Michael Sharpe skewered by @JohntheJack on Twitter

Snow Leopard said:
Right. It's ultimately about reducing costs for insurance companies and public health systems by not spending money on exhaustive testing and interventions for people with symptoms only medically explained after exhaustive testing and expensive interventions by highly trained specialists medically unexplained symptoms. Michael Sharpe has spoken at length on this topic for a long time.
Click to expand...
 
There's conflict of interest and then there's bias, they're different but overlapping constructs.

People tend to think of pharma money when they think of coi, but that's only one type of coi. We should be much, much more concerned about other, more subtle ones. They are really messing with our science.

The concept of COI is that you have a greater chance of a future reward if the outcome goes one way than if it goes another.
- If you design a study so that you will only get funds to complete it if the initial results go a certain way, then that's a COI.
- If one particular outcome would reduce your chances of getting other grants in the future (e.g., the DWP won't give you any more money), then that's a COI.
- If your next job promotion depends on whether you publish the study in a prestigious journal, and the likelihood of that is greater if you get a certain outcome (which it often is), then that's a COI.
- if you stand to get more fame, royalties, paid travel, book deals, or more therapy clients if the outcome goes a certain way, then that's a COI.

Its not a COI if you get the benefit irrespective of study outcome.

A COI does not occur just because you believe in one outcome as more likely than the other. At the core of a COI is probable future reward beyond the research itself. Belief can still cause bias though, massive bias, but that's different from a coi.

So a COI causes bias, but not all bias is due to a COI.
 
Bias is fascinating. Looking at Psychology's replication crisis, I've noticed the sources of bias are different across different fields.

In Social Psychology, the bias is caused by people wanting to get published in high profile journals and get their names in the papers. Sexy, eye-catching stuff gets rewarded. Usually the outcome has to go a certain way for that (ie. there's a definite COI). So you get "Sad people really do get the blues: Feeling sad selectively enhances attention to short wavelength light" (no lie! although the real title was not exactly that).

In Clinical Psychology, the bias is caused by strong theoretical allegiances to particular approaches or views. So you get "Training depressed people not to look at sad pictures relieves their depression" (because CBT for depression is based on the idea that negative attentional biases and negative thoughts cause depression). This one also not a lie! Sometimes, there's a COI here too (if you stand to get more clientele if the outcome goes a certain way), but sometimes its just sheer belief that causes the bias.

Both the studies I described above were absolute bullshit when you gave them even a cursory look over. Noone bothered to. Because often the readers, citers and promoters of the works have their own biases which just happen to align. "Sad people really do get the blues" sells good copy. "Positive attentional training eases depression" appeals to every psyc wanting to show that the ideas underlying CBT for depression are evidence-based.

We just tend not to spend too much time looking at the results that went the way we thought they would.
 
Esther12 said:
They did declare their insurances COIs in the Lancet paper, and there was still very little scrutiny from the academics first looking at the paper.
Click to expand...
But if those COI were not declared to the participants at the start of the trial when they signed up to it, and then once the trial results were published they find COI are declared, that's akin to some kind of breach of contract.
 
chrisb said:
It does seem strange. In another discipline every third year undergraduate will learn about subconscious ex post facto reasoning. How can leading researchers be unaware of potential for bias and the consequent need for disclosure?
Click to expand...
chrisb said:
Decision first. Reasons later.
Click to expand...
And in the sense that a contract legally committed to was then changed under the feet of those engaging in that contract prior to its completion, and without any notification to those participating in that contract, so thereby denying them a legal right surely.
 
Barry said:
And in the sense that a contract legally committed to was then changed under the feet of those engaging in that contract prior to its completion, and without any notification to those participating in that contract, so thereby denying them a legal right surely.
Click to expand...

Quite.

Also the PACE practitioners who carried out CBT/GET, were they told of the conflicts of interest?

The GP's who referred patients?

Th ethics committee members?
 
General question, does anyone know what happens when researchers apply for any MRC ( or other ) grant, do they not have to declare any COI right there, at the start, to ensure they are eligible for the grant in the first place?

So in the case of PACE might there be evidence of disclosure of COI right at the start of the process?

And if not why not?
 
Last edited:
Is itjust me, or are those MRC guidelines defective. On second thoughts, I know I am.

1.4.4. on p9 provides for disclosure of "Unremunerated involvement with any organisation named in the grant or which might benefit from the outcome of the research that are not mentioned in the application..."

It should surely provide elsewhere for remunerated involvement, but I cannot see it. But there is a catch all general provision.
 
Woolie said:
At the core of a COI is probable future reward beyond the research itself.
Click to expand...

I assume there is also a concept of past reward implying future reward. So if I have done work for insurance companies in the past then I am likely to do so in the future assuming good status follows from good results.

The thing I am wondering about whether institutions should declare CoIs. For example Barts/QMUL have a ME service that does one particular treatment and their revenue stream for this is perhaps tied to the results (and perhaps liabilities if it fails in terms of employment contracts). QMUL was the principle sponsor for PACE and I believe legally responsible.
 
Adrian said:
The thing I am wondering about whether institutions should declare CoIs. For example Barts/QMUL have a ME service that does one particular treatment and their revenue stream for this is perhaps tied to the results (and perhaps liabilities if it fails in terms of employment contracts). QMUL was the principle sponsor for PACE and I believe legally responsible.
Click to expand...

Yes, and the COI of the DWP who gave a considerable grant to PACE?
 
Daisymay said:
Yes, and the COI of the DWP who gave a considerable grant to PACE?
Click to expand...

But only if they have the ability to influence the trial. Giving money to a cause is not a CoI. A cause taking money from someone with alternative interests could be.

My point about QMUL is that as an organisation (and legal entity) they were highly involved in the trial (as primary sponsors and employers of one PI) as well as an interest in the outcome (as a treatment centre and through the research assessment process that leads to additional money). How can an organisation who are in such a position also be responsible for governance and yet hide data.
 
Adrian said:
But only if they have the ability to influence the trial. Giving money to a cause is not a CoI. A cause taking money from someone with alternative interests could be.

My point about QMUL is that as an organisation (and legal entity) they were highly involved in the trial (as primary sponsors and employers of one PI) as well as an interest in the outcome (as a treatment centre and through the research assessment process that leads to additional money). How can an organisation who are in such a position also be responsible for governance and yet hide data.
Click to expand...

Yes I see.
 
Daisymay said:
General question, does anyone know what happens when researchers apply for any MRC ( or other ) grant, do they not have to declare any COI right there, at the start, to ensure they are eligible for the grant in the first place?

So in the case of PACE might there be evidence of disclosure of COI right at the start of the process?

And if not why not?
Click to expand...
Also, is there nothing built into a trial protocol (including ethics etc), dictating what the correct procedure is if COI arise during the course of the trial? This stuff should be laid down in formal legally binding procedures, not random scribblings on the back of a cigarette packet.
 
You must log in or register to reply here.
Back
Top