Florida puts spin on virus data
Positive slant misleads public on the pandemic
Florida has obscured the true extent of its COVID-19 pandemic by using a misleading measure of positive cases to justify reopening schools and businesses, state data indicates.
While Florida has publicized that its “positivity rate” has regularly fallen below5%, other health organizations are publishing data that shows the rate maybe dramatically higher.
Independent experts, including Johns Hopkins University, consistently calculate Florida’ s positivity rate at 10% or higher, twice the recommended level for widespread reopening.
Florida itself calculates another version of the rate — not widely publicized — that shows the pandemic isworse than state officials have championed. Those figures show that the rate has never fallen to the 5% threshold, the South Florida Sun Sentinel discovered.
The result: As coronavirus cases again threaten to rise in Florida, with schools and businesses fully open, it is becoming increasingly difficult to answer a simple question: What percentage of people tested positive for COVID-19 on any given day — and are we justified in going about our business without worry? Or did Gov. Ron DeSantis put people at risk when he allowed businesses to return to normal?
“Doing what we’re doing now, I think positivity numbers are just about useless, because it’s completely opaque who is getting tested and why,” said Thomas Hladish, a scientist at the Univer-
sity of Florida’s Emerging Pathogens Institute who has done pandemic modeling for the Florida Department ofHealth.
“It absolutely does lull people into a false sense of security,” said Rebekah Jones, a former Florida coronavirus data scientist who was fired in early May for alleged insubordination, after claiming the state had encouraged her to fudge statistics. Jones now publishes her own coronavirus statistics.
Measuring the pandemic
In simple terms, “positivity” measures the percentage of coronavirus tests that come back positive. It traditionally has been considered one of the best ways to determine how much the corona virushas spread through the community.
Data scientists acknowledge that the rate can be determined in at least three different ways, all of which they say are justified. But in twoof the three cases, Florida’s positivity ratewould not meet the common standard for reopening society: below 5% for 14 consecutive days.
Method 1: The rate published by Johns Hopkins University measures how many people tested positive on a given day out of all people tested in the state for the first time. By that measure, the state’s positivity rate has not trended below10% since the middle of June — twice as high as the state claims.
On Tuesday, Johns Hopkins listed the rate at 11.7%, while Florida was reporting 4.99%.
Method 2: The number of positive tests is compared with the total number of tests on one day. This is the measure recommended by the World Health Organization. The organization says this rate shouldbebelow5% before reopening.
The Florida Department ofHealth publishes the rate, but only in daily reports posted on the Department of Emergency Management’ s website, separate from the Department of Health dashboard that the public commonly monitors. The rate also is published only for the entire state, not individual counties.
The last time that rate fell below 5% for 14 days was four months ago, according to data from the Florida Department of Health. Over the past two weeks, it has averaged 5.87%.
Method 3: This calculation is similar, but itexcludes anyonewhohas testedpositive in the past, even though people with previous negative tests are still included. That makes the method problematic, some experts say, because the number of people retested on a daily basis can be large, thus skewingthe ratedownward.
Nevertheless, this is the figure that Florida publicizeseach day and publishes ona county-by-county level. It’s theoneDeSantis continually touts in arguing that it’s safe to reopen schools, restaurants, theme parks, museums and stadiums. The rate has remained below 5% since the second week of September, according to Department ofHealth data.
The three rates have broadly tracked each other for the duration of the pandemic, rising and falling at similar times. But the state’s preferred rate is consistently lower than the others and far below the rate calculated by Johns Hopkins.
JasonSalemi, anassociate professor at the University of South Florida’s College of Public Health, said Florida’ s preferred positivity rate is “more controversial. You tend to not see it in a lot of other places.”
Other large states do not publish as much raw coronavirus data as Florida, but they appear to present their positivity rates in simpler terms.
New York, for example, publishes a chart showing the percentage of people testing positive, based on the more widely accepted method recommended by the World Health Organization. California shows a 14-day average of positivity and appears to use a similar method.
Salemi said he still finds Florida’s preferred rate useful to determine “what happened today among people who never tested positive before.” But some experts find the statistic faulty because it appears to compare two fundamentally different variables: people who tested positive versus tests thatwere negative.
“Thatis not a real percentage, because you’re using non-identical variables.” said Jones. “It’s like dividing ice creamconesby shark attacks.”
Representatives from the governor’s office, the Department of Health and the Florida Department of Emergency Management declined to explain the reason for their continued use of a possibly misleading statistic, saying simply that it was the “best” measure.
AlbertoMoscoso, spokesman for the Florida Department ofHealth, said in July that the state counts multiple negative tests for the same person because “a negative test only reflects the individual’s status at the moment the specimen was collected.”
“Since a person who tested negative previously remains susceptible to the virus, the subsequent test is still included in the denominator of the positivity rate calculation,” he said.
Unreliable results
Some experts are beginning to question whether positivity rates are even a usefulmeasure.
Theratecantell youabout the spread of the virus, but
only under certain ideal conditions, experts say. Levels of testing must remainconstant; the criteria for who gets tested must be relatively random; and you can’t mix different kinds of tests, as Florida does.
The state combines PCR tests, or molecular testing that is consideredmore reliable, with antigen testing, the rapid-results tests that are less accurate.
Positivity is useful to measure whether the state is doing enough testing, but not so much for measuring spread, said Jennifer Nuzzo, lead epidemiologist for Johns Hopkins’ COVID testing initiative.
“It’s often talked about as a measure of how many infections are out there, but youcould have a verydifferent number depending on how much testing you are doing,” she said.
Florida’s daily testing numbers have come down sharply fromtheirJuly peak, when the state was testing between 90,000 to 100,000 people a day. Currently, the state tests between 40,000 to 70,000 people a day.
Days with a lot of testing tend tohave lowerpositivity rates, and days with fewer tests tend to see spikes.
Hladish, the scientist at UF, advocates for randomized testing of individuals in order to figure out a true positivity rate.
“Randomly administer tests to 1,000 people a week, and you report that percentage every week,” he suggested. “It doesn’tmatter
whetherthey tested positive before or not, but you’llhave a representative idea ofhow prevalent the infection is in the population.”
Alexis Madrigal, deputy editor at The Atlantic, and a co-founder of the COVID Tracking Project, notes that wider use of antigen testing, which is becoming more available, will further disrupt the reliability of the measure.
Recently, Florida has said it is deploying millions of rapid tests as it opens schools, and antigen tests have been widely used by professional and college sports teams.
“Antigen tests are going to just fry thiswholething. And no one is really sure what to do about that yet,” Madrigal said.
“The metrics that we’ve come to rely on like test positivity are in for a substantial weirding if antigen testing rolls out in a major way without a workable plan for getting that data back to state health departments.”
All of this might leave the public confused about what numbers are most important. Which should youwatch to judge whether Florida’s coronavirus outbreak is starting to spiral out of control?
All of them, Nuzzo says. “Something to realize is that whatever the metric is, there probably won’t be a magic number,” she said. “I tend to pay attention to the trends in the numbers, more so than the numbers themselves.”