Thread by @phl43

As I have argued before (cspicenter.org/blog/waronscie…), this kind of chart is totally fake, because the methods used to estimate the basic reproduction number of SARS-CoV-2's variants are unreliable, but it's worth pointing out that basically everything else on that chart is fake. 1/n

We don't really know what the basic reproduction number of a virus is, and in fact the question doesn't even make sense, because the basic reproduction number of a virus is not an intrinsic property but depends on the context such as population density, culture, etc. 2/n

For instance, take measles, which is often presented as the most transmissible infectious disease with a R0 of 15 as in the chart above. But where does that number come from? 3/n

Here is a chart that shows the distribution of estimates in 18 studies. As you can see, not only are the estimates all over the place, but they don't even cluster around the range of values that are usually cited as "the" R0 of measles. 4/n thelancet.com/journals/lanin…

Even if you look at estimates that were based on data from the same type of source, region, period, etc. the range of published estimates is huge and the median estimate is often quite different from the commonly cited value of 15. 5/n

It's also amazing to read papers estimating the basic reproduction number of the 1918 strain of H1N1, which caused the Spanish flu pandemic, casually assuming that you can use estimates of the generation time of current strains for that. 6/n

But as we have seen with SARS-CoV-2, different strains of the same virus can have very different generation times. And as I pointed out in the past (cspicenter.org/blog/waronscie…), estimates of the reproduction number are very sensitive to the assumptions about the generation time. 7/n

And I was just looking at how sensitive they are to the *mean* of the generation time distribution, but actually they are also pretty sensitive to its *shape* (royalsocietypublishing.org/doi/10.1098/rs…), so the sensitivity analyses I did in that post actually understate the uncertainty. 8/n

It's the same thing with the infection fatality rate. Not only is it no more an intrinsic characteristic of pathogens than the basic reproduction number, despite what the kind of chart above suggests, but estimates are often all over the place and extremely unreliable. 9/n

For instance, the infection fatality rates of MERS-CoV and SARS-CoV-1 are probably vastly overestimated, because we miss a lot of infections that are either asymptomatic or mild enough not to be detected. 10/n

People are probably going to defend this type of charts by saying they are useful to give people a rough picture, yada yada yada, but honestly I don't really see what good they do concretely, while I see very clearly how they mislead people 🤷‍♂️ 11/11

P. S. The NYT chart at the top of this thread is actually better than most in that at least it makes clear that there is a lot of uncertainty about SARS-CoV-2, but no virus in that chart should be represented as a *point*.

They should all be associated to areas and, although SARS-CoV-2 is more recent, there is actually more uncertainty about the properties of most of those viruses than about those of SARS-CoV-2 if you ask me, because we have far more, higher-quality data about it than about them.