The mystery of Manaus

 The mystery of Manaus

Those of you who expected our troubles to somehow magically melt away on December 31^st 2020, welcome back to reality! But take heart in knowing that you weren’t alone: it must have been one of the most widely held irrational expectations of our lifetimes. So, your disappointment is shared by many, many normal people around the world.

 

Anyway, there’s a lot of bad news floating around for this time of the year. There is one particular piece of information that has caught the attention of both academics and the lay press alike: the finding that (apparently), even when over 75% of people in the city of Manaus in Brazil had already been infected with the COVID-19 virus, people were continuing to die in large numbers. People are alarmed, and respectable publications are suggesting that “herd immunity may not be possible”! (https://www.newscientist.com/article/2265696-covid-19-outbreak-in-manaus-suggests-herd-immunity-may-not-be-possible/) Academics have rushed to offer explanations as to why this is happening. (https://doi.org/10.1016/S0140-6736(21)00183-5) The fear of course is that if herd immunity doesn’t hold as a concept, vaccines, which work by inducing herd immunity, may be useless! And to add to this, there is the spectre of virus variants which may wreak more havoc. But I think (it’s just me now, but I am sure more will come around sooner or later) people are getting ahead of themselves here. So dear reader, I am going to try and paint a less bleak picture for you.

 

Making sense of Manaus

The whole thing started with the publication of a study titled “COVID-19 herd-immunity in the Brazilian Amazon” on a preprint server in September last year. (https://www.medrxiv.org/content/10.1101/2020.09.16.20194787v1) The objective of their study then appeared to have been to determine the herd immunity threshold for the COVID-19 virus (the proportion of the population that needs to be infected/immune, for the epidemic to be controlled). The authors estimated that COVID-19 seroprevalence in the city of Manaus reached 66% in June 2020 (meaning that two-third of the population had already been infected and therefore were immune). They then concluded that the “final epidemic size” was 66% (implying that this was the herd immunity threshold, corresponding to an R₀ of 3). So far, more or less consistent with what we knew at the beginning of the pandemic. (http://randomramblings2018.blogspot.com/2020/03/irrational-optimism-in-time-of-covid-19.html)

 

But they presumably continued to collect more data, and found that by their calculations, 75% of the city’s population was infected by October. By the time they got their findings published in a proper journal, the epidemic in Manaus had taken a turn for the worse. It seemed that herd immunity was not achieved even when 75% of the population was infected! And they had to remove any reference to herd immunity from the title of their paper! (https://science.sciencemag.org/content/371/6526/288)

 

For us to make sense of this study, we need to understand how population seroprevalence is estimated. Ideally of course we should check the blood of everyone in the population for antibodies to the virus. But since this is never possible, epidemiologists collect samples from a randomly identified group of a manageable number of people in the community. The idea is to choose a sample that is very similar in all respects to the overall population. Unfortunately, these authors didn’t take this approach. They chose instead to test blood from voluntary blood donors, and extrapolate using statistical approaches, to estimate the population seroprevalence. Now, blood donors are mostly young and male, and (therefore, you might add!) can hardly be considered representative of an entire city’s population. But what has gone largely unnoticed, are the findings of another study from Brazil (also published around the same time), where researchers estimated seroprevalence from samples collected in a random fashion from the community. This study also reported figures for population seroprevalence in Manaus during two time periods. (https://doi.org/10.1016/S2214-109X(20)30387-9) For comparable periods, the population seroprevalence in this study was just one-third and one-fourth of that inferred from the blood donor study! For example, in June the seroprevalence was estimated to be just about 15%, whereas the blood donor study reported a prevalence of 65%! In all fairness, during normal times, people in general, and journal editors in particular, would have been very sceptical about these numbers. But as we have seen earlier, it’s not been a stellar year for science. (http://randomramblings2018.blogspot.com/2021/01/2020-and-rise-of-half-baked-science.html) But I do hope that people will soon recognize these large discrepancies in the data and come to their senses.

 

More prudence, less panic

Some people (who get their kicks out of classifying and naming all manner of things) would have us believe that how each one of us reacts to information during the pandemic, depends on the dozen or so “COVID-19 personality types” that we may belong to. (https://doi.org/10.1057/s41599-020-00679-5) In the spirit of the (not so) New Year, I shall indulge these folks. Our scientists it seems, have succumbed to merely being Worriers (“who stay informed and safe to manage their uncertainty and viral-induced fear”) and Realists (“who recognise the reality of its harm and adjust their behaviours”). I realize that this may be a good thing when advising governments so that our leaders don’t become Deniers (“who downplay the viral threat, promoting business as usual”) and put lives in danger. But we should also want our scientists to rise above these pedestrian personality types, and get back to exhibiting prudence and pragmatism. Unfortunately, maybe because of the urgency of the situation, these have been in short supply. But dear reader, be assured that efficacious vaccines (a triumph of science in 2020) can induce herd immunity, and rein in the pandemic.