Bureau of Disinformation

The Institute for Health Metrics (IHME) puts out models for the epidemiological data of various illnesses, but in particular COVID-19. IHME is not part of nor associated with the White House (WH), although the WH has used their models to justify certain policies (or lack thereof). IHME's models are unreliable and have been revised several times since this pandemic started, having grossly underestimated the mortality rate. It seems that simple sources of data, such as nursing homes and meat packing plants, would be clear targets, but were apparently misses in their early models. A report from Vox earlier this month questions the reliability of the IHME model. IHME is now forecasting 134,00 deaths by August, which just looks blatantly wrong based on what I am seeing. And yes, I know they include a range, which is really what they should stick to, not a definitive forecast (see below).

Just look at the current figures and math:

• we have 80K deaths so far, and we are not quite at mid-May.
• It took us roughly 2.5 months to reach this point, and we still have growth in the number of overall COVID-19 cases.
• We still do not have a good hold on the actual numbers as testing is not wide-spread, consistently applied, or reliable. It seems that the only place in the US that is testing regularly, and even there has problems, is the WH.
• Even if we had the entire nation under lock-down it would take around the same amount of time to come down the curve, with roughly the same overall number of case fatalities, in other words 80K x 2=160K deaths.

However, the entire USA is not under lock-down. 24 states have eased or never had restrictions in place with close to 120 million people, so given 3 weeks we will see further outbreaks and increased mortality. Just take a look at how things are getting back to normal in beautiful Castle Rock, Colorado (I guess this is that "Colorado Rocky Mountain High").

This will also be the case even in states with lock-downs in place, as people start to relax their guard with Summer weather, due to poor messaging and lack of enforcement of public health NPIs. Social habits will be hard to break with and keep physical distancing measures in place.

The IHME forecast is 134,000 of COVID-19 deaths come Aug. 1 and is a gross underestimate, and by that around 20% off the mark.

Given all of what I mentioned above, I am going to out on a limb here with my forecast is 160,000+ come Aug. 1. I include the + as I think the range is much higher.

Let me give you my calculations:

• We are at 80K deaths right now; see my math above, but I would add to this differentials in deaths from pneumonia between monthly averages from year to year, capturing undisclosed data that was not previously reported, e.g. homeless shelters, prisons, and other congregate care or institutions.
• Let's look at the math from another point of view. If we look at the data out of Italy, which is only now starting to open up after 2 months of a fully national quarantine, it has averaged about 500 deaths per day since coming off their peak. Italy has a population of 60 mill. v.s 330 mill. in USA. The US peak has been 3,000 per day (we are not at peak). Doing the simple math here, the next 30 days with a full lock-down would see on average 1,500 per day x 30 days= 45,000 deaths. The month after (month 2) would likely see 500 deaths per day x 30=15,000 deaths. 45K+15K= 60,000 min. additional COVID-19 case fatalities through July only. If we add additional deaths for the full month of July and take into account easing of mass quarantine measures, then again I come to a forecast of 160,000+ deaths by Aug.
• Let me pose a final calculation from a different point of view. If we look at case infections solely and forecast a doubling of cases by early July based on the case infection growth we are seeing; you can see straight line growth, which will likely continue with greater testing and spread (as referenced above), so we grow in COVID-19 cases from in 1,332,609 US (as of 5-11-2020) to 2,660,000 cases or so. Taking into account a lag in fatalities (3 weeks or so) and based on the 6% mortality rate (an assumption that healthcare capacity continues to be strained), this would arrive at 79,956 additional deaths from COVID-19 or a total of 160,000.

So those are three different methods, with different points of view. I hope that I am wrong, and if I am wrong then I will donate $100 to whatever charity IHME would like me to donate to. If I am right, then I would kindly ask that IHME do the following:

• make it clear that your models are based on incomplete data and that it has weaknesses, put that up front and center. Identify potential weaknesses in your forecasting models.
• stop including a definitive forecast and use a probabilistic forecast with only the ranges that they currently publish (in other words, get rid of that dotted line in the middle that is used by the WH for forecasting).
• don't make statements with definitive forecasts, keep it to the ranges.
• include alternative methods, different specialists, and question the results from several different points of view. The foundational question I always ask is whether the result makes sense.

Last point here is that the WH has been using these models to downplay how serious the pandemic is and to make grandiose statements about how well they are doing. A little knowledge is a dangerous things. Losing over 100K Americans to a pandemic is a major failure, and providing a model that helps this administration paper over their mistakes and missteps is very dangerous.

Andrew Boyarsky is President of Pinnacle Performance Management, a business continuity and disaster recovery consultancy focusing on SME, non-profits, universities. He is also a Clinical Associate Professor in the Management and Systems MS at NYU and in the Emergency Management Graduate Program at the John Jay College of Criminal Justice, a part of the City University of New York. He is also host of the podcast Riding the Wave: Project Management for Emergency Managers.