'Playing games with numbers': Florida COVID-19 test backlog worse than state says

[mia]

Florida is significantly under-reporting the state’s COVID-19 testing backlog, a blind spot in the data that could obscure the pandemic’s size and hamper efforts to decide when it’s safe to end restrictions such as social distancing — even as Gov. Ron DeSantis touts the state’s transparency when it comes to coronavirus.

On its public website, the Florida Department of Health says about 1,400 people statewide are waiting for their test results. But that’s an undercount, the department acknowledged in response to questions from the Miami Herald. And it’s likely a massive one.

That’s because the state only reports the number of Floridians waiting to hear back from state labs, not private ones — and those private labs are completing more than 90% of Florida’s tests. The state website doesn’t say that its figures exclude the vast majority of pending tests for the novel coronavirus....

“The public needs up-to-date information on what’s going on,” he said. “By playing games with the numbers, the state isn’t giving an accurate picture and it knows it.”

https://www.miamiherald.com/news/coronavirus/article241882491.html

[SummerSnow]

1. Anything for dear leader

[rzemanfl]

2. Right, and Florida only has 4.6% unemployment.

Because it is fucking near impossible to apply for compensation.

[mia]

3. Pulitzer Prize-winning author is leading this investigation

NICHOLAS NEHAMAS
Nicholas Nehamas is an investigative reporter at the Miami Herald, where he was part of the Pulitzer Prize-winning team that broke the Panama Papers in 2016. He and his Herald colleagues were also named Pulitzer finalists in 2019 for the series “Dirty Gold, Clean Cash.” He joined the Herald in 2014.

DANIEL CHANG
Daniel Chang covers health care for the Miami Herald, where he works to untangle the often irrational world of health insurance, hospitals and health policy for readers.

[gibraltar72]

4. No number out of Florida should be given credence.

[Igel]

5. He's saying what the Covid Tracking Project has said the last month.

1. The reported positives are just the positives that have been reported. Most tests take 24-48 hours to run, so that's always meant the numbers were skewed. When testing was ramping up and people said, "Look at today's numbers--there's no increase!" they implicitly ignored what was already obvious. The increase showed up Sunday and Monday, as expected.

And not all labs report. It's why the CDC stopped doing some reporting (and was accused of hiding the truth by not reporting false numbers as accurate. Can't win with some people.)

2. Not all states or labs report negatives, and this started in early March. Even states where almost all the positives are reported, some states stopped reporting negatives weeks ago. And very often negatives are a majority of the tests run.

3. You can't know how many tests have been run and any assumption that we do know is so obviously wrong as to count, in my opinion, as intentional lying. It's hard to estimate how bad our knowledge is. But if there reported (as last I checked worldometer's site) 7,777 tests/1 million population, it's clear that the number is larger, perhaps much larger, than 7,777/1 million.

And, my own #4, if the false negative rate is as high as 30%, all the test result information is already low-quality data, even before what I said above. You get 10,000 positives? Congratulation, you sent home more than 3,000 sick people telling them they weren't infected. Enjoy your false certainty.

These things are only a surprise to experts who really haven't been paying attention--in fact, it's almost like they've avoided paying attention in some cases, or been too distracted with venting to think. Which calls into question their self-appointed expert status.

Saying, "They tried to trick us" is great for outrage, readership, self-appointed messiahship. But it's just more manipulation. But it's user-friendly, because it justifies having overlooked information that was quite public, besides being obvious, and forming wrong conclusions. We like claiming we've been victimized and the other guy's a bully.

It's why those who actually do know have been publishing to tell people that what we need to monitor isn't the horribly flawed yet ubiquitous reported case number--I've repeated what people far wiser than me have said, but it's obvious. What we need to look at are hospitalizations. And even those aren't going to be spot on, because some people will avoid the hospitals, for whatever reason--money, access, transportation, fear, immigration status, religion, whatever, and that'll produce an undercount. On the other hand, if you look at diagnosis upon admission, those COVID tests take 48 hours for results, so a lot of people admitted for COVID don't have it when they're admitted.

Even deaths are closer to the truth than reported cases (and many of those not tested are just "not tested today"--remember, it's true that we don't have elections (today), but that doesn't mean we do not have elections; a body that's not tested (today) is likely to be tested (tomorrow), and the death count doesn't say how many died today, it says how many were *reported* today). Still, some COVID deaths are overlooked, just like some people who die with COVID would have died without it, and those are included. Do they balance? How could you even come up with accurate counts for either? All that's left from that kind of argument is the solid, absolute facts produced by fact-free suspicion.

This is a great opportunity for those who say how great liberals/progressives are in dealing with uncertainty and nuance to do precisely that. In every science, esp. those involving people, there's uncertainty in every measurement. Some systemic, some random. Remember all that crap from high school science where people said, "When am I ever going to use this?" as justification for not learning something? Here we are.

[mia]

7. "Even deaths are closer to the truth than reported cases...."

Maybe so, but deaths that haven't been linked to COVID testing aren't included in the statistics.

[Yo_Mama_Been_Loggin]

6. Kick