I didn’t expect this to be a series of posts, but there is news since I wrote post number III.
There is a partial change of subject, because the news concerns the Oxford/AstraZeneca adenovirus based vaccine and not the mRNA based vaccines which I discussed before.
In the UK they decided to delay the second dose for 3 months in order to get a first dose in as many people as possible as quickly as possible. This is roughly what I proposed in post I. Since then, Israeli HMOs have noted a significant rate of infection (detected by PCR screening) two to three weeks after the first dose the Pfizer/BioNTech vaccine (link to same post number III and post number II). After 3 weeks, they injected the second dose, so they can’t follow further.
Now the UK has new data. As in Israel, these are comparisons of vaccinated people to the general population not a randomized trial with a control group. The evidence strongly supports the UK approach of delaying the second dose.
First the estimated effectiveness of one dose is 75%, which actually slightly above the estimate of effectiveness of two doses in the phase III trial (62%). Second, the estimated effectiveness of two doses 3 months apart is even higher (82.4%). This suggests that the delay increases the effectiveness of the booster shot (the three week delay is due to the fact that, for other viruses, it has long been known that booster shots before 3 weeks have passed are not effective — 3 weeks was the shortest reasonable interval).
I think the similarity of 75% and 82.4% provides relatively solid evidence that it is not important to give the second shot quickly. There are many possible biases in a study of case vs general population, but it is much harder to see how they bias the estimate of the relative magnitude of the effects of first and second doses. Clearly the estimated benefit per dose is vastly greater for the first dose, so it is hard to argue against delaying the second.
Importantly, there was no clear evidence that protection from one dose waned during the 3 months.
Finally, in addition to the headline data on symptomatic cases, there are estimates of the probability of conversion to a positive result by PCR (this is not the primary outcome measure for any phase III trial but is the outcome measured in Israel which helps explain the difference from phase III trials). In the UK AstraZeneca case, the probability of PCR detected virus decline 67%. This is close to the decline in symptomatic disease. This is reassuring as it would be very bad for herd immunity to give people a vaccine which made them become asymptomatic carriers.
Obviously UK authorities are claiming that the new data prove they made the right choice (authorities always claim that). But this time, they might have a point.