[BugsJustFindMe]

> That indicates that paxlovid actually works

It had no measurable effect in the measured group. You can't say that it works based on that. If it had a measurable effect in that group, you would expect to see improvement in the treatment group between the two scenarios.

If (65-69 no treatment) == (70-74 no treatment) and also (65-69 no treatment) == (70-74 treatment), then (70-74 treatment) == (70-74 no treatment) by the transitive property of equality.

Now the reason for (65-69 no treatment) == (70-74 no treatment) may be its own mystery, but we expect to see at least _something_ happen anyway between (70-74 no treatment) and (70-74 treatment) if the treatment had some benefit for that group, and apparently they didn't see that.