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

How about "interventions designed to induce response bias" as name for the phenomenon?

The bias operating here is not unusual. It's just what's called response bias. What is unusual is that the interventions appear to be specifically designed to induce response bias (perhaps not intentionally, but by naively continuing to modify an intervention until it gives the "best" results).

 

I've done what I often do unfortunately, and fallen into the trap of using an engineering term that does not really work in normal usage. In electronics, for instance, a bias voltage simply means a voltage that provides an offset to some other voltage, shifts its value in effect, "biases" it to a new, modified value.

But although I may have used the word bias a bit unfortunately, the essence of what I've been saying is still valid. The bias we normally speak of still involves the shifting of outcome values to some degree, whatever other connotations there may be that go along with the word.

• With treatments like CBT, GET, etc it is the shifting, modification of beliefs/perceptions that the treatment primarily sets out to achieve.
• When we speak of bias in clinical trials of these interventions, it is the modification of participants' beliefs/perceptions that we are usually talking about.

So they are essentially the same thing: The very bias that CBT, GET, introduce into the results of clinical trials of them, is much the same thing as the outcome the treatment is intended to achieve.

So when trialling a treatment whose very purpose is to shift beliefs perceptions, if you then assess the results of that treatment by only measuring the shifting of participants' beliefs/perceptions, you need to be damn sure to not pretend those results mean anything beyond that.

 

Yes, me too.

Very much so. CBT is a powerful tool. Abuse of any powerful tool can be harmful, which is why you don't put them into the hands of children

I totally agree with this.

I don't believe it is too strong. Manipulation can imply motive, but by no means automatically does. You can manipulate fabric, metal, etc, and in that sense just means to alter. Though of course if anyone wishes to read it the other way then they can do.

 

In which case the dominant bias operating here is not response bias. All manner of things might influence how someone responds to a questionnaire, but that does not automatically mark it as being response bias surely?

Response bias is about responding to a questionnaire, survey, etc. in a way that is contrary to what you actually think/feel/believe at the time; your response does not truly align with your cognitions.

But the bias we are talking about here is that someone's cognitions have been messed with in such a way that at the time of filling the questionnaire they really do think/feel/believe differently to what they previously might have, even though that might change back again at some point in the future. The response could be aligning with the cognitions, even though those cognitions have become skewed. Any response bias would be additional to that, and not the same thing.

So the key point here is that it is a person's beliefs/thoughts/etc that have been targeted by the intervention, and deliberately skewed from what they were, by that intervention.

 

I haven't read all the posts in this thread so someone might have focused on this point already. In addition to encouraging people to re-interpret their symptoms, part of the influencing involves telling participants that the approach is evidence-based and/or already shown to work in previous studies and so on. I don't know if that is standard in other CBT-like approaches or specific to research in this illness.

 

This point suggests we have two sources of bias in the self reporting, one where the intervention is successful in its attempt to change the participant’s world views (or for the cynic, brain wash them) and one where it encourages them to lie to conceal their personal failure to benefit from a known successful intervention.

 

Yes, that is so weird! Trialling a treatment, the point being to gather evidence of its efficacy, but a component part of that treatment being to impress upon participants that the treatment is already evidence based. Talk about incestuous logic.

 

Yes, you can imagine a 15 stone person answering a survey question asking their weight. If they say 15 stone then no problems. But they might think "sod that!" and respond with 13 stone, in which case that is response bias - they knowingly skew their response at odds with what they know to be true. But you could also imagine such a person has been brain washed into thinking they are not as heavy as the scales say, and that the scales always lie by 2 stone over. In which case that is not a response bias, because their response aligns with what they believe. But the notion of cognitive manipulation bias (if we were to adopt that term) has kicked in.

 

I like this. Or maybe "treatment-induced cognitive bias"

 

Yes I like that also. I think it important to include the 'cognitive' bit, because it is the shift in cognitions as an outcome of the treatment - beliefs, perceptions, etc. - that this kind of bias is rooted in. It's kind of spooky, because it is the treatment itself that potentially skews people's perceptions away from truth and reality. In fact as I write that, it seems gross to even call it a treatment, more of a mistreatment.

I would imagine CBT will have been originally formulated to shift people's perceptions away from an abnormal unhealthy state, towards a more normal healthy state. But as we see it is a tool that can be abused.

 

How about science illiteracy? People are frankly too generous here, this is a long-standing problem that has always plagued research of all types. This is just one of the last remaining space where science illiteracy is encouraged and promoted but it's the same: researchers who want to prove themselves right instead of doing actual science. That's just not understanding the letter AND spirit of the scientific method. However it manifests itself, this is no difference than the proverbial tipping of a scale to influence a measurement, instead here there is no actual measurement and everything is made-up.

Of course since the whole point of EBM is to side-step the scientific method it's very awkward but the outcomes speak for themselves, avoiding the scientific method because it's too rigorous leads to stagnation, at best, and disaster when the wrong hands meet self-serving motives. But frankly it's become obvious to me that EBM in its current form serves absolutely no purpose unless those massive flaws are corrected and everything is done all over again.

(Too much to ask I know but when you jerk around for decades no one should be surprised that nothing useful comes out of it)

 

What about treatment-inherent cognitive bias?
All sorts of treatments might induce cognitive bias through more general means but it seems that what is wanted is a category for bias where its induction is inherent in specific forms of treatment?

 

Yes. The intended outcome of these treatments is to shift cognitions, it is not merely a side-effect of the treatment. So I think "treatment-inherent cognitive bias" is closer to the mark.

Though that description could still encompass treatments where messing with cognitions was an inevitable secondary effect, even though not the primary intended outcome of the treatment.

Edited 2nd para, because I'm confused by whether the naming of this sort of bias should encompass, or not, inevitable secondary (i.e. side) effects that lead to cognitive-shift bias.

 

Treatment - response cognitive bias.

If you respond to the treatment then it is inevitable that cognitive bias is a part of that response.

 

In addition we have another bias - perhaps motivation might be a better word.

Patients often need to keep these clinics and therapists onside. It is often the only source of support they will have when dealing with employers, the Benefits Agency etc. In addition letters written by the clinic to the GP can influence the ongoing relationship between patient and GP.

Consciously or not, this will amplify the bias effect, especially as treatments are unblinded.

 

Not sure about that one @Invisible Woman, for two reasons.

The word "response" here will be muddling two meanings of that word. Response bias is nothing to do with response to treatment, but response to a question. I know you know this, but many would conflate things.

And also that the cognitive shift, once in place, is there whether any questionnaire is undertaken or not. The bias does not occur at the point of responding to the questionnaire, but has become intrinsic to the patient's cognitions.

 

I take your point about the use of the word response but that's why I said treatment-response bias - not just response bias as per the questionnaire. At that point in the treatment you might genuinely believe that is an accurate picture of how you're doing - even though it's not.

It's an inbuilt design feature (if not a deliberate one) that any response to treatment will affect your cognitions about the treatment and how it has affected you, even when that doesn't match reality.

 

Or we could just call it what it is.....loading the dice. Loaded dice bias?

 

I realise that in my posts here about treatment-inherent cognitive bias (or whatever we eventually decide to call it) I may have been giving the false impression I believe it to be the sole source of bias from treatments like CBT, GET, etc, rather than response bias. But that is not what I believe at all!

I believe that treatment-inherent cognitive bias will be a major (perhaps main) source of bias for these treatments, given how they skew cognitions. But I also believe that response bias is an additional and very significant form of bias with these treatments too. I believe both are significant.

• To a significant degree people's cognitions will be skewed (screwed?), so that when answering questions as truthfully as they can in accordance with their skewed beliefs, there will be this treatment-inherent cognitive bias occurring. Not a response bias, because their responses align with what they have been brainwashed to believe.
• But also to a significant degree, people will additionally feel obligated to lean on what they perceive to be the truth, for reasons of loyalty, personal investment in treatment etc, etc. Response bias in other words.

So I think there are two significant forms of bias in play here, that are all too easily misconstrued in aggregation as just the one form of bias, response bias. I imagine there is a grey area in between, and ends up leading to serious confusion about what is really going on, and conflation of the two.

 

In a way that sums it up perfectly.