Games with Models with a Moving Target, 5/1/20 Data


Multiple signs are starting to point to inflated COVID-19 statistics.  In no particular order:

  • The New York Times published an article seeming to imply that the COVID-19 death toll is much, much higher than estimated because total deaths are up.  This is a clever twist that will ensure that governors are never questioned about the wisdom of their reactions to the disease, because every death that is a consequence of the shutdown (like despair-driven suicides and overdoses) will actually count toward the total of the virus, thus validating their decisions.
  • The Daily Wire reported on suspicions among funeral directors that COVID-19 is being put on death certificates when it wasn’t actually the cause of death. A source in Rhode Island tells me that this is happening here, with families expressing bewilderment that the disease is being blamed for their loved-ones’ deaths.
  • WPRI seems to have confirmed my suspicions that Rhode Islands new method of counting COVID-19 hospitalizations entails counting people who “have been admitted to the hospital for unrelated reasons — but test postive for the disease” (in WPRI’s words).  That is the direction in which my own correspondence with a Dept. of Health was headed.  As the spokesman told me, “What we are now getting is reports based on lab confirmation of COVID-19, regardless of where the person is in the hospital system.”  I haven’t been able to confirm, but “where the person is in the hospital system” seems like it might include anything, including quick out-patient visits.
  • USA Today has reported that the federal government is creating incentive for hospitals to classify patients as COVID-19 illnesses by reimbursing more for Medicare patients if the coronavirus is part of the mix.

Thus, today, the number of hospitalizations continues to climb whereas two days ago it was drifting down.  As I suggested yesterday, we just can’t trust these numbers any longer.  For instance, while the top-line number suggested for deaths in Rhode Island looks like it increased by 13 people, more than half of the increase is attributable to revisions of earlier numbers, going back weeks.  In actuality, the one-day increase is listed as six people.


Some good news, however, can be found in the decreasing number of patients in intensive care (presumably despite the broader definition for when they count as COVID-19 patients).  Additionally, the rate of increase of total cases continues to go down.  In fact, the 14-day infection rate (by which I approximate how many people each person with the disease is infecting, on average) is almost to one, below which the disease is fading completely away.

  • Cases:
    • Projection for 5/1: 8,802
    • Actual for 5/1: 8,962
    • Projection for 5/2: 9,132
  • Hospitalizations:
    • Projection for 5/1: 322
    • Actual for 5/1: 352
    • Projection for 5/2: 341
  • Deaths:
    • Projection for 5/1: 272
    • Actual for 5/1: 279
    • Projection for 5/2: 284

  • Mario

    The retreat into conspiracy theories was sadly predictable. The numbers aren’t being artificially inflated, I don’t know what reason you could come up with for why the governor would even want to do that. I would think she’d want to look successful instead of like a Trump-level failure, but there’s probably a 5-dimensional chess plan behind it all that I can’t conceive of. I shudder to think what the reaction will be when we finally get a good statistical analysis of the deaths, and the numbers get pushed even higher.

    Not to mention, although I will mention, that I was always saying that the hospitalization number was too low. Even your numbers said the same thing: there were patients inexplicably unaccounted for. We don’t need to come up with a reason to pretend that the number is too high now, because we knew at the time that it seemed way too low. Now it’s just not.

    The idea that the people being listed as hospitalized as just quick out-patient visits is almost plausible, but the numbers don’t back it up. You can see the admissions and discharges for each day, even without being able to follow a single patient you can work out how long they are there. The average stay has to be close to 10 or 12 days, and that has been pretty consistent from the beginning.

    The increase of hospitalizations was a shame, but not completely unexpected. The size of the increase does concern me, although this time it was because my admissions number was too low, which under the old numbers was the only one I could estimate well. I still stand by the idea that this is the peak, and have tomorrow at 342 (I’m showing 30 admissions, 35.5 discharges and 4 deaths, just to show you where my head is). I’m going to go with 9 deaths for tomorrow; my spreadsheet says 6 but these periodic testing booms causes sudden unjustified drops in that number that I’m having trouble correcting for. The testing is ridiculously inconsistent lately, and the case number highly depends on the that, so it’s tough to predict. I will say +290 if it’s another average day, +360 if we get a good day.

    • Justin Katz

      Well, sadly, your turn toward condescension was not something I would have predicted. There’s no conspiracy needed; much of this has to do with the typical way in which government creates incentives, such that the ability of the public to assess the crisis becomes secondary to other objectives like bringing in maximal federal aid.

      Simply put, by the standards I have been applying and that I think should be applied, it is all but acknowledged fact that the numbers are being “artificially inflated.” Specifically, I don’t really care if a healthy teenager with a broken arm is found to have otherwise undetectable COVID-19 because he is screened for it upon admission. The relevant questions are whether COVID-19 is overwhelming our hospitals and how harmful it is. People who are in the hospital for some other reason are not additional hospitalizations, and people who are not so sick that they had to be hospitalized for the illness are not evidence that it is especially harmful.

      Of course, they do give some sense of how widespread the virus is, but that just brings us back to the questions of appropriate strategy and the objectives that justify giving the governor unprecedented power.

      • Mario

        I apologize that I was rude. It goes back to something I said years ago, perhaps not to you, about getting your news from bad sources. You can try to make sense of this however you want, but when the ideas come filtered through James O’Keefe and Laura Ingraham you should take a step back and reevaluate. I know, I know, all of the media is bad, but some are especially so, and always when what they are saying is what you want to hear.

        I encourage you to look through the actual numbers. They aren’t inflated. Even if some people are discharged quickly, that just makes the definite cases even worse. They are there for an average of 10 days. Do you know how bad things have to be for someone to be there that long? I went in for a stroke and came out after four. And then you have to remember that a lot of the worst patients are being treated at nursing homes for a good portion of the time, if those ever reach a hospital at all.

        Medicaid and Medicare fraud would be nothing new. There are already incentives to boost your numbers, and, for that reason, there are also processes in place to catch hospitals that try to cheat. The cost of getting caught is a lot higher than the payment boost they would see, and the people that would have to make those changes are not the people who would see any personal impact, certainly not immediately. And even if they weren’t caught by the Feds, insurance companies are also right there, doing the same job, and have the same risks. Cheating is exceedingly unlikely.

        • Justin Katz

          The Daily Wire article is one of four outside sources to which I link, and on that one, I’ve corroborated the suspicions with a source in Rhode Island.

          Nothing I’ve written suggests that those who are hospitalized specifically for COVID-19 have an easy time of it. What I’m saying that the number I think to be the most important is the one we were getting before… namely, who is in the hospital FOR this virus, not WITH this virus.

          • Mario

            I’m not sure what difference that makes. Any patient that has the virus, even if they have no symptoms, has to be isolated just like anyone else. They will use up the same amount of extra gowns and masks, etc. just like the other patients, because it isn’t the symptoms, it’s the danger they pose. We know it can take days to develop, it can take a sudden bad turn, and it has seemingly random effects on people. If they are sick enough that they have to be in the hospital for days even with no virus symptoms, and they have to be treated as the infectious carriers they are, and they could take a bad turn at any moment, I don’t see a difference between them and any other COVID-positive patient.

            And the distinction between “for” and “with” makes me just as uncomfortable as when the media was quick to point out that anyone who died had “underlying medical conditions,” as if to say “well, someone died, but they didn’t really count.” Because I know that if I were a patient , my death would be dismissed just as easily, as if it were just a slight acceleration of my demise. I’m not saying that anyone who has the virus and gets hit by a bus should count, but the number of excess deaths happening during a global pandemic aren’t just a coincidence. I expect that RI will have around 400 official deaths by the fall, but the number will probably be closer to 600 once the excess deaths are tallied. You can quibble with the definition of a COVID death, but that latter number will be the more accurate one.