Editor’s Note: Kent Sepkowitz is a CNN medical analyst and a physician and infection control expert at Memorial Sloan Kettering Cancer Center in New York. The views expressed in this commentary are his own. View more opinion at CNN.
The Florida version of the Covid-19 pandemic continues to puzzle. Many had anticipated that the death rate there would be extremely high, given the state’s late and initially disorganized outbreak response and its elderly population. More than 4 million Floridians are 65 or older, a vulnerable age group.
Then there are the beaches, which drew huge, heedless crowds during spring break in March, and again more recently, when restrictions were lifted for locals.
But the numbers and rates of infection and death, including on the state’s Department of Health Covid-19 website, continue to be all over the place. And, as with so many Covid-19-related issues right now, it appears that politics has given us flexible “facts” that allow for a range of conclusions and arguments.
Let me explain: The basic task of assessing the impact of Covid-19 should be simple, involving just two metrics: the number of deaths and the number of cases. But in Florida, these numbers – though presented with a high-gloss friendliness – are difficult to follow if you dig an inch below the surface.
As reported in The Washington Post, the governor’s office and state medical examiner have been at odds over counting deaths. First came the governor’s well-publicized attempt not to name affected nursing homes; after weeks of refusal he relented in late April. Next has been an argument over the state’s definition of a Covid-19 related death, which public health experts say guarantees a substantial under-count.
In Florida, citing Covid-19 as the official cause of death requires a positive diagnostic test for the virus. The US Centers for Disease Control and Prevention use a different standard, counting probable cases of people who had the clinical syndrome, whether or not a lab test confirms it. This is also how the CDC reports seasonal death tolls from influenza and other respiratory illnesses during less politically charged times.
This distinction matters a great deal – particularly in a nursing home population where testing was not easily available early on.
A third group of cases is also being included by a number of countries: the number of excessive deaths in a given period compared with the same period in previous years. In general, mortality rates vary only slightly year-to-year, making the impact of a pandemic easy to discern. Counting cases this way would include Covid-19 victims who never sought medical care as well as people who suffered heart attacks and strokes, avoided hospitals due to fear of infection, and died without treatment at home.
A recent CDC report has shown the profound impact of the latter two groups in New York City. Over two months, 13,831 cases in New York were lab-confirmed, an additional 5,048 were probable (with no lab confirmation) and another 5,293, of the total of 24,172, or 22%, died above the number expected on years of historic numbers.
In other words, New York’s overall number increased by almost 40% when you include the kinds of cases that Florida does not report among its 2,096 deaths.
Understanding the direction of new diagnoses is similarly confusing. The Florida Department of Health dashboard shows a clear downward slope for new diagnoses, emergency-room visits by people with complaints suggesting Covid-19 infection and the overall rate of positive test results.
Here, Florida shares a major limitation with other states. It does not identify which populations it tests, making results basically uninterpretable. For example, only 0.7% of almost 6,000 Major League Baseball employees were found to have the Covid-19 antibody, but in three other settings that predictably sit at the top of the New York Times “hot-spots” and largest-outbreak lists – meat-packing plants, prisons and nursing homes – 10% to 50% of persons tested positive.
But data for these three groups in Florida is incomplete. For example, both JBS and Tyson have meat-packing plants in Florida and neither has reported outbreaks or testing results.
Florida has a high incarceration rate, with about 100,000 prisoners at any given time. As of May 20, the Florida Department of Corrections had diagnosed active infections in 12% of 9,690 persons tested. Excluding more than 1,181 tested persons without results yet available (a large problem in a pandemic), the positive rate is actually 14%. The number of cases among the thousands still not tested will in large part define the severity of Florida’s Covid-19 epidemic.
Importantly, the corrections department also found 241 infections in staff members, although the number tested was not listed. Florida employs 24,000 people in its correctional facilities. Infected members of this group could rapidly contribute to the spread of the virus in surrounding towns.
The amount of Covid-19 in Florida’s 691 nursing homes, where 71,000 residents live, also is not yet well-characterized.
Thus far, more than 700 deaths (1% of all residents) have occurred in this population, accounting for more than half of all Covid-19 deaths in Florida, according to a recent report in the Tampa Bay Times. However, the extent of testing and the number of cases overall are unknown. Similarly, there is very little information on the employees of these facilities, who, like prison workers, live in surrounding communities and may bring infection home with them.
Get our free weekly newsletter
Furthermore, the untimely and unwelcome removal of the lead data scientist from the Florida Department of Health team, Rebekah Jones, has raised concerns that the numbers may be inaccurate. Florida Gov. Ron DeSantis called the situation “a non-issue.”
For example, on May 12, about a week after her removal, 240 cases in nursing homes were deleted from a department report, lowering the case count from 1,907 to 1,667. This represents more than an 11% decrease. The daily increase of cases since then has been more modest as well, possibly another disturbing signal.
This apparent reworking of data may have a straightforward explanation. Data does wiggle and waggle as more is learned and new definitions are applied. But given the political tension nationally as well as the dismissal of a key member of the data team, for now, Covid-19 reports from Florida will need to be handled with care.
