A quick Sunday case data round-up. First, the age-group growth rate charts continue to show a declining trend in short-term growth. I’ll start with the 0-30s today:
and then the 30-60s, also with a clear declining trend:
and lastly the 60-90s, now mostly holding at a fairly steady growth rate (which is better than where they were a few days ago, with accelerating growth). (ctd)
I think the question is no longer: is growth falling? But rather: what will it do next? Does it keep going down (great news), stabilise at a new level (less good, but still better than it was), or bounce back up (bad news)? Mesdames et messieurs, faites vos jeux….
Taking a quick look at the regional data, it’s clear that (as others have noted, and I flagged as a concern over 2 weeks ago), the North East is now the region with the highest case rates – and still concentrated in the 10-30 year olds:
And looking at growth rates rather than case rates, the North East remains in the lead (which is not a good place to be). I’m not particularly liking the trend in the South East either.
Finally, just keeping an eye on the male:female split, in case the football makes a difference. Case rates continue to be more male-biased than normal, and particularly in the 15-40s (and, more surprisingly, the 75-90s). But there’s been no big shifts in recent days. /end
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I’m not going to do a very long model update thread, because it’s a busy week at work and I do actually need to get some sleep at some point. But I thought you might appreciate some headline conclusions from the latest version, as it might help inform the debate re. 19 July 1/n
I’m also not going to give you a policy recommendation, because it’s a genuinely tough call, and as usual I’m going to take it to the wire before getting off the fence (and I’ll stop the mixed metaphors there, before I do myself a nasty – if imaginary – injury). 2/n
So as usual I’ve updated to all the latest data and made a few other adjustments in an attempt to get as close as possible to a genuine “central view” i.e. as likely to be wrong on the upside as the downside. This week, those changes have included: 3/n
I know the cases and admissions data looks bad again today, but I’m going to stick my neck out and say: I think we might have passed the current peak of growth rates, and so R could be about to head down again. Here’s the recent growth trends in the 30-60 year olds: 1/
And here’s the same graph for the 0-30s: 2/
The trends in the older age groups (60-90) are less good – but *if* we think a lot of these cases are direct infections from younger people, then we’d expect these curves to follow the shape of the younger groups, just a bit behind. So if those come down, these should too. 3/
Today we got @PHE_uk’s latest weekly surveillance report which provides their age-stratified hospital admissions data. I can use this to extend my analysis of the hospitalisation ratio, as follows:
This shows that in the most recent week (admissions in 21-27 June), the ratio has dropped further, to around 1.5%. I can also extend my “fixed age mix” curve, in orange (see thread below for explanation of this) – which has dropped by a similar amount.
This implies that the most recent drop is “real” i.e. it’s caused by falls in hospitalisation ratios *within* each age group, not by changes in the age mix of cases. We can see this if we look at the hospitalisation ratios for each age group, as per the chart below:
@ThatRyanChap@declamare@whippletom I know you didn’t really ask for this, but I had to go back and check my intuition – and luckily, it’s not wrong. But the exercise was useful because it exposes a couple of interesting dynamics – and helped me understand why we can’t see the impact of delta on hospitalisations.
@ThatRyanChap@declamare@whippletom So here goes: imagine that pre-vaccine we have a population with 1000 cases per day, and 100 hospital admissions per day. So a hospitalisation ratio of 10%. Now let’s suppose that we administer 1 dose of a vaccine that stops 25% of infections, and 80% of hospitalisations
@ThatRyanChap@declamare@whippletom Now we’ll have 750 cases and 20 hospitalisations per day, for a ratio of 2.7%. And let’s apply the second dose, which stops 80% of cases and 95% of hospitalisations. So we’ll have 200 cases and 5 hospitalisations, for a ratio of 2.5%.
Obviously today’s case data is horrible again, and admissions look like they might be picking up. But there is a small piece of hidden good news: we might be at or near the peak of growth rates. And some bad news in the older age groups (sorry). Details in thread below... 1/7
The good news is that growth rates in the under-60s look like they may have stopped growing. Here are the under-30s: as you can see, only the preschool kids (0-4s) still have a rising growth rate, with other age groups looking flat or even with slightly falling growth. 2/n
And here are the 30-60s, with an even clearer pattern. This might seem like small comfort: after all, growth isn’t yet falling, let alone cases. But if a zero second derivative is the best thing available, I’ll celebrate that – and it’s better than the alternative. 3/7
I’m not going to comment on today’s case data, because the message hasn't changed, and I don’t want to spoil an otherwise positive evening. But case data only really matters if it causes bad medical outcomes, and here the news may be a bit better. 1/n
The ratio of hospitalisations to cases has been dropping over the last few months – mostly due to vaccines. (the main effect of vaccines is to stop people getting infected, but for those who do get infected, they also reduce the chances of going to hospital or dying). 2/n
[note: to calculate this ratio, we need to compare hospitalisations to cases a few days earlier, and there’s some debate as to how long a lag to use. I’ve used lags from this recent ONS study ons.gov.uk/releases/coron…, but I get similar results with different assumptions] 3/n