Donald Trump says 99.98% survive the virus. Other numbers float around. I thought I’d dig in and find out the real answer.

The simplest possible answer is derived as the ratio of deaths to cases in the United States. As of 10/13, there have been 204,265 deaths and 7,669,759 deaths. This gives us a death rate of 2.66% of cases.

There are a number of problems with this number:

  • “Cases” in this metric is positive tests. All evidence points to this being a subset of actual cases, since we have some asymptomatic people as well as people with mild symptoms who don’t bother getting tested. We’ll talk about how to deal with this later.
  • A lot of conspiracy theories posit that deaths aren’t accurate. One popular video says that shadowy forces are bribing doctors nationwide to falsify cause of death. I’m going to discount this and all other data falsification ideas (also – data is being falsified by red/blue states to make the pandemic less/more severe looking. Also – death certificates list comorbidities so 90% of deaths were really caused by their asthma or obesity). I’m going to discount all of these and go with the data we have.

So all this said, let’s assume we can go with this 2.66% fatality rate. There’s a correction I’d still make: this rate has been dynamic over time for one or more of several reasons:

  • COVID patient demographics has been changing. Initially, it got loose in a lot of assisted living facilities where the death rates were super high.
  • Testing has become more available. Initially we were only testing a small percentage of cases because tests were unavailable. This situation has improved over time. The effect is that early on, our tests caught a much lower portion of our cases, and triage meant it caught only the worst cases.
  • Treatments have gotten better. People talk about Remdesevir or other treatments; the studies show only a very nominal impact of these. However we may be smarter in ICUs about how we treat people, e.g. we may take longer to ventilate people and keep them on their sides longer, helping more recover.

So how has the fatality rate been changing over time?

Figure 1 – death rate over time

What we can see in figure 1 is two important eras in the death rates. During the initial months, we had a total death rate of 6% – a horrifying number. As things have settled out, we’ve reached a death rate of less than a third of that – 1.7% – and appear to be very stable there. Remember this change might not be because we have a magic treatment regimen – it’s also because different people are getting the virus and we’re testing more people.

1.7% may still not be the true death rate we’re seeing, because we may be missing many sick people. There are estimate, but they are very crude and it’s been too soon to really get good estimates. However, we can do some scenarios.

  • if we’re missing half of the cases, then the true death rate is 0.85%
  • if we’re missing 3/4 of the cases, the true death rate is 0.43%

This improvement is nice, but it’s still much worse than the flu. The flu has a fatality rate of 0.1%, plus we have vaccines to use to control its spread. Even with no vaccine, COVID is four times as deadly as the flu. To believe it’s “just like the flu” you have to believe we are missing 95% of the cases, which is a much higher estimate than any study has given.

How much worse than the flu? In 2019, the US had 35M flu cases and 34,000 deaths – fitting the 0.1% fatality rate. Our staggering success with the flu results from broad distribution of a vaccine and immunity from past seasons, which keep the number of deaths so low – around 10% of the US population got it.

COVID-19, on the other hand, would be very different. With no vaccine and no immunity, it would rip through our population until around 60% of us had been infected – around 210M people – and we hit herd immunity. at a conservative 0.43% death rate, that would mean we’d have 903,000 deaths before it slowed down.

How can we avoid this?

  • First, there’s social distancing and masks to keep the case number down. Fewer people infected = fewer people dead.
  • Second, we can get better at treating COVID. If we can get that death rate down to 0.1%, then fewer of the people who do get infected will die.
  • Third, we can further shift the demographics of who gets sick. Protect the assisted living facilities, keep obese people home, let older people stay in their houses. This will further decrease our actual fatality rate, resulting in fewer deaths.
  • Finally, we can get a vaccine. This is the longer term way we can get to herd immunity without 900,000 deaths.