In this blog for people affected by chronicA health condition marked by long duration, by frequent recurrence over a long time, and often by slowly progressing seriousness. For example, rheumatoid arthritis. pain and those who support and work with them, Dr. Amanda C de C Williams discusses the findings of her team’s latest Cochrane ReviewCochrane Reviews are systematic reviews. In systematic reviews we search for and summarize studies that answer a specific research question (e.g. is paracetamol effective and safe for treating back pain?). The studies are identified, assessed, and summarized by using a systematic and predefined approach. They inform recommendations for healthcare and research. on psychological therapies.
Page last checked 19 April 2023
Many people with chronic pain (or persistent pain, a term some prefer) feel misunderstood and offended when their GP or pain specialist suggest that they need ‘psychological treatment’: “The pain is real – it’s not in my head”. But psychological methods of treating chronic pain or, rather, treating the problems associated with chronic pain have been used for over 50 years. They are based on understanding the importance of the individual’s beliefs about what is causing the pain, fears about pain worsening over time, conflicting advice from clinicians, friends and family, and disappointment with successive failures of medical attempts to resolve the pain.
Psychological methods are not about ‘thinking positive’ or ‘mind over matter’, but about achieving a deeper understanding of pain and what affects it, about changing habits of thinking and of lifestyle: becoming (ideally) a person who, despite pain, lives a full life with confidence in managing that pain.
We recently updated our Cochrane systematic review ‘psychological therapies for the management of chronic pain (except headache) in adults’. Given how long these treatments have been around, this is not our first version of this review, but it is by far the biggest. This has 75 studies while our last version, in 2012, had 42. Since we started, NICE (National Institute for Health and Care Excellence) has developed guidelines for chronic pain in the UK, to be published early next year, which review some of the same evidence. The question of what works, and what doesn’t work, is important, given the size of the problem. A systematic review and meta-analysisThe use of statistical techniques in a systematic review to combine the results of included studies. Sometimes misused as a synonym for systematic reviews, where the review includes a meta-analysis. of UK studies (Fayaz et al. 2016) provided an estimate that 10-14% of the adult population has moderately to severely disabling chronic pain.
What sort of psychological treatments were reviewed?
We investigated the three main types of psychological treatmentSomething done with the aim of improving health or relieving suffering. For example, medicines, surgery, psychological and physical therapies, diet and exercise changes. for chronic pain. By far the commonest type was cognitive behavioural therapy (CBT), while behavioural therapy (BT) and acceptance and commitment therapy (ACT) had much smaller bodies of evidence. We also reviewed several treatments that drew on other methods from psychological therapies outside pain, but they were too diverse to combine for analysis. We did not review studies of mindfulness since, although it can be used as a psychological therapy, it is not a mainstream treatment.
We chose studies where trained psychological therapists delivered treatment, because applying the specific methods to a range of people with pain requires understanding of psychological principles and the skills to help people adapt methods to their own needs. Treatment had to be delivered face-to-face: internet and phone treatments are reviewed elsewhere (Eccleston et al. 2014). We excluded small studies because there is a higher riskA way of expressing the chance of an event taking place, expressed as the number of events divided by the total number of observations or people. It can be stated as ‘the chance of falling were one in four’ (1/4 = 25%). This measure is good no matter the incidence of events i.e. common or infrequent. of unrealistically positive results with these (Ioannidis 2005).
What was treatment compared with?
For each type of treatment, we separated comparisons with waiting list controls or treatment as usual from comparison with an ‘active control’. Treatment as usual means just continuing any ongoing treatment but not starting anything new, so it could include painkillers, physiotherapy, medical consultations, or nothing (like waiting lists). This is a useful comparison because it resembles patients’ experience of choice: try this treatment or carry on as before. The active control involved engagement in something new, which could be an educational programme, exercise routine, or support group with other people with chronic pain, or combinations of these. This comparison tests whether the treatment has effects that are more than simply being involved in a studyAn investigation of a healthcare problem. There are different types of studies used to answer research questions, for example randomised controlled trials or observational studies. and sharing experiences with other people with pain.
What benefits of treatment were we looking for?
We chose outcomesOutcomes are measures of health (for example quality of life, pain, blood sugar levels) that can be used to assess the effectiveness and safety of a treatment or other intervention (for example a drug, surgery, or exercise). In research, the outcomes considered most important are ‘primary outcomes’ and those considered less important are ‘secondary outcomes’. of pain, disability, and distress. These were assessed by questionnaires completed by the person in pain at the beginning and end of treatment, and at follow-up times for some studies. We separated effects at the end of treatment from effects at follow-up from 6 to 12 months after the end of treatment.
We were also concerned about adverse effects (harms), often overlooked by those running trialsClinical trials are research studies involving people who use healthcare services. They often compare a new or different treatment with the best treatment currently available. This is to test whether the new or different treatment is safe, effective and any better than what is currently used. No matter how promising a new treatment may appear during tests in a laboratory, it must go through clinical trials before its benefits and risks can really be known.. Furthermore, dropout from trials can mean that people feel worse or are disappointed by the treatment, and that can be detrimental to their wellbeing. Dropout from treatment is often described in terms of patient choice, but could also be considered to indicate treatment failure, or unacceptability. A small number of those who continue treatment may become more distressed by the end, or may suffer incidental worsening of pain from the activity elements of treatment. We need to know more about these risks in order to inform patients fully when they are offered treatment.
What did we find?
We searched up to April 2020 and found 75 studies, 34 of which were in our previous review: that represented 9401 people completing treatment or control. While many people in the trials had back pain, 19 trials were for people with fibromyalgia, nine for rheumatoid arthritis, and five for osteoarthritis. Most recruited people from hospital or community clinics, and two from care homes for older people. The mean age of people in the trials was 50, and two-thirds were women, similar to many clinical populations.
Cognitive behavioural therapy (CBT)
CBT was used in 59 studies. Twenty-nine (with over 2500 people in all) compared CBT with treatment as usual/waiting list and showed on average slight benefit for pain, disability, and distress at the end of treatment. Fewer studies reported people’s scores at follow-up but average effects were somewhat smaller. We have moderate certainty for most of these findings. Twenty-three trials (with over 3000 people in all) compared CBT with an active control (described above), and showed little or no benefit at end of treatment or follow-up for any of the three outcomes.
Behavioural therapy (BT)
BT was used in eight trials, using several different methods in varying combinations, against both types of control. The certainty of evidenceThe certainty (or quality) of evidence is the extent to which we can be confident that what the research tells us about a particular treatment effect is likely to be accurate. Concerns about factors such as bias can reduce the certainty of the evidence. Evidence may be of high certainty; moderate certainty; low certainty or very-low certainty. Cochrane has adopted the GRADE approach (Grading of Recommendations Assessment, Development and Evaluation) for assessing certainty (or quality) of evidence. Find out more here: https://training.cochrane.org/grade-approach was mostly low or very low; trials were small and we had uncertainty in our findings. This is clearly an area that needs better quality, larger trials.
Acceptance and commitment therapy (ACT)
We found five trials of ACT that met our criteria. This was a surprise as there are several systematic reviewsIn systematic reviews we search for and summarize studies that answer a specific research question (e.g. is paracetamol effective and safe for treating back pain?). The studies are identified, assessed, and summarized by using a systematic and predefined approach. They inform recommendations for healthcare and research. in the literature and ACT has been rapidly and enthusiastically adopted by many psychologists treating chronic pain. Most of the evidence was of very low quality, and we have no certainty in our findings, which were very mixed, with some apparent benefit but without any consistent pattern across outcomes and times of assessment.
It seems that ACT has been propagated less on evidence than on practitioner enthusiasm and good results from uncontrolled trials (which compare the same people’s scores before and after treatment) or very small controlled trialsA trial in which a group (the ‘intervention group’) is given a intervention being tested (for example a drug, surgery, or exercise) is compared with a group which does not receive the intervention (the ‘control group’). that fell below our minimum size. One of the advantages of using the Cochrane Collaboration’s rigorous methods is that it can expose these anomalies, where the clinical impression differs substantially from the evidence. On this basis, we recommend large trials of ACT, evaluated using the same outcomes as other psychological treatments, such as CBT, rather than acceptance-specific scales, and run by personnel without a strong personal allegiance to ACT.
The other surprise in our findings were four CBT trials with results very substantially better than all others. One we thought was due to statistical error, but could not get clarification from the authors: we excluded it. The other three were by the same author group: we enquired to find what it was about their treatment that yielded such very large benefits, with very low dropout even at follow-up. Their description resembled many other trials in our set but we got no response to further enquiries. Since we could not be certain of the authenticity of these results, and a (sensitivityA measure of a screening or diagnostic test’s ability to correctly detect people who have the disease.) analysis showed the three trials to inflate beneficial average effects, we also excluded them.
What do our findings mean for someone with chronic pain?
When offered psychological treatment for chronic pain, the person with pain is unlikely to have a choice of the types described above, because local provision will vary. The best choice on the basis of evidence is CBT: the benefits may be small, but we have reasonable confidence that these are real. For other treatments – BT, ACT, or psychotherapies – the evidence is not of sufficient quality to conclude one way or the other, so treatment should be treated as experimental, with careful evaluation of its effects, including possible harms.
Join in the conversation on Twitter with @CochraneUK and @CochranePaPaS or leave a comment on the blog. Please note, we will not publish comments that link to commercial sites or appear to endorse commercial products. We welcome diverse views and encourage discussion but ask that comments are respectful and reserve the right to not publish comments we consider offensive.
Dr. Williams has nothing to disclose.
You raise important points, but they do not invalidate the review.
First, subjective states such as pain and distress cannot be measured other than by self-report, and proxy measures of these (such as by clinicians or carers) are rightly seen as no substitute. Pain and distress are high on the priority list of outcomes for people with chronic pain (from a US survey which needs replicating in the UK: Turk et al. 2008), and we have compared them with outcomes most often used in trials (Beale et al. 2016). Cochrane involves lay reviewers in processing the protocol and the review, asking them to consider the suitability of outcomes selected; they commented on being pleased to see adverse events taken more seriously in our review.
For disability, which arguably could be assessed by more objective methods, there is no satisfactory single method, and metrics such as work/employment status misrepresent those who are not working/employed for other reasons than pain. Work/employment is important in its own right but cannot represent the wide range of personal, social, creative and vocational activities that are valued very differently by people with pain, and are affected differently by pain. We do raise in our review the underuse of automated activity monitors to supplement assessment of activity, but I know of no assessments of caring, recreational and social activities that do not rely on self-report.
Second, we have separately addressed change in health care use (Pike et al. 2016; Williams 2016). Cochrane reviews are already large, and adding further outcomes has costs to readability. Additionally, healthcare use is of major interest to funders and commissioners of care, but is rarely mentioned (except in terms of quality, not countable episodes) by people with chronic pain as an outcome that matters to them.
Beale M, Cella M, Williams ACdeC. Comparing patients’ and clinician-researchers’ outcome choice for psychological treatment of chronic pain. Pain 2011;152(10):2283-6.
Pike A, Hearn L , Williams ACdeC. Effectiveness of psychological interventions for chronic pain on health care use and work absence: systematic review and meta-analysis. Pain 2016;157(4):777-85. doi: 10.1097/j.pain.0000000000000925
Turk DC, Dworkin RH, Revicki D, Harding G, Burke LB, Cella D, Cleeland CS, Cowan P, Farrar JT, Hertz S, Max MB, Rappaport BA. Identifying important outcome domains for chronic pain clinical trials: an IMMPACT survey of people with pain. Pain 2008;137:276–85
Williams, ACdeC. Corrigendum to: Effectiveness of psychological interventions for chronic pain on health care use and work absence systematic review and meta-analysis, by Pike et al. PAIN 2016;157:777–785. Pain 2017;158:1398-9.
Thanks for this. I wasn’t trying to invalidate the review. Neither was I suggesting that objective outcomes should replace the subjective measures of pain and distress which are obviously high on the priority list of people living with chronic pain. I am not sure of the relevance of looking at what outcomes are most often used in trials. Why not use the results of US survey, or even replicate it, perhaps on a smaller scale with your lay reviewers. When treatments cannot be blinded, it weakens the evidence of effectiveness of these treatments to rely solely on self-reports of pain and distress. I see that assessments of of caring, recreational and social activities also rely on self-report, but can be objectively checked at least to some extent. It would have been interesting and useful to include them in the review, even if only to illustrate they are rarely used in trials, or need improvement and validation as useful outcome measures.
Both Turk 2008 and Beale 2016 suggest that pain/distress are just one of many aspects that are important to patients. This is not about either/or, but the importance of focusing on all outcome measures that are relevant to patients.
Secondly, it must be acknowledged, that answers on patient reported outcome measures (PROMs) are subject to a wide variety of biases, and that only true double blinding can control for these biases. (True blinding can only be confirmed if participants are actually asked which treatment group they were allocated to)
Hence Caroline Struthers’ suggestion of the importance of objective outcome measures in addition to self-reports. Only when both objective and subjective measures improve in concert can we be sure that the results are not reflecting biases in questionnaire answering behaviour, rather than a true improvement in health.
Lastly, too many researchers in the field confuse statistical measures of content validity of PROMs (such as factor analyses, Cronbach’s Alpha) as somehow representing overall validity. Despite (a) the social context of a clinical trial is quite different to the studies where content validity is assessed (b) Construct validity is simply assumed and rarely if ever tested against important objective outcome measures (criterion validity). (c) Face validity of specific PROMs themselves are rarely if ever tested (and compared) using culturally diverse groups of patients.
It needs to be acknowledged, that the generalisability of PROMs in unblinded clinical trials is limited and the quality of conclusions suffers as a result. Only by acknowledging this, will investigators of future trials aim to improve methodology to better meet patient needs. Rather than simply choosing methodology that is easy to bias to get the results that the investigators seek.
In this review, there is the usual reliance on subjective outcomes in unblindable trials. Because psychological therapies are delivered and usually also evaluated by psychologists, there seems to be an acceptance that relying on outcomes measured using self-report questionnaires is OK. This is what happens in the trials, and the triallists’ decisions are reinforced by the choice of outcomes in reviews. In this review, the only objectively measured outcome is drop outs which may be related to adverse events.
The reviewers briefly acknowledge in the Implications for research section that “… there remains a plethora of heterogeneous measurement tools for subjective experiences, with varied content within domains, and few behavioural measures supplementing self-report”. Later in that section they report “Assessment of treatment benefits in terms of reduced costs, in health and social care, and incurred by patients, is generally lacking”.
Presumably, the results indicating some benefit of CBT in this review would not be supported by objective measures. If they weren’t measured, we’ll never know. Surely it’s crucially important to point out that trials which don’t objectively measure outcomes which matter to people who suffer from chronic pain – such as ability to work, study, perform caring responsibilities, take part in recreational and social activities – cannot provide reliable evidence that they might be beneficial.
As with the Cochrane reviews on Exercise therapy and CBT for Chronic Fatigue Syndrome, this “general lack” of objective outcomes in unblindable trials is not flagged up as a significant problem with research in this area.