Thread by @JamesWard73

Model Update Klaxon: the good thing about publishing model results is that everyone tells me why they think the model is wrong – mostly constructively, and quite often I agree with them. So I’ve changed a few things, and it’s really good news. (thread)

A few links to previous threads here for anyone who's wondering what I'm talking about:

https://twitter.com/JamesWard73/status/1364568662468919297?s=20

https://twitter.com/JamesWard73/status/1364009555068338176?s=20

https://twitter.com/JamesWard73/status/1363607776203845634?s=20

https://twitter.com/JamesWard73/status/1363125673372635138?s=20

https://twitter.com/JamesWard73/status/1363127593734651906?s=20

Most of the updates today are to do with children, which hasn’t been a big focus of my modelling, as I was much more worried about what was going on with the over-50s. But it’s starting to become more relevant as we begin to ‘flirt’ with the herd immunity threshold (HIT). 3/n

Before we go there: there was near-universal consensus in the replies, including from some people whose judgement I know and trust, that I have my starting level of immunity on 4th Jan 2021 (at 18%, plus 2% infected) slightly too low 4/n

So without any great science, I’ve nudged this up by 5%, so it now starts at 23% plus 2% infected, or 25% in total. That could still be wrong (either way), but at least it feels more central in the range of believable values now. 5/n

Unfortunately it doesn’t make as much difference to the “plateau” level of immunity as you might think: after vaccination, we’re at 64% (instead of 62% previously). So we’re still a little way short of the HIT, and the model still predicts a (slightly smaller) fourth wave. 6/n

Now to the kids: there is good evidence that children (and particularly, pre-teenage children) are, on the whole, less susceptible to catching and passing on covid. At the moment my model doesn’t account for this, they’re just treated the same as adults. 7/n

So I’ve introduced an assumption that under-13s are 60% less susceptible to covid; this is equivalent to a susceptibility reduction of 45% for all under-18s, which I think is consistent with the literature (but happy to be corrected). 8/n

With this assumption, we get our immunity levels up to 68% after vaccinations are done, which is tantalisingly close to the HIT, and we have a very tiny fourth wave emerging in 2022: 9/n

Could we do any better? Well, we know that safety trials are underway for vaccines in children – for example nihr.ac.uk/news/first-chi… (and similar ones for Pfizer and Moderna). So, assuming good results, we could vaccinate them too. 10/n

Given what we know about susceptibility, I think it’s more likely we would start with teenagers, and that’s where most of the benefit is. So I’ve assumed a first dose in June/July, and a second dose in Sept/Oct, for all 13-17 year olds who want it – with a take-up of 70% 11/n

Does that get us over the line? Well, still not quite, but post-vaccination immunity is over 70%, within spitting distance of the HIT, and with that level of immunity in the population there’s really no prospect of a material fourth wave. 12/n

So that’s really good news, then? Well, yes, it’s the first time that my model’s central case has predicted no fourth wave – so worth celebrating. But I wouldn’t be a corona centrist if I didn’t take this opportunity to remind you of a few things that could still go wrong: 13/n

To start with, there’s still quite a lot of uncertain variables here: I might have my starting value for R0 too low, or my assumptions on vaccine take-up in the under-50s, and effectiveness vs. transmission after 2 doses of AZ, too high. 14/n

Secondly, as we look over longer timescales, it becomes more relevant to think about how immunity might change (and, probably, reduce) over time. And of course we could get variants that partially ‘escape’ the transmission immunity we’ve built up. 15/n

And finally, as we’ve noted before, covid risks are not spread uniformly across the UK. If at a national level we’re poised precariously on the HIT, that means that in some areas we’ll be above it, and some below. So in reality, … 16/n

while there might not be a significant national fourth wave, it’s still likely that there would be regional or local outbreaks in those areas that have higher natural R0 (e.g. due to deprivation / population density), lower vaccine take-up, or lower infection-based immunity. 17/n

So there’s still lots of things that could cause us problems in the next few months, but I’m increasingly confident that we’re on a track to beating this, and restoring something like normality in the summer. And that’s my happy thought for today. /end