Reluctantly, under pressure, Florida disclosed COVID-19 data. What the numbers tell us
READ MORE
A Numbers Game
The Miami Herald analyzed both public and non-public data in order to illustrate the spread of the novel coronavirus in Florida, examine reopening benchmarks and look at what comes next.
Expand All
Floridians flattened the COVID curve. Then, amid upbeat talk, the numbers began to rise
The Florida COVID-19 data said one thing while Gov. DeSantis sometimes said another
Reluctantly, under pressure, Florida disclosed COVID-19 data. What the numbers tell us
‘Leading indicators of a resurgence,’ COVID-19 numbers trending up since reopening
As Florida’s government built a plan to contain the effects of a global pandemic, it elevated one tool above all others: data.
Since announcing its first positive case of coronavirus in Florida on March 1 — after weeks of tracking suspected cases but not telling local hospitals, much less the public — the state has generated reams of information, much of it displayed on a public dashboard that is updated daily.
But unlike the way the state has handled other infectious diseases, the COVID-19 records have been incomplete, changed without explanation, dropped from the Department of Health website without warning, or don’t match the public narrative advanced by the governor.
Nonetheless, the state has received praise.
The COVID Tracking Project, a data collection website run by volunteers, awarded Florida’s data quality an “A.” Dr. Deborah Birx, a member of President Donald Trump’s coronavirus task force, has given it plaudits.
Among those who have raised questions about the reliability of the data is the employee — dismissed a month ago — who helped build the dashboard that displays it.
Trust in data is crucial to getting the public to comply with government guidelines for how to behave in a crisis, said Leonard Marcus, co-director of the National Preparedness Leadership Initiative at the Harvard T.H. Chan School of Public Health.
“It’s difficult to get the compliance of the population if they don’t believe what they’re being told, or if they are getting different versions,’’ he said, adding that unreliable data also hampers effective decision-making and problem-solving.
“If there’s discrepancies in the numbers, if there are controversies about who is a COVID fatality and who’s not, that only foments confusion,’’ he said.
“It makes it more difficult to get the compliance necessary from the population. And from a leadership perspective, that’s one of the weaknesses of what we’re seeing going on right now across the country.”
Florida COVID-19 data is presented primarily by way of the following three sources, all published by the Florida Department of Health:
▪ A “Data and Surveillance Dashboard.” Among its features are updated tallies of total cases, total deaths, breakdowns by county and ZIP Code, exposure source, number of people tested, and percent of tests that come up positive.
▪ A separate portal offering “case line data.” It has detailed, downloadable information, minus names, on all 69,000-plus COVID-19 cases in Florida.
▪ A summary document, covering more than 1,000 pages, of statewide COVID-19 trends that also aggregates reports on nursing home and prison deaths.
Before it was finally included in the reports, the nursing home and prison data had to be pried loose by the news media. For weeks, the Miami Herald and others demanded that the state release data on COVID-19 deaths at nursing homes and assisted living facilities. For weeks, Florida refused, even as other states were making the data public so families could make informed decisions.
The Herald had its law firm, Holland & Knight, draft a public records lawsuit. After receiving formal notice of the pending case, the governor’s office attempted to quash the lawsuit by applying pressure on the law firm, using as leverage the fact that Florida too was a client.
The state capitulated after another law firm was hired and several other news organizations joined as plaintiffs.
News organizations again had to enlist lawyers to force the state to release testing data on individual nursing homes and ALFs, including visits by National Guard “strike teams.”
Although the information was eventually turned over, showing hundreds of ALFs seemingly reporting no testing whatsoever of staff or residents, the Department of Health has refused to answer detailed questions about the data.
It was one of many clashes between journalists and the governor’s office, which once barred a Herald reporter from a news conference after she requested that organizers provide more room for social distancing.
The administration of Gov. Ron DeSantis has boasted of its transparency with data. But in addition to withholding important information on deaths and infections at nursing homes and deaths at specific state prisons for weeks on end, the state continues to withhold information about Florida’s testing backlog and has not publicly disclosed when early cases began appearing in the state.
David Bruns, spokesman for AARP of Florida, said data gaps about long-term care facilities, and the state’s failure to acknowledge them, have real impact on families. But, he said, the state had come “a very considerable distance’’ from where it started.
“A family member trying to make decisions has most of the information that they need, but not as it relates to testing,’’ he said.
The University of South Florida has filled some of the gaps. Using the Department of Health’s data, USF researchers have created a data hub to archive much of the state’s COVID-19 data. They also created a map of every nursing home in the state that has had a positive COVID-19 case. It details how many residents and staff have tested positive at each home, and how many people associated with the home have died. But the website also notes the omissions in the data, such as the list of dates where no data has been recorded.
Here is a summary of some of the state’s data and questions that have been raised:
Positive cases
Every day, DOH updates its COVID-19 Data and Surveillance Dashboard and separately sends out a news release listing the “new cases” it has added since the last update.
The numbers don’t match because the timing isn’t coordinated. The email highlights the “new cases,” while the dashboard reflect positive cases by date, and also gives a total. The dashboard data doesn’t report active cases. Rather, it is a cumulative look at how many people have gotten COVID-19 since the state began testing and tracking the information on March 1.
Therefore, the number of people currently sick with COVID and the number who have recovered is not available.
Unlike some other states, Florida does not count anyone who has not been tested but that a doctor determined was a “probable” case of COVID-19. This has the potential to produce an undercount in total cases, hospitalizations and COVID-related deaths.
Testing
Testing data provided by the state is opaque. Florida does not provide information on backlogs from private labs, which analyze 90 percent of the tests. (The Herald reported on a massive state backlog in April.) Health experts say this distorts the daily count and undermines efforts to promptly detect community spread of the virus.
Initially Florida began calculating the “positivity rate” by dividing positive tests on a given day by total test results on the same day. But the state then shifted and began dividing first-time positive individuals by total tests. The first-time group is smaller, since people who test positive are often later retested multiple times to see if they have shed the disease. That change would lead to a lowering of the positivity rate. The change was executed on April 25 — four days before the governor announced that 64 of 67 counties would relax social distancing requirements and reopen. DeSantis pointed to the decline in the positivity rate as a justification.
Sometimes the numbers quoted by the governor don’t match what the reports say.
On May 5, DeSantis wrote in a tweet that “May 4th represented the best testing for Florida since the start of the pandemic. Florida received 23,884 test results, which yielded only 589 new Florida cases and the lowest statewide positivity rate — 2.61% — on record for new cases.”
However, data from the DOH’s case line portal showed there were actually more positive results recorded on May 4 — 609, to be exact.
There is another issue with the testing numbers. On the dashboard this past Thursday, the state listed a total of 1.307 million tests conducted, with a positivity rate of 5.3%. There is no disclosure of whether these numbers account for people who have been tested multiple times and how that might distort the positivity rate.
Death data
Deaths reported on the DOH dashboard include only Florida residents and exclude people who died in Florida but who were from other states, even if they were infected in Florida. For a while, the deaths attributed to each county included only people residing in that county, even if others were infected and died there. But that changed at some point.
Case line data show that as of Thursday, June 11, a total of 2,938 people had died in Florida as a result of COVID-19. Of those, 90 were non-Florida residents and one person’s jurisdiction was listed as “not diagnosed/isolated in FL” — meaning a Floridian diagnosed outside the state. The dashboard reflects only the number of Florida residents who have died, which was 2,848 as of Thursday.
The DOH deaths also don’t match those provided by medical examiners, who count all deaths in Florida, regardless of whether the deceased is a resident of the state. Deaths counted by medical examiners are listed by the county where the person died. Conversely, the state counts Floridians who died out of state and the medical examiners don’t.
Medical examiners receive information on deaths before DOH does. For example, on Friday, April 10, Miami-Dade Mayor Carlos Gimenez said that the county medical examiner had tallied over 100 deaths total from COVID-19 in the county. But dashboard figures did not reach that level until the following Monday, three days later.
Why would county-level numbers not be in sync?
“An individual who has tested positive for COVID-19 and later died is counted as a death in the county they reside in,” said DOH spokeswoman Olga Connor in an email to el Nuevo Herald. So it’s possible that some of the people who had died in Miami-Dade weren’t county residents and were accounted for on the dashboard in a different county.
Finally, researchers have reported discovering several counties that were “missing” tests — while, at the same time, other counties had a greater number of positive and negative results, when lumped together, than the overall number of tests reported.
Hospitalization numbers
The number of people admitted to the hospital by day is an important measure that public health experts use to determine how severely the virus has hit a community. It also is one of the benchmarks the Centers for Disease Control and Prevention says should be used to determine if a community is ready to reopen.
That information is not provided on the state’s dashboard, which only provides a cumulative number for hospitalizations in the state.
The total number of hospitalizations is also likely an undercount because many people who were hospitalized for COVID-19-like symptoms weren’t tested early in the pandemic.
How many people are currently hospitalized? DeSantis has mentioned the number in several of his daily briefings, but that information isn’t offered to the public and DOH says it doesn’t “have a figure” other than the cumulative total.
Rebellion over data
In May, Department of Health data manager Rebekah Jones, who played a critical role in maintaining the data dashboard, was fired after going public with her concerns about the agency’s commitment to “accessibility and transparency.” She told the Tampa Bay Times that after media requests about the publicly available COVID-19 data, she was asked to remove fields indicating when patients had begun experiencing symptoms and to exclude rural county data.
Jones said she was discharged for declining to fudge data.
The governor said Jones was fired for “insubordination.”
Following Jones’ firing, and reports that the Department of Health failed to disclose early on that the virus had arrived in Florida, state Sen. Jose Javier Rodriguez, a Miami Democrat, asked Senate President Bill Galvano to convene a joint committee of the state House and Senate “to seek all related information and communications regarding Florida’s COVID-19 data collection.”
“The panel should have the ability to review any backlogs in reporting, evaluate analyses and reporting, interview witnesses under oath, and engage experts to fully understand the issues and to make recommendations,’’ Rodriguez wrote.
Galvano rejected the idea.
“I feel strongly that the time is not right to begin an intrusive investigation that does not have a clear starting point or ending point for action at an indeterminate time,’’ he wrote in a letter to Rodriguez. “This global pandemic has been evolving, recommendations from the federal government and public health experts have been changing, and there are many dimensions to both the pandemic and the state’s response.”
Agriculture Commissioner Nikki Fried, the lone Democrat in statewide office, has also raised questions about the reliability of the data emerging from the administration.
“As people are relying on this information to make personal decisions, the least we can do is get them accurate information,’’ she said after confronting DeSantis about the issue at a Cabinet meeting. “And we’re just not seeing that, and it’s very dangerous.”
The governor did not respond.
For her part, Jones has now created her own website that expands on the one she created for the state. It uses the publicly available DOH data but also shows how each county does at meeting the criteria for reopening as set forth by the CDC and the White House task force.
Her dashboard also presents data the state collects but doesn’t promote on its dashboard, including the percent of available hospital beds in each county and the number of people tested by county. On the “impacts” page, it details unemployed by county, and provides details, minus names, about the people whose deaths have been linked to COVID-19.
Miami Herald staff writers Sarah Blaskey, Ben Conarck, Daniel Chang and Nicholas Nehamas contributed to this report.
Mary Ellen Klas can be reached at meklas@miamiherald.com and @MaryEllenKlas
This story was originally published June 12, 2020 at 8:00 AM.
