Thread by @JamesWard73

When I tweeted my initial reaction to the PHE data release last night, I promised you some model scenarios to help understand the impact of B.1.617.2 on the roadmap, and in particular whether opening up on 21st June still looked possible. 1/

https://twitter.com/JamesWard73/status/1396232456127361036?s=20

I’ve done some analysis, and I’ll warn you in advance it’s a bit of a mixed bag: at the more optimistic end of assumptions, things look not too bad. But at the more pessimistic end, we’re back to facing a mid-sized exit wave, which calls the timing of Step 4 into question. 2/

Before that, a quick word on assumptions: while the PHE data last night was very helpful, it still leaves quite a lot of questions unanswered, and some of the data is confusing, or appears to contradict things we thought we knew. See for example: 3/

https://twitter.com/DevanSinha/status/1396575235114209282?s=20

But there are some lights emerging in the mist. The data on secondary attack rates points to B.1.617.2 spreading 50-60% faster than B.1.1.7. And while this could be distorted by local factors in the hotspots where 617.2 is prevalent, this is starting to feel more solid now. 4/

So let’s take that as our starting point. The variable I want to explore is the *source* of that gain i.e. is it an innate change in transmissibility (R0), or is it a side-effect of escaping immunity (not just vaccines but maybe also prior infection), or a mix of both? 5/

Hence I will consider 3 main scenarios, all with a transmission gain of +55% (B.1.617.2 vs. B.1.1.7). In the first scenario, the gain is all down to increased R0, with no immunity escape. In the second scenario, it’s all escape and no R0 gain. And the third is a 50:50 mix. 6/

The first scenario is quite similar to one I looked at last week, where I showed that even scenarios with 60% higher R0 for the new variant were potentially manageable with the retention of baseline controls after 21st June, and vaccination of teens. 7/

https://twitter.com/JamesWard73/status/1393932001271554049?s=20

But as the model is always changing to stay up-to-date with new data, let’s re-run that to be sure. First, here’s the results of increasing R0 for the new variant, with no change to current plans (and no baseline controls after the end of July). 8/

So that’s a big wave in the autumn, nearly as bad as December/January, and with around 34,000 additional deaths. But with baseline controls retained indefinitely (or until we can achieve higher levels of immunity via booster vaccinations), we get this: 9/

That looks a bit less worrisome, although there’s still around 16,000 deaths in the wave. If we decide it’s OK to vaccinate teens (or can find another 3 million adults to vaccinate in July), then we can go further and squash the wave nearly completely: 10/

(It’s worth noting that vaccinating teens in July may not be possible if we have run out of our initial order of Pfizer, and won’t get more until Sept – as @PaulMainwood has flagged. But there are ways this could happen – see his thread for detail) 11/

https://twitter.com/PaulMainwood/status/1394223295797547008?s=20

OK but given the new data on vaccine escape, we don’t actually think the whole transmission gain for B.1.617.2 is likely to be just an uplift in R0. So, what happens if it’s actually all down to immunity escape? To analyse this, I can force my model to assume an escape.. 12/

… which creates a 55% transmission gain in the short term. This requires very pessimistic assumptions re. the properties of the vaccines with the new variant, and re. protection from prior immunity– so much that I find them hard to believe. But let’s go with this for now. 13/

With these assumptions, we again get a large wave – in fact larger than the pure R0 gain scenario, with peak hospitalisations c. 50% higher than in January, and around 42,000 deaths. 14/

Again, we can try retaining baseline controls after June, which definitely helps: 15/

But even if we are able to vaccinate teenagers in July, we’re still left with a moderately large wave, within NHS capacity but entailing c. 22,000 deaths: 16/

The one thing that would kill this wave off is deferring the Step 4 changes on 21st June, until such time as we’re able to rebuild our immunity levels (e.g. via booster vaccines, maybe ‘tuned’ to the new variant). But clearly that’s not the outcome we really wanted: 17/

Before we start debating policy implications, let’s look at the middle scenario, which is the one I consider most likely. Here the 55% gain in transmission is a combination of vaccine escape driving a 25% increase, and R0 gain of a further 25% (note 1.25*1.25 = 1.55 ish) 18/

Again with no policy response we get a large wave, similar in peak hospitalisations to January, and with around 38,000 deaths: 19/

With baseline controls extended indefinitely, it gets better: 20/

And with teenagers (or another 3 million adults) added, better again. But still with a non-trivial 13,000 or so deaths in that exit wave. 21/

This time, if we want to kill off the wave, we don’t need to defer Stage 4 until we’ve completed booster vaccinations in late autumn– we just need to defer it until the end of August / early September, by which time we should have given nearly all adults their second dose. 22/

So, what does this all mean for policy, and particularly for the roadmap?
Well, first note that in none of my scenarios was there any need to reverse Step 3, nor was there a dangerous explosion before 21st June (in some cases, R went up to ~1.3, and then came down again). 23/

In the more pessimistic scenario (where the gain in transmission is mostly/all down to immunity escape) there was a potential case to defer Stage 4 until the late autumn, once a booster vaccination campaign could be completed. 24/

In the more likely scenario (with a mix of R0 gain and immunity escape), deferring Stage 4 to early September, and then reverting to baseline controls, was sufficient to suppress the exit wave. 25/

Of course, we don’t have to do that: we could decide to open up on schedule, and accept the likely numbers of deaths and hospitalisations that would result. I expect different people will have different reactions to that trade-off, and I’m not going to give a view now. 26/

Staying in Step 3 for a few more months might seem OK to some: after all, if you don’t spend much time in bars and nightclubs, and aren’t that fussed about foreign travel or large stadium events, life probably feels relatively normal. 27/

But if your livelihood depends on indoor hospitality, travel or entertainment, and you note that a good proportion of the deaths that would result from Step 4 are amongst those who have chosen not to take the vaccine, you might see the trade-off a little differently. 28/

So, I expect this debate is going to get intense over the next few weeks, in the run-up to decisions in early/mid June. Of course, we may get more data before then that pushes us more clearly in one direction or the other. 29/

(please read the following very quickly, in the manner of the “small print” at the end of a TV/radio advert: It may turn out that my assumed 55% transmission gain is too high or low, or that immunity escape is larger or smaller than we currently think. My model doesn’t… 30/

…account for local hotspots, nor for waning immunity, which may become significant in the longer-term predictions. And other models (e.g. Warwick, Imperial) will give you different results – most likely more pessimistic ones, if previous experience is anything to go by. /end