Update: this has turned into a stupid, needless, mess. Which frankly seems to be AstraZeneca’s pandemic brand so far. It turns out that the company’s press release (as discussed below) is apparently more of an interim read than reflective of the final data. The NIH took the extraordinary step of stating its concerns about this late last night, and there’s all sorts of fallout today – see this article at the Washington Post for more. It seems clear that the final efficacy figure will be lower than the 79% quoted from the initial press release, and it seems clear – at least to me, and to a lot of other people – that handling the data in this way was a deliberate attempt to make that number look better. Anyone should know better than to try this, but especially a large, experienced, multinational drug company whose data handling during this vaccine’s development has already been the subject of a lot of (justified) criticism. I can’t think of a benign explanation, so unless we would like to assume that the company is so breathtakingly incompetent than they did this bizarre move for no reason at all, we’re left with bad ones. This is the sort of thing that gets high-level people submitting resignations and “pursuing other interests”. Isn’t it?
This morning brings news of new trial results of the AZ/Oxford vaccine, with results from the US, Chile, and Peru. This one has been a much more complicated and difficult story than it really should have been – at least, I think that’s how it’s going to look years from now. As you’ll likely remember, the initial trial results (from the trials in the UK, South Africa, and Brazil) were. . .well, they were a mess, and a good part of that mess was the way that they were presented to the world. Later results in a smaller group that had a longer dosing interval cleared things up a bit, but in recent weeks the focus (rightly or wrongly) has shifted from the efficacy numbers to the safety ones.
Now the new trial has shown 79% efficacy against symptomatic disease, and says that the vaccines showed 100% efficacy against severe disease and hospitalization. There’s always room to argue about that latter number, because the number of such cases is not large and the statistical power involved is invariably going to throw up a wide confidence interval on that 100% number. There were over 32,000 participants, with one-third of them in the placebo group. In this case, the press release mentioned 141 total symptomatic cases, and it’s come out this morning that there were 5 severe cases in the 10,000+ placebo group, versus zero in the roughly 21,000-patient treatment group. It’s worth remembering that the 95% confidence interval for the both-doses-completed group in the original trial data was 41% to 75% efficacy, so these new numbers are higher than people were expected based on that. But as mentioned, I would be careful about trumpeting a solid 100% protection in severe disease, for this or for the other vaccine trials – that said, that protection is obviously very strong, and that of course is good news.
What about safety? The review board specifically looked for thrombotic events of all kinds (as well they might after the recent reports from Europe), and saw no vaccine-related signals. But remember, this was about 21,000 people, so you’re not going to pick up something that’s real but happens at a very low level. And that’s where the arguing is, how many such events there have been in the vaccinated population, what different sorts of cardiovascular problems have been seen, whether they are above the levels you would expect to see in a general unvaccinated population, and (finally) just what those background levels really are. The dust is still in the air about all this, but I would not have expected this trial to help resolve these issues, which are down in the per-million-patients level.
If only the original trial data could have been as clear a readout as this one has been! Instead we had questions about overall efficacy, questions about subgroup analysis, about single doses versus two, half-doses versus full, intervals between doses, etc. All that is on AZ and Oxford. Now, they’ve had help muddying things since then, for sure. For example, a German bureaucrat misread the trial data and based on that a newspaper report spread the idea that the vaccine was only 8% effective in patients over 65, which was completely wrong. The latest arguments about safety have not always been conducted at the most honorable level, either, with accusations of nationalism and favoritism flying around. And back out in the world of real data, it appears that the vaccine is significantly less effective against the B.1.351 variant (although to be sure, the other vaccines are also less effective to some degree against it as well).
No, the going never would have been smooth, but it didn’t have to be like this. There will be books about the way vaccines were developed and launched during this pandemic, and there’s enough material for one just on this vaccine in particular. But the latter chapters are still being formed: this vaccine should continue having a big role in knocking down this pandemic around the world. Today’s data give more confidence that it can do so.