One problem around the concept of “herd immunity”, and why it's so badly understood, is that there is such variability in its definition. Basically it's “the proportion of the population that needs to be immune so that the epidemic does not grow exponentially” BUT … •1/20
… nearly every single term in this definition can be interpreted in different ways, so we get a incredible gamut of different “herd immunity” thresholds. It's not just that the “1 − 1/R₀” formula is simplistic, it's that the whole concept is slippery to define! •2/20
Problem 🄰: people will adapt their behaviors to respond to the epidemic, both spontaneously (because they are afraid) and in an manner organized by the authorities. So the growth dynamic of an epidemic (reproduction number, contagiousness…) aren't constants! •3/20
So do we mean “the immune proportion such that the epidemic stops growing exponentially with the current behavior” or “…with the behavior of the status quo ante”? The former definition is fairly silly: it means that a lockdown causes herd immunity! •4/20
(Because as soon as the epidemic stops growing, with that definition, you have herd immunity. Even if 0% are immune!) But the latter is also fairly silly: the behavior of the status quo ante is ill-defined and irrelevant, why should we care about it? •5/20
I think the right definition should be with behavior changes that we can count on in the long run that do not impose undue and unacceptable burden on society (so lockdowns are out because we can't stay in lockdown indefinitely, but putting an end to shaking hands isn't … •6/20
… but of course what exactly is acceptable as a long-term change and what isn't is very much in the air: probably we can wear masks in public forever but can we keep restaurants closed forever? public transport? schools? forbid all public gatherings? hard to tell!). •7/20
So anyway, remember when you are talking about herd immunity threshold or reproduction number that you are assuming some kind of behavior of society and that this needs to be qualified. These figures are measures of society as a whole, not just of the epidemic. •8/20
And because of what I have just said, the herd immunity threshold, suitably defined, can be as low as 0% (arguably it is so in South Korea, at least if we assume they can keep their current social regimen forever). But maybe not a useful concept as such! •9/20
Problem 🄱: population is deeply heterogeneous. Reproduction and hence the herd immunity threshold are vastly different in urban areas and in rural areas, but they are also different among all sorts of different socio-economic categories. Aggregate figures mean nothing. •10/20
I mentioned this in detail in the thread threadreaderapp.com/thread/1249738… (and the various other threads it links to), so I won't get any further into this in this thread. But this explains why “1 − 1/R₀” just isn't right, even targeting R₀ for population behavior. •11/20
Problem 🄲 (related): the threshold doesn't depend just on which proportion is immune but actually WHO these people are. If people having many contacts are immune this is of much more value than others. Again, something I mentioned previously in threadreaderapp.com/thread/1241745… •12/20
The net effect of problem 🄲 is that herd immunity comes sooner when the disease itself confers immunity (because it will hit people with many contacts before the others) than when this immunity comes from a vaccine. (This doesn't mean vaccines aren't great, of course!) •13/20
Problem 🄳: “immunity” is itself a slippery word. The naïve definition assumes all-or-nothing immunity. But immunity can be partial and it can be short-lived (or medium-lived). It can depend on symptom severity. We know very little about what happens for Covid-19. •14/20
We do know that at least SOME form of immunity is TYPICALLY acquired by patients having recovered from the disease, at least for SOME time: time.com/5810454/corona… presents a good summary on this. •15/20
But there are some signs that Covid-19 immunity isn't always perfect and/or permanent. How bad is this? Nobody knows. And “how bad” doesn't have the same meaning for an epidemiologist (who cares about the overall dynamics of the population) and for individual people: … •16/20
… e.g., if, say, 10% of people don't acquire any lasting immunity, it's very bad for these 10%, but it won't change the herd immunity threshold too much because 90% of people still develop good immunity. •17/20
So when we put together problems 🄰,🄱,🄲,🄳 (and possibly others I didn't think of), “herd immunity” becomes a very fuzzy concept indeed, ranging from “already attained in many places” to “possibly impossible to reach ever”. •18/20
What is certain, however, is that the more people are immune (and it seems plausible — although even this is not certain! — that barring a vaccine they can only become so by infection) the easier it will be to deal with the epidemic. •19/20
So it's not like there's a magical threshold that is either reached or not and which changes everything: every little bit helps. But a big dichotomy (and question) is whether natural immunity is a significant part of the strategy used to fight the disease, or not! •20/20
