Biopsychosocial rehabilitation for repetitive-strain among working-age

Biopsychosocial rehabilitation for repetitive-strain injuries among working-age adults. Scand J Work Environ Health 2000;26(5):373-381. The objective of this study was to determine the effectiveness of biopsychosocial rehabilitation for upper-limb repetitive-strain injuries among working-age adults. Studies were identified from electronic bibliographic databas-es, reference checks, and consultations with experts in rehabilitation. Four blinded reviewers selected randomized controlled and controlled trials. Two experts evaluated the clinical relevance of the findings. Two other reviewers extracted the data and assessed the main results and the methodological quality of the studies. Finally, a qualitative analysis was performed. Only 2 studies satisfied the criteria. They were both considered to be low-quality trials. The clinical relevance of the included studies was also unsatisfactory. The level of scientific evidence was limited, showing that hypnosis as a supplement to comprehensive treatment can decrease the pain intensity of acute repetitive-strain injury in short follow-ups. There appears to be little scientific evidence for the effectiveness of biopsychosocial rehabilitation with respect to repetitive-strain injuries.

The term repetitive-strain injury (RSI) is often used to describe upper-limb disorders caused by overuse (1). Common acute RSI cases usually recover rapidly. Prolonged pain tends to develop into a combination of physical, psychological, and social disabilities. For some people the pain persists, and the disorder can result in chronic work disability similar to that observed for low-back pain. Therefore physical rehabilitation for musculoskeletal disorders has been combined with psychological, behavioral, and educational intervention (2)(3)(4). The conceptual basis for this kind of rehabilitation lies in the biopsychosocial (biological, psychological, and social) model of illness (5), which is nowadays applied in inpatient and outpatient rehabilitation. Biopsychosocial rehabilitation is often a multidisciplinary treatment program, which requires substantial staff resources and financial resources on the part of the heath care system. The indirect costs burden the employers, insurance companies, and patients as well. Therefore, there is a need to determine whether intervention that accounts for the majority of costs has a significant impact on long-term outcomes (3).
Some meta-analyses on the efficacy of multidisciplinary biopsychosocial pain treatment have been published (6,7), but they lack information on RSI. T o gain evidence on the effects of biopsychosocial rehabilitation on, especially, RSI, we have written this review from data gathered for a more general systematic review (Karjalainen et al, an unpublished manuscript of a systematic review) on the effectiveness of biopsychosocial rehabilitation with respect to common musculoskeletal disorders among working-age adults. The objective of the present 1 Finnish Institute of Occupational Health, Helsinki, Finland.
systematic review was to investigate the effects of biopsychosocial rehabilitation in the treatment of upper-limb RSI among working-age adults.

Types of studies
Only randomized controlled trials and prospective concurrent controlled trials on biopsychosocial rehabilitation were considered. However, it was agreed that, should there be 3 or more randomized controlled trials, only randomized controlled trials would be included. Studies reported in English, Dutch, Finnish, Swedish, Norwegian, German, French, or Spanish were included.

Types of participants
Working-age adults (18 to 65 years of age) suffering from upper-extremity RSI were included. Excluded were patients with acute trauma, neoplasm, and inflammatory or neurological diseases.

Types of intervention
The inpatient or outpatient biopsychosocial rehabilitation programs had to be biopsychosocial; they had to consist of a physician's consultation plus psychological, social, or vocational intervention or some combination of the three. Thus trials in which rehabilitation was exclusively or predominantly medical were excluded. For example, a program consisting solely of medical treatment and physiotherapy would not have been included, but a program with medical treatment and a workplace visit would. Electromyographic (EMG) biofeedback as the only component of physiological intervention was excluded.

Methods
The methods used for this systematic review on RSI were based on current recommendations (1 1-13). The clinical relevance of the trials was assessed using a set of criteria based on the recommendations of Shekelle et a1 (14) for the critical reading of clinical literature.

Study selection
Four reviewers (KK, AM, RR, HH) independently selected the trials for the systematic review. We used the latin square method, whereby each of the 4 reviewers selected the studies from among half of the papers so that the assessments of each reviewer overlapped with another's assessments. The reviewers were blinded as to the journal and authors. From the title, keywords, and abstract they determined whether or not the study met the inclusion criteria regarding design, subjects, and intervention. Disagreements were discussed in consensus meetings, and, if they remained unresolved or if it was unclear whether the article should be included, the full article was retrieved for final assessment. The same reviewers, still blinded as to authors and journal, then assessed the full text of the article to make the final decision about whether the study met the inclusion criteria. Any disagreements were again discussed in a consensus meeting.

Data extraction
Two reviewers (MvT, BK), blinded as to authors, journal and institution or origin, independently extracted the data using a standardized form.

Evaluation of effect for individual studies
We analyzed the data with MetaView 3.1 using the fixedeffects model and weighted mean differences. The mean differences and 95% confidence intervals were calculated comparing the used interventions according to the predefined outcomes.
We considered a study to have a positive outcome if biopsychosocial rehabilitation showed statistically significant improvement when the results were compared with those of the reference forms of intervention for at least one of the outcome measures listed earlier. A study was judged negative if the biopsychosocial rehabilitation was less effective than the reference forms of intervention, or it was judged neutral if there was no statistically significant difference between the intervention groups with respect to the assessed outcomes.

Clinical relevance assessment
Two experts in the field of rehabilitation (AM, HH) assessed the clinical relevance of the extracted studies by evaluating whether the patients, health care setting, and interventions were described precisely enough to allow inferences about the clinical applicability of the results and the clinical relevance of the measured outcome. On this basis, they decided whether the evidence was sufficient for clinical recommendations to be made.
The following 3 questions were used to assess the clinical relevance: (i) are the patient characteristics and treatment settings described well enough to decide whether they are comparable to those you see in your own practice, (ii) are all the intervention(s) described well enough to allow you to provide the same for your own patients, and (iii) were clinically relevant outcomes measured? Each question was given a plus (+) if the methodological criterion was fulfilled, a minus (-) if the criterion was not fulfilled, and a questionmark (?) if data were not available or not sufficient.

Methodological quality assessment
Two reviewers (MVT and BK) independently assessed the methodological quality of the trials. A consensus method was used to resolve disagreements.
We used a list of criteria that has been used in other Cochrane reviews in the field of back pain (15) and selected only items reflecting the internal validity of the trials. Equal weights were applied to all 10 criteria, which have been described in detail in appendix 1.
The maximum overall score was 10. A randomized controlled trial was considered to be of high methodological quality if at least 5 out of 10 items received a positive score and if the study had no fatal flaws. Examples of fatal flaws were withdrawals by more than 40% of the patients, total or nearly total nonadherence to the protocol, or very poor, unadjusted comparability in the baseline criteria. Any trial with fatal flaws in the study design or execution was considered inadequate and excluded from the analysis.

Qualitative analysis
The qualitative analysis involved using a rating system with 4 levels of scientific evidence for the effectiveness of biopsychosocial rehabilitation (13). The results were considered contradictory if the overall conclusions regarding the effectiveness in different studies were contradictory or inconsistent for the same diagnosis. The 4 levels were as follows: level A: strong research-based evidence: generally consistent findings in multiple highquality randomized controlled trials; level B: moderate research-based evidence: generally consistent findings in 1 high-quality randomized controlled trial and 21 lowquality randomized controlled trials or generally consistent findings in multiple low-quality randomized controlled trials; level C: limited research-based evidence: 1 randomized controlled trial (either high or low quality) or inconsistent or contradictory evidence in multiple randomized controlled trials; level D: no research-based evidence: no randomized controlled trials.

Description of the studies
Our extensive literature searches identified only 3 articles, which were reports on 2 trials (4,16,17), that focused on upper-limb RSI. Another trial (18) was identified through contact with the Cochrane Musculoskeletal Injuries Group. The articles of Spence (16,17) were later excluded due to their lack of a biomedical component in the rehabilitation scheme. We did not find any concurrent controlled trials that could be included in this review. Thus 2 trials studying RSI in the upper extremities have been included (4,18).

Characteristics
Moore (18) examined 32 patients, with a mean age of 35.4 years, recruited from the occupational medicine department of a local medical center in California. All the patients had recently diagnosed RSI (carpal tunnel syndrome, wrist tendinitis, de Quervain tenosynovitis or cervical myofascial dysfunction) with an onset of symptoms within 8 weeks. They had no previous medical treatment for RSI, no associated medical diseases and no significant psychiatric illnesses, were nonsmokers, and their employment required operation of a keyboard. All the patients were involved in comprehensive treatment in the department (including medication when indicated, referral to physiotherapy focusing on ergonomic evaluation, self-management, soft-tissue flexibility, and strength). The study explored the extra effect of hypnosis with biofeedback and autogenics (a form of autohypnosis using self-suggestion), provided once a week for 6 weeks, compared with the results for waiting-list referents (table 1).  imum of 10 points in the methodological quality assessment scheme described in appendix 1. The results of the quality assessment for individual items are presented in table 2.

Clinical relevance
Moore (1 8) satisfactorily described the intervention and the setting from which the patients were recruited, but the proportions of patients according to a more specific diagnosis (eg, epicondylitis, wrist tenosynovitis, carpal tunnel syndrome) within RSI remained unclear. In addition, the study did not report what type of treatment the waiting-list reference group received while it waited for the same treatment as the studied intervention group. The only outcome measured in the study of Moore (18) was the intensity of pain in 6 weeks of follow-up. Spence et a1 (4) measured relevant outcomes but did not adequately describe the setting, patient characteristics, or intervention.

Results
Outcome Moore (18) recorded pain, measured by a 10-cm visual analogue scale (VAS), in both groups. After 6 weeks of follow-up, there was a greater and statistically significant reduction in the mean decrease in pain rating in the intervention (hypnosis) group than in the reference group (mean difference -3.6 cm, 95% confidence interval -5.1 to -2.0 cm).
According to the assessed outcomes (Self Monitored Pain Index and WHYMPI) in the study of Spence (4), the 2 biopsychosocial interventions were not more effective than the 2 reference interventions after 8 weeks of follow-up. After 6 months, data on the waiting-list referents could not be used, but in the comparisons between the other groups no statistically significant differences were found, although there was a trend showing that applied relaxation could have been more effective than any of the other forms of intervention. The sample size was too small to show a statistically significant effect, however.

Grading of the scientific evidence
The large heterogeneity of the study population, the interventions, the outcomes, and the follow-up times of the 2 studies indicated that statistical pooling would have been inappropriate, even if possible. According to a qualitative analysis, the evidence of the effectiveness of biopsychosocial rehabilitation for RSI was graded as limited. This limited evidence showed (i) a positive effect of hypnosis combined with comprehensive treatment in comparison with comprehensive treatment alone after 6 weeks of follow-up and (ii) no differences in effect between applied relaxation, EMG biofeedback plus applied relaxation, and the waiting-list conditions after 8 weeks and 6 months.

Included studies and results
RSI of the upper extremites is usually considered to be associated with repetitive tasks at work. Therefore, workplace visits with an ergonomic evaluation could be expected to be included in biopsychosocial rehabilitation, as was done in the study of Moore (18). However, because a worksite visit was also part of the control intervention, the effectiveness of worksite visits could not be assessed. We think that relying on the positive result of hypnosis on pain intensity after 6 weeks of follow-up is debatable, and it would be interesting to see whether a positive effect would be maintained for a longer period after treatment.
Behavioral treatments are a fundamental feature of biopsychosocial rehabilitation for chronic musculoskeletal disorders. However, there were no studies comparing behavioral or biopsychosocial programs with nonbiopsychosocial programs. The study of Spence et a1 (4) could have provided evidence of the effectiveness of different types of behavioral treatments for RSI for use in decision making with respect to combining effective behavioral treatments with other rehabilitation programs for RSI patients. However, the results did not favor using applied relaxation or EMG biofeedback plus applied relaxation in rehabilitation, but they did indicate a need for larger studies.

Author
Year 4 1 a Each quality item was rated positive (t) if the methodological criterion was fulfilled, negative (-) if the criterion was not fulfilled, and unclear (?) if data were not available. The maximum score was 10; studies with a total of >5 were considered to be of high methodological quality.

Definition of repetitive-strain injury
The definition of RSI encompasses a range of upper-extremity disorders, which are often manifested during manual handling tasks. If the diagnoses vary, as in the 2 studies included in this review, it may be difficult to generalize the results to a setting in which the spectrum of the diagnoses is different. We think that a single diagnostic entity would be a far more fruitful starting point as an intervention study for RSI. Unfortunately, we did not find any studies of this kind for this review. However, we think that biopsychosocial rehabilitation may have a role in the treatment of any diagnosis included in RSI, when we are dealing with working age adults. In the acute stage, intervention in the workplace may speed up recovery, and, if the pain becomes chronic, psychosocial rehabilitation may help the patient cope with the situation.

Definition of biopsychosocial rehabilitation
In this review, we required each intervention to be executed by a professional of the discipline in question. A psychologist should be involved with psychological or behavioral treatment, but, on the other hand, a social worker, an occupational health nurse, or an occupational physiotherapist could perform social intervention. An occupational health nurse or physiotherapist specialized in the field of occupational health care could provide the vocational intervention. We considered that a physician needed to be part of the rehabilitation procedure, in order to enhance the reliability of the diagnosis, and a physician should have made a differential diagnosis to exclude diseases that did not belong to the study domain. Furthermore, we considered treatment provided by physiotherapists or nurses as relevant from the viewpoint of biomedical rehabilitation. We defined biopsychosocial rehabilitation as rehabilitation that contains at least 2 biopsychosocial components. Therefore, programs that included a physician's intervention and traditional physiotherapy were not included. Trials were not excluded if any psychological, social, or vocational intervention was used in addition to physiotherapy.
To consider EMG biofeedback as a behavioral treatment is controversial. In this review we excluded studies in which EMG biofeedback was the only component of psychosocial rehabilitation. The study of Spence (4) was included because the intervention for the biopsychosocial rehabilitation groups contained applied relaxation provided by a psychologist and applied relaxation combined with the EMG biofeedback.

Validity of the review
A challenging part of conducting a systematic review is finding all relevant papers. We identified only 1 relevant trial by searching the electronic databases. Because our original literature search did not focus especially on RSI, it is possible that we may have missed some studies on RSI. One additional study was identified through contacting the Cochrane Musculoskeletal Injuries Group. However, we believe that we found most of the existing relevant papers.
The study selection was primarily based on abstracts. To ensure that every relevant study was selected, we retrieved the full article for further assessment if there was any doubt about the study fulfilling the inclusion criteria. Retrieving full copies of all articles would have been extremely time consuming and expensive. Furthermore, the electronic database search also produced many irrelevant articles for our purposes, such as case reports and letters.
To avoid bias, we blinded the reviewers with regard to authors, institute of origin, journal, and acknowledgments. At the study selection stage blinding demanded much work, but it may have reduced any selection bias.
When assessing the methodological quality of the individual trials, we only included the items reflecting internal validity (15,19). We did not quantify the importance of each validity item, as the weights can be considered arbitrary.

Clinical relevance of the included studies
Intervention studies are intended to be applicable in clinical practice. Assessment of the applicability of the results demands that patient characteristics and interventions be described precisely enough for readers to make clear inferences about which patients can be treated, how they can be treated, and by whom (or in what setting). Pertinent50utcome measurements are also important for clinical relevance.
In both of the randomized controlled trials included in this review, patients on waiting lists served as a reference group, and their use posed a drawback in that waiting-list referents can only provide shost-term results. The study of Moore (18) was designed to have only 6 weeks of follow-up, and therefore the use of waiting-list patients was understandable. Still, clinically, a longer follow-up would have been more relevant, even for acute RSI patients. Spence et a1 (4) concentrated on chronic RSI patients, and, therefore, the value of a 4-6 weeks waiting-list reference group is questionable because data could not be used in the 6 months' comparison.

Methodological quality of the studies
The low methodological quality was mainly due to an insufficient description of the method of randomization and a lack of blinding. Blinding patients and therapists in randomized controlled trials on rehabilitation is usually not possible, but a feasible alternative is to evaluate baseline expectations of both patients and therapists for the intervention and reference groups.
The information reported in the articles was insufficient for 8 out of 20 (40%) rncthodological quality items. This rate of insufficiency may hopefully improve in the future, now that many journals have accepted the CONSORT statement on the quality of reporting randomized controlled trials (20).

Level of the scientific evidence for the effectiveness of intervention
Methods for structured systematic reviews are constantly evolving. Using different methods for grading the evidence may lead to different conclusions.

Proposed studies
Any trials that would unequivocally fulfill the criteria of truly biopsychosocial rehabilitation could not be included in this review. We think that, at the acute phase of RSI symptoms, it would be important that the rehabilitation focus on worktasks, as well as biomedical treatment.
In the chronic stage more comprehensive rehabilitation paying attention to biomedical, psychological, social, and occupational factors may be needed for the patients to continue working. In the near future, we would like to be able to investigate these types of trials, which would fulfill the methodological and clinical requirements of a high-quality study. The study population should be large enough (at least 50 patients for each treatment) to be able to show statistically significant differences in clinically meaningful changes in relevant outcome measures. For example, at least a 15-mm reduction in the intensity of pain, as measured with a 0-100 mm VAS, or a statistically significant reduction in days on sick leave due to RSI. Because RSI is a heterogeneous group of disorders, to study a single diagnosis (eg, lateral epicondylitis) would be preferable. Biopsychosocial intervention could be compared with usual care or with some treatment that is commonly used but has not been shown to be effective. The pertinent follow-up of the patients is dependent on the duration of the disorder and on the duration, content, and intensity of the intervention. Arbitrary recommendations for follow-up may vary for acute patients from 6 to 12 months and among chronic pain patients from 1 year to a couple of years.

Concluding remarks
Two trials, both assessing the effectiveness of a very specific intervention, could to be included in this systematic review. We found limited scientific evidence showing that hypnosis, combined with comprehensive rehabilitation, has some positive effect in the treatment of recently diagnosed RSI and that applied relaxation has a trend towards a positive effect in comparison with EMG biofeedback combined with applied relaxation, EMG biofeedback alone, or 8 weeks of being on a waiting list when chronic RSI patients are being treated. Drawing clinical conclusions about the magnitude and duration of the effect of any intervention for RSI was hampered by the lack of high-quality studies. We believe that no health policy decisions concerning the effectiveness for biopsychosocial rehabilitation for RSI can be made at present.
There seems to be an obvious need for good-quality randomized controlled trials on comprehensive biopsychosocial treatment programs for RSI, especially for those that would include workplace intervention. Focus on a single disease entity would enhance the generalizability of the results. In addition, more exploring studies on the effectiveness of specific components of these programs would be valuable.