July 2020

“Herd Immunity”

Excerpts taken from “Letter on Covid-19 – 2020” by David Capital Partners, LLC, Chicago

A person in blue scrubs writing on glass



STATEMENT #3. Let’s start with COVID-19. How has the pandemic peaked without a vaccine or herd immunity?

Many believe the only two paths out of the pandemic are either (1) a vaccine or (2) “herd immunity.”

We see this as a false choice. In fact, we believe the most likely outcome is a third and different path: that C19 has reached its “disease break point” in the US/Europe such that population-level spread is now in inexorable decline.

So let’s discuss how outbreaks end.

First, a definition. What is “herd immunity”? A population reaches herd immunity when a sufficient percentage of its members have specific resistance (are “immune”) such that the disease dies out. Specific resistance can be acquired either by recovering from the disease or via a vaccine.

The percentage required for herd immunity varies with a disease’s contagiousness. To measure contagiousness, epidemiologists use the mathematical term R0 (pronounced “R-naught”). R0 is the theoretical number of people who contract an illness from each infected person, assuming no existing specific resistance in a population.

For influenza (the flu), which has an R0 between 1.5 and 1.8, the threshold for herd immunity might be 45-50%. For COVID-19, which we think has an R0 of 2.5-3.0, the herd immunity threshold may be 60-65%. For measles, a highly-infectious disease with an R0 in the teens, more than 95% must have specific resistance to reach herd immunity.

One fact about herd immunity, however, is too often glossed-over: the herd immunity calculation estimates the theoretical threshold at which it is mathematically inevitable the disease will go extinct.

That is a lofty goal. So lofty, in fact, that almost no disease ever achieves it. Why? Because in practice, spread of a given disease collapses far before a population ever reaches herd immunity.

A good example is the flu. Influenza mutates easily, so each year a new vaccine (a flu shot) is required which attempts to “guess” what the flu will look like that year. Sometimes it hits. Sometimes it misses.

But a curious thing happens when the flu shot misses. That year’s strain of the flu explodes, gets a lot of people sick – and then spread of the strain collapses when around 10-15% of the population is infected. The seasonal flu never comes within spitting distance of reaching the 45-50% level required for herd immunity.

And this doesn’t apply just to modern seasonal illnesses. The 20th-century’s greatest pandemic (the Spanish Flu of 1918) probably had an R0 just above 2.0, so the herd immunity threshold was likely 55-60%. But historians estimate just 20% of people had been infected when the Spanish Flu’s spread suddenly collapsed. Philadelphia saw peak deaths in mid-October 1918, but by mid-November the disease was effectively gone from the city. Spread of the Spanish Flu peaked and plunged in weeks, without ever reaching herd immunity.

So how do we explain this?

The answer: there is not one, but two levels of population “immunity” to consider.

First, herd immunity: the level of specific resistance in a population required for a disease to fully disappear.

Second, the disease break point: the level of specific resistance in a population at which spread of a disease collapses. The disease break point is generally one-third or less the threshold required for herd immunity.

The difference between the two can be understood with basic math. Herd immunity is a theoretical calculation. It relies on mathematical assumptions. For example, the herd immunity calculation assumes homogeneity of actors and outcomes: that infections, specific resistance, social graphs, and individual susceptibility are all equally distributed. If the R0 is 1.8, each infected person transmits the disease to exactly 1.8 people. If 1% of a population has specific resistance, then 1% of each person’s social graph has specific resistance too.

The problem is obvious. These assumptions simply do not reflect the real world. Actors and outcomes are neither homogenous nor equally distributed. In fact, the opposite is true. Some people have highly-connected social graphs (once infected, they are likely to infect many more) while others have relatively few connections (and may not infect anyone else). Some sub-populations are highly vulnerable (e.g., nursing homes) while others are highly resistant (e.g., young people). Actors and outcomes are not equally distributed. They have tremendous variance.

The disease break point model uses graph theory to better explain how outbreaks evolve in practice. The model assumes actors and outcomes are not equally distributed – and in fact assumes they are concentrated in certain individuals and sub-populations. A node  with a well-connected social graph is more likely to be infected early and to transmit the disease widely. Once recovered, however, the “immune” node becomes a dead-end for future disease spread. The system spikes and then collapses far quicker than a herd immunity model (a homogenous approximation model) would predict as these “super-spreaders” become “super-suppressors.”

For COVID-19, the implications are powerful. If C19’s R0 is 2.5-3.0 and its herd immunity threshold is 60-65%, then the disease break point would be only 15-20% specific resistance (a population’s precise disease break point likely varies somewhat due to differences in susceptibility and social graphs).

Our research indicates Europe and the US reached this disease break point in March and April, respectively. We believe spread of COVID-19 in these geographies has peaked and is now in irrevocable, sustained decline.


***The links from EdViews.org are all inactive now because of the death of the webmaster. It may be possible to locate these articles by using THE WAY BACK TIME MACHINE: https://help.archive.org/help/using-the-wayback-machine/

***It is also possible to locate some of these articles by searching for them on DuckDuckGo.