UPDATED: A Big Drop-Off in Games with Models, 5/18/20 Data

COVID19-hospitalizationsandprojections-051820-featured

As my daily COVID-19 hospitalization chart for Rhode Island shows, today’s data brought a major drop-off of about 10% from the day before.

Even better, hospitalizations obviously lag infections, and the data itself is two days behind, and new cases overall are down, too.  Yesterday, only 121 new cases emerged, even though the number of total tests was within the range that we’ve been seeing.  That’s the smallest increase in cases since April 4.  This is despite the fact that we’re now well into the incubation period for anybody who might have caught the virus when things began opening up a bit more.

A question therefore emerges:  What “facts and science” led the governor to decelerate her reopening plans?

Interestingly, one metric she’s said she’s following is the R-naught value, which is a measure of how many people each infected person spreads the disease to.  A value of one means the disease is not increasing; less than one means it’s receding. The governor stated that the value is right now “1 or just under 1.”  However, if the average infection period is 14 days, the numbers of positive cases suggests an R0 of 0.32.

UPDATE: The website rt.live uses a sophisticated formula to estimate when people’s infections began and to adjust for the number of tests conducted each day.  This obviously gets back to the problem I posed at the beginning of this exercise, that every new assumption could actually move an estimate farther from the truth.  For example, correcting for a reduction of all testing and thereby devaluing a drop in positive tests could ignore the fact that people aren’t being tested because they aren’t sick.  In any event, the website puts the R value at 0.85 for Rhode Island, which is in the top half for the country.

News reports give no explanation for the governor’s change of plans beyond stating:

“You don’t know the impact [of changes] until you have observed them for a couple of weeks,” Raimondo told reporters in explaining the current timeline: two weeks for any new cjanges to take hold and then another two weeks to study them.

It doesn’t take much skepticism to wonder whether the governor had expected the improvement of the numbers to go more slowly, naturally spacing out the phases.  But this head fake on the schedule damaged her ability to keep imposing rules, and if things continue to get better, more people will wonder how she has the authority to set regulations more than 30 days out when she has no idea what’s coming.

After all, her statements about continuing economic restrictions are coming at a huge cost to Rhode Islanders.  “Better safe” for her could be “sorry” for thousands of others.

COVID19-hospitalizationsandprojections-051820

(See here for my original methodology and here for a subsequent modification I made.)

Projections versus actuals (date of report).

  • Cases:
    • Projection for 5/18: 12,799
    • Actual for 5/18: 12,795
    • Projection for 5/19: 12,919
  • Hospitalizations:
    • Projection for 5/18: 258
    • Actual for 5/18: 236
    • Projected for 5/19: 224
  • Deaths:
    • Projection for 5/18: 504
    • Actual for 5/18: 506
    • Projection for 5/19: 511


  • Mario

    I haven’t been keeping up with my predictions because of my own illness and the fact that my hospitalization number desperately needed retooling, but I agree that things are improving very quickly. I don’t blame the governor for going slowly, I expect that we’ll still be seeing COVID deaths throughout June even if nothing bad happens to reverse the trends, so how much risk are you comfortable taking when we’re so close to victory? It’s bad enough that it might come back in the fall even if it’s completely eliminated now, and I expect it to be completely eliminated. On our current trend, by the 1st of July, I think the state will be completely threat-free — no deaths, no cases, nothing (I actually have the last death on June 21st right now, but it was later not too long ago, and I don’t think there’s anything wrong with giving myself some padding).

    The idea that we have to open up and just except the constant threat as the new-normal had some justification a while ago, but it looks increasingly reckless when I can actually see an end to this, and so close. Obviously, I could just be wrong about it entirely, but I see what I see. I don’t know how it is for the rest of the country, but for RI this is working better than anyone expected it to, and I think we need to see it through to the end.

    So I expect 135 new cases, only 204 hospitalizations (but I have no confidence in that) and 10 deaths. I have the total topping out at 636 deaths by the time this is over. I want to say “but it could be higher” except that number has been steadily dropping each day, so I don’t know.

    • Justin Katz

      I’d note that we always must defend against our tendency to build narratives and presume knowledge. The fact is that, as we sit here, we do not know what the future holds. Yes, this trend gets to zero, but maybe it gets to zero while reopening. Or maybe we do all this economic suffering (and don’t doubt that people are really and truly suffering and dying over our response to the virus) only to declare victory, leave the house, and bump into a traveler from elsewhere who kicks off a fresh round.

      How much risk I am willing to take is a question for me and my family to answer for ourselves, not for me to answer for the other million of my fellow Rhode Islanders.

  • ShannonEntropy

    Models / Schmodels …. this virus acts the same everywhere it hits: rapid, even exponential rise in new cases / deaths for 6 – 8 wks; then a rapid dropoff in same

    As for you guys quaking in your boots about reopening, look at two countries with radically different approaches. The UK — in almost total lock-down — has a per 100K population death rate of 52.45

    Sweden OTOH took a much more relaxed, open, “let’s get herd immunity” approach. Their per 100K death rate: 36.31

    Which country wore it better ??

    https://coronavirus.jhu.edu/data/mortality