AstraZeneca kind of messed up its virus announcement.

It got one of either 62%, 90% or 70% efficacy. After a mistake with the dosing of a small number of trial participants, scientists found there was a better result in that group. Generally with clinical trials a protocol is pre-registered and results are reported against that. Any 'findings' outside of that protocol should be taken with a large pinch of salt.

AZ only registered one dosage regimen. The problem with the type of subgroup analysis here is that there are many ways in which it can result in spurious results. If you analyse enough subgroups eventually you will find one where the drug worked better simply due to chance. This is OK - you can control for this - but the results are only valid once they have been confirmed thought further trials.

Maybe we shouldn't do science by press release.