I've been staying off social media, but I stumbled across this & it made me physically ill. I'm sorry; I don't know how to let it go by w/o comment.

TL; DR - This round-up is not what engagement with science looks like. (1/x) https://twitter.com/jwarnerwallace/status/1254455250909618177

It is certainly possible to have a discussion about uncertainty in both estimate of statistics and epidemic modeling. Was there a lot of uncertainty in the CFR estimate of 3.4 from early in the outbreak? Absolutely. That was not secret at the time. (2/x)

I had conversations in January about possible sources of statistical bias in that estimate, such as overrepresentation of severe cases. Whether that uncertainty was widely understood in the general public is a valid question, but let's not pretend the experts were clueless. (3/x)

There have been subsequent lower estimates from larger data sets. There have also been higher estimates, such as those coming out of Italy. There is still uncertainty about the true CFR, and how it varies within different subgroups and risk categories. (4/x)

All told, our current understanding of the CFR for COVID-19 still does not put it on part with the flu. 0.7%, the lowest figure cited in that roundup, is still 7x higher than the 0.1% figure often mentioned for flu. (5/x)

And that is fatality among cases. Every year, many people do not get the flu at all because of their own immunity and/or that of those around them. No one has preexisting immunity to this virus, so more people are at risk of becoming cases in the first place. (6/x)

As for the commentary on epidemic modeling, yes it is possible to find examples of predictions which were not accurate. But any real discussion of such issues needs to dig into those predictions and the underlying models and find out why they were wrong. (7/x)

For starters, did the model assume no interventions to limit spread? Are you comparing those predictions to post-intervention outcomes? Keep in mind that the mere act of publishing the initial model predictions is a form of intervention, as it can modify behavior. (8/x)

We can also explore whether the predictions were based on uncertain estimates from limited data of key parameters, whether the model neglected critical features of transmission, and whether the cited prediction was a known outlier even at the time. (9/x)

Similarly, if you are going to mention that COVID-19 cases have peaked (where? Different locations are at different points in their outbreaks), it is relevant to discuss the impact our interventions have had to cause the decline. (10/x)

As for the obesity commentary, I don't know what else to say other than just because obesity is a risk factor doesn't mean you get to reduce all related suffering & death to a one-liner to go out on. (11/x)

In short, for all the hand wringing about the reputation of science in this roundup, I don't see how it is helping. (12/x)

So, @jwarnerwallace, you follow me. I have a PhD in the biology of infectious diseases and have been in & around public health for 20 years. If you want to have a conversation about the science of our present moment, I'm available. (13/13)