Thread by @ElGee35

Quick thread on the fatality rate (IFR) of COVID-19 and why it continues to be a misleading indicator of impact. Key points:

-We have no cognitive intuition of what a "bad" IFR is
-We anchor to the flu "IFR"
-This creates blindness to (1) spread density (2) non-fatal damage 1/12

As discussed on yesterday’s show about long-term health, there are multiple layers of impact to assess:

-fatalities
-health of survivors
-psychological damage (e.g. health care workers, family)
-economic disruption

Let's start with fatality 2/12

podcasts.apple.com/us/podcast/lon…

The exact fatality rate is unknown. Many have estimated a range...But what's a "low" fatality rate for an infectious disease?

It depends!

If you want to predict population deaths, you need to know
-immunity rates
-speed of transmission
-effect of medical care 3/12

The average person has no reference point for a "low" *infectious* disease IFR.

We *do* have an intuition for how dangerous the flu is. We've had it. We don't know young people who died from it.

And we all know the estimated flu fatality rate by now: 0.1% 4/12

So our minds "anchor" to that 0.1% IFR and start to judge fatality from that intuition.

Under 0.1% is "weaker" than flu.
Over 0.1% is "stronger" than flu.

Since flu is so non-threatening, we think that rate has to be MUCH higher (5x? 10x? 20x?) for COVID to be a problem. 5/12

But remember, population impact isn't ONLY determined by IFR.

Unlike the flu, we have no prior immunity to COVID (that we know of). It spreads faster. So the per capita fatalities will be different, even with the *same* IFR! 6/12

Based on differences in transmission & population immunity, a 0.1% IFR for COVID would yield about 7-10 times more deaths per capita than the flu.

Again: The SAME IFR could be about 10x deadlier in the population without interventions. 7/12

7-10x deadlier might not sound too bad to some, but there's no time frame included. More infections in a shorter period of time (temporal density) means:

-Your entire office can be extremely sick at once
-Your local hospital can be overwhelmed

Not typical from the flu. 8/12

Crossing these thresholds that will temporarily cripple businesses and drown hospitals will, in turn:

-drive up fatality rates from COVID *and* from non-COVID conditions
-worsen non-fatal outcomes of COVID and non-COVID conditions.

(Depending on effect of medical care.) 9/12

These kinds of effects aren't captured by IFR. They can be seen in numbers like all-cause mortality, hospital admissions, etc.

So by focusing on IFR -- and being anchored to the flu number of 0.1% -- we miss the fatal and non-fatal population impact. 10/12

This cognitive effect also explains how mass graves can pop up & COVID can be the leading cause of death in April while seeing estimates of IFRs that "feel low."

Flu/pneumonia is the 8th-leading cause of death already. Something 12x deadlier would be the leading cause.
11/12

PS: There are other nuances to flu data that we lack intuition for: eg Reported flu deaths are a fraction of estimated flu deaths. (Estimating flu IFR is really tricky.)

But this thread is about cognitive focus on IFR and anchoring to 0.1% as an intuitive reference. 12/12

#epitwitter or medical experts, please amend/correct any issues that need it. Thanks! @EpiEllie @maiamajumder @CT_Bergstrom @mebottazzi @nataliexdean @trvrb