayed home to avoid the virus. It is unclear whether the decline was enough to offset the increase in deaths from natural causes, but that was within the range of possibilities the analysis revealed.

The findings illustrate the complexity of measuring the virus’ impact, given the many ways it has reshaped daily life.

“There are indirect and direct consequences of a pandemic like this,” said Mark Hayward, an expert in mortality statistics and professor at the University of Texas at Austin.

“Of course we know the direct consequences,” he added. “Some of the other ones are poorly measured.”

The analysis considered five weeks of data published by the Centers for Disease Control and Prevention for March 22 through April 25.

It found between 1,358 and 1,831 more deaths than expected from natural causes in Florida.

The CDC’s provisional count of coronavirus deaths in Florida during that period was 1,162.

Expanding to all deaths — including homicides, suicides and accidents — the analysis found between 805 and 1,486 more deaths than expected.

The data does not break down deaths from non-natural causes. But Quast and other experts said they believed motor-vehicle accidents and homicides fell below normal levels, pulling the total number of deaths down.With limited testing obscuring the true number of coronavirus cases nationwide, data on deaths has been key to understanding the scope of the epidemic.

In Florida, the health department’s count of COVID deaths now exceeds 2,100.

Increasingly, experts have been looking at a broader measure called excess mortality. It compares the total number of deaths over a certain period to how many were expected based on historical averages.

The measure proved important after Hurricane Maria devastated Puerto Rico in September 2017. The government’s initial count of deaths caused by the storm was 64. But research found there had been at least 1,139 more deaths than expected.

The government later revised its figure to 2,975.

The coronavirus epidemic is far larger in scale than Hurricane Maria and a fundamentally different kind of event, said Jeffrey Howard, an assistant professor at the University of Texas at San Antonio and one of the authors on the study.

“But both will produce excess deaths,” he said.

Recent analyses have found that’s true of the current epidemic. Both publications identified an above-average number of deaths nationwide in March and April — and concluded the surge was only partly explained by the COVID deaths that have been publicly reported.

The disparity has been most pronounced in the hardest-hit parts of the country. In New York City, for example, some 24,000 more people than usual died through May 9. The count of coronavirus deaths through that time was 19,926.

The analysis Quast conducted was based on the same data, which comes from death certificates and is typically delayed a few weeks.

The analysis compared the number of deaths each week to the average number for that same week over the last six years. The historical counts were adjusted for population growth.

It also produced a more-conservative number by calculating a margin of error based on the historical counts and using the highest number within the margin. The two numbers form the range of expected deaths.

Experts who reviewed the work said the extra deaths likely had a variety of causes. Some were almost certainly misattributed COVID deaths, Howard said.

“The testing for coronavirus is not 100 percent accurate,” he said. “You have false positives and false negatives.”

Eunkyung “Muriel” Lee, an assistant professor in the health sciences department at the University of Central Florida, said others may have happened because the epidemic kept people with chronic conditions from getting the appropriate treatment.

Hospitals in Florida may have admitted or kept fewer patients than usual in anticipation of a surge of COVID patients, experts said.

People in Florida may have also been afraid to seek care. It was reported that emergency room visits across Florida dropped 50 percent since the epidemic began. In Broward County, the number of people who died by the time rescue workers arrived doubled from last year.

Quast said the larger trend contradicts the narrative that many COVID deaths would have happened regardless because the people who died were elderly or had existing health conditions.

“This view is not, at least completely, accurate,” he said.

Some experts say there could be fewer excess deaths moving forward as physicians and medical examiners get better at identifying the virus.

But they also caution that there could be another wave of cases as the state reopens. At the same time, the number of non-natural deaths could return to normal as drivers return to the road and violent crime picks up.

“If there’s a large return to normal traffic and normal interactions, it could be a double whammy,” Quast said.

About this analysis

The analysis in this report was conducted by Troy Quast, an associate professor in the University of South Florida College of Public Health who specializes in health economics and health policy. Earlier this year, he published research in the journal Drug and Alcohol Dependence that found drug overdoses were underreported in national statistics.

To examine excess deaths in Florida amid the coronavirus epidemic, Quast pulled data from the National Center for Health Statistics, an arm of the Centers for Disease Control and Prevention. He calculated the expected number of deaths by taking the number of deaths in each of the last six years, adjusting for population growth and determining the average.

Quast also used the historical data to calculate a 95 percent confidence interval and determined the highest number within the margin. The approach was similar to the one used to estimate excess deaths after Hurricane Maria.

The estimate of excess deaths was presented as a range between the two figures. As a point of comparison, the newspaper also presented the CDC’s provisional counts for coronavirus deaths for Florida.

The analysis focused on five weeks of data from March 22 through April 25. Data through May 16 was available online, but the most recent information was largely incomplete.