For quite a while now, we have been badgered by “models” of what the coronavirus is projected to do in our state and city.
The most quoted model is the Institute for Health Metrics and Evaluation (IHME) at the University of Washington. These are the same folks who have been predicting that soon we in Florida would run out of hospital beds and ventilators and a good chunk of our people.
Thankfully for us, their latest “modeling” predicts that in just 12 days, by April 21, Florida will “peak” in our rate of fatality from the virus, reaching 242 deaths per day. Well, that is very nice of them – only if you look at the actual curves from which they extrapolate what will happen in just thirteen days, what you actually see is a zig-zag that has no pattern whatsoever.
It’s worse than that. The data points on their curve seem to show zero fatalities one day, close to 100 the next and 0 the day after that.
In other words, some student must have been plugging numbers into a computer as he received them and never bothered to check if the data was current, improperly bundled over time as reports came in, or just plain made up.
As per the saying now made famous by Sanjay Gupta, “all models are wrong; some are useful.”
Here’s what I think the folks at the University of Washington did.
First they took historical data from states or countries that were thought to be “more advanced” than Florida; then they plugged in the population of Florida and one other variable. Yes, I said it: only one other variable.
The only other variable, other than population, is the date at which each state put in place a “stay-at-home” order. In the case of Florida, that was April 1, which the IHME folks consider too late in the cycle to constrain the spread of a disease which, under normal circumstances, multiplies exponentially until the now-famous “apex” is reached.
Here’s why the model has no application whatsoever to Florida.
Of the approximately 20 million people who live in Florida, almost one quarter live in Dade and Broward, which are the two most populous counties. They are also the most urban, and the ones that have the largest cruise ports and busiest airports. If Florida were to follow the New York model, we could rightly expect to have the bulk of the fatalities in our state.
But those two counties were among the earliest to impose either stay-at-home orders or the equivalent: “safe-at-home.” Not only that, but those are the hottest counties in the state and the COVID-19 fellow seems to dislike heat.
And so Miami Dade and Broward have had, in total, a bit over one-hundred deaths, which is about one-fifth of what New York City has in a single day.
How the folks from the state of Washington – where the virus has wreaked havoc, particularly among seniors in nursing homes – get the idea that in 12 days, Florida will go from a total of 300 deaths, occurring during the last month, to 242 per day is beyond any extrapolation I have ever seen. (I call it “uber-extrapolation.”) This is particularly doubtful when the daily confirmed cases have flattened or dropped for about a week now, despite fierce competition from municipal governments to test people at increasing number of drive-through sites and even at home.
As I write these lines, Miami-Dade County has enormous capacity for hospitalization of new patients. We have gone from 36% to 43% bed capacity in one week, and added temporary hospitals with another 750 beds, plus an entire building at the central Jackson facility that was about to be demolished, but has been kept intact, just in case.
Well, it’s good to be prepared, just in case the IHME folks are right, with their “uber-extrapolation.” But I do want them to know that I have a mnemonic device for their “IHME” acronym, which is “I Hate My Estimates.”