Armchair epidemiology—or why I’m a better runner than Eliud Kipchoge

Over the last week or two, pre-emptive postmortems on the continuing worldwide shutdowns due to Covid-19 have begun to metastasize. Their authors stand in judgment over these shutdowns, claiming to present incontrovertible facts that prove that the shutdowns were unnecessary.

These postmortems get published in widely-read, respected periodicals like The Telegraph, The Hill, and even the Wall Street Journal. The way they’re written, they sound like so much Serious Thought.

Today, I want to suggest that they’re the opposite. They’re rife with serious epidemiologic errors, and—perhaps more damningly—they conceal insidious privilege and a touch of implicit bias.

Buckle in. This is long. There are even footnotes.

First, let’s talk about running. (I promise this will become relevant).

I’m not a runner. I hate running. Running is what Dante forgot when he was making up punishments in the Inferno.

Since gyms are (appropriately) closed here in Massachusetts, however, I’ve taken to running around the city like a madman with a mask on. It feels even worse than I’d ever imagined.

Earlier this week, I ran hill sprints, which I wanted to do about as much as I want to see the Lion King remake. Hill sprints mean picking a hill, running up it as fast as you can, and then walking back down hoping you don’t throw up on the way. Here’s how it went:

Suck it, Kipchoge

Look at that best pace! That’s faster than usual for me, and I felt really good about it. I mean, the unofficial marathon world record holder, Eliud Kipchoge, runs only a 4:34-minute mile.

Obviously, I’m a better runner than Kipchoge. Right?

Hold that thought.

All the postmortems make similar arguments against shutdowns, so let’s just focus on this one, published in The Hill. Its author, a neuroradiologist and fellow at Stanford’s Hoover Institution, produces five “facts” about the coronavirus and the US response to it.

He sounds like he knows what he’s talking about, too. After all, his Serious Thoughts are backed by a medical degree. Unfortunately, this pandemic has created a cadre of Armchair Epidemiologists — people with just enough understanding of public health to be dangerous — and I’m worried our author might be one.

On to the “facts”.

“Fact” 1: The overwhelming majority of people do not have any significant risk of dying from Covid-19.

Oh boy, where to start with this one? This is just horribly misleading. It’s correct—but only in the most superficial way possible.

First, let’s examine the supporting data the Armchair Epidemiologist supplies:

The recent Stanford University antibody study now estimates that the fatality rate if infected is likely 0.1 to 0.2 percent, a risk far lower than previous World Health Organization estimates that were 20 to 30 times higher and that motivated isolation policies.

“WE’VE BEEN OVERREACTING!” armchair epidemiologists all around the nation concluded after the Stanford study was issued. They said the same thing when New York State released its seroprevalence numbers.

“This infection is no worse than the flu!

—Problem #1—

The fundamental issue here is that the Armchair Epidemiologist confuses infection fatality rate (IFR) with case fatality rate (CFR). This may seem arcane, but bear with me for a second.

When the New York seroprevalence study was reported, there were 15,500 deaths in the state, out of a total of 263,500 confirmed Covid cases. These are the numbers behind the official mortality estimate of 5.9 percent. [1]

More specifically, this is the case fatality rate:

CFR = Number of deaths ÷ Number of confirmed cases

Then, New York performed a non-random sampling of asymptomatic New Yorkers and found that almost 14 percent of them had antibodies for the virus that causes Covid. [2]

“Well, would you look at that!” armchair epidemiologists exclaimed. “Fourteen percent of New York is 2.7 million people, not 263,500! That means the virus only kills 0.6 percent of the people it infects!”

Hoping you wouldn’t notice, they subtly swapped denominators. They moved from talking about the case fatality rate to talking about the infection fatality rate:

IFR = Number of deaths ÷ Number of infections

When our Armchair Epidemiologist concludes that the virus only kills 0.1 percent of people, he’s calculating an IFR.

But so what? If flu “only kills 0.1%” also, then isn’t the Armchair Epidemiologist right? Aren’t these two viruses the same? Haven’t we been over-reacting the entire time?

Don’t touch your face!

Well first of all, whatever the denominator is, it still translates 15,500 deaths. In good public-health–speak, that’s an imperial crap-ton of mourning families.

That aside: Remember my claim that I was a better runner than Eliud Kipchoge?

It’s patently ridiculous. You know it. I know it.

But why it’s patently ridiculous is key to understanding our Armchair Epidemiologist’s error. See, that 4:23 pace my watch logged is an instantaneous pace. And it’s vanishingly easy to sustain a 4:23 pace over just a couple of meters.

Kipchoge sustained his over 26.2 miles! If you really want to decide who’s the better runner (it’s Kipchoge), you need to compare me to him in the same race. You need to compare us across the same denominator.

The Armchair Epidemiologists doesn’t. He compares Covid’s IFR to influenza’s CFR.

That’s like calling Blake Bortles the best quarterback in the entire NFL but only comparing him to Ryan Leaf.

Or it’s like saying I’m a better runner than Kipchoge.

In either correct comparison—IFR for IFR, or CFR for CFR—Covid is still far deadlier than the flu.

—Problem #2—

There’s more than just denominatorial trickery going on here. The Armchair Epidemiologist hopes you don’t notice that he’s played fast and loose with the numerator too.

See, to go from the reported CFR to his IFR—from a mortality of 5.9 percent to 0.6 percent in New York—requires a pretty hefty assumption: that we’re perfectly capturing the number of deaths.

We aren’t. Not by a long shot. We’re undercounting Covid deaths.

Source: New York Times

In the IFR equation:

IFR = Number of deaths ÷ Number of infections

the Armchair Epidemiologist would like you to believe that the denominator has gone up, while the numerator has stayed the same.

In fact, both numbers are likely to be higher than the original reports. So even if we (erroneously) compare Covid’s IFR to influenza’s CFR, Covid might still win. We can’t say yet.

—Problem #3—

None of this matters anyway! The Armchair Epidemiologist misses the point altogether. The whole reason we locked down wasn’t because of the IFR. It was because of the CFR.

Think of the virus like a car. Every new car has its fuel efficiency report plastered to its window. But once you take the car home, its actual fuel efficiency will depend on how you drive it. Do you abruptly speed up and slow down? Do you only drive on gravel roads? Do you stay in second gear for as long as possible? Well, you’ll never get a fuel efficiency like the manufacturer reports.

The efficiency numbers in the window are a measure of the car’s inherent capabilities. What determines how often you fill up gas, however, isn’t those numbers. It’s the actual fuel efficiency—in your hands, on your roads, with your driving style.

And that’s what matters to your pocketbook.

The IFR is the manufacturer’s idealized number. The CFR is what happens when the car leaves the lot.

See, no virus exists in isolation. It acts in an interplay with the health system. That’s what determines how many people actually die. Not the idealized IFR. It doesn’t actually matter how many hidden infections there are. [3]

What matters is how many people will use the health system, how overworked will it get, and can it cope? Evidence from country after country has shown that this virus overwhelms health systems, that they can’t cope, and that, when that happens, mortality—actual mortality—skyrockets:

CFR over time for countries with and without overwhelmed health systems. Source: Our World in Data

Compare the CFR over time in countries where the pandemic never got out of hand (South Korea, Taiwan), with countries where it did (Italy, Spain). The IFR for the virus should be the same everywhere. But that’s not why there are mass graves on Hart Island in New York.

“Fact” 2: Protecting older, at-risk people eliminates hospital overcrowding.

Boy, that sounds lovely. Let’s start with his supporting data again:

In New York City, an epicenter of the pandemic with more than one-third of all U.S. deaths, the rate of death for people 18 to 45 years old is 0.01 percent…

Young adults and children in normal health have almost no risk of any serious illness from COVID-19.

I’m sure the children with multisystem inflammatory syndrome likely due to Covid would beg to differ. Oh yeah, and they’re from New York City, where the Armchair Epidemiologist has claimed that children have “almost no risk of serious illness.”

While it’s true that children do better—thank God!—the Armchair Epidemiologist plays fast and loose with this fact. He concludes that, since Covid mortality is lower in children and healthy young adults (which it is), then they’re safe (which they’re not).

He concludes that, since the majority of hospitalized patients are older (which they are), then younger, healthier people should be able to go about their daily business while we “protect” those most at risk.

(Conveniently, he doesn’t tell us how to protect them, but we’ll get back to that in a second.)

The Armchair Epidemiologist seems to have forgotten that this virus spreads before it’s symptomatic. Older and higher-risk people can get it from their younger, lower-risk sons, daughters, cousins, and friends—and no one would be any wiser.

Right now, there’s no way to protect at-risk people without a lock-down. Temperature checks don’t help if the virus spreads for 5 days before someone gets a fever. Neither do symptom questionnaires.

Widespread, regular testing of asymptomatic people would help, but can you imagine an armed REOPEN protestor willingly submitting to a government-mandated nasopharyngeal swab?

It’s pretty terrible:

“Fact” 3: Vital population immunity is prevented by total isolation policies, prolonging the problem.

Oh, herd immunity, that anti-lockdown golden child! Herd immunity’s gonna save us, if we could just get out of its way!

First: Herd immunity is awesome. I’ve talked about it before. It’s what keeps measles away—or, at least, it did until antivaxxers decided they’d muck things up.

The concept is pretty simple: all viruses need susceptible hosts to spread. If you can surround an infected person with enough immune people, then he can’t start a nationwide epidemic because there’s no one for the virus to jump to. Soon, the infected person fights off the virus — or he dies. Either way, the virus’s marauding days are over.

The problem is, right now everyone is susceptible to Covid, and herd immunity requires the opposite to be true:

Herd immunity requires lots of immune people

For herd immunity to work, basically almost everyone’s got to be immune.

Even if the high New York seroprevalence numbers are correct, then at least three-quarters of us are susceptible, not immune. Without a vaccine, herd immunity will either never happen, or it’ll happen at too high a cost.

  1. It’ll never happen: There’s a reason we have vaccines for things like smallpox and measles. Natural herd immunity is vanishingly rare. Even for viruses that have been around for millennia, the human race has never established herd immunity. [4]
  2. It’ll never happen, part 2: As far as we know, we’ve never developed herd immunity for other coronaviruses. Why should this one be different?
  3. It’ll happen at too high a cost: Let’s be generous. Let’s pretend CFR doesn’t matter and the mortality for the Covid virus is 0.1 percent, like the Armchair Epidemiologist claims. To establish herd immunity will mean something like 7.5 million people dying. Are we willing to sacrifice more people than all of Hong Kong, just in the hopes that we develop an immunity that the human race has never developed before? [5]

Fact 4: People are dying because other medical care is not getting done.

Ok. This one’s actually true.

“Fact” 5: We have a clearly defined population at risk who can be protected with targeted measures.

This is basically “Fact” 3, said a different way, and it’s where you really start to see the privilege and implicit bias that tinges the Armchair Epidemiologist’s Serious Thoughts.

Since he hasn’t told us how he intends to protect at-risk people, and since no serious scientist has come out with plans to do so, all we’re left with is two options: isolation or killing lots of people in the hopes that we achieve herd immunity.

Because he wants to protect at-risk people, I can’t imagine the author is saying he wants them to die, so let’s take herd immunity off the table.

What’s left—although he’s not brave enough to say it — is isolation. The only thing the Armchair Epidemiologist can mean by “protecting” high-risk people is isolating them. The people most at risk—the older and the sicker among us—they’re the ones the Armchair Epidemiologist would like to bear the economic brunt of lockdown, so the rest of us can get our hamburgers.

It gets a bit worse, though. Because what the Armchair Epidemiologist has neglected to mention is that older people aren’t the only ones at risk of dying from Covid. You know who else is dying in record numbers? People of color and people in poverty.

Let’s process that. Either the Armchair Epidemiologist is willing to lose a whole bunch of older people, poorer people, and people of color in the hopes that the rest of us will develop herd immunity—

Or he wants to isolate (sorry, “protect”) them, pushing them further into poverty, so he can get his hamburger.

These would be huge sacrifices others would have to make, but the Armchair Epidemiologist is ok with that.

Many of us warned that armchair epidemiologists would do this, way back at the start of this pandemic.

We knew that these sorts of articles would be published once the nation approached blue:

Welcome to blue (credit unknown—if you know, please reach out!)

We warned that people would start writing Serious Thoughts, claiming that there simply hadn’t been enough dead people to warrant shutting down their golf. (Aside: how many dead people would be enough?). We warned that people would forget that, even with containment measures, 73,000 people have so far died in America.

But we underestimated how unabashed they’d be. We underestimated how fearless they’d get in suggesting that others sacrifice so that they can go back to work. And we underestimated how effectively they’d hide their privilege behind Serious Thoughts and erroneous epidemiology.

And we underestimated how easily we’d fall for it.


[1] As of May 6, 2020, the case fatality rate in New York State sits at 7.8%.

[2] New York’s most recent seroprevalence study suggests an infection prevalence even higher than the original 14%.

[3] I mean, it does. To virologists. Not to pandemic planning, though.

[4] Of course, this is biased — perfectly effective herd immunity means eradication of a virus. That’s how smallpox was eradicated. It’s possible we’ve established perfectly effective herd immunity to thousands of viruses we’ve never heard of, because they no longer exist.

[5] If it’s 0.6 percent, like the initial New York numbers suggest, then that’s 45 million people, the size of Argentina.

O’Brien Chair of Global Surgery, Royal College of Surgeons in Ireland | Global surgeon | Decision analyst | Climber | 3x American Ninja Warrior Competitor

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store