This is my new favourite graph of the week: case growth rates down (or roughly flat) over the last week in nearly every age group. There’s just one fly in the ointment: rising growth in the 90+, which is not the group I would choose for that pattern.
Here’s the more familiar graphs by age group, starting with the 0-30s:
then the 30-60s:
And finally the 60-90s, who are now (thankfully) showing clearer signs of slower growth in most groups. (although, as before, there’s a risk that delayed test reporting might be exaggerating the effect on the most recent day or two)
If we look at the raw case data for the 20-24s (the group with the highest case rates, and in late May /early June the highest growth as well), there’s an intriguing sign that we may already be at the peak of cases (not just peak of growth). 🤞
Of course there’s still plenty of ways for this to go wrong (not least with the football ongoing), but so far I’m encouraged by this. I’m also intrigued by what exactly is pulling the growth rates down. I’d love to believe it’s a herd-immunity effect, led by the 20-24s…
…where we may finally have reached a “wall” caused by the combined pressure of vaccinations and prior infections (and perhaps assisted by heterogenous mixing effects). But that wouldn’t explain why growth peaked at the same time in nearly every age group…
…and also in nearly every region (which you wouldn’t expect, if this was mainly driven by immunity). That synchronicity points to a more external driver e.g. a behavioural reaction to a government announcement (but what??) or the weather. Views welcome….
apologies - have just realised that chart was mislabelled, it's actually the weekly growth (with 0.6 meaning 60% growth, for example), not the 5-day growth rate. but the point was more about the synchronised peak not the values on the y-axis, so will leave it there.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
When I wrote my thread before Christmas (linked below) I promised you a postscript on the interaction between vaccines and NPIs, so here it is. And it’s good news – the combined effect is more powerful than I expected. 🧵 1/25
Let’s start with what I expected. In an endemic model, with no vaccines, there is essentially a “required” rate of infection which is needed to keep immunity levels topped up at their equilibrium (herd immunity threshold) level. 2/25
NPIs can reduce this equilibrium level, and hence also the required rate of infection – although as we’ve seen in other threads, the % reduction in the infection rate is usually less than the % impact of the NPIs on transmission. 3/25
So Covid Twitter (and Twitter in general) seems to be dying… with just a few limbs still twitching. To be fair I haven’t been helping much this year, having held true to my New Year’s Resolution to spend time talking to my family rather than on Excel and Twitter. 🧵
So many thanks to those (including but not limited to @BristOliver@PaulMainwood@kallmemeg@john_actuary@chrischirp) who have kept things going this far. But I haven’t completely given up the modelling thing, and there’s a couple of new things I’d like to show you.
So for old times’ sake, and as an early Christmas present, here goes with probably my last ever Twitter thread (and yes, I know it would probably work better on Substack, and maybe I should replicate it on Mastodon &/or Post, but just for today, let’s remember Twitter as it was).
@TAH_Sci@i_petersen@karamballes@chrischirp@CathNoakes@MichaelSFuhrer@MichaelPlankNZ I agree our biggest concern right now should be booster take-up in 65+, but I’d be a bit more open to the case for “clean air” interventions. It’s possible that the business case for doing this in some settings actually does stack up, and imo we should be investigating this, …
@TAH_Sci@i_petersen@karamballes@chrischirp@CathNoakes@MichaelSFuhrer@MichaelPlankNZ … while being very realistic about the costs and benefits. If we’re happy to start with a basic model (and then refine it) then there’s really only three things we need to work out: 1) what % of transmission do we expect to be interrupted in the specific settings it is used?
Multiplying 1) and 2) gives us the overall % reduction in transmission. And then we need to convert that into a reduction in medium-term prevalence. @MichaelPlankNZ gives us the formula here:
really nice analysis from @Jean__Fisch here using the SIREN data to imply changing patterns in the amount of protection that previous infection is giving in each wave. A few points to note:
- infection seems to give strong protection through the autumn 2020 and the Alpha wave
- this reduces (through the combined effects of waning, immune escape and impact of vaccinations) for the Delta wave
- and for the first Omicron (BA.1) wave, there appears to be no benefit from prior infection. I don’t believe this is literally true: more likely there was a…
…small benefit but it may be offset by demographic confounders eg those more likely to be exposed in one wave are also more likely to be exposed in the next
- the good news is that the protection recovers in the later Omicron waves, suggesting that BA.1/BA.2 infection does…
I’ve been looking at this question the other way around, but it’s still very gently encouraging I think. The case curve is continuing with exponential growth (straight line on the log plot) for much longer than we would have liked. We might have expected it to curve over by now
…as the effect of growing immunity to the latest variant starts to bring the R number down. We’ve had a couple of false dawns already (what do we call these, they can’t be “dead cat bounces” because we’re still going up… so maybe “live cat slumps”?), but still it keeps rising.
I *think* what’s going on here is not that immunity has suddenly stopped working, but rather that the effect of growing immunity is being offset by something else. The obvious candidate (as explored in Oliver’s thread) is the upwards pressure from continued variant mix changes.
1. From a mathematical perspective, the arrival of new variants is a bit like waning host immunity, and has much the same effect (i.e. immunity gets lower). It arrives in a slightly different way (at the same time for everyone, rather than gradually across the population)
...which will affect the short-term dynamics, and mean that we’re more likely to get new peaks and troughs, rather than settling into a more stable equilibrium. But viewed over periods of several months or years, the impact on total infection rates will be similar.