Minnesota Trends

Our state health department has started doing briefings less frequently and the Governor isn’t showing up much at all.  The reason is that despite all the hype, there is no evidence that we are “climbing the curve” as the Governor always liked to say, in fact, we appear to be gently sliding down it.  (But don’t give up yet Governor, there is always hope that the protests could create that mythical “surge” of cases.)  Yesterday’s briefing was notable again for the evasion of certain clear questions.  One thing that is really puzzling is why they won’t release antibody test results.  They keep blaming it on test uncertainty, but that is crap.  Give us the results from the tests that are approved by the FDA, and there are a lot of those.  Give us the results weighted according to the sensitivity and specificity of the tests.  Give us the results and tell us they are not trustworthy.  But give us the results, because at a minimum they will be directional.  When you don’t release them and make lame excuses it looks like you are hiding something.

The funniest evasion, of course, was again in regard to questions about Wisconsin.  The response about why there is no uptick at all in cases there was a repeat of the science is different in different places answer.  Seriously, the virus knows the difference between Minnesota and Wisconsin or between a person who lives in Minnesota and one who lives in Wisconsin.  If we were talking differences between Minnesota and Texas or Florida or California, I might buy some reason for a difference, climate or population mix, although properly adjusted for health conditions, doesn’t seem to be an ethnic difference.  You have to stop with the absurd answers to questions and just admit that what has happened in Wisconsin pretty clearly demonstrates that you don’t need a lockdown to keep things under control.

So what has happened with cases and hospitalizations and deaths in Minnesota; what is worth noting? Our testing has risen pretty dramatically. If you take the average number of tests on the seven days ending with and including June 9, there were a little over 10,000 tests a day. If you take the seven days ending with and including May 9, it was a little under 4000 tests a day. So in a month, our testing is running about 2 1/2 times more. What has happened with positive results? I am just starting with cumulative tests and cumulative positives on April 15, moving every few days and ending on June 4 to allow plenty of time for lagged results. The cumulative numbers may not completely match up but they are close enough to give good direction.

On April 15 the cumulative positive to cumulative total tests was 5.3%. On April 20, it was 6.3%; April 30, 10%; May 5, 10.7%; May 10, 10.8%; May 15, 11%; May 20, 10.8%; May 26, 10.1%; May 31, 9.2%; June 2, 9.2%; and June 4, 8.7%. So the positive rate appears to have peaked. That may reflect broader testing and testing criteria, not a diminished spread of the epidemic or fewer cases. Just looking at the table of new cases by day, it would appear that we peaked in mid-May and are seeing a decline, although a steady number of new cases per day. It would be very helpful to know how many of new cases by day turned out to be asymptomatic or mild, but the state doesn’t give us that information, and it may be hard to obtain, as you would need to check back on people with no symptoms at the time of testing for a couple of weeks.
Trying to identify active cases is harder. This would technically be the difference between total active cases on a day and those cases identified as being out of isolation, or no longer infectious. The state, however does not provide historical data on its out of isolation numbers. A reader has tracked that and very kindly shared it, and it appears to have plateaued in recent weeks. For example on June 10, there were 4194 “active” cases. On June 3, it was 4701; on May 25, 7499; on May 15, it was 4737; and on May 10, 6145. So it appears that this number has gone up and down a fair amount. Not sure what would account for that or how the Department determines that someone is out of isolation. I suspect there is a fair amount of measurement error.

Hospitalizations is a complicated metric to use. There can be a lot of variability in why someone is admitted to a hospital and I suspect that, just as is the case with other diseases, physicians differ greatly in their admission rates for coronavirus patients. Some are going to be more cautious and admit people who may not really need a hospital. Some are going to keep people out of the hospital to the extent they can, because hospitals actually aren’t safe places–lots of infections, coronavirus and others, occur in hospitals. And patients probably vary in how hard they push to be hospitalized. So there isn’t going to be a clear line defining the relationship between a person’s condition and hospitalization.

In addition, just looking at daily census, or how many people are in the hospital for coronavirus on a given day, doesn’t give you much useful information. You would also want to know daily admissions and daily discharges or deaths occurring in the hospital and at least average length of stay. The admissions number would be the best marker of trends. And even that is complicated because some patients get infected when in the hospital for other reasons. Some people are pretty sick, get to the hospital, get tested there and learn for the first time that they have coronavirus. The length of stay would be informative about trends in seriousness of cases as well as capacity management. And it would be very useful to see trends in death in the hospital and place of discharge. With the concern about skilled nursing facilities, I wouldn’t be surprised if discharge to SNFs has decreased. And if people are reluctant to be discharged to a SNF, they may end up in the hospital for a longer time. And while I hate to suggest it, hospitals are just masters at manipulating payment systems to maximize revenue. Keeping a coronavirus patient longer likely increases revenue.

So given those caveats, what do we see in Minnesota’s hospitalization numbers? If you just eyeball the chart on the website, hospitalizations and ICU use within hospitals both seem to have slowly risen to a peak around the end of May and are now declining. It is very hard to know exactly how to relate this to case numbers. As noted above, some people probably aren’t even tested til they get to the hospital. So figuring out what the lag from date of infection, or date symptoms appeared or date of testing is not really possible with the numbers we have. I have tried some analysis of cases to hospitalizations assuming different lags, centering around 5 days, the presumed average incubation period, or 10 days, which is maybe how long it takes for the illness to get bad enough to hospitalize someone, and I am not sure I can tell much. I was mostly interested in trying to see if I can figure out if fewer cases are severe enough to warrant hospitalization. As best I can tell at this point, the ratio of hospitalizations to cases seems fairly steady.

Finally, deaths, which everyone fixates on. The number of deaths on a daily basis has definitely plateaued, although it also jumps around from day-to-day. In ten-day blocks, starting on June 10 and working backwards, the average over that ten days was 19.6 deaths per day. For the ten-day period ending May 31, it was 23.1 deaths per day. For the ten days ending May 21, the average per day was 21.8 deaths. So the peak appears to have been that last week in May, at least for now, which is the same peak we have seen so far for hospitalizations. And the percent of deaths among those in group living settings, such as nursing homes and assisted living, has stayed around 80% for weeks. It seems to be slightly lower over the last two weeks. At some point we may see the effect of the concerns about nursing homes. People may end up dying at home rather than go to such a facility. And the state’s plan to reduce cases and deaths in group residential settings may finally be producing some results as well.

When I look at all these numbers, what I see is a fairly clear rollover in the epidemic.  And this is occurring notwithstanding the substantial changes in the number of contacts people are having.  More businesses are open, people are just moving about more.  This is completely contrary to what we were told would occur.  We were told that cases would inexorably keep increasing no matter what, along with deaths.  It could be seasonality, but it isn’t yet clear to me that there will be strong seasonality.