New preprint on the Oxford/AstraZeneca vaccine & B.1.1.7 ("UK" variant). Includes data on the people diagnosed with Covid-19 up to Jan 14 (so a month of extra data than their recent preprint on single dose)... 1/n papers.ssrn.com/sol3/papers.cf… (HT @joelving) ...1/n
...The vaccine reduced efficacy (more on that later). They were able to sequence the virus for 256 out of the 499 people sick with Covid-19...2/n
...The analysis is based on the UK phase 2/3 trial, where participants had regular swabs. So it includes the people they detected who had asymptomatic & symptomatic Covid-19....3/n
...Self-swab home tests were used. People who got unblinded in January so they could join the vaccination program if they'd been on placebo are excluded from analysis at the point they were unblinded....4/n
...The low-dose-first group are included, & they assessed viral loads in swab results & blood samples, as well as attempting to sequence the virus...5/n
...People had multiple samples. This diagram shows how they arrived at the group for their analyses. Out of 499 people, they were able to sequence 256 (N=cases), but with exclusions, that ended up with a subset of 179 out of the 499 people (36%). ...6/n 
...The variant started appearing in this group in November, & increased in proportion across Dec/Jan. Altogether 28% were infected by the new variant, but that rate would clearly be higher if it was counted only for the months in which people were being infected by it...7/n
...People infected with the variant were symptomatic at about the same rate as those infected by the original virus. Vaccine efficacy rates they calculated for the 2 groups are an estimate of the difference, but with most infections not sequenced & counted, that's all ...8/n
..Because this is a subset of the trials' data (just 1 trial), & it's the one with the low-dose group (which also happened to have higher efficacy related to interval of dosing), vaccine efficacy rates are considerably higher than they are for standard dosing across all data..9/n
..This excerpt from Table 1 shows the difficulty with calculating vaccine efficacy when the variant status of most of the people infected isn't known. When you look at all, & symptomatic status separately, the main thing is a substantial % point drop in efficacy, but...10/n
...with so much missing data &c, I think it's risky to say by how much. If vaccine efficacy against non-B.1.1.7 is in fact quite high (& I think it's too soon to know), even a substantial drop would leave protection still high. (Neutralizing antibodies were also lower) ...11/n
...Press releases for another 2 (Novavax, J&J) suggested there could be a drop in efficacy. This preprint, though, is the most data released for a vaccine on this & it points in the same direction. It ups the stakes on suppressing outbreaks. /12
...PS...The authors' interpretation is more positive than mine - there's a good report from that perspective here: nytimes.com/2021/02/05/wor…
...Important omissions from ⬆️: just the symptomatic infections from table 1 for the whole group. Also shows impact of the missing data. After exclusions, it was 74.6% (42-89%) for B.1.1.7; for non-B.1.1.7, 84.1% (71-91%). A lot of uncertainty.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
