Ann: Mesoblast Phase 3 CLBP Trial Completes Enrollment, page-150

I think you are missing one important detail. The intervention is not assumed to work, the P2 trial has proved it does work at the dose being supplied in the P3 trial, 6 million cells. That dose already comfortably exceeds the required minimum performance.

However unless the P2 dose of 6 million cells was repeated in a P3 trial as the rules stand it could not be released onto the market, that is not inefficient design on the part of MSB as I interpret it. It is unknown if the addition of HA will make any positive difference but it may, the placebo arm is fairly essential to for that alone as I see it, and I have never designed one of these trials. If it does not exceed the minimum % improvement compared to Placebo the fall back is MPC-06-ID on its' own, presuming the performance mirrors the previous trial.

The trials are not run to test cognitive bias. I believe they are looking for a better result than just MPC-06-ID on its' own.

This is the info on the P2 trial results released to the ASX
http://www.mesoblast.com/clinical-trial-results/mpc-06-id-phase-2

Note that at least some of the participants have durable results out to 3 years now.

If you don't want to look at the rest of the info here is just one of the charts from the P2 trial.