New COVID cases in Wisconsin thru May 17, captured from Google/Wikipedia
Values updated 5/17/20:
People seem to think that the primary election in Wisconsin led to only 50 folks getting infected, which if true would be pretty minor. But it’s much worse than that. The graph above shows that 2 weeks after the election — in good agreement with the known incubation period for COVID 19 (see below) — the trend of new cases, which had been going down, suddenly turned up. As of 5/17/20, the average after April 20 is 296 cases per day. So that’s not 50 but about (296-160)*25 = 3392 cases plausibly assigned to the election. At Wisconsin’s currently fatality rate of 4%, that would be about 136 new deaths attributable to the election. Of course, without tracing, we can’t absolutely prove that these cases were the result of the election, but that conclusion does fit the known behavior of the virus and the measured data. And it’s important to note that, without a vaccine or treatment, we may all eventually have to deal with the virus, so maybe these folks were just infected earlier than they could have been. But the longer we can wait, the better chance of improved treatments and perhaps prevention.
Nice job by the Supreme Court, condemning others while protecting themselves by delaying oral arguments in April.
5/17 added remarks:
Several folks have commented correctly that this data does not prove that the increased case load resulted from the election or only from the election, nor was this asserted. We need detailed testing and tracing of the whole infected population, which no state in the US currently has. But the null hypothesis — that the election had no effect — is certainly not the default for this dataset. Unlike some other possible causes noted in the comments, the election is a single-day event with a very large number of people participating. Other influences on infection and transmission have more diffuse effects. The sudden change in the appearance of the data is sensible if connected to a single large event, harder to explain if resulting from gradual changes in social behavior.
It’s also worth noting that the current value for the incubation time of COVID 19, from infection to clear symptom expression, is about 4-5 days (from a talk by Prof. Ho of Columbia University yesterday May 16), so the delay above is longer than the time it took for a person to get infected and show symptoms — though we should remember that testing and detection are already 1 cycle after transmission, since our testing is so limited. If the detection above is really related to the election, it is rather the expansion of the infection by about 2 transmission cycles (using a multiplier of 3, that would be a 9-fold expansion on the number of folks directly infected). Given the limitations of our testing so far, this seems plausible, and similar delays between actual introduction of the infection and observation of large case numbers have been noted in a number of outbreaks.
Comments are closed on this story.