CORRESPONDENCE The PACE trial of treatments for chronic fatigue syndrome: a response to WILSHIRE et al (2019) Sharpe, Goldsmith & Chalder

This raises serious questions for the journal. The reuters article would have had to take this response into account had there been simultaneous publication. Are they able to respond to the very obvious suspicions?

 

It also brings into question The Lancet's fast-track process, whereby trials that have lodged a protocol can then undergo an expedited review. This makes no sense if crucial elements on the protocol are then amended at publication.

Has this issue been raised with The Lancet?
[eta - apart from me banging on about it on Twitter that is]

 

An excellent analysis @Carolyn Wilshire. It looks as if you have been nobbled. The Sharpe, Goldsmith, Chalder piece would have looked ridiculous alongside this. You might have to re-submit it to J Health Psychology!

 

Sharpe once claimed that the reason Lancet fast-tracked it is because they pre-registered the protocol.

He also said it's not an issue that they modified their protocol from how it was registered because they prefered it this way (I can't believe this is an acceptable answer... SMH).

So the question of fast-tracking and the weirdly inexplicable enthusiasm of Horton for an unblinded, non-controlled trial with self-reported outcomes that did not adhere to their protocol really needs a thorough explanation, especially given the pre-planned media tour that Horton was looking forward to and the SMC seemed to have been preparing for well in advance.

As always it just confirms that the trial was a formality, would produce the results they wanted no matter what and that there was political will to make it skip unhindered to the finish line. This is everything science is not about, it's a massive perversion of the entire point the scientific method.

 

There was some crying foul by Sharpe and Vogt about why it took so long for their letter to be published.

Now this companion response is somehow held up separately.

So... this is weird and suspicious.

 

I'm not sure that was the case though. They made some things damn hard for themselves. It was massively over-complex and was a logistical nightmare, as the TMG minutes recount. They were clearly trying to cover all bases and not really thinking of what the participants were going to have to put up with as a consequence.

 

Oh, I don't mean it was easy. It's pretty hard to cheat in a way that seems legit and you're right about the TMG minutes showing how much trouble they had keeping the appearance of doing a serious trial. Reality really didn't want to conform to their expectations.

But the conclusion was reached decades ago and nothing in between would have affected that.

 

... and if you include the recovery paper, are also consistent with regression to the mean ... all results including basically doing nothing have the same outcome.

 

In good science this ends all their claims. Since they wont accept that ...

I still think the deliberate use of SD on SF36PF data, given the PDW 2007 paper, is evidence of deliberate scientific misconduct and needs to be formally investigated by independent and qualified investigators.

 

Ditto.

 

I have always thought this. Plus not disclosing in the papers the fact that 13% met the fraudulent "normal range"/"recovery" thresholds. No one has held them to account for the fact that this was in no way a "normal range" given their population data. It is a seriously overlooked point.

 

Sharpe seemed particularly sensitive to this point.

 

Such a great response with many well reasoned points that ought to be obvious.

One point in particular that never ceases to amaze me is where they accuse the reanalysis of 'not using an a prirori analysis plan'. So... their change is fine because they simply felt it was better this way, but if anyone else does it (which was not even the case) it is bad? I mean, yes it is bad for all the reasons mentioned as a response to THEIR changes, but come on. You cannot have your pie and eat it too.

 

My favorite is dismissing the reanalysis because it only used partial data.

Because they withhold the data.

I don't understand how an editor finds that acceptable. Completely ridiculous.

 

I like how they claim to prefer their own paper over the Wilshire one when the Wilshire one was just anaylised according to the original protocol.

 

Update: since I last wrote there has been much frantic action by the BMC editor to hurry our response through. I don't think there was any intention to the delay, after all it was the editor who invited me to write a response in the first place. BMC's interests are best served by hosting a discussion that presents both sides of the debate.

Thanks to @Jonathan Edwards, @Trish and all others who said nice things about our response. Its pretty softly softly, and I expect some of you might have wished we'd made some points more strongly. I do feel much more strongly than what I said in the response, but I wanted to take the scientific high ground by addressing only the science. I feel this strategy may pay better dividends in the long term.

 

@Carolyn Wilshire I think we will all be pleased to here what you say. Do you think the editor could be prevailed upon to give an indication as to the length of prior warning of the date of publication which he would normally expect to give to an author?