Fine-tuning Estimated Recoveries and Estimated Active Case Count

Posted on May 17, 2020

Back on April 15, I wrote about a statistical discrepancy I noticed, namely that the number of recoveries being reported by state agencies was much lower than we would expect, given what we know about average incubation period and average time from first symptoms to death for COVID-19.

We know from various studies that the average incubation period (time from virus exposure to emergence of symptoms) is 5 days.[1][2][3] We know that the accepted average time from symptom emergence to death for COVID-19 is 18 days.[1]

Since recoveries are a key factor in determining how many active cases there are (Active Cases = Total Cases – Deaths – Recoveries) , it’s important that the recoveries be accurately tracked. However, it doesn’t appear that recoveries have been accurately tracked at all from the beginning of the outbreak in the US.

My original assumption was this: the date on which a state reported a COVID-19 case would represent the day that symptoms first emerged. Therefore, if we were to skip ahead 18 days from that reported case date, the patient will have either died or recovered, on average. As I learned last week, when comparing 7-day average trends for cases and deaths, it turns out that my assumption was not correct. Here’s a look at what I found:

As we can see, the average time from cases being reported to the death being reported is 6 days. So, cases are not being reported publicly until very late in the disease timeline for each patient.

Because of this, the estimated recovery model assumption needs to be changed from accounting for 18 days from case reporting to death to the actual 6 days from case reporting to death. When we change this assumption, we find that the estimated active case count is 3/10,000 people, compared the reported active case count of 30/10,000 people. It’s a massive discrepancy that needs to be publicly acknowledged and addressed.

Are you wondering why this matters at all? Well, the fear of the people has been such a massive factor and massive tool that our leaders have exploited over the last 2 months. Obviously 1 million active cases nationwide is much more concerning than 113,000 active cases. But 1 million is not even close to the true number, and people need to know and understand this so that we can all start taking bigger, bolder, more reasonable steps to get back to normal in all areas of the country.

P.S. Here’s a way to visualize the estimated active case count per 10,000 people.