Today’s installment in “Florida officials conveniently published a lot of bogus data inflating coronavirus cases” now calls into question one of the primary arguments to keep K-12 schools closed. And after correcting the wild numbers that generated so many hysterical headlines and caused school districts to postpone school re-openings, the Department of Health tried to keep everything hush-hush.
An error by the Florida Department of Health produced a COVID-19 positivity rate for children of nearly one-third, a stunning figure that played into the debate over whether schools should reopen.
A week after issuing that statistic, the department took it back without explanation. The next weekly report on children and COVID-19 showed the rate had plunged to 13.4%.
The department blamed a “computer programming error” for the mistake, in response to questions from the South Florida Sun-Sentinel. Experts said the change and the failure to explain it to the public calls into question the state’s data at a time when accurate and trustworthy information is crucial to a society grappling with an unprecedented health crisis.
“It’s unacceptable to publish information that changes so dramatically that it warrants explanation, and then to not provide any explanation,” said Jason Salemi, associate professor of epidemiology at the University of South Florida College of Public Health in Tampa. “I’m trying to get an understanding of why the number changed so much, what underlies it — and can we trust this new number.”
The 31.1% positivity rate for children was provided in the health department’s July 10 pediatric report, a weekly summary posted on the state health department’s COVID-19 website. The new report, dated July 17, showed the sudden decline in the rate. The new report also showed what appeared to be a massive increase in testing. The number of children tested rose from 54,022 in the earlier report to 173,520 in the later one.
The state blamed the erroneous children’s statistic on a failure to include thousands of negative test results, which artificially inflated the positivity rate.
“Initially, there was an error in the pediatric reports in which a certain number of negative test results were not included,” the department said in an email. “That error has since been corrected and the current pediatric report available on FloridaHealthCOVID19.gov reflects the most up-to-date data available regarding pediatric COVID-19 cases.”
“It was a computer programming error specifically linked to the production of the pediatric data report. As a result, a subset of negative pediatric test results were unintentionally excluded from the pediatric report.”
This is the second time in a matter of weeks that negative test results were suppressed, making it look like there was a higher incidence of the coronavirus as testing increased by multiples.
I don’t trust anything local officials – or the CDC, for that matter – put out anymore. They received so much criticism early on for their decisions to classify anyone who had common cold or flu symptoms as having the coronavirus when testing was insufficient, then for counting anyone who died after testing positive as a coronavirus victim to the point of absurdity (suicides, motorcycle accidents, people with pre-existing terminal conditions).
Then all transparency on data was gone. The data is now just a black box. We’ve been reduced to journalists asking very narrow questions during the occasional press conference about individual data points, and the media largely has zero economic incentive to do anything but hype the numbers for the clicks.
I would point out (as someone who knows a bit about data science) that if this “programming error” only affected one report, it is highly unlikely that it was a programming error. It’s also remarkable that there are no programming errors that understate cases. These errors miraculously seem to default in the direction of doom.