"Got ME? Just SMILE!" - Media coverage of the SMILE trial…..

[JohntheJack]

Thanks for you hjelp. I really appreciate it.

We are trying to get an article in a journal in Norway and are checking out a lot of things.

This article with comments are really informative
From this https://theconversation.com/researc...rogram-help-youths-with-chronic-fatigue-84769
Here, a comment from Dan Clarke says
That there was no significant difference between groups for their original primary outcome? That they appear to have failed to release results for school attendance figures verified by the participants school, instead relying on self-reported attendance which could be more prone to problems with biased reporting? That data on school attendance was missing for a third of the LP group at 12 months?

The fact that data on school attendance was missing for a third of the LP group at 12 months, is it a fact? Does anyone knows where that statement comes from? Is it reliable and can be used?

This is the paper.
http://adc.bmj.com/content/early/2017/09/21/archdischild-2017-313375

The original primary outcome was school attendance. Check out Table 3 'secondary outcomes': 2.6 (SMC) 3.2 (SMC + LP) at 6 months. Then look at school attendance N for 12 months for SMC+LP is 34 (51 in group at start of trial).

 

[Benji]

@JohntheJack Thank you so! much.

 

[Wolfiness]

Yes, I do. Simon and I have had an email exchange on this.
…
I have had something like a dozen senior people get back to me saying that my point about blinding and subjective endpoints is not right - including in fact I think James Coyne.

What is their argument please?

 

[Jonathan Edwards]

What is their argument please?

They don't really give an argument. I think they are just terrified that people will discover that most psychiatric research is crap.

 

[Benji]

Can I ask again, about the SF 36 physical function data. They go up from 53/56 up to 70-80.
Is that much? Is a increase of 10 much better quality of life?

 

[user9876]

This is the paper.
http://adc.bmj.com/content/early/2017/09/21/archdischild-2017-313375

The original primary outcome was school attendance. Check out Table 3 'secondary outcomes': 2.6 (SMC) 3.2 (SMC + LP) at 6 months. Then look at school attendance N for 12 months for SMC+LP is 34 (51 in group at start of trial).

There is an issue with the figures quoted in the paper in that they seem to be self reports from people who chose to report rather that figures from the school that they say they had permission to collect in the statistical analysis plan,

 

[user9876]

Can I ask again, about the SF 36 physical function data. They go up from 53/56 up to 70-80.
Is that much? Is a increase of 10 much better quality of life?

Its worth reading the questionnaire.

On this link questions 3 to 12 form the physical function subscale. The UK version will have slightly different wording

https://www.rand.org/health/surveys_tools/mos/36-item-short-form/survey-instrument.html

But the thing is that it is self reported so tell people to believe they can do stuff and they get better and they may well give better scores. A real increase of 10 (i.e. 2 question answer changes) could be significant but it could be a slight change over a threshold. I have a theory (based on looking at peoples answers) that when someone scores in the mid points of the scale their answers can change with little real improvement but the edges are less sensitive.

 

[Benji]

There is an issue with the figures quoted in the paper in that they seem to be self reports from people who chose to report rather that figures from the school that they say they had permission to collect in the statistical analysis plan,

Yes; that is one of the obvious things. They chose to move away from the objective measure, and use questionnaire instead. And not explaining why. Not very impressive. This is one of our major points.

Also, I think there might be a point in pointing out that the missing data, which are set to some kind of mean data, could hide harms. So that this study can't be used to prove LP is safe.
Just as PACE couldn't prove that especially Get is safe, when the raw data showed that one third missing data on walking test 52 weeks. And some 10+% of shorter walk distance at 52 weeks that week 0

 

[Esther12]

Can I ask again, about the SF 36 physical function data. They go up from 53/56 up to 70-80.
Is that much? Is a increase of 10 much better quality of life?

If there were not problems with response bias, etc, that seems a significant difference (my hunch, based on little research) so imo that's not an area worth prioritising for criticism.

 

[Esther12]

Also, I think there might be a point in pointing out that the missing data, which are set to some kind of mean data, could hide harms. So that this study can't be used to prove LP is safe.

I was surprised how low their drop out rate was, considering how quackery LP is (they were getting some data from a high % of participants, even if self-report school attendance seemed to be missing a lot). It does sound like they put a lot of effort into chasing people up.

 

[user9876]

Yes; that is one of the obvious things. They chose to move away from the objective measure, and use questionnaire instead. And not explaining why. Not very impressive. This is one of our major point

So what I was meaning is that the school attendance data quoted appears to be self reported given the numbers being quoted. But in their statistical analysis plan they specified that they had permission to collect this from the schools. Hence all the data (or most if people left school) should be there.

There is another issue on the sf36 which is they were phoning those who failed to return the questionaire to get answers over the phone which will have added additional reporting biases.

 

[Benji]

I was surprised how low their drop out rate was, considering how quackery LP is (they were getting some data from a high % of participants, even if self-report school attendance seemed to be missing a lot). It does sound like they put a lot of effort into chasing people up.

In table 3 it seems that a lot of n's are around 30 out of 50 people. 60% is not much?

 

[Esther12]

In table 3 it seems that a lot of n's are around 30 out of 50 people. 60% is not much?

That's true. I've not got their paper in front of me, but I thought that for the primary outcome it was relatively high. Also, my judgement will have been shaped by my expectation that lots participants would have said "what is this BS?!" and then never had any contact with them ever again.

 

[user9876]

That's true. I've not got their paper in front of me, but I thought that for the primary outcome it was relatively high. Also, my judgement will have been shaped by my expectation that lots participants would have said "what is this BS?!" and then never had any contact with them ever again.

If people were not responding they sent out a smaller set of questionnaires and then phoned and got answers to the SF36-pf and possibly one or two others. So the imbalance may indicate the level of chasing up.

 

[Wolfiness]

They don't really give an argument. I think they are just terrified that people will discover that most psychiatric research is crap.

Oh come on!?
Prof. Wessely's book is currently on sale for 40p at Amazon. Maybe a few of us could get together and split the cost to check it out?
https://www.amazon.co.uk/gp/aw/d/0198526423/

 

[SamanthaJ]

This probably isn't important in the context of the many, many things wrong with this trial, but it bugs me.

I've previously seen a couple of parents say that attending the Bath clinic was useful for getting the schools off their backs.

In a well-designed trial (ha!) with school attendance as an outcome measure, shouldn't the concessions and support sought from the schools be the same for every participant? Is there any mention of this? If the therapists and doctors involved in the trial are also in contact with the schools, this role strikes me as something that could be used to influence attendance figures, through securing either more helpful, or stricter arrangements for some children. This could even happen unintentionally if there's no rule about it. Although since the schools' own figures were not sought anyway, this might not be very important.

 

[NelliePledge]

Oh come on!?
Prof. Wessely's book is currently on sale for 40p at Amazon. Maybe a few of us could get together and split the cost to check it out?
https://www.amazon.co.uk/gp/aw/d/0198526423/

I don't think it would flush very well

 

[Sean]

I don't think it would flush very well

I ain't letting that paper anywhere near any of my precious bodily orifices. Never know where its been.

Yes; that is one of the obvious things. They chose to move away from the objective measure, and use questionnaire instead. And not explaining why. Not very impressive. This is one of our major points.

I would describe it as straight corruption of the scientific process. It is both an intellectual and moral failure.

 

[Snow Leopard]

Can I ask again, about the SF 36 physical function data. They go up from 53/56 up to 70-80.
Is that much? Is a increase of 10 much better quality of life?

An increase of 10 is not necessarily very much. The SF-36 is also not a linear scale - some movements on some questions have much greater impact on quality of life than the same "score" increase on points of other questions.

I was surprised how low their drop out rate was, considering how quackery LP is (they were getting some data from a high % of participants, even if self-report school attendance seemed to be missing a lot). It does sound like they put a lot of effort into chasing people up.

Only the true believers participate in studies like this - I mean if you're not gullible, why would you participate in a trial of LP anyway?

 

[anni66]

Remember LP itself " screens" for suitability and within a trial setting this would be vital to facilitate " efficacy" of the process