Lots of questions still bouncing around on vaccine efficacy vs B.1.617.2, so here are some follow-ups to our Saturday morning story:
Thread follows, and @SarahNev and I published a new story last night covering all the details including transmissibility: ft.com/content/e71471…
Following our original story, PHE later published more detailed data disaggregated by vaccine.
That data shows our pooled figure of 7% relative drop in two-dose efficacy against B.1.617.2 vs B.1.1.7 comprised a 6% drop for Pfizer, 10% drop for AstraZeneca. Very little difference
Similarly, the 35% relative drop in efficacy after one dose was virtually indistinguishable between the two vaccines.
A lot of attention has been given to the fact that Pfizer’s relative drop was from 93% to 88%, while AstraZeneca’s was from 66% to 60%, markedly lower figures.
While the attention on 60 & 66 is understandable, it needs interpreting in its proper context:
First, let’s think about what question the PHE study set out to answer: how much lower is vaccine efficacy against B.1.617.2, compared to B.1.1.7?
(We already know how the vaccines perform against B.1.1.7. We’ve been witnessing their impact in the UK for several months)
And the answer: after two doses, roughly 6% lower for Pfizer, roughly 10% lower for AstraZeneca.
The fact AZ is 60% effective whereas Pfizer is 88% effective is worth noting (and I’ll address that in a minute), but not relevant to the question of relative reduction vs B.1.617.2.
Onto the much-discussed figure of 60% efficacy after two-doses of AZ:
As PHE explained in their study (and we reported last night in our story), that figure shouldn’t be taken as a 'final' estimate of two-dose VE, nor should it really be compared directly to the 88% for Pfizer.
Both of those are for the same reason: the later roll-out of second doses of AZ means less time had passed between second dose and symptom onset for the people in the AZ sample compared to the people in the Pfizer sample. In other words, less time for protection to build.
In addition, second dose sample for AZ is smaller than Pfizer, so more uncertainty around 60 & 66 than around 88 & 93
Of course, that uncertainty could resolve in either direction, but experts we spoke to including report’s authors believe it’s likely 60 & 66 are under-estimates
So overall takeaway from this preliminary vaccine efficacy data remains:
• First dose protection looks be around one-third lower against B.1.617.2, both for Pfizer and AstraZeneca
• Second dose protection holds up pretty well, both for Pfizer and AstraZeneca
Happily, today’s update of the cases data remains consistent with a solid vaccine effect in areas where B.1.617.2 is dominant.
Cases among older, mostly-double-dosed age groups remain low & stable, while rising among less-vaxxed groups
And remember, most UK jabs are AstraZeneca
Onto transmissibility: the most significant data point from yesterday’s reports was on secondary attack rates (num of new cases among contacts / total number of contacts)
The data here suggest B.1.617.2 spreads around 50% more easily/widely, but as ever, it comes with caveats:
Part — maybe a very large part — of this will be down to innate, biological characteristics of B.1.617.2, but it can also be influenced by social and behavioural settings where it has spread, and by characteristics of contacts such as vaccinated status (HT @AdamJKucharski)
So on transmissibility we remain in a largely similar place to last week:
• B.1.617.2 does spread more easily
• How much more easily varies by location/cluster, and different clusters have different contexts
• So still very hard to know how much is innate vs context-specific
Today’s update of the cases data once again paints a mixed picture:
Case rates in Bolton show signs of stabilising, positivity rate is falling. But numbers in Bedford, Blackburn, Bury still rising.
Certainly evidence here for higher transmissibility, but again context-specific.
But — again paraphrasing @AdamJKucharski — sitting around waiting for cast-iron evidence doesn’t have a great record over the last 18 months.
Pandemics come at you fast, so decisions often have to be made with less or weaker evidence than we would ideally have.
A paper presented to SAGE earlier this month makes the point that while vaccinations remains remain our way out of all this, keeping a lid on transmission while more people get vaccinated will also help. This seems pretty uncontroversial. It’s vaccines *and*, not vaccines *or*.
I’ve talked before about how restrictions, as well as vaccines, have been key to the UK’s declining rates in recent months
Even with second doses providing good protection against B.1.617.2, we shouldn’t pretend they’re the only tool available to us.
So that, really, is the big question right now. What is the *and* that we add to vaccines?
So far it’s been surge-testing in hotspots. There are signs that may already be helping. Giving more agency to local public health teams may also already be helping.
Other options include local lockdowns, more support for isolation, regular rapid testing. None of these implemented yet.
Next level of course would be altering reopening timeline, maybe postponing "full reopening" on June 21. Question is what triggers each additional step?
So there we are. Lots of stuff there but hopefully worth the time to paint a comprehensive picture.
Enjoy the remainder of your Sunday, folks.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I would ask people to read what follows carefully. Vaccine efficacy is a nuanced topic and the numbers here need to be interpreted in their proper context.
The data, which the FT has seen, suggest first dose offers around 35% less protection against symptomatic infection with B.1.617.2 compared to B.1.1.7, but after two doses the relative drop is only 7%.
The figures are based on pooled data from the Pfizer and AstraZeneca vaccines
Cases continue to climb in Bolton, Blackburn & Bedford, (known B.1.617.2 hotspots), though rate of acceleration in Bolton has slowed slightly, and test positivity there is flat, suggesting surge testing is playing a role in 📈
Continued increase in those areas would be a concern, and it’s worth noting rises in neighbouring Bury and Burnley too.
B.1.617.2 not believed to be dominant in either of these areas, but Sanger data is now 12 days old so that may have changed.
Nonetheless, the same chart shows that high prevalence of B.1.617.2 is by no means a guarantee of prolonged resurgence.
Hounslow, Nottingham and Sefton are all places where B.1.617.2 is believed to be dominant, and yet their upticks reversed and now appear as brief blips.
First, today’s Sanger data on variants at local level. On the surface, this doesn’t look good. Cases of non-B.1.617.2 are in decline, but those red peaks are the variant sending overall rates climbing
To state the obvious: that pattern is not what we want to see, and if things keep going in that direction, we could see national cases rapidly climb again.
But there are a couple of reasons to pause before assuming we’re going to see those peaks steepen and proliferate.
First, the Sanger data is sequences with specimen date before May 8.
We’ve got more days of data since then. It’s not broken down by variant, but it can show us what’s happened to total cases in those areas since May 8.
Answer: growth rates have slowed, in some cases reversed
If B.1.617.2 does prove to be more transmissible, we may need to keep things like indoor masking and tests for entry to large events for a while beyond June 21.
My thoughts:
• No, that wouldn't mean "lockdown never ends". No, it wouldn't be curtailing freedom. It would be a very small compromise for a very large benefit
• As vaccination rates continue to climb further into the year, those few remaining measures could then be eased
• It's still entirely possible any transmissibility advantage will prove to be much smaller than some have estimated. In that event, June 21 plans can proceed as planned
Here's the latest from @AnnaSophieGross and I on the Indian variant, digging into how serious a threat it is currently believed to pose ft.com/content/eb158a…
One thing I think it's important to be clear about is that public health officials and epidemiologists are looking extremely closely at this stuff.
To the extent that policy has not followed one particular path or another, it's because the people looking at the evidence have yet to determine where it points. This is the scientific method, not complacency.
It’s early days, but there are signs that the vaccines may be working against the Indian variant B.1.617.2
Resurgences in Bolton & Blackburn are so far confined to younger people. Cases remain low & flat among the mostly-vaccinated older population.
(We must also note that in the past, the higher levels of social mixing you typically see among younger people have led cases to rise among them first before climbing in the older groups, so vaccines are not the only thing that can cause this. We need this pattern to hold)
What about in India, where the variant originated and is believed to be dominant?
Age-stratified data on cases & deaths here is very patchy. But what little there is also hints at a vaccine effect: share of cases/deaths taken up by the elderly (the most vaccinated) is falling 📉