Thread by @BillHanage

Some recent credible studies from the UK suggest that the variant B.1.1.7 which was first detected there is not only more transmissible but associated with greater severity, measured by deaths. A thread 1/n

The studies on which this is based are summarized in this document. Notably, it is not founded on one study, but several. They incorporate different data, methods and different biases and they are all imperfect, but they point in the same direction. assets.publishing.service.gov.uk/government/upl… 2/n

The upshot is that people infected with B.1.1.7 are about one third more likely to die than similar people infected with the pre-existing viral lineages. There is a LOT of uncertainty around that number. For a lot of reasons. But some increase (even if small) seems legit 3/n

This is consistent with early statements that there was not a large change in virulence. You would not expect to detect a 30% increase from the data available early on (I seem to remember saying there might be some change but not enough to be detected by the current data) 4/n

Of course, if a large number become infected, then a 30% increase really really matters in terms of the absolute numbers of deaths. Instead of the 400k + deaths in the US, we would have 520k+ were they due to a virus like this (<-toy example please don't quote out of context) 5/n

More confusing what is meant for hospitalizations, especially because of lags in data reporting. I think we will know more soon. But I think it is reasonable to think it increases the probability of severe disease in general, rather than death conditional on severe disease 6/n

(which has its own impact on healthcare. One of the more shortsighted elements of pandemic reporting is the emphasis on deaths. Hospitalizations, and the long term effects on those hospitalized and indeed not hospitalized matter too) cidrap.umn.edu/news-perspecti… 7/n

We should recall that for most, the risk of severe disease, let alone death is not high. The reason to avoid infection is to break transmission chains and avoid the risk of the long term consequences of infection. Linking to this again, but worth it cidrap.umn.edu/news-perspecti… 8/n

There are some things to shake out. I think that the quite rapid escalation on healthcare in December in the South East of England is not fully accounted for in this, although people have tried. I think that remains to be fully teased out. 9/n

Studies like this are hard to design. Ideally you compare like with like, but how do you do that in a situation where exposure to a virus is changing week by week and healthcare is being compromised at the same time as a consequence 10/n

Overall though I think this is more than enough reason to sit up and take notice – this doesn’t need to be more virulent to be worrying. And if it *is* more virulent, even a bit, that’s a very nasty cherry on top. 11/end