Thread by @JamesWard73

So it’s time for me to come off the fence re 19th July. I’m in favour of:
a) Opening up on 19/7, but with stronger mitigating baseline measures than the govt is proposing
b) Opening up vaccinations to 12-17s as supply permits, starting with the vulnerable.
Long🧵follows...

I’m very conscious that many people will disagree with something I’ve said there, and possibly all of it. And I have good friends on all sides of these debates, who I don’t want to lose. So it may help to say: just because you oppose (or support) vaccination of teenagers…

…doesn’t mean that I think you’re a psychopath. And just because you might support extending regulations for another few weeks, doesn’t mean I think you’re a “bedwetter”, or uncaring about those who have suffered during lockdown. So I’d be grateful if you…

…could do me the return favour of not thinking I’m a genocidal maniac because, after a great deal of thought, analysis and heart-searching, I think it’s probably the best of a series of unattractive options for us to proceed to Step 4 on 19th July.

My logic for this is as follows:
1. I think the summer wave probably won’t be catastrophic
2. Although there could be some benefit from delaying, it’s unlikely to be large, and it could well make things worse in the winter.
I’ll expand on these reasons, using the model results

As you’d expect, I’ve refitted the model to latest data, which isn’t a big change from last weekend, but given the faster-than-expected reduction in case growth, this produces slightly lower estimates for infections over the summer.

However, I’ve also adjusted my hospitalisation assumptions, as I think I slightly over-fitted to the low admissions data from 2-3 weeks ago, which is starting to reverse now. As a result, my new base case is slightly improved overall, with a peak of ~9k hospitalisations per week:

You’ll note that with a less intense summer wave, we re-open the door to a small wave in the spring of 2022, but we can dispose of that by vaccinating teenagers (or finding a similar number of adults not previously vaccinated) in the autumn:

This is my central case, and it’s certainly possibly for things to be better, or worse. If the government’s attempts to portray July 19 as “Freedom Day” and removal of nearly all legal restrictions leads to an abandonment of cautious behaviour, we might get something like this:

…which has a peak of hospitalisations around twice the size (~19k per week), although the total number of hospitalisations is only increased by ~20%, and deaths from 11k to 13k. So not good news, but in the scale of what has come before, also not catastrophic.

What happens if we delay? In the base case, pushing Step 4 back to 6th Sept would give us this, with a slightly smaller summer wave (peak hospitalisations ~8k per week), but a larger one in the spring. Total infections, hospitalisations and deaths are larger than with no delay

Again, we can reduce –but in this case not eliminate– that spring wave by vaccinating teens in the autumn. It’s possible we could go further than this with booster vaccinations, but since we also have waning immunity to contend with, I’m not sure how much net benefit to expect.

If we take the downside scenario, and apply a Step 4 delay to that, then (with vaccination of teenagers) we get this. It’s about the same number of total infections, hospitalisations and deaths, but it does reduce the peak hospitalisations from ~19k to ~8k.

So in the base case, a delay could make things worse in terms of total infections, hospitalisations and deaths – although it does reduce the peak NHS load. In a downside scenario, delay looks a bit more attractive – it reduces peak load without any penalty on total outcomes.

BUT I am very nervous about using the model to predict what happens in the winter or next spring – there are so many factors here that are unmodelled or highly uncertain– including seasonality, waning immunity, booster vaccinations, and the potential arrival of new variants.

So I think the best I can say is: it’s unclear whether a delay helps or not. It’s likely to reduce peak NHS loads, but it could well (depending on what else happens) make things worse overall in terms of total infections, hospitalisations and deaths.

And that’s why I don't support a delay. If it’s equally likely (based on current info) to make things better or worse, and the predicted peak loads on the NHS are survivable, then I don’t see a compelling case for it. And at this stage, I think the case needs to be compelling.

I was vocal in supporting an extension from 21st June to 19th July because all the modelling (including mine) suggested there was a significant reduction in risk from doing so – and I believe that was the right call. But this time around there isn’t that same strength of logic.

Let’s address some of the potential objections to that argument. Firstly, as @ChrisCEOHopson @rupert_pearse and others have vividly documented, the NHS is already under strong pressure from the combination of covid, urgent non-covid and elective demand.

So adding ~9k hospitalisations per week could still provoke a crisis, and will certainly lead to the postponement of elective care, which is already well behind schedule from the impact of the last year+ of the epidemic. But honestly, I don’t see a better alternative.

If delay could mean more hospital admissions overall (or at least, is not clearly expected to produce a material reduction), then we’re trading a likely (slight) reduction in peak load for a potentially longer period of disruption. Which isn’t really a strong case.

Similarly on LongCovid, every infection is bad news, and I would much rather people gained immunity via vaccination than by infection. But if delay could just as easily lead to *more* infections as it does to fewer, then I don’t see the compelling case for pushing things back.

There is also the argument that more spread means more variants, and that spread in a partially-vaccinated (single-dosed) population is particularly dangerous in terms of evolutionary pressure. Frankly I’m not qualified to comment on the second part of that,

…but we’re already committed to a summer wave of some form (unless we’re willing to lockdown hard right now), and delay is not confidently expected to materially reduce the total number of infections over the next several months. So again, why wait?

I said at the start that I would apply stronger baseline measures than the government is proposing – what did I mean by this? I believe it would be sensible to maintain the mask mandate in crowded spaces such as on public transport. I would also propose...

…some form of Covid certification (i.e. 2 doses of vaccine or negative test) for nightclubs and indoor events – particularly given the Dutch experience in recent days. There are also holes in the current approach that aren’t too late to close e.g. on #SupportToIsolate.

My main reason for proposing these measures isn’t the direct effect they will have (which some may dispute) – it’s the signal they send to the wider public that this isn’t really Freedom Day when everything returns to normal, it’s Step 4 on the roadmap…

…and at some point there will be a Step 5 when things really can go back to normal. But that’s not quite yet. By doing this, the objective is to reduce the risk of a more explosive summer wave (my downside scenario), without significantly limiting people’s freedom.

I also said at the start that I would support the vaccination of teenagers. As we’ve seen from the scenarios, there are some cases (but not all) where this has a material impact on the progress of the epidemic and could save a few thousand lives –

…so that’s one clear reason for supporting it. But I also think there’s a strong equity argument here: Pfizer has already been approved by MHRA for safety and efficacy in over-12s, and we will have plenty available in September when the new order arrives.

So why would we not make this available to teens who want it? No-one is suggesting making it compulsory, and the (small) risks should be clearly explained; maybe we shouldn’t promote it in the way we have for adults, but I don’t see why we shouldn’t make it available.

I’m also conscious that my model lacks full age-stratification, and even in the scenarios where it predicts little impact from vaccinating teens, a more sophisticated model might well show that it reduces or eliminates the risk of a further school-led wave in the autumn.

The strongest counter-argument for me is that the vaccine is needed more in other countries. But we’re talking relatively small amounts (3m people) and we will have a lot more excess supply that we can donate to others, so I believe we really can and should do both.

We should start with offering vaccine immediately to all clinically-vulnerable teens (and those that live with highly vulnerable adults e.g. immunosuppressed), and then move down the age groups as supply permits – per @PaulMainwood’s analysis I think this means we may

…be able to offer vaccine to 16-17 year-olds during July/August, but the 12-15 year-olds would probably need to wait until September when our new Pfizer order arrives. It’s possible this could ramp up sooner if Moderna deliveries accelerate.

And that’s it. I don’t envy those in government or on JCVI trying to make these decisions, but if I had to provide advice, this would be it. Open up on 19th July, with stronger baseline measures. And open up vaccination to teenagers as supply permits. /end

PS this is an extension to the thread for those wondering: how can a delay make things worse? There’s essentially two reasons for this: firstly, the summer wave (+ vaxxing teens) will slightly overshoot the current herd immunity threshold, but that turns out to be pretty much

…exactly what we need to get through the winter with normal behaviour (which will increase HIT). If we reduce the size of the summer wave, we’re holding back infections that are likely to re-emerge later. And if we’re below winter HIT we get another wave and more overshoot.

And it’s that second round of overshoot that causes the downside of delay. The second (less significant) effect is that a winter/spring wave will have a slightly higher % of older, more vulnerable people in it, because we’ve made more progress by then in vaxxing younger groups.

At present the case rates are skewed very strongly towards younger groups, but after vaxxing the 18-30s (and maybe teens) with both doses, that effect will reverse to some extent, and the pattern of infections will become more evenly spread across the ages.

This means that a winter wave could produce more hospitalisations and deaths than the same number of infections in the summer. (there are other reasons why that could be, e.g. if people are more frail in the winter, or NHS under more pressure, which I haven’t modelled). /endPS

PPS one more thing - and sorry, I realised I missed out one point (it's easy for me to forget things that are obvious to me, but not others). Step 5 (full return to normality) is assumed to happen in the model on 1st January 2022. So that's part of why we get a wave in spring..

...2022 (rather than any other time). And it's possible we could optimise this further e.g. by delaying Step 5 to the spring, when seasonality helps us, or by doing it in smaller stages which might reduce overshoot (and hence deaths / hospitalisations etc). But it has to...

...happen at some point (unless you think we can sustain mask-wearing and Covid checks indefinitely, even with low case rates - which I don't). And I'm conscious that even if we planned a later/more gradual Step 5, we can't necessarily control how compliance with these measures

...will evolve. And there's risks from new variants / waning immunity etc. to consider as well. So I think it's a reasonable modelling assumption, but I should have made it a bit more explicit for those who weren't familiar with previous iterations of the model. /endPPS