SMILE patient cohorts

Not that it matters much. The whole trial is rubbish because it completely fails to control for bias affecting subjective outcomes.

 

In fact it probably took control of the bias rather well!!

 

The only way to improve on LP is a therapy that attempts to convince patients that answering questionnaires of the kind used in clinical trials in the most positive way possible regardless of the truth will cure them thanks to (insert neuropsychobabble here).

Since this is so weird, they also need some way to filter out participants that don't believe in this.

When this is achieved, astonishing positive results for any condition and problem are guaranteed.

PS: I even have an idea how to justify this weird idea. That thoughts create the brain's reality but especially so when expressed in words or text and communicated to others in the most committed way, such as speaking publicly in front of a large audience, or when being examined by a professional in position of authority, or in persistent messages like a questionnaire.

 

It's an old ruse, used by a boss of mine in the 1970s. If patients answer the questionnaires right they stay on the consultant's clinic list. Otherwise they get to see the registrar. The cure is the magic of the consultant. It worked very well until colleagues cottoned on. These days colleagues are too busy looking after their own problems to notice.

 

Isn't this what neuro-linguistic programming is about?

 

Uhm.. yes? It is not optimized for clinical trials though. Forget telling your disease to stop while making gestures and standing in a magical circle. These rituals and modalities are outdated. The most powerful way to change the brain is clearly to give the best score on health related questionnaires and telling healthcare personell how powerful the treatment is.

 

Thank you, I can check the actual data. I didn't know it was released. I also did not know about Intention-to Treat and Pre-Protocol Analysis. I'll have to have a look into those.

It looks like that is explaining where some of my confusion is coming from. The tables and figures are referring to ITT. This intuitevely makes sense in drop-outs, but less sense if cohort swapping occured, although it looks to include that scenario. Again, have to read about it. Still confused about 11 being subtracted from their allocated cohorts, still looking into if 11 patients were not included or swapped and included and I guess eventually seeing if ITT is actually being applied correctly. Have to look at these numbers with a different lens now.

 

Thanks to @Trish most of what I’ve said looks fairly explainable.

What it looks like now is the patients did swap cohorts, but were counted in their randomized cohort for ITT analysis. The graphs and tables refer to their randomized cohorts minus the consent withdraws and not-followed-ups. I’m now assuming these 11 are genuine not-followed-up patients rather than cohort swappers. If they weren’t, ironically, that would be the new problem.

As for appropriate ITT analysis, there may be some problems with large proportion of the patients cross over, but I don’t know if 11 or 12 reaches that threshold. The SMC cohort looks a little on the low side for this type of analysis, but I don’t know enough to say. A separate analysis can be performed with just the patients that completed their respective randomized treatment (per-protocol), although whether that is necessary or not seems subjective.

>CPMP guideline states that in a noninferiority trial, the full analysis set based on the ITT principle and the PP analysis set have equal importance and their use should lead to similar conclusions for a robust interpretation

I did look at the raw data, but was unable to say definitively what happened because there is no information on which treatment was completed for each patient. I feel a little silly for wasting anyone’s time, so sorry if that was the case.

 

Absolutely. It's good to be diligent.

 

I think this is important, especially given it is taken such a lot of discussion to make sense of what actually happened. Either this was a very hasty poor write up, despite the fact that the authors had an opportunity to redraft following negative scrutiny or it suggests the authors themselves do not have the niceties of good experimental design and how to eliminate bias clear in their own minds.

What is worrying that this paper’s author list includes members of the Bristol Randomised Trial Collaboration and the lead author of a recent Cochrane paper addressing risk of bias in randomised trials (https://www.s4me.info/threads/rob-2...in-randomised-trials-2019-sterne-et-al.11025/), suggesting these problems may be more wide spread than just Bristol ‘CFS’ research.

 

@Trish @Barry @Peter Trewhitt

Thanks. It's great to have a place where we can have this kind of discussion and dig a little deeper as patients but still try to find a balance. I didn't mind look into it and I don't mind being wrong. That is part of the process. I'm grateful that people helped me out.

 

One last thing. Isn't it strange that the SMC cohort had 4x the 'not-followed-up' of SMC + LP? (9 to 2).

After all, it's included in the LP arm and you would expect dropouts after LP. If people most often drop-out after are doing worse, why were so many doing worse in the "easier" cohort?

From the raw data, 10 SMC patients with missing 6 month spf data and 5 SMC + LP spf data had slightly different average baseline fatigue scores: 58 to 54 by my count. Not a huge difference.

---------
tinfoil question; Could the authors have more aggresively followed up on the SMC + LP cohort?

 

Why Dreampop, you are almost as suspicious as me

 

Or given the way LP works, would the LP cohort not be " feeling better" and have the need to remain engaged. Is that not how brainwashing works?

 

I was describing the problems with EC's work to a friend when I realised something. Bristol is behind training social workers and the police that children with ME are suffering emotional abuse so that child protection orders should be given - that it is the parents causing the illness.

Yet the SMILE trial uses the LP to teach children that it is their own fault they are ill and they can cure themselves by thinking right - so the children are at fault for being ill.

Makes no sense.