This appears to be an appeasement to get people to back off, but as
@JohnCB says, it's not what they were asked to do. It's a compromise, and it's weird that they believe a compromise to be in any way sufficient.
I don't think it's intended as a compromise, so much as an attempt to release their narrative (the changes didn't matter) before patients release a competing narrative (the changes had a large impact). Journals also often won't publish something if the authors have already made it public somewhere, so they might be hoping that they can persuade journals not to publish an outsider analysis of the same data.
They've not released the recovery data, and that's what will cause them the most trouble. If this was an attempt to get their presentation of pre-specified analyses out before others got the data they'd have included recovery outcomes imo. I think that this is an attempt to avoid some of the blow back from the scientific community that they're expecting when they appeal.
I think it's a sign that they won't appeal. A Tribunal appeal at this point could only be on the interpretation and application of the law. They would not be examining evidence, and the (lack of) publication of the data certainly is a matter of evidence, not law. It is completely impossible that they would be able to introduce new evidence at this point.
Their "publication" also won't work as a substitution for releasing the data as ordered by the court. The court order was clear regarding what QMUL must do if they don't appeal, and anything short of that would be in contempt of the court, and cause them extremely serious legal problems.
So I expect we'll be seeing the actual data within in a week or so.
So the already marginal advantage of CBT & GET over SMC dropped by approximately a third when calculated by the more rigorous method.
I think it's more accurate to say that their post-hoc changes to improvement criteria resulted in improvement figures being inappropriately inflated by nearly 50%. The numbers are more impressive that way as well