Interventions that manipulate how patients report symptoms as a separate form of bias

When we discuss the problems with trials on cognitive behavioral therapy (CBT) and graded exercise therapy (GET) we usually mention things like lack of blinding + subjective outcomes or the lack of a credible control group etc. These are methodological weaknesses that are generally regarded as leading to high risk of bias.

But as most of you know, the problems with trials on CBT and GET go further because the intervention consists of influencing/manipulating how patients view and report their symptoms. Usually, code-words are used for this like self-efficacy, reduced symptom focussing and catastrophizing, tackling fear-avoidance etc.

Many have already highlighted this issue but I was wondering if we should argue for a separate risk of bias domain for interventions that influence how patients report their symptoms. I think that might make the problem clearer to those who haven't taken it seriously thus far.

Perhaps we should try to expand the scope to behavioral interventions outside ME/CFS. 'Pain education' is another example I can think of that seems to do the same thing. I suspect most forms of CBT suffer the same problem, although to a lesser extent than in ME/CFS.

It would be good if we could collect examples from other fields (including alternative medicine and quackery).

 

Some hypotheses of how functional neurological disorders arise involve abnormal attention towards symptoms.

 

Not sure whether this is actually needed.

Clinical trials just should apply scales that are both truly validated and relevant for the illness.

I think that there exist some good health anxiety and depression scales , but they are validated to measure symptoms of health anxiety and depression. Why apply them as measures to investigate treatment benefits in an illness that has nothing to do with the health issues for which they were validated?

It might be OK to use them to check for comorbidity or side effects but that has to be specified in the protocol.

And just don't include items that are based on unproven assumptions about symptoms or certain illnesses like central sensitization, anormal attention, fear avoidance etc.

There might be a point in highlighting that even validated measures for non-specific health issues/ illness like the SF 36 quesionnaire, or maybe some good pain or fatigue scales, still are especially prone to manipulation when applied alongside certain therapies. However, if they were used only together with reliable objective outcomes, that would shoe the problem anyway?

Edit: On a second thought, I think it's a good idea anyway, as Michiel said, to make the problem clearer. But perhaps needed to include criticsm on the widespread problem of per se inapt measures, e.g. the Chalder Fatigue Scale, depression and health anxiety scales, for ME/CFS and other illnesses, particulaly those labeled as 'MUS'?

 

IBS is another condition that has the same issues.when it comes to how patients view and report symptoms.

 

Is there any literature specifically about this? If not, there should be. It seems to me that the CBT/etc industry has been premised on a self-perpetuating body of work in which researchers and practitioners have figured out how to game the system and generate a methodologically flawed but politically expedient "evidence base" for what they are doing. Then you can't question any of it without questioning all of it. And since reversing course on this is deemed unacceptable and politically undesirable, it all continues. It has come too far to turn back.

As the OP says, part of the mechanism is undoubtedly the promotion of these questionable concepts, labels, and buzzwords such as "fear avoidance", "health anxiety", and so on, which are uncritically accepted by many (and are politically popular), and constitute a framework to be inevitably pushed on the participants of such trials. This bias should absolutely be dealt with head-on as there is so much manipulation, priming, etc that occurs through the use of this framework. Practitioners and researchers may even buy into the concepts themselves and come to believe in their validity. They may genuinely feel they are "educating" people instead of manipulating and priming them. Courses are taught to masters and PhD students when they "learn" about these concepts so they too can become practitioners; they may consider this perfectly natural as it's something they learned on a course. It may come to seem like "common sense", rather than a conceptual and practical framework loaded with bias, pseudoscience, and junk terminology designed to produce "correct" convenient outcomes in trials.

When you put this together in a research environment with the use of subjective outcomes, lack of blinding, controls, etc, you have a sure-fire recipe for generating said flawed "evidence-base". The problem seems rampant and out of control. It hardly only affects ME/CFS patients, seems to be a global problem affecting this kind of research and practice. Is anyone writing about and documenting this as a more global phenomenon? I am unable to read much.

It seems that people I have known just accept this as "the way it is", despite being aware of the problem, and simply do their own (unrelated) thing/research/etc. It has seemed to me for a while that a lot of the problems that people concerned about ME/CFS have had in getting their message through about methodology are traceable to the fact that this empire of flawed "evidence" cannot be allowed to fall.

 

It might not be worth including this into risk of bias tools because interventions that influence how participants report subjective outcome measures are likely going to be totally flawed unless they use objective outcomes.

I meant more that we should point out that this is a separate form of bias that is not addressed in risk of bias tools like the new one made by Cochrane. In such cases, saying that patients and therapists were not blinded is a serious understatement of the risk of bias involved. The risk of bias is not comparable to say, two surgical interventions that are being compared without blinding.

 

I agree that this is another level of bias that does not appear to be taken into account.
I am personally against any use of 'tools' though. It irrational and as long as it is considered OK the system will be manipulated.

Using a tool works like this:

1. Gather reliable evidence for factors affecting bias from all possible sources.
2. Derive a set of rules that seems to cover those factors and put it in a tool.
3. Use the tool.

A better approach is:

1. Gather reliable evidence for factors affecting bias from all possible sources.
2. Use it, case by case.

Since in the first method 2 can only be at best a rough approximation to 1 it has to be better to use 1.

The practical reason for having a tool is to make it possible to employ people with no real understanding of clinical trial psychology to apply some rules. I think that should not happen.

 

My devious plan is to make the algorithm of the tool so complex and realistic that only those who understand clinical trials methodology know how to use it and so nobody sees the benefit of using it anymore.

No seriously, I think it might be useful to name and treat this as a separate form of bias that is not addressed in risk of bias tools. Lack of blinding and an active control intervention does not fully address it. It would be good if we could come up with a catchy name ".... bias" and some examples from outside ME/CFS research.

Yes, I remember a trial of hypnosis for IBS that was promoted rather uncritical in the Belgian press.

 

Worth doing but a lot of work. But yeah none of the issues with how bad BPS research in ME is are unique to research on ME, it's wherever psychology has a say that all standards just completely collapse down to zero and bias is actually not only good but necessary. Basically the entire concept of psychosomatic medicine is completely dependent on this kind of deceitful manipulation.

And this playing with language features heavily. I absolutely hate this framing of "self-efficacy", which we know very well means not trying hard enough but they just pretend they don't mean that, even though we all know they do and so do they. It's all out in the open, but that's because those practices are beloved, otherwise psychology has basically no useful role to play (so far) in medicine. Which would be improvement, they went live even before they had an alpha build and that's their own problem. Nothing wrong with starting from scratch, but that's hard to sell.

 

Yes, a name would be useful (assuming one does not already exists that we've not come across). How about something like indoctrination bias?

 

I thought you were serious- such good idea, although I was a bit shocked you admitted to it.
But seriously, names for this bias: something like 'psychosocial bias'? or 'cognitive behavioural bias'? or even 'cognitive manipulation bias'? Or something more allegorical like 'Animal Farm bias'? (maybe not so seriously).

 

Or an acronym?

IMPRESS bias (Interventions that aim at Modifying how Patients REport SymptomS)

(not serious.)

I still think it's not just 'one' bias -- it's an entire subdivision of clinicians' distorted perception of certain illnesses and symptoms that implies a complex of biases, including the application of completely inappropriate measures etc.

@JohnTheJack's "inherent' I think would fit but might need an addition to be understood?

Edited for clarity.

 

Perhaps one needs to differentiate the risk for bias in in interventions that aim at modifying how people perceive and thus report symptoms?

1) In health anxiety or depression, I think some of the symptoms actually are how patients perceive symptoms.

It might still be difficult to differentiate if there is a real treatment effect or patients just answer what they learned will please the therapists and investigators. So perhaps in this case intervention inherent risk of bias is rather the risk of amplifying other risks of bias?

2) Symptoms and illnesses that can't be influenced by altered perceptions. The intervention is completely inappropriate in this case. But why would participants let themselves manipulate by the therapist?

Perhaps largely the same applies as in 1) -- and there is only an intervention inherent 'amplifyer' of other, well-known risks of biases? (Suggestiveness, expectations of both patient and therapist, wanting to please the therapist etc.?)

I'm sure others know the correct terms for the kind of biases I mean.

Apologies if I'm being either too trivial or inaptly complicating things.

 

Instruction bias?

When research participants are instructed to report an improvement in health or instructed to view their health in more positive terms.

 

One could also view it as defect in study design rather than as form of bias. Then it's perhaps easier to describe:

It is especially important to use outcomes which are not affected by the participants perception of their health when testing an intervention which aims to improve health by modifing the participants perception of their health.

 

marketing tricks bias by creating deliberately loaded self report forms .