Multidisciplinary treatment for chronic pain: a systematic review of interventions and outcomes

Objectives. To provide an overview of the effectiveness of multidisciplinary treatments of chronic pain and investigate about their differential effects on outcome in various pain conditions and of different multidisciplinary treatments, settings or durations. Methods. In this article, the authors performed a systematic review of all currently available randomized controlled trials (RCTs) fulfilling the inclusion criteria, by using a recently developed rating system aimed to assess the strength of evidence with regard to the methodological quality of the trials. Results. Compared with other non-disciplinary treatments, moderate evidence of higher effectiveness for multidisciplinary interventions was shown. In contrast to no treatment or standard medical treatment, strong evidence was detected in favour of multidisciplinary treatments. The evidence that comprehensive inpatient programmes were more beneficial that outpatient programmes was moderate. Fibromyalgia and chronic back pain patients tended to profit more substantially than patients with diverse origins or chronic pain diagnoses. No evidence was found that treatment variables, such as duration or programme components, were influential for the success of the intervention. Conclusion. A standard of multidisciplinary programmes should be internationally established to guarantee generally good outcomes in the treatment of chronic pain. Our results highlight the lack of quality of design, execution or reporting of many of the RCTs included in this article. Future studies should more specifically focus on differential effects of treatment components and patient variables, allowing the identification of subgroups, which most probably would profit from multidisciplinary pain programmes.


Introduction
Chronic pain symptoms cause major medical and socioeconomical problems in industrialized countries due to high direct and indirect costs and are the most common cause of long-term disability in middle-aged people [1]. A great variety of treatment strategies suggest difficulties to treat these patients effectively. Knowing that chronic pain and disability are not only influenced by somatic pathology, but also by psychological and social factors, multidisciplinary interventions for chronic pain have become more accepted in various comprehensive approaches and have rapidly increased in number over the last few decades [2][3][4]. These are currently based on a cognitive-behavioural principle aimed at reducing disability through the modification of both cognitive processes and environmental contingencies. While cognitive treatment is aimed at modifying maladaptive cognitions on pain and its control, operant-behavioural treatment is designed to support healthy behaviours by reinforcement of those behaviours and through withdrawal of attention from pain behaviour. Timecontingent instead of pain-contingent drug use may be a part of this strategy as well, as is the involvement of the spouse. A third approach focuses on the physiological response system and aims at reducing muscular tension by providing the patient with a model of the relationship between tension and pain and teaching him/her relaxation techniques. It is mostly combined with cognitive techniques. A further common method is 'the graduated activity exposure or pacing, which is an operant-strategy used in the management of chronic pain conditions, to enable patients to control exacerbations in pain by learning to regulate the activity and once a regime of paced activity is established, to gradually increase their activity level' [5].
A comprehensive treatment approach for chronic pain patients includes one or more of these four methods combined with therapies such as physiotherapy, pain management by medication, patient education and ergonomic training. Multidisciplinary treatment has been acknowledged in the past few decades and now finds further expansion [6]. It has been evaluated in many studies and some reviews do exist, but they have their specific limitations.
For those reasons, the aims of this systematic comprehensive review on multidisciplinary treatment of chronic pain first is to give an overview on multidisciplinary treatment for chronic nonmalignant pain in general, second, to compare the results for different pain diagnoses and third, to find out whether a conclusion may be drawn about the efficacy of different kinds of multidisciplinary treatments, settings or durations.
The secondary search strategy was performed by contacting experts in this field, screening of references of the RCTs included and relevant reviews.

Abstract selection and eligibility criteria
In order to optimize agreement between the two reviewers (L.S. and V.T.), all assessment tools were independently pretested using a few studies and comparing the results. After this pilot stage, L.S. and V.T. inspected the titles and abstracts of all the references retrieved by our search strategy. L.S., V.T. independently assessed the abstracts of relevant papers using a structured form to determine whether the inclusion criteria were fulfilled. In doubtful cases, the article was retrieved in full length and evaluated before making any decision. In case of uncertainties, a third reviewer (H.S.) was consulted.
RCTs were exclusively included. The original study had to deal with adult patients (>18-yrs old) with chronic non-specific musculoskeletal pain (e.g. chronic low back or back pain, FM). At least one study group had to be treated in a multidisciplinary approach in a group setting. To rank as a multidisciplinary treatment, at least three out of the following categories of psychotherapy (PS), physiotherapy, relaxation techniques, medical treatment or patient education, vocational therapy, needed to be part of the programme. At least 2 of the 12 following domains had to be covered: pain, emotional strain, quality of life, disability, coping, physical capacity, return to work, sick leave, use of medicaments, use of the health care system, pain behaviour or subjective overall success. A follow-up (FUP) of at least 3 months had to have been conducted. The studies had to be published in full length in any language and no publication date restrictions were made. To note, we focused, as mentioned in the introduction, on cognitive-behavioural, operant, psychological response system and graded exposure pain management programmes, excluding work-hardening programmes, which have partially the same contents but are otherwise weighted and have generally other primary outcomes.

Methodological quality assessment and levels of evidence
Even though there is still limited empirical evidence [19] of a relationship between specific methodological criteria and bias, it cannot be excluded that methodological flaws, which affect the internal validity of a study may introduce some bias in its results.
All trials selected were judged according to a 10-item checklist by two independent reviewers (L.S., V.T.) to describe the methodological quality. 'Assessing the quality of trials in the field of this systematic reviews is faced up to differences regarding pharmacological trial (e.g. influence of experience of the care givers, blinding of the patients not always possible) and therefore specific instruments should be used' [20]. Hence, a recently developed checklist to evaluate reports of non-pharmacological trials (CLEAR NPT [18,21]) was utilized to assess the methodological quality of the studies included in this systematic review. This checklist was specifically developed to assess the reporting of RCTs assessing non-pharmacological treatment [18,21]. Many validity questionnaires include the items about comparability of the different groups at baseline and eligibility criteria. As those items are not part of the CLEAR NPT, indeed, we decided to introduce two supplementary items [(11) Comparability; (12) Eligibility criteria]. To draw a conclusion on the quality of evidence, we followed the criteria of the modified GRADE quality assessment, as described elsewhere (Table 1) [22].
We based our conclusions on the effectiveness of the various therapeutic interventions and on the strength of scientific evidence using a rating system with four different levels based on the quality of the studies ( Table 2) [17].

Data extraction
Two reviewers (L.S., V.T.) independently extracted data according to a pre-defined protocol and a final version of the data extraction was developed by consensus. The majority of the studies measured various outcomes and our decision about primary and secondary outcomes was somewhat arbitrary. In accordance with the literature, we considered the following domains as primary outcomes: psychological strain, disability in everyday life, health-related quality of life and pain, as well as more appropriate coping strategies, which seem to account for these changes [23]. Physical capacity, return to work rate, sick leave, the use of the health care system, medication, pain behaviour, quality of sleep and other domains (e.g. subjective improvement) were considered as secondary outcomes. Furthermore, we extracted data regarding duration of the multidisciplinary pain programme (weeks and hours), type of interventions of the pain programmes and treatment components, setting and follow-up length.

Determination of success
Most chronic pain patients have a long clinical history of more or less successful treatments and the goals of therapy have to be realistically adapted to each individual situation. A multidisciplinary treatment was considered as successful if it was more effective than a control treatment [treatment as usual (TAU), waiting list control (WLC), placebo (attention control) or a treatment that did not fulfil our criteria for a multidisciplinary treatment (e.g. either physiotherapy, PS or relaxation techniques solely)]. The higher effectiveness had to be demonstrated in at least two out of the five primary outcomes, or at least in one of the primary and two of the secondary outcomes.

Description of included studies
The number of patients of the studies included, varied between 15 and 214 (median ¼ 86), totalling 2407 patients. The size of the individual treatment groups varied between 3 and 10 patients, but was mostly between 5 and 7.
The duration of the programmes varied between 4 and 15 weeks for outpatient programmes over 15-135 h (median ¼ 31 h) and between three and eight weeks for inpatient programmes over up to 200 h (median ¼ 150 h). Based on the available data, the median duration of all treatments was 45 h. In order to obtain a better comparability, we tried to classify the multidimensional treatments into treatments with cognitive-behavioural approaches (CBT) and operant-behavioural approaches (OBT), although the authors called it integrated or multidisciplinary group therapy.

Central elements of multidisciplinary therapy
As study settings, populations, interventions and control groups were heterogeneous, we decided not to pool effect sizes in a metaanalysis. Details of the intervention administered were made in all reports (100.0%, Table 4). The 27 studies comprised of 74 groups including 39 with multidisciplinary treatment regimens, 20 with non-multidisciplinary treatment strategies and 15 with WLC or TAU (Table 3).
A medical doctor was part of the team in eight studies [31,32,34,35,39,40,44,47]. His/her task was mostly the adaptation and/ or reduction of the medication, as well as information about the patho-physiological processes of chronic pain.
Patient education was often an integral part of the therapy. In 16 studies, some sort of patient education was conducted [24-27, 30-33, 37, 39, 41, 44-46, 49, 50, 56]. Other elements that were part of the therapy were ergonomic training [25,32,35,37], vocational therapy or occupational therapy [24,27,30,34,35 Table 4 shows the items of the CLEAR NPT of the 27 studies included without FUP studies and the additional analysis studies. The generation of allocation sequences was considered adequate in 77.8% of the trials and only in 25.9% of the reports, the treatment allocation was concealed. Based on the information available in the text, we judged the care providers' experience or skill in each arm being adequate in 48.1% of the included studies, though the information were rather scarcely reported. The participants' adherence was assessed quantitatively in just 33.3% of the included reports. Blinding was adequately reported for the participants in only 11.1% of the studies, for care providers in 11.1% and for the outcome assessors in 25.9%. When the blinding criterion was not fulfilled, co-interventions were the same in each randomized group in 16.7% of the studies. Withdrawals and losses to follow-up were the same in each randomized group in 25.0% of the studies. In most of the papers included, there was insufficient information to make a decision for the items 6.1-6.2 and 7.1-7.3 ('Unclear' 75.0-25.0% and, respectively, 75.0-29.1%).
No specific methods were used to avoid ascertainment bias (0.0%). The FUP schedule was the same in each group in almost all studies (92.6%). The median of the length of the FUP is 12 months. An intention-to-treat analysis was calculated in 37.0% of the articles. The baseline comparability was fulfilled in 25 articles (92.7%). The same results were shown for the declaration of the eligibility criteria (92.7%). Table 3 shows the overall design quality of the studies included. Only six studies were ranked as high-quality studies [30,33,35,38,39,61] according to the GRADE definition (Table 1) [22].

Comparison of multidisciplinary treatment vs WLC or TAU
Fifteen studies comparing multidisciplinary treatment vs. WLC or TAU [24, 25, 27, 29-32, 37-40, 44, 46, 49, 50] showed strong evidence that a multidisciplinary treatment is superior to a standard medical treatment or WLC (Table 5). Thirteen studies reported positive results [25, 27, 29-32, 37-40, 44, 49, 50], and two did not demonstrate positive results [24,46]. Results of long-term FUPs were not available for this comparison in all studies, as patients from waiting lists often entered the treatment programme after the post-assessment, due to ethical reasons. The differences after treatment were maintained at FUP in those studies where results were described.

Comparison of multidisciplinary treatment vs other control group treatments
Fifteen studies comparing multidisciplinary treatment vs nonmultidisciplinary control group treatment (e.g. physiotherapy with discussion group, patient education) were identified [26, 28, 29, 34-36, 38, 40-43, 46-48, 61]. Together they showed moderate evidence that a multidisciplinary treatment is more effective. In five studies, the results indicated no significant difference between the groups [28,35,41,46,47]. Where success was recorded, it was maintained at FUP (Table 5).

Comparison inpatient vs outpatient programmes
Four studies directly compared inpatient and outpatient programmes [28,34,44,50]. Three of them demonstrated moderate evidence for superior long-term effects of intensive inpatient programmes. One study showed no differences [28]. Notably, the inpatient programmes were much more intensive than the outpatient programmes ( Table 3).

Comparison of effects for groups with different pain diagnoses
There is moderate evidence that a multidisciplinary programme is more effective than no treatment or non-multidisciplinary treatment for chronic back pain patients. Six of seven studies comparing it with a WLC or TAU had positive results [25,27,32,37,38,48], as well as the 8 of 11 studies comparing it with another treatment showed moderate evidence that a multidisciplinary treatment is more effective [26, 32-34, 38, 42, 43, 48]. In five studies, no differences were shown between the groups [24,28,35,45,61].
In FM, there is moderate evidence that a multidisciplinary programme is more effective than no treatment. Three studies [30,31,49] showed positive results for a multidisciplinary treatment vs a WLC, on the other hand two studies did not show any difference [29,46].
In two studies, the comparisons with other treatments did not show any difference [29,41]. Only two studies showed a superiority of the multidisciplinary group [36,39].
There was limited evidence that a multidisciplinary programme for mixed chronic pain patients was more effective compared with TAU or WLC [40,44,50]. No difference were shown for other treatment strategies [47]. 3/27 11.1 (6.1) If participants were not adequately blinded were all other treatments and care (i.e., co-interventions) the same in each randomized group? b 4/24 16.7 (6.2) Were withdrawals and lost to FUP the same in each randomized group? b 6/24 25.0 (7) Were care providers or persons caring for the participants adequately blinded? 3/27 11.1 (7.1) If care providers were not adequately blinded were all other treatments and care (i.e., co-interventions) the same in each randomized group? b 4/24 16.7

Comparison of different multidisciplinary programmes
Four studies compared different kinds or duration of multidisciplinary treatments [33,42,45,49]. There is no evidence that a special kind, duration or setting of multidisciplinary treatment as described in the evaluated studies is superior to any of the other study regimens (Table 5).

Success in connection with measurements
The range of instruments to assess the various domains of interest is very broad. In fact, in each domain, 6-12 different instruments were administered. There is no tendency that special domains or certain instruments show successful results more often and are more sensitive than others (

Discussion
This article provides the most current and comprehensive review of the existing evidence of the efficacy of multidisciplinary pain programmes and represents an unique evaluation with a detailed overview of the outcome instruments and intervention in multidisciplinary pain programmes. With reference to our first aim, it seems that a minimum standard of multidisciplinary therapy can be currently established from these data, namely ideally: specific individual exercising, regular training in relaxation techniques, group therapy led by a clinical psychologist (1.5 h) per week, patient education sessions once a week, two physiotherapy treatments per week (CBT) for pacing strategies, medical training therapy and neuro-physiology information given by trained physician.
The efficacy of such programmes is not only better than standard medical treatment, but also better than other non-multidisciplinary treatments. Therefore, the set-up of multidisciplinary programmes for chronic pain patients appears to be reasonable and patients should be referred to adequately specialized institutions, instead of being sent to various individual medical specialists sequentially.
Altmaier et al. [24] FUP in: Patrick et al. [54] OBT þ CBT ¼ TAU OBT þ CBT ¼ TAU OBT þ CBT ¼ TAU improvements maintained Nicholas et al. [43] CBT > attention control CBT > attention control -Burckhardt et al. [29] Lomi et al. [57] CBT  [50] CBT in > CBT out > WLC CBT in > CBT out -Basler et al. [25] CBT > TAU CBT > TAU -Keller et al. [37] CBT þ OBT > WLC Improvements maintained -Rose et al. [45] Individual Gowans et al. [31] CBT þ OBT > WLC CBT þ OBT > WLC -Bendix et al. [28] No results in the article CBT þ physical training ¼ physical training -Jensen et al. [32] FUP in: Jensen et al. [53] No results in the article CBT þ OBT, CBT, PT ¼ TAU CBT, PT > TAU (women) CBT þ OBT, CBT > TAU (women) CBT þ OBT > CBT, PT > TAU (women) Soares et al. [46] CBT > PE ¼ WLC CBT ¼ PE ¼ WLC -Turner-stokes et al. [47] CBT ¼ individual PS CBT ¼ individual PS -Jousset et al. [34] No results in the article PT þ OT þ medical treatment > PT PT þ OT þ medical treatment ¼ PT -Cedraschi et al. [30] No results in the article CBT þ OBT > WLC -Lemstra Olzynski [39] CBT þ OBT þ PT > WLC CBT þ OBT þ PT > WLC -Kä ä pä et al. [35] CBT þ OBT ¼ PT CBT þ OBT ¼ PT ->: first group has significantly better results than the second group; ¼: no significant difference between the two groups. Bold shows significant results in at least two of the primary outcomes or in at least one primary and two secondary outcomes In relation to our second aim, the results seen in patients with mixed chronic pain are definitely less beneficial as compared with the promising studies with FM and chronic back pain patients, and should be a question of further investigation. FM as well chronic back pain are different but share some similarities. In fact, both musculoskeletal disorders are strongly associated to a behavioural component, i.e. fear avoidance, over-under activity, passive coping strategies, etc. Additionally, we observed that both diagnostic groups have maladaptive beliefs about the explanation of the pain (catastrophizing behaviour, structural damage, kinesiophobic disturbs, high level of depression, distress).
Our third aim was to assess different kinds of multidisciplinary programmes. Intensive inpatient programmes seem to be more effective, which is consistent with the findings of Guzman et al. [7]. Such programmes may be justified for patients with more severe disabilities. Regarding treatment components or duration, there is no evidence for a superior effect of a special treatment regimen. However, a final conclusion cannot be drawn due to the low number of studies comparing this aspect.
The overall methodological quality of the studies was found to be rather low. Some requirements, such as the blinding of care provider and patients, may not be met by multidisciplinary therapy. Other requirements, such as coverage of the method of randomization or concealment of treatment allocation, were insufficiently reported. An important point to consider is the small study population in some investigations. As a consequence, some studies were underpowered and some effects may not have been detected. For physicians it is fundamental to apply the evidence from systematic reviews only if the results are judged as clinically relevant and applicable. Thus, in accordance with the criteria recommended from Malmivaara et al. [62], we can state that generally the papers included are to be considered as clinically relevant and applicable.
Our systematic review is (as any review or meta-analysis) bound to publication bias and we cannot exclude that we may have missed some relevant trials, despite the fact that we used a highly sensitive search strategy, we did not have any language restrictions and consulted an experienced librarian, as recommended in Crumley et al. [63]. We did not apply a quantitative pooling of effect sizes but decided to summarize the findings by strength of evidence. Regarding the large heterogeneity of the studies, this seemed to us the more appropriate way to report the results. The decision to include or exclude some articles fulfilling the inclusion criteria, but not with the main focus on pain programme, is questionable. Our decision was based on the content of the programme and depending on the primary outcome measurements.
Multidisciplinary treatments are effective, but it is still not known which treatment components are really important and whether all patients (with different diagnoses, age, duration of pain, social background, etc.) would profit from all components. Future studies should compare different methods, settings and durations of multidisciplinary treatments and examine their connection with patient characteristics in more detail in order to detect differential effects. In order to achieve these demanding goals, multicentre studies may be useful. Further studies are needed to establish determinants or prognostic indicators of success, and to also define the therapeutic potential for a successful rehabilitation. As an upshot of this systematic review, we would recommend a stronger observance of methodological guidelines and the use of internationally accepted outcome measures in order to make studies more comparable, due to the extensive heterogeneity among the outcome measurements. An important task for the future will be the realization of more costbenefit analyses in order to see which of the treatments are really worth being carried out. Health care insurances should finance and promote high quality of pain programmes that fulfil the minimal recommendations mentioned, representing the state of the art for multidisciplinary pain programmes.
In summary, this work may be helpful, especially for practising physicians in their daily work, in setting priorities more on disabilities and health-related quality of life in the treatment of chronic pain patients and also for researchers to optimally plan the outcome measurements and intervention modalities of future clinical trials.
Disclosure statement: The authors have declared no conflicts of interest.