Dr Fauci is one of the national leaders in the pandemic, but occasionally he says things that I have really disagreed with. One of these times is his recent statements suggesting we need to get up to 90% of the country vaccinated before we reach herd immunity before backing off to 70-85%, which I still think are far too high based on best current knowledge.
I appreciate his efforts to try to urge all Americans to take the vaccine once available, but the numbers he quotes simply do not seem to follow the current scientific knowledge. Perhaps he is trying to exaggerate the number in an attempt to convince more to consider taking the vaccine, but if that is the case I disagree and think it is better to try to transparently explain the current state of knowledge as completely as possible to the people.
In a blog post back In September (look at later part of blog post discussing R0 and herd immunity), I described several articles discussing herd immunity, whose main points were that
Based on the estimated R0 rate of 2.5-3.0 for SARS-CoV-2, the estimated herd immunity threshold is 60-67% assuming equal susceptibility and exposure across society.
Mathematical models show that this threshold comes down in communities with variability in exposure and susceptibility, which describes our current reality better than equal assumptions.
Herd immunity is reached by a combination of vaccination and natural infection, so the proportion of infected and recovered individuals with strong immune memory reduce the proportion needing vaccination.
Some other facts I support on a recent blog post urging careful thought in vaccination distribution:
in spite of claims in media reports that only 15% or less of USA has been infected, the preponderance of evidence is that at least 20%, likely >25%, and possibly >30% have been exposed, and this number is sharply increasing during this winter surge with unprecedented infection levels across much of the USA.
in spite of an inordinate focus on anecdotal evidence of reinfections, these are many orders of magnitude less than what one would expect if most lost immunity at 4, 5, or 6 months, with at least 130,000, 50,000, or 30,000 reinfections expected in those cases but only dozens to hundreds observed.
This fact, along with the growing literature on lasting antibody levels, B-cell memory to produce new antibodies, and T-cell memory to produce rapid immune response to reexposure, mean that there is considerable evidence that a vast majority of recovered individuals maintain vigorous immune response for >6 months, and leading many immunologists to believe it is likely >12 months, which is what would be required for annual vaccination regimens.
The two blog posts linked above make the case for these points, making arguments based on certain clearly stated assumptions that I support with cited scientific literature and data, and making quantitative arguments that I try to lay out in a clear, concise, and transparent manner. I leave these detailed arguments in the blog posts above and don't repeat them here to keep this post more concise -- if you are skeptical about these points (which you probably should be), I encourage you to check the past blog posts and evaluate if you agree with my logic or not.
Based on these facts, it seems clear that the proportion vaccinated to reach herd immunity is far less than the 90% or even 70-85% claimed by Dr. Fauci. If we conservatively assume equal susceptibility and exposure and just 20% infected, recovered and still immune in 2021, with a vaccine with 90% efficacy, we would likely only need 45-50% vaccination until herd immunity was reached (assuming initial vaccine distribution is focused on those without existing immunity), and if 30% have existing immunity, this number drops to 35-40%, and if we account for heterogeneity of exposure and susceptibility and are smart about prioritizing those with higher susceptibility and propensity for exposure, then this number could decrease even further. It is plausible that we will see substantial reduction of transmission and move towards herd immunity once we vaccinate 80-100 million Americans or so.
Given the uncertainty of durability of immunity after recovery or vaccination, the uncertain efficacy of vaccines in preventing asymptomatic infectious disease, and the potential for some new variants to emerge and resist vaccination, it is wise to urge as many as possible to become vaccinated.
But to claim herd immunity won't be reached until 70%-90% are vaccinated does not seem accurate based on our current best knowledge, and produces an unnecessarily pessimistic viewpoint of our time until the pandemic is brought under control.
I'm still trying to understand where he gets this number from. Maybe there is a projection that R0 has increased from 2.5-3.0 based on the emergence of a new variant in the UK that appears to spread faster -- this bears watching -- but projecting a much higher R0 based on the data presented to date is a big leap since (1) I have not seen convincing evidence delineating which proportion of the fast spread is causally linked to mutations of that variant and what proportion is related to the environmental factors that are producing rapidly accelerating transmission all over the world right now, and (2) assuming that this variant will take over the pandemic is at best a projection right now - it is possible but too soon to tell.
Also, it is true that if vaccination prioritization ignores previous infection, this would delay herd immunity further, especially if supply and distribution challenges remain problems well into 2021. However, even if so, by the time we get to half vaccinated, half of those who were infected and recovered would still reduce the remaining susceptible population and bring the required number down, unless one assumes recovered people have no lasting immunity, which is a point that seems popular (perhaps to counter the misguided Great Barrington Declaration?) even though it is not supported by the accruing scientific evidence.
With due respect to his expertise and position, I would simply argue that basic quantitative reasoning based on the best current knowledge of SARS-CoV-2 based on the scientific literature to date suggests the threshold is lower, maybe considerably so.
The potential of us moving out of epidemic spread phase of the pandemic by the summer should energize us to remain vigilant as we see the light at the end of the tunnel, while an overly pessimistic viewpoint, if not accurate, may backfire and cause more to lose hope and stop doing the things we need to do to suppress spread until the vaccines take us to herd immunity. Transparent communication of emerging truth is in the long run the best policy for informing the public about the pandemic.
UPDATE 1/5/21: Looking at this blog post after 1+ weeks, I realize I came on too strong at Fauci's claims. While I think the substantive points made in the post above are valid, I reacted too strongly based on my perception and judgment that he was completely dismissing any natural immunity from previously infected and recovered individuals (perhaps motivated by opposition to misguided ideas like the Great Barrington Declaration) and that he might be purposefully overstating the threshold in an attempt to persuade more people to consider taking the vaccine, which flies in the face of my philosophy that full, objective disclosure of the best state of knowledge is the best strategy, independent of what falsehoods one fears it might reinforce.
But two key factors I did not account for enough that might raise the threshold substantially:
While I think it is a mistake to dismiss the contribution of natural immunity after recovery in the calculation, it is also possible to overplay its effect which I have done in my arguments above. While I stand by my claim that there is strong evidence for durable natural immunity for many infected and recovered, there is also good evidence and expert knowledge that asymptomatic cases tend to have less durable immunity than more serious cases, as reinforced by this recent Nature paper. Given that the substantial number of infections that are asymptomatic or only showing mild symptoms, it is not clear whether the natural immunity from these individuals is enough to remove them from the susceptible population in 2021, especially those who were infected earlier in the pandemic. Thus, the % infected and recovered clearly needs to be discounted when trying to compute herd immunity thresholds -- while they should not be completely discounted, but should be substantially downweighted. My discussion above downplays the uncertainty of durability of immunity too strongly.
It is starting to look like this new UK variant may really be more transmissible than the SARS-CoV-2 we have experienced thus far in the pandemic. Last week, I was extremely skeptical about this, thinking it likely that the claims were based on non-causal analyses that failed to account for the confounding environmental and seasonal factors contributing to increased transmission all over the world, independent of any new variant. However, more analyses have come out since then that demonstrate ~50% greater transmission while adjusting for these factors, and given that this variant went from 5-10% prevalence in London and southeast and eastern England in early November up to 80% by mid-December, it is becoming difficult to find any other plausible explanation for this other than increased transmissibility afforded by its characteristic mutations and the conclusion that this variant will soon displace others as the dominant variant around the world. I am working on a longer blog post investigating these issues and what the available data suggest. But if indeed transmission is increased by 50%, this does raise the R0 to the 3.5-5.0 range that would suggest higher herd immunity thresholds than the 60-70% corresponding to an R0 of 2.5-3.0 that I assume in my arguments above.
These two factors could raise the herd immunity threshold considerably, and in light of that Fauci's revised claim of 70-80% vaccination necessary for herd immunity make more sense to me -- I should not have come down so hard on it. It still could be quite a bit lower depending on the degree of natural immunity from recovered individuals and the effect of heterogeneous susceptibility and exposure, but not with the certainty with which I implied in the original post above.
It may be a while before we reach full herd immunity and end the epidemic growth phase completely, but one positive mathematical principle to keep in mind is that the herd immunity threshold is more of a continuum, in that the removal of susceptible individuals from the population as we move towards herd immunity suppresses the transmission rates substantially. Thus, even if herd immunity would not be fully reached until we reach somewhere near 2/3-3/4 vaccinated, we should see sharp reductions in the community spread rates by the time we reach the halfway point towards that ultimate threshold, which is a realistic goal by late Spring and Summer of 2021 if we can work out the distribution and supply issues. Achievement of the full herd immunity threshold will depend on broader public acceptance of the vaccine, which hopefully will greatly increase by the summertime if observational data suggesting vaccine safety after hundreds of millions of vaccinations is sufficient to convince many of those who are currently hesitant.
Agreed that the calculation of the estimate of +0.4 in R for the variant is subject to some confounding. On the other hand, the United States is undersampling by a lot and has no plan or campaign for stratified sampling with randomization. So suggesting that existing counts of deaths/proportions of positive cases in the United States give an actual picture of the pandemic is too strong.
True but that would lead me to caution that the infection rate could be higher not automatically presume it. And it is not clear whether there is really evidence of the causative effect of the characteristic mutations on infectiousness, so I think questions remain about whether this variant is indeed more prone to spread
Estimates are that R is increased by 0.4 for the new variant. Less specifically, there is also concern that at least in "island pockets" (e.g., sets of immunocompromised people) mutation rates can run much higher than the 1-2 surviving mutations per month we have seen thus far.
For the United States I'd feel better about your estimates if we had a well running and systematic sequencing campaign and robust national test and trace. However, instead, if there were a new variant which had a higher R, we would not know until a country that has such a sequencing campaign detected it.