Modeling Coronavirus in Minnesota

As I explained in a post yesterday, you really shouldn’t be placing a lot of trust in models about coronavirus infection rates or death rates.  We simply don’t have enough information for their outputs to be anything more than than very rough guesses with very wide ranges.  Yet they are being used to justify extreme mitigation measures that have locked down the economy, resulting in the loss of at least 20 million jobs so far.  An article in one of our local papers gives a little more insight into just how bad the modeling is and in what a dishonest manner the results have been used.   (Star Tribune Article)   The apparent lead modeler, Stefan Gildemeister, is quoted in the article and some details are given that I don’t believe were promoted to the public before.

Governor Walz told Minnesotans that 74,000 people would die if we didn’t put this stringent economic lockdown in place, a lockdown which has resulted in at least 300,000 Minnesotans losing their jobs.  What he didn’t tell people was that those measures would at most save 24,000 of those deaths.  If his numbers, based on what I am going to generously characterize as insane modeling, were right, if he really believed them and thought if he didn’t shutdown the economy 74,000 Minnesotans would die, then even with the shutdown, 50,000 Minnesotans will die.  I want to be sure people get that number–50,000 people would still die, nothwithstanding the extreme shutdown.  He certainly didn’t make that clear.  Probably because his lockdown might have been greeted with a very different reaction if he had.  (Once again, I want to be very clear, there is no way in hell that under any circumstance 74,000 people were going to die in Minnesota.)

And the 24,000 deaths that are supposedly saved are said to be due to not overwhelming hospitals, ventilators and other health resources.  Unless the state releases the actual model, and it should, it is hard to know what assumptions underlie it, except for comments in articles like this one in the Star Tribune.  I would note that the model definitely is using infection rates which are much higher than appears likely to exist or to be occurring in the real world.  There are comments regarding the transmission rate, but I have pointed out before that this rate is actually pretty useless, what counts more is actual infection rates.  The model, or at least the output of the model that is apparently being used, is also using hospitalization and ICU rates that are too high.  And the average days expected per person in the ICU is far higher than actual experience would justify.  Minnesota has substantial health resources and can create more quickly.  I think the odds of running out of resources is pretty low.  So I don’t think it at all realistic to say there would be 24,000 deaths if the economic shutdown were not in place.

There are another observations I could make about both the model and some of the comments in the article, but I am frankly depressed by both the scare tactics being used and duplicitousness by which the shutdown order was justified.