Post-Hospitalisation COVID-19 Rehabilitation (PHOSP-R): A randomised controlled trial of exercise-based rehabilitation, 2025, Daynes et al

no, it is not. in my view, any change should be explained.

 

Shouldn’t changes from the protocol have a further ethical approval as well as being reported and explained in any write up?

 

Ethical approval probably depends on the type of change.
Reporting and explaining in publications should be mandatory.

 

I'm not sure that changing a reference would automatically require new ethical approval. I think it would likely depend on how substantive it is. I haven't looked at the protocol at this point but do we know if/what changes occurred? But yes, if the protocol had a different MCID cited, they should definitely at the least explain in the paper. Saying that new research has come out I think would be considered a perfectly normal explanation (unless they themselves cooked up and published a shady new analysis in order to get a better advantage in the trial itself).

 

This could be done through a sensitivity analysis that would show the impact of any changes on the results.

 

Thanks for highlighting (the flaws with) this study David.

Minor issue:

I got lower drop-out rates: 11/56 (20%) in the face-to face group and 17/63 (27%) in the remote group.

Strange that they don't report the data for the primary outcome of the control group. I suspect that the control group decreased based on the data that they do report.

• The face-to-face rehabilitation group improved from 285 m to 312 m. So that is an average increase of 27 meter, lower than the MCID.
• In the remote group the The remote rehabilitation group improved from 353 m to 388 m so an increase of 35 meter, similar to the MCID.

But for the comparison with the control group they report increases of 55 and 34 respectively.

That probably says it all. Quite surprising and sad that these patients weren't able to improve more.

 

Had a look at the protocol which said:

So 50m was originally viewed as the minimum important difference?

Reference 14 is:
Minimum clinically important improvement for the incremental shuttle walking test | Thorax
Which says:

 

The paper says they added multiple factors in the statistical model which weren't mentioned in the protocol:

The protocol says they would correct for the false discovery rate (FDR):

But I don't see this in the paper.

 

I was going with the numbers they used in the per-protocol analysis: "140/181 participants were included in the per-protocol analysis, 40/56 (71%) face-to-face, 38/62 (61%) remote, 60/62 (98%) usual care completing 75% of the intervention and the follow-up measures." From this, it would mean that 29% and 39% in the face-to-face and remote groups, respectively, were lost to follow up. It's true that in the consort diagram they cite the numbers you mentioned. But that doesn't make a lot of sense. If those people completed the study, why were they not included in the per-protocol analysis? Those not included in a per-protocol analysis would be considered lost-to-follow up. Can anyone make sense of this for me? I'm not much of a statistician, so I can't figure it out.