I published an article yesterday in which I estimated that the total number of people who will have died from Covid-19 by April 26, 2020, could be as high as 213,504. A friend pointed out to me that there was another, potentially more accurate way of predicting deaths.
Yesterday, I estimated the true number of people who were, or had been, infected with the SARS-CoV-2 virus in the USA on March 29, 2020, not just the confirmed cases, by dividing the total number confirmed deaths on March 29 (1,668) by an updated and more realistic mortality rate (0.5%). Then the expected total number of deaths, not accounting for interventions, 28 days later (by April 26, 2020) could be estimated by multiplying that number by 2⁷ (=128) to account for seven doublings over the seven four day periods. This produced the estimate of 213,504 (= 1,668 / 0.005 * 128 *0.005). Note that the mortality rate cancels out, and all that matters is the rate of doubling, which was every four days before the shelter-in-place orders.
An alternative approach is to start with the number of confirmed new deaths on March 29: 425. Divide that by the mortality rate to get the number of new infections roughly sixteen days (incubation + course of illness) before: 425 / 0.005 = 85,000. Then double that four times (2⁴ = 16) to estimate the number of new infections on March 29: 1,360,000. Then wind it forward by…
