Jared Heyman

Apr 1, 2020

4 min read

On your odds of dying from coronavirus

Humans are notoriously poor at assessing risk. There are neurobiological reasons for this, and since my brain is just as handicapped as anyone else’s when it comes to risk assessment, I like to compensate by running numbers in a spreadsheet. If you’ve read any of my previous posts about coronavirus, you’ll find this to be the common thread.

There has been much conversation about the threat the virus may pose to the health of society, initially from a clinical perspective and now increasingly from an economic, sociological, and psychological perspective. In this post, I’d like to take things down to the most individual and intimate level by answering the question, “What are my odds of dying from coronavirus?”

This is more than an academic exercise because these odds can impact everything from how much anxiety we feel about the virus to whether we decide to visit the grocery store. I hope that these numbers provide a bit of sanity in what currently feels like a crazy and uncertain world.

Before I share my analysis, I’d like to define some important terms. This first is Case Fatality Rate (CFR) which refers to the percentage of known cases of coronavirus that end in death. The second is Infection Fatality Rate (IFR) which refers to the percentage of known and unknown infections that end in death. If we knew about every infection, then CFR would equal IFR, but we never will. We’re pretty good at tracking at coronavirus deaths, so the numerator of these equations is consistent, just not the denominator.

CFR estimates, especially in the early days of an outbreak, are often overly scary and misleading. For example, on March 3rd, the World Health Organization reported:

Globally, about 3.4% of reported COVID-19 cases have died.

At first glance, one might think this means that 3.4% of people who were infected with the virus died, but that’s not at all the case. The word ‘reported’ here is key.

In recent weeks, we’ve been able to follow the early results of population-wide COVID-19 testing efforts, which is providing a clearer picture of how many virus infections there might be in certain countries. South Korea and Iceland have been doing an especially good job of this, the former in terms of the sheer number of the tests (approaching 400,000) and the latter in terms of the percent of their population tested (approaching 5%).

Each of these countries has been testing a mix of suspected COVID-19 infections and random people, the latter being especially important to discover mild and asymptomatic cases. In the US, by contrast, we’ve mostly been testing people that show up at hospitals, so our CFR numbers are pretty misleading.

There’s also the issue that the vast majority of tests these days are PCR tests, which are good at catching active infections but not people who have already contracted the virus and recovered (often without ever noticing symptoms). New serological antibody tests are doing a better job catching recovered cases as well, which are important additions to the CFR and IFR denominators.

I ran an analysis to estimate a coronavirus IFR based upon the most recent testing results data from South Korea and came up with about 0.1%. This is accounting for estimated new infections that haven’t yet been hospitalized, hospitalized patients who aren’t yet in critical condition, and ICU patients who haven’t yet died but likely will.

Dr. John Ioannidis, a professor of medicine, epidemiology, population health, and biomedical data science at Stanford, estimates an IFR in the US of 0.125%.

Dr. Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases, said recently that he expects 100,000 to 200,000 Americans could die of COVID-19. This sounds like a lot, and it is, but if you assume 60% of our population eventually gets the virus before herd immunity kicks in, it implies an IFR of 0.05% to 0.1% given a population size of 330M.

Another thing to consider when estimating your personal risk of dying from this virus is your age since fatality rates vary wildly by age category.

Combining we know about likely IFR ranges for COVID-19 and how CFRs vary by age category, I added to my analysis the calculated odds of someone in a given age category dying from the virus under various IFR scenarios:

Here’s how to read the numbers: I’m 42 years old, so if we assume an IFR of 0.1%, my odds of dying from coronavirus if I’m infected are about 1 in 7,188. My 6-year-old niece has about 1 in 460,000 odds and my 94-year-old grandfather has about 1 in 155 odds.

My IFR estimates above may be overly optimistic (hey, I’m an optimist) but if you want to assume a more pessimistic IFR, you can see how that impacts the odds as well. I should also note that many other risk factors exist beyond age, including smoking and pre-existing medical conditions, so your actual risk could be higher or lower than these estimates.

The purpose of this post isn’t to belittle the threat that coronavirus poses to our society, nor the pain of its victims and families, nor the heroic actions of American healthcare workers who are battling this virus on the front lines. It’s also not a statement on how the virus should be addressed at a policy level, nor the moral aspects of viral spread. These are all important things but beyond the scope of this post.

My sole intention is to explain the difference between CFR and IFR, and to help individuals estimate their personal fatality risk mathematically rather than emotionally.