As I said in the previous comment to hutan: how do you believe studies like this could be better?
It's not that hard to think of things that might make a better study. Perhaps the people that you should be asking, dundrum, are the researchers who are currently producing these low quality studies. First, they need to accept that they are currently producing studies that are a waste of everyone's time.
Here are some ideas, off the top of my head:
Researcher equipoise and bias management
Not all of the researchers on the team should be biased to getting a positive result for the treatment. Ideally, the trial would be run entirely by people whose careers/reputations/ financial income would not be affected by the result. Then, there should be bias management wherever possible. So, for example, the protocol should be written and posted/published in advance of the study, and it should be followed.
Data analysis could be undertaken by people with equipoise and blinded to the interventions associated with the groups. The data should be open access, so other researchers can examine it.
Participant selection
Participants should be selected as much as possible in a way that does not select for people with a strong belief that the intervention will work. This can be difficult, as there is typically bias in people who are referred to a particular clinic and even more in those willing to take time to participate in a study. Recruitment of volunteers by posters or online ads is very bad though. Going to GPs and asking for their help can be good, so long as the researchers have not cherry picked GPs who believe in the therapy. Whatever is done, it is important to quantify the entire population and show the selection ratios for the study.
People should be selected that have had the pain/problem for a sufficiently long time, to eliminate those who only have a temporary problem. People should be selected who have the pain/problem to a sufficiently severe degree that it is really clear that they have a problem. Ideally, there should be some functional impact as a result of the problem e.g. unable to walk, unable to walk at a normal speed or for more than a specific distance, a low level of daily activity, unable to work...
Participants could be asked about their beliefs that the treatments will work, before the intervention starts, so that relationships between that pre-existing belief and responses (and drop outs) can be assessed.
Study size and duration
The size needs to be large enough. What that means will vary, but there probably should be at least 30 people in each treatment. The study needs to last a sufficiently long time. That normally means more than 6 months after the last contact with the therapist.
Controls
There should normally be a control intervention that is plausible to the participants. There are ways to do that, especially with pain because there are actual treatments (painkillers).
You could chop the psychological intervention up into its component parts and see the effect of each one. So, if you are teaching pain distraction techniques, goal setting and meditation, with 'tea, biscuit and group chat', or whatever, you could have groups doing just one of the components, or two, or the whole lot. You could have separate therapists for each component, so they wouldn't know which participants were getting the supposed 'therapeutic package'. If all of the groups (with one or a combination of treatments) had similar results, then you could conclude that nothing did anything much.
Outcomes
Somehow, I've lost some of what I wrote for some of the headings and all of what I wrote for this one. So I'll just jot down some of the points I remember.
This idea that because pain is a subjective feeling, the outcomes can only be subjective is complete nonsense.
Objective outcomes are possible, because a condition typically has a functional impact.
I agree that the 6 minute walk test isn't useful for mild ME/CFS, or any condition where people can walk at normal speed for six minutes. But, it can be ok for some conditions. Activity levels, measured with monitors. If sleep is affected, then some measure of sleep fragmentation, measured with a monitor. Use of drugs for pain relief.
Subjective measures as adjuncts can be ok - e.g. a simple measure of change in quality of life. Subjective measures of pain or fatigue should be taken as close to real time as possible e.g. with an app 'what is your pain level right now?', rather than something like 'what level has your pain been over the last month?'
There should be time between the intervention and all the enthusiasm and hope that goes with it, and the primary outcome assessment.
@dundrum, Do you agree that the study that is the subject of this thread is not good enough to be useful evidence about pain reprocessing works, other than, at best, it's scarcely better than an open label placebo? What do you believe a good study would look like?
