Thread by @davidrliu

Let’s estimate the total # of people in the US who have been infected by #COVID19 #SARSCoV2 as of Apr 4, and the number of US fatalities that will result from these infections. To avoid testing coverage uncertainty, we will use fatality data, rather than positive test data. (1/9)

Importantly, the # of fatalities that we end up estimating from the infections through Apr 4 turns out to be independent of the infection fatality rate we use, as shown below. The infection fatality rate only changes our estimate of how many people in the US are infected. (2/9)

As of 10 pm US eastern time Apr 4, there were 8,476 reported US deaths from #COVID19 (gisanddata.maps.arcgis.com/apps/opsdashbo…). Let’s assume an infection fatality rate of 1%. While there has been some debate about this number, most estimates are around 1%. (3/9)

We also need to know how long it takes an average fatal #COVID19 case to go from infection to death. We will use 18 days from onset of symptoms to death (medrxiv.org/content/10.110…) +5 days from infection to symptoms (annals.org/aim/fullarticl…) = 23 days from infection to death.(4/9)

So 8,476 deaths by Apr 4 corresponds to ~847,600 total US infections as of Mar 12, 23 days before Apr 4. How many infections were active as of Mar 12? Assuming an average US doubling time of 3 days before Mar 12 (before most stay-at-home orders), 814,228 cases were active. (5/9)

If those 814,228 active cases as of Mar 12 doubled every 5 days on average until Apr 4, there would be 19.75 million US infections from Mar 12 to Apr 4. Adding back the pre-Mar 12 infections makes 19.78 million total US #COVID19 infections as of Apr 4, or 6.0% of the US. (6/9)

The total number of fatalities from 19.78 million US infections is 197,800 deaths. Note that if we used, for example, 0.66% as the IFR instead of 1%, the # of estimated deaths doesn’t change at all, but the total # of infected in the US grows to 29.97 M (8.6%). (7/9)

These estimates are sensitive to the average doubling time from Mar 12 to Apr 4. If doubling time was 6 days instead of 5, then 3.5% of the US is infected, which will cause 116,400 deaths. If the doubling time was 4 days, 13.3% of the US is infected, causing 438,500 deaths. (8/9)

Importantly, these estimates are only for total infections as of last night. Unless future infections drop to 0 as of today, US infections and deaths by the end will only be higher. Thus continued #socialdistancing is critical. Thank you @michaelzlin for analysis feedback. (9/9)