Wes Pegden Profile picture
13 Sep, 19 tweets, 5 min read
This commentary on herd immunity thresholds is making the rounds today. There is nothing new in this article, and parts of its commentary are quite misleading. It is worth thinking about why a commentary like this is widely shared by scientists.

I worry that the ways in which the article is misleading is, in fact, precisely why it is widely shared.

E.g., they sow doubt for infection controlled immunity, suggesting that "several rounds of re-infection" might be needed to generate robust immune responses. 2/
This is not a reasonable level of uncertainty to apply in an article that also discusses vaccine-induced herd immunity as an inevitable eventuality.

Although they explain some of the mechanisms which can lower HIT's, they also say misleading things, e.g. that HITs are only... 3/
lowered if we expect that "it is always the same set of individuals that are potential super-spreaders." It is possible to stretch these statements into truths—in reality, heterogeneous contact models assume there is some durability in the *propensity* of people to interact,...
something which is very hard to question with a straight face—but phrasings are chosen here which, at the very least, are easily subject to misinterpretations by anyone who might have something to learn from this article.

Most problematic is that the authors suggest there's..5/
"little evidence" HITs for COVID-19 might be below 50%. It's particularly inappropriate to suggest this while failing to mention the only high-profile work I am aware of which has attempted to actually make any HIT predictions for particular countries.6/

We have a situation where the only people who have tried to predict HITs from the data have concluded that they are much less than 50% for several European countries, and then other people ignore this work, offer no competing predictions, and say we lack evidence.

... the question of what HITs we should expect for COVID-19, the article also adopts a very misleading approach to analyzing the mortality which would result from immunity levels significant enough to control the epidemic. They ignore the significance of age in both... 8/11
transmission patterns and mortality, use a crude IFR range of .3%-1.3%, and suggest that herd immunity would inevitably (in their most optimistic scenarios, where we don't usually need multiple rounds of infection for robust immunity!) result 500k-2.1 million deaths.

that it would seem reasonable for a reader to conclude that these authors believe that having 500k or fewer fatalities from HI is as unlikely as having 2.1 million or more.

Producing even half-reasonable estimates like this for COVID-19 requires, at a minimum, a basic 10/11
age-stratified epidemic model. Alarmist over-estimates are not an acceptable substitute.

We should not tolerate distortions of science, even when they could be used to support our favorite policy positions.

Let me mention just one more anti-scientific aspect of this piece, which has featured prominently in many attempts to dismiss suggestions that HIT's can be low.

The authors of this article point out that there is a boat, where 70% of people became infected. 12/
In particular, they bring this up to attack what can be fairly characterized as a "strawman"; namely that 50% of humans are perfectly immune to infection by COVID-19, in the sense that they would never test PCR+ in any situation.

But the huge impact of heterogeneity on HITs..13/
does not depend this kind of "perfectly immune" population.

Biological variation in susceptibility affects HITs to the extent that the *probability* an individual is infected at a given exposure level varies at all from individual to individual. As with contact patterns, ...14/
heterogeneity in susceptibility should be our expectation; perfectly equal infection susceptibility would be a remarkable feature of the human species which would require some explanation.

Even if the strawman has no relevance, does the boat with 70%+ still tell us something?15/
Any epidemiologist who studies herd immunity thresholds would tell you that we should not expect HITs or final epidemic sizes to be the same in hyperlocal environments like boats, meatpacking plants, or nursing homes and towns, cities, and whole countries. Indeed, ... 16/
the HIT of COVID is a property not just of the infectious agent itself, but of our society and its interaction patterns; indeed, this is why it we should also not expect HITs to agree in all places and be careful about comparing, say, Manhattan with the country of Portugal 17/
(let alone comparing a country to a boat). 18/18
I agree with this suggestion; I certainly do not mean to "subtweet" anyone and would welcome a discussion with @SCauchemez and/or Fontanet about this article.

The same goes for those retweeting it, e.g. @michaelmina_lab, @AdamJKucharski, etc.


• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Wes Pegden

Wes Pegden Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @WesPegden

25 Sep
Age-targeted strategies are often interpreted in extreme ways: "cut off", "perfectly segregate", etc, some part of society, while everyone else "gets on with normal life".

Obviously in this formulation, skepticism is merited. Nothing is perfect, and "cutting" things off...1/9
sounds hard!

But this is an absolutist interpretation of ideas which should, like most, be interpreted quantitatively. The same misinterpretation could be given for all current strategies, by suggesting that masks, restrictions, etc., aim to "perfectly" stop all transmission. 2/
In fact, the goal of current strategies is to have a quantitative effect on transmission, reducing it for all groups essentially "as much as the market will bear".

In practice, this has involved implementing policies which begin harshly (lockdown), and gradually relax. 3/
Read 9 tweets
20 Sep
A quick note on this thread by @joel_c_miller before bedtime.

Joel is summarizing a nice argument coming from random graphs to suggest that increasing transmission rates among low-risk groups cannot be good, unless accompanied by other decreases. 🧵 1/9

This argument is valid if comparing two scenarios with constant transmissions. It is not valid if we expect (as I think we do!) that transmission patterns will eventually increase.

In particular, it is worth noting, that... 2/9
even in the simplest single-population models with time-varying transmission rates, epidemic sizes (and thus mortality) can be decreased by increases in transmission.[🤯]

Time dynamics make coupled systems complex, and intuitive reasoning about the effects of changes is tricky.
Read 12 tweets
17 Sep
In our paper w/ @ChikinaLab we
1) account for preferential mixing in age groups
2) do not require perfect isolation of at risk groups
3) do not assume that we can "turn off transmission" (indeed, we assume transmission gradually reverts to normal levels)1/
Joel is also concerned that aged-care facilities could support an epidemic (not localized outbreaks at individual centers) on their own without any involvement of younger individuals (staff/doctors/etc). 2/8

An earlier discussion on this ended abruptly:
I think its fair to say this is an unusual viewpoint.

Threads like this which dismiss age-targeted strategies out of hand inevitably work with the most simplistic and absolutist version of an idea. I have not seen an argument that we cannot have a quantitative... 3/8
Read 8 tweets
24 Aug
In light of the most credible reinfection case to date (👇) there has been discussion of what this means for herd immunity from infection or vaccines.

Quick thread with a simplified explanation of the basic concept of population immunity.🧵 1/8

What does it take for an epidemic to die out, so that a people are no longer being infected in significant numbers?

When first confronting this question, it might seem like the answer is that nearly everyone has to be protected from infection, so that any infected person is...2/
unlikely to infect anybody else.

However, it is not required that each infected person infects 0 new people for an epidemic to be unsustainable, we need only that each infected person infect less than 1 person on average, so that new infections are outnumbered by recoveries. 3/8
Read 8 tweets
18 Aug
🧵I talk a lot about the mistake people make thinking it is always safer to err on the side of extreme action.

Here is a simple thought experiment to illustrate this point.

Imagine you are given a magic wand which, if you wave it, will reduce COVID transmission by... 1/11
90% for 2 weeks. The magic wand has no side effects, and it can only be used today.

Should you wave it?

It doesn't sound like a hard question. The magic wand only confers benefits, it seems, and no costs. What could be the problem?

Surprisingly, in some situations... 2/11
waving the magic wand can result in more COVID mortality and morbidity than not waving it.

The reason is that even in the simplest epidemic models, where all individuals are identical, the final size of an epidemic is not monotone with respect to a time-varying pattern... 3/11
Read 11 tweets
15 Aug
Is there a seasonal component to the COVID-19 mortality rate?

A thread with some figures with @ChikinaLab on what seems to be a question of underestimated importance. 🧵 1/8

Slightly longer version here: math.cmu.edu/~wes/aus.html

Much of the coverage of summer second waves has noticed the reassuring feature that the mortality rate seems to be lower. Possible theories explaining this include different age-distributions of cases, or that the effect is actually an illusion from more testing. 2/
Another explanation for the effect could be that there is a seasonal effect on mortality rates. There is some precedent for the plausibility of this (👇) and it's important to note that it is unclear what we should expect to be driving the effect... 3/

Read 11 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!