Don't forget that the BioNTech/moderna trials weren't really comparable - they relied on self-reported symptoms to trigger checks for infection. That is likely to lead to a higher apparent efficacy at the same underlyng efficacy.
By now it's clear that for all three vaccines regardless of the exact efficacy of the vaccine, the chances of getting a severe case is greatly diminished. It's at least plausible that getting a vaccine reducings the severity of an infection in general; specifically that it might push what would otherwise have been a mild case into the asymptomatic (but infected) category. I don't know how large that group is, but given how many mild cases there are, it could be quite sizable. And those cases would not have been included in moderna and biontech/pfizer's efficacy numbers, but would have been included in the oxford/AZ trial's numbers.
So while I'm no expert in any of this, I think it's at least fair to say that it's not trivial to compare the efficacies reported from the two mRNA vaccines to that of the oxford/AZ vaccine, because of this difference in methodology. Including vs. mostly excluding asymptomatic cases will affect the numbers.
The point being? All of these trials have their limitations; the oxford/AZ trials may have had bumps (but nothing all that serious sounding), but they also have extra data in the form of asymptomatic cases. The comparison isn't straightforward.