Thread by @JamesWard73

A few thoughts on yesterday’s PHE data: overall, the report is a bit of a mixed bag. Of the three significant bits of news, one is positive, one negative, and the other neutral. But you might not get that impression from the reaction here on Twitter or in the media. 1/n

Let’s take the negative first: we now have data suggesting that the new variant (Delta) is more likely to lead to people being admitted to hospital – perhaps 2.5 times more likely than when infected with the old Alpha variant. 2/n

https://twitter.com/kallmemeg/status/1400489338425810951?s=20

That’s clearly bad news, and will have an impact on model projections for Step 4. But it’s not necessarily a complete disaster: if we can control the spread of the virus, then it doesn’t matter what the hospitalisation ratio is, because very few people will be catching it. 3/n

Now the positive: PHE’s estimate of Delta’s Secondary Attack Rate (SAR) has come down, from 67% higher than Alpha in last week’s report, to +50% in this week’s. That might not sound like a big deal, but it’s probably the figure we care about most. 4/n

https://twitter.com/kallmemeg/status/1400495544078946311?s=20

The SAR tells us how likely someone is to infect one of their close contacts, so it feeds directly into our estimate of R. We have other analysis (e.g. from @alexselby1770 and @TWenseleers) suggesting that Delta is spreading 70-75% faster than Alpha, which appeared to fit… 5/n

… neatly with PHE’s +67% estimate from last week. But if the SAR is only 50% higher, it re-opens the question as to whether there are other environmental effects which could be causing the remaining ~15% of the apparent transmission gain; if so, that would be good news. 6/n

And finally the neutral: there was no new data on vaccine effectiveness in this week’s PHE report. But they did update Table 4 which contains the data that I previously used to do a back-of-the-envelope calculation of vaccine effectiveness. 7/n

https://twitter.com/kallmemeg/status/1400492226099167235?s=20

And having re-done the calculations with the updated data, I see no significant changes to what I reported below. I need to put a big caveat on this: I don’t have the granular dataset that PHE does, and so my calculations make various assumptions... 8/n

https://twitter.com/JamesWard73/status/1398012064942235648?s=20

…about the distribution of risk which could be wrong, or have been shifted by Delta. They also don’t match PHE estimates from last week for VE vs. symptomatic infection from Delta after 1 and 2 doses (see below). But there are reasons to think those may be underestimated, 9/n

…and even if they’re not, then the fact that my estimates are unchanged with new data might suggest the PHE analysis will also be unaffected. Which implies the vaccines are still working well after 2 doses, and even 1 dose should give good protection vs. hospitalisation. 10/n

So, how am I feeling after all that? Well, to be honest, not much different to how I was before yesterday. The bad news on hospitalisations is balanced by the hope that the transmission gain / SAR might be a bit lower, and there’s no real change on vaccine efficacy. 11/n

Some may challenge why I’m not more concerned about the hospital data. And it’s clearly bad news. But for me, it all comes back to this old thread by @AdamJKucharski about the trade-off between transmissibility and severity. 12/n

https://twitter.com/AdamJKucharski/status/1343567425107881986?s=20

For us in the UK right now, a doubling of the hospitalisation rate might be less bad news than an extra 15-20% on R, because of how close we already are to the herd immunity threshold. So for example here’s my model with a 50% increase in R, and doubled hospitalisations: 13/n

And here it is with no change to hospitalisations, but R up 75% instead of 50%. Which would you prefer? [note to build those scenarios I’ve assumed a vaccine escape which drives a 25% transmission gain, with the rest being increased R0 – which is then dialled up as needed] 14/n

Of course, this is very dependent on the starting conditions, and there are other scenarios where higher hospitalisations is worse news than increased transmission (and other countries will be in different places to the UK). But my point is: let’s focus on the numbers... 15/n

… that really make a big difference to the outcome, and not over-react to one single piece of data. We need to look at this in the round, and consider the range of plausible outcomes, given the (still significant) uncertainties in the data and in our analysis of it. 16/n

So, where does this leave us with regard to opening up on June 21st? I’m not honestly sure, and of course we don’t have to decide just yet. I’ll try to run some new model scenarios at the weekend, which might help to point us in a particular direction. 17/n

But my guess is that, barring some unexpected data from PHE next week, the plausible range of outcomes will still include both a small exit wave, as per @andrew_lilico’s model yesterday (note this was before the new data on Delta) 18/n

https://twitter.com/andrew_lilico/status/1400379339548631041?s=20

...and a much more severe wave, threatening NHS capacity, and with much higher rates of hospitalisation and death (e.g. as per my second scenario above). Based on previous form, it is likely that the SPI-M models will lean towards the latter outcome. 19/n

And that will leave us with a very tough call: whether to proceed as is, delay to September to get more second doses in, or take a short delay to gather more data (it’s unlikely to stop the wave). I’ll let you know my thoughts on this as they evolve. 20/n

Finally, big thanks to the PHE team including @kallmemeg @meera_chand and many others, working tirelessly to analyse the data and make it available for us all so quickly. And apologies I won’t be around to respond to questions on this thread – I'm focusing on family today. /end