UK COVID data, week starting 20th March
The rollercoaster ride of all the trouble kept her inside
(9am 23/3/23) I skipped COVID Substack updates last week. There’s only so many ways you can say “it’s going up still but slower than before”, there’s less data than there used to be, and of course 3 years on from lockdown there’s a general feeling that this stuff matters less than it used to. So maybe I’ll move to fortnightly or more infrequent updates.
Anyway, ZOE has been acting very strangely this week, with daily incidence ping-ponging between about 100k and 130k with no apparent rhyme or reason. The overall trend is probably up, but it feels hard to have confidence in that right now in the circumstances, so time to ignore that measure again I think.
And of course this is the final week of the ONS prevalence survey. I think it was a fantastic resource in its time which helped answer a lot of important questions, but equally I understand that it didn’t come for free and there are other calls on the public purse. Anyway, we’ll just need to get used to figuring things out with fewer numbers than before (of course, we’ve been doing this anyway, dating back to the dashboard going weekly and so on). It’s not like there aren’t other useful indicators out there still.
In terms of variants, XBB.1.5 continues to increase in market share. However on logistic scale, where you’d expect a straight line if it had a constant growth advantage, the rate of takeover seems slower than it was (though the wider confidence interval on the final point makes it hard to be sure if it has peaked in share yet).
Coming up in second place we have XBB.1.9.1. But you can see that the green and red lines are more or less parallel lately, which suggests that XBB.1.9.1 doesn’t have much of an advantage over the currently dominant XBB.1.5.1. Alex estimates it as a 2% or so daily advantage, suggesting that it’s unlikely to cause a serious wave over and above the current one. (This doesn’t mean that there won’t be a wave to come - as usual it’s a race between boosters and waning, though hopefully spring, Easter holidays and warmer weather will help - but it seems unlikely to come from this particular source).
Of course there’s always another variant coming down the line, and the next one people seem to be excited about is XBB.1.16.1, based on how it is doing in India. I’d say it’s possible, but similar arguments would have led you to predict a big BA.2.75.2 (“Centaurus”) wave that never showed up over here, and while it’s under 1% in the UK I’d prefer to wait for a bit more data before getting too excited.
Anyway, the key thing probably remains the hospital admissions data, which has come in fits and starts a bit lately. There was a plausible peak a few weeks ago, and then a return to growth, and so while this week maybe shows the beginnings of a peak ..
it’s probably best to not get too excited yet, and just think of it as “maybe flattish, at worst growing slowly”. The overall level is about 1,000, which is certainly higher than we might like, but equally lower than even the BA.1, BA.2 and BA.5 peaks of early 2022.
Here’s another view in terms of growth rate - you can see it’s not even been as fast as four week doubling lately.
(9pm 23/3/23) So much for my optimism about being able to keep on top of all this. With the news that COG-UK will be closing, there won’t be a way to track UK variants, and so most of the analysis I’ve done in the last 15 months or so wouldn’t have been possible. So I think this may be the end of the road for me and regular COVID graphs. Just for completeness though, here’s a suggestion of a possible peak in the dashboard case data as well.
(12.20pm 24/3/23) The final ever ONS is showing an increase, though (I guess for reasons of sample size) the confidence interval is wider than it’s been in recent weeks, so it’s hard to be completely sure about the narrative here.
