Coronavirus: Infection Fatality Rate Estimates for the United States

How many people die when infected with coronavirus? That is one of the current debates. And I'll be looking at various estimates, but before doing so let's look into the three different types of percentages that are discussed:

1. Mortality rate: percentage of total population that die from a disease (including those who don't get the virus at all).

2. Infection fatality rate (IFR): percentage of those who got the disease and died. This includes some level of estimate as it will include people who never got tested.

3. Case fatality rate (CFR): percentage of those who tested positive for the virus and die.

Based on what I've read, I've always went with the idea that the case fatality rate was at 1%.

Recently, I read a couple more recent estimates.


CDC Estimates

The CDC recently came out with estimates that got a lot of push back. Reason wrote:

According to the Centers for Disease Control and Prevention (CDC), the current "best estimate" for the fatality rate among Americans with COVID-19 symptoms is 0.4 percent. The CDC also estimates that 35 percent of people infected by the COVID-19 virus never develop symptoms. Those numbers imply that the virus kills less than 0.3 percent of people infected by it—far lower than the infection fatality rates (IFRs) assumed by the alarming projections that drove the initial government response to the epidemic, including broad business closure and stay-at-home orders.

If I understand the numbers correctly, the CDC has an estimated CFR % of 0.4% while their IFR % is less than 0.3%. That CFR is far below what I always assumed.

The article admits that these rates seem unreasonably low for states such as New York and New Jersey; however, they argue it looks reasonable for other areas of the country:

But the CDC's estimate looks more reasonable when compared to the results of antibody studies in Miami-Dade County, Santa Clara County, Los Angeles County, and Boise, Idaho—places that so far have had markedly different experiences with COVID-19. We need to consider the likelihood that these divergent results reflect not just methodological issues but actual differences in the epidemic's impact—differences that can help inform the policies for dealing with it.

University of Washington's School of Pharmacy Estimates

The Hill discusses:

The study, conducted by the Comparative Health Outcomes, Policy and Economics Institute at the University of Washington's School of Pharmacy, found that 1.3 percent of those who show symptoms of COVID-19 die, an infection fatality rate that is 13 times higher than a bad influenza season.

"COVID-19 infection is deadlier than flu — we can put that debate to rest," said Anirban Basu, a health economist at the University of Washington who authored the study.

If the infection fatality rate is accurate, and if the coronavirus continues spreading at current rates even before most states open their economies and relax social distancing restrictions, COVID-19, the disease caused by the virus, could claim between 350,000 and 1.2 million American lives by the end of this year, Basu found.

So here they come up with an IFR % of 1.3%. This would indicate a much higher CFR % than my 1% assumption. Basu is basing his estimates on deaths based on 20% of the population getting the virus by the end of the year. My bet is that the baseline is 860,000 deaths by end of the year. That would be the total population of 330.8M * 20% * 1.3% = 860K.

I will admit I was a little confused by how the article is written due to my understanding of the 3 types of percentages. My understanding is that IFR would apply to people who got the virus, but never showed symptoms. Yet, the first sentence I quote mentions that the study estimates that 1.3% of those who show symptoms will die, but then it goes on to say this is the IFR. Reading the study abstract they go with something called an IFR-S. The "S" means symptoms.

Disputes with the CDC numbers

As you can see, there is a big range with the CDC coming up with an IFR of less than 0.3% while the University of Washington is coming up with 1.3%. That's a big range though they are probably slightly closer as they're using different rates.  Buzzfeeds reports on some disputes with the CDC numbers:

New CDC estimates of coronavirus death rates look suspiciously low and present almost no data to back them up, say public health experts who are concerned that the agency is buckling under political pressure to restart the economy.

. . . Other estimates in hot spots outside the US are also higher than the agency’s deadliest estimate.

In Spain, a massive antibody survey of more than 60,000 people put its overall fatality rate at around 1.1% to 1.3%. In Italy, researchers estimate that 1.2% of all cases have resulted in death, and in France, 0.8%.

. . . A preliminary analysis of more than two dozen studies from Europe, China, the US, and elsewhere, conducted by Meyerowitz-Katz and colleague Lea Merone, suggests that the overall infection fatality rate is between 0.5% and 0.78%. Even the lower end of that range is higher than what the CDC says is its “best estimate” for the rate, which is about 0.26%. 

Who knows if one should be including China in your estimates for an IFR %. I wouldn't.

Buzzfeeds is writing something that immediately came to my mind when the CDC numbers came in far lower than my assumption: is the CDC under pressure to low ball their forecasts? Also, what should be concerning is that the CDC apparently hasn't released their data for how they came to their estimates. Hopefully, that data eventually gets released so that there is some peer review conducted.