Physical conditioning as part of a return to work strategy aims to improve work status for workers on sick leave due to back pain. This is the second update of a Cochrane Review (originally titled 'Work conditioning, work hardening and functional restoration for workers with back and neck pain') first published in 2003, updated in 2010, and updated again in 2013.
To assess the effectiveness of physical conditioning as part of a return to work strategy in reducing time lost from work and improving work status for workers with back pain. Further, to assess which aspects of physical conditioning are related to a faster return to work for workers with back pain.
We searched the following databases to March 2012: CENTRAL, MEDLINE (from 1966), EMBASE (from 1980), CINAHL (from 1982), PsycINFO (from 1967), and PEDro.
Randomized controlled trials (RCTs) and cluster RCTs that studied workers with work disability related to back pain and who were included in physical conditioning programmes.
Two review authors independently extracted data and assessed risk of bias. We used standard methodological procedures expected by The Cochrane Collaboration.
We included 41 articles reporting on 25 RCTs with 4404 participants. Risk of bias was low in 16 studies.
Three studies involved workers with acute back pain, eight studies workers with subacute back pain, and 14 studies workers with chronic back pain.
In 14 studies, physical conditioning as part of a return to work strategy was compared to usual care. The physical conditioning mostly consisted of graded activity with work‐related exercises aimed at increasing back strength and flexibility, together with a set date for return to work. The programmes were divided into a light version with a maximum of five sessions, or an intense version with more than five sessions up to full time or as inpatient treatment.
For acute back pain, there was low quality evidence that both light and intense physical conditioning programmes made little or no difference in sickness absence duration compared with care as usual at three to 12 months follow‐up (3 studies with 340 workers).
For subacute back pain, the evidence on the effectiveness of intense physical conditioning combined with care as usual compared to usual care alone was conflicting (four studies with 395 workers). However, subgroup analysis showed low quality evidence that if the intervention was executed at the workplace, or included a workplace visit, it may have reduced sickness absence duration at 12 months follow‐up (3 studies with 283 workers; SMD ‐0.42, 95% CI ‐0.65 to ‐0.18).
For chronic back pain, there was low quality evidence that physical conditioning as part of integrated care management in addition to usual care may have reduced sickness absence days compared to usual care at 12 months follow‐up (1 study, 134 workers; SMD ‐4.42, 95% CI ‐5.06 to ‐3.79). What part of the integrated care management was most effective remained unclear. There was moderate quality evidence that intense physical conditioning probably reduced sickness absence duration only slightly compared with usual care at 12 months follow‐up (5 studies, 1093 workers; SMD ‐0.23, 95% CI ‐0.42 to ‐0.03).
Physical conditioning compared to exercise therapy showed conflicting results for workers with subacute and chronic back pain. Cognitive behavioural therapy was probably not superior to physical conditioning as an alternative or in addition to physical conditioning.
The effectiveness of physical conditioning as part of a return to work strategy in reducing sick leave for workers with back pain, compared to usual care or exercise therapy, remains uncertain. For workers with acute back pain, physical conditioning may have no effect on sickness absence duration. There is conflicting evidence regarding the reduction of sickness absence duration with intense physical conditioning versus usual care for workers with subacute back pain. It may be that including workplace visits or execution of the intervention at the workplace is the component that renders a physical conditioning programme effective. For workers with chronic back pain physical conditioning has a small effect on reducing sick leave compared to care as usual after 12 months follow‐up. To what extent physical conditioning as part of integrated care management may alter the effect on sick leave for workers with chronic back pain needs further research.
Physical conditioning as part of a return to work strategy to reduce sickness absence for workers with back pain
Review question
We reviewed the evidence about the effect of physical conditioning as part of a return to work strategy in people with low back pain. We found 25 studies.
Background
The main goal of physical conditioning as part of a return to work strategy, sometimes called work conditioning, work hardening or functional restoration and exercise programmes, is to return injured or disabled workers to work or improve the work status for workers performing modified duties. Such programmes may also simulate or duplicate work or functional tasks, or both, using exercises in a safe, supervised environment. These exercises or tasks are structured and progressively graded to increase psychological, physical and emotional tolerance and to improve endurance and work feasibility. In such environments, injured workers improve their general physical condition through an exercise programme aimed at increasing strength, endurance, flexibility and cardiovascular fitness. We wanted to discover whether physical conditioning was more or less effective than usual care and other types of interventions like exercise therapy.
Study characteristics
The evidence was current to March 2012. We analysed 17 comparisons of physical conditioning as part of a return to work strategy. Some trials examined physical conditioning in addition to care as usual versus care as usual only, and others compared physical conditioning to other types of interventions such as standard exercise therapy. Participants had either acute back pain (duration of symptoms less than six weeks), subacute back pain (duration of symptoms more than six but less than 12 weeks), or chronic back pain (duration of symptoms more than 12 weeks). Participants were followed for anywhere from three weeks to three years. We divided physical conditioning into light or intense, depending on its intensity and duration.
Key results
Results showed that light physical conditioning has no effect on sickness absence duration for workers with subacute or chronic back pain. We found conflicting results for intense physical conditioning for workers with subacute back pain. Intense physical conditioning probably had a small effect on reducing sick leave at 12 months follow‐up compared to usual care for workers with chronic back pain. Involving the workplace, or physical conditioning being part of integrated care management may have had a positive effect on reducing sick leave, but this needs further research.
Quality of the evidence
The quality of the evidence ranged from very low to moderate. Although 16 of the included studies were well designed and had no major flaws, some studies were poorly conducted and the small number of participants in most studies lowered the overall quality of the evidence.
Back pain is a major health and economic problem for society. In Western countries, the reported point prevalence varies from 12% to 30% (Anema 2009). Whether back symptoms are attributed to work, are reported to workers' compensation systems, lead to healthcare‐seeking behaviour, result in time off work, or any combination of these scenarios depend on complex individual psychosocial and work organizational and social security factors. People with physically or psychologically demanding jobs may have more difficulty working when they have back pain and so lose more time from work, but that can also be the effect rather than the cause of their pain (Waddell 1999). Nevertheless, most workers with back pain, their employers, and insurers agree that the goal of managing back pain is a timely return‐to‐work following back pain‐related work disability.
Physical conditioning as part of a return to work strategy, variously called work conditioning, work hardening or functional restoration and exercise programmes, aims for return‐to‐work or improvement in work status for workers performing modified duties. Such programmes either simulate or duplicate work or functional tasks, or both, in a safe, supervised environment. These tasks are structured and progressively graded to increase psychological, physical and emotional tolerance and improve endurance and work feasibility (Lechner 1994). In such environments, injured workers learn appropriate job performance skills in addition to improving their physical condition through an exercise programme aimed at increasing strength, endurance, flexibility and cardiovascular fitness. Work hardening programmes are individualized, work‐oriented activities that involve clients in simulated or actual work tasks. Work conditioning is a programme with an emphasis on physical conditioning, which addresses the issues of strength, endurance, flexibility, motor control and cardiopulmonary function (Lechner 1994). Functional restoration refers to any intervention aimed at restoring a reasonable functional level for activities of daily living, including work (Bendix 1996).
These programmes differ in their goals from other programmes, such as patient care management, multidisciplinary treatments, pain clinics, standard medical care or physiotherapy, which aim to reduce symptoms, pain intensity, use of medications and health services, and to increase global improvement and quality of life (Guzman 2001; Guzman 2002), physiological outcomes such as range of motion and spinal flexibility (Hayden 2005), or behavioural outcomes such as anxiety, depression and cognition (Ostelo 2000). Recent years have shown a development towards more involvement of the workplace in interventions aiming for return to work for various musculoskeletal disorders including back pain. A systematic review by Carroll 2010 reported that stakeholder participation and work modification are more effective and cost‐effective for returning adults with musculoskeletal conditions to work than other workplace‐linked interventions, including exercise.
Physical conditioning as part of a return to work strategy is characterized by some form of structured exercise or advice about exercise based on the idea that inactivity due to avoidance of painful activities can lead to so‐called ‘deconditioning syndrome’, which in turn can lead to more pain from attempts to move joints that are stiffened and muscles weakened by disuse. Central to these programmes is the notion that as physical and functional capacities improve, so will the person’s capability of returning to work. The programme may be comprised of actual or simulated work tasks, or interventions addressing individual and work‐related psychosocial factors that may play an important role in persisting symptoms and disability, or both (Waddell 1999). This intended work outcome or job‐attached status to the pre‐injury employer is important for a successful outcome with these physical conditioning programmes (Schonstein 1999).
Maintenance of a job‐attached status to the pre‐injury employer is often best accomplished by the provision of suitable modified duties (Voaklander 1995). The effectiveness of modified duties has been studied and comprehensively reviewed (Krause 1998; Loisel 2005) and results indicate that the provision of suitable duties facilitates return‐to‐work, reduces days lost due to injury, and is cost‐effective. Accordingly, this review documents work conditioning programmes that include the availability of modified duties in the back pain management plan.
This review focuses exclusively on workers with back pain who are either off work or are at risk of being off work due to reduced work capacity, and evaluates the effectiveness of work hardening and functional restoration in improving their work status. This review is the second update of a Cochrane review first published in 2003 that summarised the evidence on the effectiveness of physical conditioning programmes for workers with back and neck pain (Schonstein 2003; Schonstein 2003a). An update was performed in 2010 (Schaafsma 2010). Results of the in 2010 updated review indicated that the effectiveness of physical conditioning programmes in reducing sick leave, when compared to usual care or to other exercises, in workers with back pain remains uncertain. In workers with acute back pain, these programmes probably have no effect on sick leave, but there may be a positive effect on sick leave for workers with subacute and chronic back pain. Workplace involvement may improve the outcome. Better understanding of the mechanism behind physical conditioning programmes and return‐to‐work is needed to be able to develop more effective interventions.
Other reviews have evaluated the efficacy of multidisciplinary back pain management programmes (Guzman 2001; Heymans 2005; Karjalainen 2000; Teasell 1996) in reducing disability related to back pain. In a review of functional restoration programmes for chronic low‐back pain, Teasell 1996 concluded that evidence to support physical conditioning was lacking. In contrast, Karjalainen 2000 reported that multidisciplinary biopsychosocial rehabilitation reduces subacute low‐back pain among working age adults, and that a work site visit increases its effectiveness. While some of these reviews have incorporated return‐to‐work in their outcome measures, none have focused exclusively on work outcomes.
The review by Heymans 2005 on back schools for non‐specific low‐back pain stated that there is moderate evidence suggesting that back schools, in an occupational setting, reduce pain and improve function and return‐to‐work status, in the short‐ and intermediate‐term. This is compared with exercises, manipulation, myofascial therapy, advice, placebo or waiting list controls for workers with chronic and recurrent low‐back pain. Back schools are defined as programmes consisting of educational and skills acquisition components, including exercises, in which all sessions are given to groups of workers and supervised by allied health professionals or a medical specialist. Unlike the interventions of this review, the components of back schools are not generally tailored specifically to job demands.
Hayden 2005 questioned whether exercise is more effective than reference treatments for individuals with non‐specific low‐back pain. They concluded that specific exercises are as effective as either no treatment or other conservative treatments for acute low‐back pain. However, exercise therapy is slightly more effective than no treatment or other conservative treatments at decreasing pain and improving function in adults with chronic low‐back pain. For subacute low‐back pain, there is some evidence that a graded activity programme reduces absenteeism, though evidence for other types of exercise is unclear. Despite the fact that exercise is an integral component of physical conditioning programmes, in our review physical conditioning programmes also needed to have a stated focus on functional job demands.
This review is unique because it addresses the specific question of whether physical conditioning that has a stated focus on functional job demands is effective in reducing sick leave and improving work status for workers with work‐related low‐back pain. As a result, some of the studies are also included in other reviews that explore the effects of specific interventions on pain, function, general well‐being and disability (Guzman 2002; Hayden 2005; Heymans 2005; Karjalainen 2000; Ostelo 2000).
In this update, we changed the title from 'Physical conditioning programs for improving work outcomes in workers with back pain' to 'Physical conditioning as part of a return to work strategy to reduce sickness absence for workers with back pain;. This new title illustrates that over the years physical conditioning has become more and more part of an integrated care programme including various modules and more explicitly involving the workplace. We focused exclusively on workers with back pain and on work status outcomes and therefore excluded neck pain and secondary outcomes such as functional status and physiological outcomes, which were included in the original review. We only included interventions that have a stated relationship with the workplace, a focus on job demands, and measured work outcomes. We reviewed new evidence available since the previous search carried out in 2008. In the results section of this updated review, a distinction is again made between workers with acute (less than six weeks), subacute (between six and 12 weeks) and chronic (more than 12 weeks) back pain. Further, we made a distinction between comparisons of physical conditioning in addition to care as usual compared to care as usual only, or physical conditioning compared to care as usual.
To assess the effectiveness of physical conditioning as part of a return to work strategy in reducing time lost from work and improving work status for workers with back pain. Further, to assess which aspects of physical conditioning are related to a faster return to work for workers with back pain.
Only randomized controlled trials (RCTs) and cluster RCTs, regardless of the language in which they were published, were included in this review.
Male and female adults (> 16 years) with work disability related to back pain who took part in physical conditioning programmes were included in this review. Back pain was defined as pain in the thoracic, lumbar or gluteal region, or a combination, with or without radiation to the lower extremities. Studies with at least 50% of workers with back pain were included. Work disability was defined as being on full or partial sick leave, or not being able to perform adequately at work due to back pain.
All workers who were accepted into physical conditioning programmes, whether they had acute (duration of symptoms less than six weeks), subacute (duration of symptoms more than six but less than 12 weeks) or chronic back pain (duration of symptoms more than 12 weeks), met our inclusion criteria.
Studies with non‐workers, or workers with specific diagnoses such as infection, neoplasm, metastasis, osteoporosis, rheumatoid arthritis, fracture, inflammatory processes or other conditions for which valid diagnoses had been demonstrated were excluded.
Physical conditioning programmes are also known as work conditioning or hardening, or functional restoration and exercise programmes. They include advice about exercise and may also simulate or duplicate work or functional tasks, or both, in a safe, supervised environment. These exercises or tasks are structured and progressively graded to increase psychological, physical and emotional tolerance and improve endurance and work feasibility. We included studies on physical conditioning programmes when they included the following three key elements:
(advice about ) exercises specifically designed to restore an individual's systemic, neurological, musculoskeletal (strength, endurance, movement, flexibility and motor control) or cardiopulmonary function, or a combination;
explicitly stated to have an intended improvement of work status; a stated relationship between the intervention and functional job demands.In addition to these three key elements, physical conditioning programmes could include components such as operant conditioning behavioural approach, pain management, back pain education, advice on return‐to‐work, workplace involvement and case‐management. The delivery of physical conditioning programmes could involve multidisciplinary teams or individual health professionals. In addition, they could be delivered in a one‐to‐one fashion or in a group situation.
Based on the intensity of the programme we differentiated between the following.
Light physical conditioning: these programmes included the three key elements and were delivered in fewer than five sessions (of one to two hours) or were described by the primary study author as a light intervention programme.
Intense physical conditioning: these programmes included the three key elements and were delivered in more than five sessions or were delivered on a full‐time basis for more than two weeks.
Work status outcomes were:
1. time between intervention and return‐to‐work;
2. return‐to‐work status in terms of 'at work' or ‘off work';
3. time on light or modified duties.
For this updated review, searches were conducted in the same databases as in the original review, for the period of June 2008 to March 2012. Searches were performed by the Trials Search Co‐ordinator of the Cochrane Back Review Group (CBRG). RCTs were identified by searching electronic databases, using the Ovid search strategy. Databases included were: CENTRAL, MEDLINE (from 1966), EMBASE (from 1980), CINAHL (from 1982), PsycINFO (from 1967), and PEDro. The CBRG Trials Register, ClinicalTrials.gov, and World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) were also searched until March 2012. In May 2013 we ran an additional search and all relevant trials were added to the 'Studies awaiting assessment' reference list.
All RCTs were included regardless of the language in which they were published. The highly sensitive search strategies of The Cochrane Collaboration were run in conjunction with a specific search for back pain and the interventions investigated (Furlan 2009). The MEDLINE search is based on the first two stages of the MEDLINE search strategy recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). This search update added terms to the original search strategy, based on the results of testing terms synonymous with work. These were added with a combination of year limits and NOT. The logic here was to retroactively search for the new terms without retrieving all of the same results that were already retrieved in the previous searches up to 2008. See Appendix 1.
This strategy was modified for EMBASE, CINAHL, and PsycINFO. Search words used for the PEDro database were: low‐back pain, backache, lumbar, thoracic, work conditioning, work hardening, functional restoration, exercise, and gym.
Two review authors (FS, KW) independently examined search results and independently applied the selection criteria to the studies retrieved. A consensus method was used to resolve disagreements concerning inclusion of RCTs. A third review author (JV) was consulted if disagreements persisted.
Two review authors (FS, KW) read all papers independently and determined eligibility according to the inclusion and exclusion criteria listed above.
Data were independently extracted from the studies by two review authors (FS and LE) using an adapted version of the pre‐designed form from the CBRG for data extraction. Discrepancies were resolved by consensus or by consulting a third author (JV). The following criteria were used in our data extraction.
1. Characteristics of study population: number of workers, gender, age and setting, country and date of the study, duration of symptoms, work status.
2. Characteristics of interventions: the content, duration and frequency of the physical conditioning programmes and control interventions.
3. Results on outcomes of interest.
We used the criteria recommended by the CBRG (Furlan 2009) to assess the risk of bias of the selected RCTs. The criteria and their operational definitions are outlined in Appendix 2. Each of the criteria were scored 'high', 'low' or 'unclear' according to the operationalisation of the criteria. Following the recommendations of the CBRG (Furlan 2009), studies were rated as having a ‘low risk of bias’ when at least six of the 12 CBRG criteria were met and the study had no serious flaws. Serious flaws were inadequate concealment of treatment allocation, a large dropout rate, or statistically significant and clinically important baseline differences not accounted for in the analyses. Studies with serious flaws or those in which fewer than six of the criteria were met were rated as having a ‘high risk of bias’. The results of the assessment are presented in the 'Risk of bias' table and Figure 1 .
The risk of bias in the RCTs was independently assessed by two review authors (LE, AB). This process was not blinded with regard to the authors, institution or journal. A consensus method was used to resolve disagreements, and a third review author (JV) was consulted for persisting disagreements. If information was absent for evaluation of the methodological criteria, the authors of the study were contacted with a request to provide additional information.
Studies expressed time to return‐to‐work as mean numbers of days off work. Rate of return‐to‐work was expressed as odds ratios (ORs). We regarded both time to return‐to‐work and the rate of return‐to‐work sufficiently alike, if they were measured at the same follow‐up time, to combine them as similar outcomes in the meta‐analysis. We converted the outcomes into standardized mean difference (SMD) because of the different scales used. We assumed that the continuous measurements in each intervention group followed a logistic distribution, and that the variability of the outcomes was the same in both the treated and control groups. Therefore, we were able to re‐express the calculated ORs as a standardized mean difference (SMD) according to the following simple formula (Chinn 2000; Chinn 2002) as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008): lnOR = π/√3 x SMD.
Pooling of studies was only considered when statistical heterogeneity was less than moderate (I² less than 60%).
Translating SMDs into daily practice and knowing what is a clinically worthwhile effect is difficult. Following the rules of thumb on SMDs from the Cochrane Handbook for Systematic Reviews of Interventions and the Cochrane Back Review Group (CBRG) guidelines (Furlan 2009; Higgins 2011), a SMD of more than 0.2 can be considered a small effect, more than 0.5 a moderate effect, and more than 0.8 a large effect.
However, in the original review it was stated that for continuous outcomes, a mean saving of 10 sick days per year was considered the smallest effect that would be clinically worthwhile, based on the assumption of the cost of providing physical conditioning programmes. For dichotomous outcomes, an OR of 0.65 was used as the smallest clinically worthwhile effect. This corresponded to a number needed to treat (NNT) of 10 when the baseline rate (prevalence of the event measured (on or off work)) was about 40%, based on the two comparisons from the original review (Schonstein 2003). An intervention that affected fewer than one in 10 people was considered not clinically worthwhile.
Following this original hypothesis, we decided that we would use an OR of 0.65 as the smallest clinically worthwhile effect. As we would recalculate all the continuous outcomes into SMDs, we also converted the OR of 0.65 back into a SMD. For this, we used the same formula: lnOR = π/ √3 x SMD. This led to a SMD of ‐0.24, which we considered the smallest clinically worthwhile effect.
We assessed the overall quality of the evidence for each outcome using an adapted GRADE approach (Atkins 2004) as recommended by the CBRG (Furlan 2009). The quality of the evidence for a specific outcome was based on the study design, limitations of the study, consistency, directness, precision of results, and publication bias. For further details see Appendix 3.
The GRADE Working Group recommends four levels of evidence; the CBRG recommends the addition of a fifth.
High quality evidence: where there are consistent findings among 75% of RCTs with low risk of bias that are generalisable to the population in question; there are sufficient data, with narrow confidence intervals; there is no known or suspected publication bias; further research is unlikely to change either the estimate or our confidence in the results.
Moderate quality evidence: one of the domains is not met; further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality evidence: two of the domains are not met; further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality evidence: three of the domains are not met; we are very uncertain about the estimate.
No evidence: no RCTs are identified that address this outcome.The clinical relevance of the studies was independently assessed by two review author (FS and AB). Clinical relevance tables were constructed using the five questions recommended by the CBRG (Furlan 2009).
1. Are the patients described in detail so that you can decide whether they are comparable to those that you see in your practice?
2. Are the interventions and treatment settings described well enough so that you can provide the same for your patients?
3. Were all clinically relevant outcomes measured and reported?
4. Is the size of the effect clinically important?
5. Are the likely treatment benefits worth the potential harms?
Regarding question 4. above, we converted the outcome data of studies into SMDs or ORs. We calculated cut‐off scores for these two effect sizes, as explained under 'Measures of treatment effect' below, to decide whether the size of the effect was considered clinically important and worthwhile. This conversion of study data and the use of cut‐off scores sometimes resulted in a conclusion that was different than the study authors' conclusions.
To clearly express the quality of the evidence and the magnitude of the effect on worker important outcomes we used the recommended statements by the GRADE Working Group in the results‐ and discussion sections.
For cluster designs we used the group estimates taking into account the cluster randomisation. For multiarm studies we used the data from both comparisons that included the physical conditioning programme.
First, we assessed which studies were clinically homogeneous, with similar populations, interventions, comparisons and outcomes measured at the same follow‐up point. Populations were considered similar if their symptoms were of a similar duration (acute, subacute or chronic).
Interventions were considered homogeneous if they satisfied the inclusion criteria (that is exercises, improvement of work status, and explicit relation to job tasks), regardless of the inclusion of extra components or modes of delivery (that is group or individual, multi or monodisciplinary). Sensitivity analyses were performed to evaluate the effects of the variable components (see below).
We compared light or intense physical conditioning, with or without care as usual, to care as usual, exercise therapy, cognitive behavioural therapy (CBT), a cognitive intervention and to a brief clinical intervention. In addition, we compared intense physical conditioning to a combination of physical conditioning with a CBT component.
Outcomes were considered similar when they measured either the time to return‐to‐work or the proportion of workers that resumed work at specific times. These outcomes were similar because they both measured the time between injury and return‐to‐work, but were expressed in either days or percentages of workers returning to work. Follow‐up was classified into four categories: short‐term follow‐up refers to measures taken closest to three months, intermediate‐term follow‐up refers to measures taken closest to six months, long‐term follow‐up refers to outcomes closest to one year, very long‐term follow‐up refers to measures taken closest to two years. For an outcome measure of return‐to‐work, the return‐to‐work should be sustainable over a longer period of time, generally at least four weeks. For this reason, we did not consider a measurement of outcome at one month follow‐up as recommended by the CBRG method guidelines.
We tested for statistical heterogeneity by means of the I² in the meta‐analysis graphs. We used the criterion of 50%, mentioned in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008), to discern the presence or absence of heterogeneity between studies. When studies were statistically heterogeneous according to the I² statistic, a random‐effects model was used, otherwise a fixed‐effect model was used.
A subgroup analysis was performed for differences in components of physical conditioning programmes such as an operant conditioning behavioural approach, pain management, back pain education, advice on return‐to‐work and a workplace visit. A subgroup analysis was also performed on the mode of delivery of physical conditioning programmes, such as multidisciplinary or monodisciplinary and group or individual exercises.
A sensitivity analysis was performed to determine whether the overall results were the same when studies with different definitions of low or high risk of bias were analysed and when studies with different types of participants on full‐time or part‐time sick leave at baseline were analysed.
The updated database search was run from 2008 up to March‐April 2012. This resulted in 358 references from CENTRAL, 544 references from MEDLINE, 725 references from CINAHL, 1114 references from EMBASE, and 22 references from PsycINFO. A separate search was conducted for the PEDro database until the end of March 2012, which yielded 13 references. Another separate search was conducted for the CBRG trial register, which resulted in 70 references. In April 2012 a search in ClinicalTrials.gov and WHO ICTRP resulted in another 46 registered trials. Duplicates of references and irrelevant articles were removed leaving 1761 references and 37 registered trials. Two review authors (FS and KW) independently assessed (by title or key words, or both) these references for appropriateness for inclusion.
After screening the abstracts of articles from the searches, seven papers were considered. An evaluation of the full text led to four new papers meeting our inclusion criteria. One paper reported a third set of follow‐up results from the same study (Roche 2007). Therefore, only the remaining three new studies were included in this updated version of the review. In the previous update we presented the data from Roche 2007 separately from the same study by Jousset 2004. In this update both articles will be referred to by only one reference: Roche 2007. The addition of three new studies (Bethge 2011; Jensen 2011; Lambeek 2010) resulted in a total of 25 studies published in 41 papers in this updated review. All data used in this review were obtained from these published studies.
We found four studies that did not specifically focus on back pain, but on workers with musculoskeletal pain (Bethge 2011; Corey 1996; Meyer 2005; Mitchell 1994). However, all three studies did have more than 50% of the workers with back pain and were therefore included in this updated review. All studies were RCTs, carried out between 1992 and 2012. Two studies had a cluster randomised design (Bethge 2011; Loisel 1997).
In 15 studies all workers were on sick leave, either part‐time or full‐time (Altmaier 1992; Corey 1996; Heymans 2006; Jensen 2001; Jensen 2011; Lambeek 2010; Lindstrom 1992; Loisel 1997; Meyer 2005; Mitchell 1994; Staal 2004; Steenstra 2006; Storheim 2003; van den Hout 2003; Wright 2005). In the other 10 studies not all workers were on sick leave, but all workers were described as having decreased ability to perform job requirements or may have had an episode of sick leave before randomisation (Bendix 1996; Bendix 1997; Bendix 2000; Bethge 2011; Faas 1995; Gatchel 2003; Karjalainen 2003; Kool 2005; Roche 2007; Skouen 2002). The average period of sick leave correlated with the duration of the back pain, between three weeks and six months.
All interventions were related to work, contained physical exercises or advice about physical exercises, and had a focus on return‐to‐work. However, the number of sessions and their content varied greatly. For example, one intervention included one session only during which the worker was examined and engaged in a discussion about working conditions, trained to use five exercises and then referred back to the general practitioner (GP) with specific recommendations (Karjalainen 2003); another included a full‐time multidisciplinary treatment of eight weeks (Mitchell 1994). We labelled four interventions as light physical conditioning (Faas 1995; Heymans 2006; Jensen 2011; Karjalainen 2003; Skouen 2002; Wright 2005); the other 21 studies were labelled as intense physical conditioning. Seventeen interventions were delivered by a multidisciplinary group (for example physiotherapist, occupational therapist, ergonomist, social worker, case manager, rehabilitation physician, occupational health physician or nurse); five interventions were delivered by a physiotherapist (Faas 1995; Heymans 2006; Lindstrom 1992; Steenstra 2006; Storheim 2003). In Mitchell 1994, it was unclear who delivered the intervention.
For more details about the content of the interventions, see Table 1 ; Table 2 ; Table 3 ; Table 4 .
| Faas 1995 | Heymans 2006 | Jensen 2011 | Karjalainen 2003 | Skouen 2002 | Wright 2005 | |
| Time span of training | 5 wks | 4 wks | 18 weeks median duration | na | na | 2 wks |
| Number of sessions | 2 per week | 1 per week | after the interview, the participant was seen at least once, and four times on average | 2 | approx. 4 hrs at the start + 6 follow up or individual sessions over a period of 1 year | 1 examination + treatment initially then 3 per week |
| Length of sessions | 20 min | 120 min | 1‐2 hrs | 1‐1,5 hr | unclear | 1 hr |
| Full time | no | no | no | no | no | no |
| group or individual | individual | group | individual | individual | both | both |
| exercises | yes | yes | yes in advice | yes | yes, advice and programme | yes |
| work related exercises | yes | yes | yes in advice | yes | yes in advice | yes |
| operant conditioning behavioural approach | no | no | no | no, although intervention was based on graded activity programme | no | no |
| pain coping/ management | no | no | yes | no | no | no |
| back pain education | no | yes | yes | no | no | no |
| ergonomic advice or occupational training | no | yes | no | no | no | advice on how to modify physical activities specific to the individual's work situation |
| return‐to‐work advice | no | yes | yes | no | no | no |
| workplace visit | no | no | yes | yes | no (not for all individuals) | no |
| therapists involved | physiotherapist | physiotherapist, occupational physician | rehabilitation physician, specialist in clinical social medicine, physiotherapist, social worker, occupational therapist, case manager | physiotherapist, physician | physio therapist, nurse, psychologist | physiotherapist |
| other aspects | written compliance contract | CAU | brief clinical examination | CAU | no | CAU being a back book and advice on how to modify physical activities specific to the indivual's work situation |
| comparison | CAU | CAU/ intense PCP | brief clinical examination only | CAU | CAU / intense PCP | CAU being a back book and advice on how to modify physical activities specific to the individual's work situation ‐ only |
| Altmaier 1992 | Bendix 1996 | Bendix 1997 | Bendix 2000 | Bethge 2011 | Corey 1996 | Gatchel 2003 | Heymans 2006 | |
| Time span of training | 3 wks | 3 wks | 6 wks | 3 wks | 3 wks | 33 days | 3 wks | 8 wks |
| Number of sessions | 2 per day | 39 hrs per week + 3x6 hrs follow up | 135 hr in total | 39 hours per week + 3x6 hrs follow‐up | total of 82.2 hours of therapy | 6.5 hr per day | up to 41 | 2 per week |
| Length of sessions | ? | na | na | na | 1‐1.5 hrs | na | 15 min‐1hr | 1 hr |
| Full time | no | yes | yes | yes | yes | yes | no | no |
| group or individual | both | group | group | group | group | both | both | individual |
| exercises | yes | yes | yes | yes | yes | yes | yes | yes |
| work related exercises | yes | yes | yes | yes | yes | yes | yes | yes |
| operant conditioning behavioural approach | yes | yes | yes | yes | yes | yes | not clear | yes |
| pain coping/ management | yes | yes | yes | yes | unclear | yes | yes | no |
| back pain education | yes | yes | yes | yes | yes | yes | no | no |
| ergonomic advice or occupational training | no | occupational therapy and ergonomic training | occupational therapy and ergonomic training | occupational therapy and ergonomic training | no | no | occupational therapy | yes |
| return‐to‐work advice | no | no | no | no | yes | no | no | yes |
| workplace visit | no | no | no | no | no | no | no | no |
| therapists involved | multidisciplinary | multidisciplinary | multidisciplinary including physician, psychologist | occupational therapist, physician, psychologist,physical therapist, social worker | physician, social worker, psychologist, physical therapist | interdisciplinary programme | physiotherapist, occupational therapist, nurse, physician | physiotherapist |
| other aspects | vocational rehabilitation | recreation activities | aerobics, recreational activities | recreation activities | no | no | no | CAU |
| comparison | PCP + CBT | CAU | exercise therapy / pain management | outpatient intensive physical training | inpatient conventional musculoskeletal rehabilitation | CAU | CAU | CAU‐only/ light PCP |
| Jensen 2001 | Kool 2005 | Lambeek 2010 | Lindstrom 1992 | Loisel 1997 | Meyer 2005 | Mitchell 1994 | Roche 2007 | |
| Time span of training | 4 wks | 3 wks | 12 wks | until RTW | 13 wks | 8 wks | 8 wks | 5 wks |
| Number of sessions | 6 sessions + 20 hrs exercise + 6 booster sessions | 4 hrs per day / 6 days a week | varying | approx. 11 with physical therapist, approx 10 self training sessions (3 per week) | 3,5 hr per day, 5 days a week | 6 hrs per day, 5 days a week | ||
| Lenght of sessions | ||||||||
| Full time | no | almost | no | no | unclear | almost | yes | yes |
| group or individual | both | group | individual | individual | unclear | both | group | group |
| exercises | yes | yes | yes | yes | yes | yes | yes | yes |
| work related exercises | yes | yes | yes | yes | yes | yes | yes | yes |
| operant conditioning behavioural approach | yes | no | yes | yes | yes | yes | yes | no |
| pain coping/ management | 2 didactic sessions on psychological aspects of pain + 2 sessions on medical aspects of pain | no | no | no | no | yes | no | no |
| back pain education | yes | no | no | yes | yes | no | yes | no |
| ergonomic advice or occupational training | 2 sessions on ergonomics | work simulation | yes | no | yes, participatory ergonomics evaluation | education in ergonomics | no | occupational therapy |
| return‐to‐work advice | yes, workplace visit + rehabilitation plan | no | yes | yes | yes | yes | no | no |
| workplace visit | yes | no | yes | yes | yes | yes | no | no |
| therapists involved | physician, physical therapist, psychologist, | rheumatologist, physical and occupational therapist, sports therapist, social worker, nurse | clinical occupational physician, medical specialist, physiotherapist | physical therapist | back pain specialist; multidisciplinary medical, ergonomic and rehabilitation staff | interdisciplinary:rehabilitation physicians, psychologist, social worker, occupational therapist, physiotherapist | unclear | specialist in physical medicine, physiotherapist, psychologist |
| other aspects | no | no | CAU | CAU | CAU | case‐manager recreational activities | no | no |
| comparison | CAU / CBT PCP + CBT | pain centred treatment | CAU‐only | CAU‐only | CAU‐only | exercise therapy | CAU | active individual therapy |
| Skouen 2002 | Staal 2004 | Steenstra 2006 | Storheim 2003 | van den Hout 2003 | |
| Time span of training | 4 wks | max 3 months | 13 wks | 15 wks | 8 wks |
| Number of sessions | 5 per week for 4 weeks + follow‐up as in LMT | 2 per week until RTW | 2 per week 26 in total | 2‐3 per week | 28 |
| Lenght of sessions | 6 hr | 1 hr | 1 hr | 1 hr | 30‐90 min |
| Full time | almost | no | no | no | no |
| group or individual | both | individual | individual | group | both |
| exercises | yes | yes | yes | yes | yes |
| work related exercises | yes | yes | yes | not clear | yes |
| operant conditioning behavioural approach | yes | yes | yes | no | yes |
| pain/ coping management | no | no | no | no | no |
| back pain education | yes | no | no | no | yes |
| ergonomic advice or occupational training | yes | no | no | yes, training had a focus on ergonomic principles and functional tasks | yes, training by occupational therapist |
| return‐to‐work advice | no | yes | yes | no | yes |
| Workplace visit | no, occasional workplace intervention | no, but intervention was at workplace | no | no | yes, if necessary |
| therapists involved | physio therapist, nurse, psychologist | physiotherapist, occupational physician | physiotherapist, occupational physician | physical therapist | physiotherapist, occupational therapist, psychologist, occupational physician |
| other aspects | no | CAU/gradually increasing exercise, GP or occupational physician if workers wanted to | CAU/gradually increasing exercise | exercises accompanied by music | contact with patients' supervisor |
| comparison | CAU/ light PCP | CAU‐only | CAU‐only | CAU/ cognitive intervention | PCP + CBT |
There was a large variety of comparisons with physicial conditioning. The effectiveness of physical conditioning combined with usual care was compared to usual care in seven studies (Heymans 2006; Karjalainen 2003; Lambeek 2010; Lindstrom 1992; Loisel 1997; Staal 2004; Steenstra 2006). Usual care was mostly provided by a primary care physician without any restrictions on treatment such as referral or prescriptions. The studies by Heymans 2006; Staal 2004; Steenstra 2006 specifically mentioned the guidance of an occupational health physician in their usual care group. One study compared a multidisciplinary intervention involving a case manager who, together with the worker, constructed a tailored rehabilitation plan in addition to a brief clinical intervention (Jensen 2011). One study compared physical conditioning with a back book, workplace advice with a back book, and workplace advice only (Wright 2005). In nine studies physical conditioning was compared to care as usual (Bendix 1996; Bendix 1997; Bendix 2000; Corey 1996; Faas 1995; Gatchel 2003; Mitchell 1994; Skouen 2002; Storheim 2003). In six studies, exercise therapy was used as the comparison (Bendix 1997; Bendix 2000; Bethge 2011; Jensen 2001; Meyer 2005; Roche 2007). In two studies a comparison was made between light physical conditioning and intense physical conditioning (Heymans 2006; Skouen 2002); in two other studies a comparison was made between physical conditioning and physical conditioning with added CBT (delivered by psychologists) (Altmaier 1992; van den Hout 2003). Two studies compared physical conditioning with CBT (Bendix 1997; Jensen 2001). Two studies compared physical conditioning with multidisciplinary inpatient rehabilitation (Kool 2005; Bethge 2011). One study compared physical conditioning with a cognitive intervention that focused on explanation of pain mechanisms, staying active, and taking responsibility (Storheim 2003).
Follow‐up times varied from outcome data taken directly after a five‐week intervention up to three years. Five studies reported on short‐term follow‐up: three months (Bendix 1996; Kool 2005; Roche 2007; Storheim 2003; Wright 2005); eight studies reported on intermediate‐term follow‐up: six months (Altmaier 1992; Bethge 2011; Heymans 2006; Meyer 2005; Roche 2007; Staal 2004; Steenstra 2006; van den Hout 2003); 18 studies reported on long‐term follow‐up: one year (Bendix 1996; Bendix 1997; Bendix 2000; Bethge 2011; Corey 1996; Faas 1995; Gatchel 2003; Jensen 2001; Jensen 2011; Karjalainen 2003; Kool 2005; Lambeek 2010; Lindstrom 1992; Loisel 1997; Mitchell 1994; Staal 2004; Steenstra 2006; van den Hout 2003); and six studies reported on very long‐term follow‐up (Bendix 1996; Bendix 1997; Jensen 2001; Karjalainen 2003; Loisel 1997; Staal 2004), varying between two and three years.
All studies reported on work status, most often measured as mean days of sick leave until follow‐up or return‐to‐work rate at follow‐up. Several studies reported on 'work readiness' or work capability (Bendix 1996; Bendix 1997; Bendix 2000; Skouen 2002), and one study reported on percentages of full‐time workability (Meyer 2005). The data from these studies were recalculated in order to allow pooling of the data. For example, data on full work days of workers were extracted from the follow‐up days to calculate the sick leave days. For studies that measured mean days of sick leave, the results were shown using the SMD. Results from studies that reported on return‐to‐work rate were shown using odds ratios (OR). There were six studies that reported hazard ratios for return‐to‐work (Heymans 2006; Jensen 2001; Jensen 2011; Lambeek 2010; Staal 2004; Steenstra 2006). These data were recalculated or the authors were asked to provide mean days of sick leave to make it possible to pool the data with other study results.
We found four studies (Altmaier 1992; Loisel 1997; Roche 2007; Wright 2005) that reported on other work status outcomes, such as part‐time return‐to‐work, return to 'light' duties, or 'therapeutic' return‐to‐work. However, only the data from Altmaier 1992 and Roche 2007 could actually be used. Wright 2005 did not provide separate data on the change to or from 'light' duties, and Loisel 1997 only mentioned in the results section that analyses with return to any work as an outcome showed no significant benefit in any group or combination of groups.
Seven out of 25 studies scored positive on all the five questions regarding clinical relevancy (Bendix 1997; Karjalainen 2003; Kool 2005; Lambeek 2010; Lindstrom 1992; Loisel 1997; Staal 2004). All studies clearly explained the type of workers that participated in the intervention. Only one study (Altmaier 1992) scored a 'no' on the question about the clarity of the intervention. Two studies scored a 'no' on: 'were all clinically relevant outcomes measured and reported' (Corey 1996; Storheim 2003). With the recalculation of data into SMD or OR we found seven studies that scored positive on the fourth question: 'Is the size of the effect clinically important?' (Bendix 1997; Karjalainen 2003; Kool 2005; Lambeek 2010; Lindstrom 1992;Loisel 1997; Staal 2004). By consensus, we considered that the fifth question 'whether the likely treatment benefits would be worth the potential harms' would be scored positive for all studies as there was no apparent harm for the worker with these type of interventions. However, we realize that this is debatable and this topic therefore needs to be further explored (see Table 5 ).
| Study ID | 1 | 2 | 3 | 4 | 5 |
| Altmaier 1992 | + | ‐ | + | ‐ | + |
| Bendix 1996 | + | + | + | ‐ | + |
| Bendix 1997 | + | + | + | + | + |
| Bendix 2000 | + | + | + | ‐ | + |
| Bethge 2011 | + | + | + | ‐ | + |
| Corey 1996 | + | + | ‐ | ‐ | + |
| Faas 1995 | + | + | + | ‐ | + |
| Gatchel 2003 | + | + | + | ‐ | + |
| Heymans 2006 | + | + | + | ‐ | + |
| Jensen 2001 | + | + | + | ‐ | + |
| Jensen 2011 | + | + | + | ‐ | + |
| Karjalainen 2003 | + | + | + | + | + |
| Kool 2005 | + | + | + | + | + |
| Lambeek 2010 | + | + | + | + | + |
| Lindstrom 1992 | + | + | + | + | + |
| Loisel 1997 | + | + | + | + | + |
| Meyer 2005 | + | + | + | ‐ | + |
| Mitchell 1994 | + | + | + | ‐ | + |
| Staal 2004 | + | + | + | + | + |
| Steenstra 2006 | + | + | + | ‐ | + |
| Storheim 2003 | + | + | ‐ | ‐ | + |
| Skouen 2002 | + | + | + | ‐ | + |
| Roche 2007 | + | + | + | ‐ | + |
| van den Hout 2003 | + | + | + | ‐ | + |
| Wright 2005 | + | + | + | ‐ | + |
From the original and previously updated review, 13 studies were excluded because the interventions from those studies had no clear relationship with the work situation or functional job demands, the majority of workers were not on sick leave at baseline, or the outcome was not return‐to‐work (Alaranta 1994; Aure 2003; Bentsen 1997; Dahl 2001; Dettori 1995; Friedrich 1998; Hagen 2000; Hansen 1993; Kellett 1991; Linton 2005; Malmivaara 1995; Moffett 1999; Niemisto 2003; Schiltenwolf 2006; Seferlis 1998; Torstensen 1998). With the new search, another three studies were excluded because the majority of workers were not on sick leave at baseline due to low‐back pain (Rantonen 2012; Whitfill 2010), or physical conditioning was not considered a structural part of the intervention (Bültmann 2009).
The risk of bias of the studies was independently assessed by two review authors (LE, AB), who used a consensus method if disagreements occurred. We used the 12 criteria recommended by the Back Review Group (Furlan 2009) to assess the risk of bias of the selected RCTs. The criteria are outlined in Appendix 2. Each of the criteria were scored 'high', 'low' or 'unclear'. Between the two risk of bias assessors, there were an average of one or two items of disagreement for every study. All disagreements were resolved after discussion. For the newly included studies, we sent the results of our risk of bias assessment to the (first) authors of the RCTs when there were answers with a 'high' or 'unclear' assessment asking them to comment on our scores, especially if the answers were 'unclear', and to provide us with additional information. We used the data on risk of bias of the originally included studies from the original review and did not ask those authors for comment.
Six authors responded to questions concerning the risk of bias tables and another four authors responded to questions concerning outcome data, leading to three changes in the risk of bias tables.
Following the advice of Furlan 2009, studies were rated as having a ‘high risk of bias’ if they had serious flaws (for example inadequate concealment of treatment allocation) or had met fewer than six of the risk of bias criteria. We found two studies (Corey 1996; Mitchell 1994 ) with a high risk of bias because they did not meet at least six of the risk of bias criteria. Another seven studies (Altmaier 1992; Bendix 1996; Bendix 1997; Gatchel 2003; Lindstrom 1992; Mitchell 1994; Roche 2007) were considered to have a high risk of bias because of unclear or no allocation concealment. The other 16 studies were considered to have a low risk of bias ( Figure 1 ).
The included studies covered the following comparisons.
1. Light physical conditioning and back book with advice versus back book with advice (N = 1) (Wright 2005)
2. Light physical conditioning versus care as usual (N = 1) (Faas 1995)
3. Intense physical conditioning versus care as usual (N = 1) (Gatchel 2003)
4. Light physical conditioning with care as usual versus care as usual (N = 2) (Heymans 2006; Karjalainen 2003)
5. Light physical conditioning and brief clinical intervention versus brief clinical intervention (N = 1) (Jensen 2011)
6. Intense physical conditioning with care as usual versus care as usual (N = 5) (Heymans 2006; Lindstrom 1992; Loisel 1997; Staal 2004; Steenstra 2006)
7. Intense physical conditioning versus light physical conditioning (N = 2) (Heymans 2006)
8. Intense physical conditioning versus cognitive intervention (N = 1) (Storheim 2003)
9. Intense physical conditioning versus care as usual (N = 1) (Storheim 2003)
10. Intense physical conditioning versus multidisciplinary exercise treatment (N = 2) (Bethge 2011; Kool 2005)
11. Light physical conditioning versus care as usual (N = 1) (Skouen 2002)
12. Intense physical conditioning with care as usual versus care as usual (N = 1) (Lambeek 2010)
13. Intense physical conditioning versus care as usual (N = 5) (Bendix 1996; Corey 1996; Jensen 2001; Mitchell 1994; Skouen 2002)
14. Intense physical conditioning versus an exercise programme (N = 5) (Bendix 1997; Bendix 2000; Meyer 2005; Roche 2007)
15. Intense physical conditioning versus intense physical conditioning with CBT (N = 3) (Altmaier 1992; Jensen 2001; van den Hout 2003)
16. Intense physical conditioning versus CBT (N = 2) (Bendix 1997; Jensen 2001)
17. Intense physical conditioning versus light physical conditioning (N = 1) (Skouen 2002)
Despite the heterogeneity of studies with respect to duration of back pain, comparison of treatment, follow‐up time and effect measure, we were able to pool some studies in specific subgroups and perform several meta‐analyses.
Three studies reported on the effect of physical conditioning on work status for workers with acute back pain. Due to different type of comparisons none of the studies could be pooled.
One RCT with low risk of bias (Wright 2005) (80 workers) reported low quality evidence that there was no difference in the reduction of the proportion of workers off work at two‐month follow‐up between light physical conditioning and GP advice with a back book compared to GP advice with a back book only for workers with acute back pain, with an OR of 0.36 (95% CI 0.13 to 1.03).
One RCT with low risk of bias (Faas 1995) (190 workers) reported low quality evidence that there was no difference in the reduction of mean days off work at one‐year follow‐up between light physical conditioning and care as usual for workers with acute back pain, with a SMD of ‐0.02 (95% CI ‐0.30 to 0.27).
One RCT with high risk of bias (Gatchel 2003) (70 workers) reported low quality evidence that there was no difference in the reduction of the proportion of workers off work at one‐year follow‐up between intense physical conditioning and care as usual for workers with acute back pain, with an OR of 0.22 (95% CI 0.05 to 1.06).
One RCT with low risk of bias (Heymans 2006) (299 workers) reported low quality evidence that there was no difference in sickness absence duration at six‐month follow‐up when comparing light physical conditioning and care as usual to care as usual only for workers with subacute back pain, with a SMD of ‐0.18 (95% CI ‐0.45 to 0.10).
Another RCT with low risk of bias (Karjalainen 2003) (112 workers) reported low quality evidence that there was no difference in sickness absence duration at one‐and two‐year follow‐up when comparing a light mobilization and graded activity programme to a workplace visit with care as usual and care as usual only for workers with subacute back pain, with a SMD of ‐0.35 (95% CI ‐0.74 to 0.03) and a SMD of ‐0.30 (95% CI ‐0.69 to 0.09), respectively.
One RCT with low risk of bias (Jensen 2011) (351 workers) reported low quality evidence that there was no difference in sickness absence duration at one‐year follow‐up when comparing a multidisciplinary intervention (that is case manager and worker constructed a tailored rehabilitation plan in addition to a brief clinical intervention) to a brief clinical intervention only for workers with subacute back pain at one‐year follow‐up, with a SMD of 0.21 (95% CI 0.00 to 0.42).
Three RCTs reported conflicting evidence on the effect of intense physical conditioning with care as usual versus care as usual for workers with subacute back pain at intermediate‐term follow‐up (Heymans 2006; Staal 2004; Steenstra 2006). We found high statistical heterogeneity when all three studies were pooled (I² = 75%). We compared the components of the three physical conditioning programmes and their style of delivery and found that Staal 2004 differed from the other two studies in that the intervention was conducted at the workplace. This RCT with low risk of bias (Staal 2004) (134 workers) reported that intense physical conditioning may be more effective than usual care at reducing the time to return‐to‐work, with a SMD of ‐ 0.42 (95% CI ‐0.76 to ‐0.08). In contrast, pooling the other two RCTs with low risk of bias (Heymans 2006; Steenstra 2006) (313 workers) resulted in low quality evidence that there was no difference in sickness absence duration between intense physical conditioning with care as usual versus care as usual for workers with subacute back pain at intermediate‐term follow‐up, with a pooled SMD of 0.13 (95% CI ‐0.09 to 0.35).
Four RCTs reported conflicting evidence on the effect of intense physical conditioning with care as usual versus care as usual for workers with subacute back pain at long‐term follow‐up. Again, we found high statistical heterogeneity when all four studies were pooled (I² = 78%). We compared the components of the four physical conditioning programmes and their style of delivery and found that Steenstra 2006 differed from the other three studies in that no workplace visit was included in their physical conditioning. Lindstrom 1992 and Loisel 1997 had an explicit workplace visit included in their programme, and Staal 2004 executed the intervention at the workplace. This RCT with low risk of bias (Steenstra 2006) (112 workers) reported that care as usual alone was more effective in reducing sickness absence duration than physical conditioning plus care as usual, with a SMD of 0.39 (95% CI 0.02 to 0.77). In contrast, pooling the other three RCTs, one with high and two with low risk of bias (Lindstrom 1992; Loisel 1997; Staal 2004) (283 workers), showed that intense physical conditioning with explicit workplace involvement plus care as usual compared to care as usual only was more effective and clinically relevant in reducing the time to return‐to‐work, with a pooled SMD of ‐0.42 (95% CI ‐0.65 to ‐0.18) ( Figure 2 ).
Forest plot of comparison: 7 Intense PC + CaU versus CaU only, subacute pain, outcome: 7.3 Time to return‐to‐work.
The pooled results of two RCTs, one with a high and one with a low risk of bias (Lindstrom 1992; Staal 2004) (257 workers), showed moderate quality evidence that intense physical conditioning with care as usual was more effective in reducing sickness absence duration compared to care as usual only for workers with subacute back pain at very long‐term follow‐up (two years), with a pooled SMD of ‐0.39 (95% CI ‐0.76 to ‐0.02) ( Figure 3 ).
Forest plot of comparison: 11 Intense PC versus multidisciplinary exercise treatment, subacute pain, outcome: 11.1 Proportion off work short‐term follow‐up.
One RCT with low risk of bias (Heymans 2006) (196 workers) reported low quality evidence that there was no difference in sickness absence duration between light physical conditioning and intense physical conditioning for workers with subacute back pain at intermediate‐term follow‐up (six‐month), with a SMD of ‐0.24 (95% CI ‐0.52 to 0.04). Both interventions were combined with care as usual.
One RCT with low risk of bias (Storheim 2003) (64 workers) reported low quality evidence that physical conditioning consisting of intensive group training with a focus on ergonomic principles and functional tasks was no more effective in reducing time to return to work compared with a cognitive intervention for workers with subacute back pain at 18‐week follow‐up, with a SMD of ‐0.10 (95% CI ‐0.59 to 0.40). The cognitive intervention was provided by a specialist in physical medicine and a physical therapist, and consisted of two consultations in which pain mechanisms were explained, reassurance was given, and advice and instructions were given on how to stay active and use muscles for demanding tasks.
One RCT with low risk of bias (Storheim 2003) (59 workers) reported low quality evidence that there was no difference in sickness absence duration between an intense physical conditioning programme and care as usual for workers with subacute back pain at 18‐week follow‐up, with a SMD of ‐0.10 (95% CI ‐0.61 to 0.41).
One RCT with low risk of bias (Kool 2005) (173 workers) reported low quality evidence that intense physical conditioning was more effective and clinically worthwhile than a multidisciplinary exercise treatment with a focus on pain reduction in reducing the proportion of workers off work, for workers with subacute back pain at short‐term (three months) follow‐up, with an OR of 0.41 (95% CI 0.22 to 0.78).
One RCT with low risk of bias (Bethge 2011) (149 workers) reported low quality evidence that intense physical conditioning was not more effective than a conventional musculoskeletal rehabilitation programme in reducing the proportion of workers off work, for workers with subacute back pain at intermediate‐term (six months) follow‐up, with an OR of 0.72 (95% CI 0.40 to 0.99).
The pooled results of two RCTs with low risk of bias (Bethge 2011; Kool 2005) (301 workers) showed moderate quality evidence that intense physical conditioning was more effective and clinically worthwhile than a multidisciplinary exercise treatment in reducing the proportion of workers off work, for workers with subacute back pain at long‐term (12 months) follow‐up, with a pooled OR of 0.63 (95% CI 0.40 to 0.99) ( Figure 3 ).
One RCT with low risk of bias (Skouen 2002) (106 workers) reported low quality evidence that there was no difference in sickness absence duration between a light physical conditioning programme and usual care at one‐ and two‐year follow‐up, with a SMD of ‐0.33 (95% CI ‐0.67 to 0.02) and a SMD of ‐0.34 (95% CI ‐0.69 to 0.01) respectively, for workers with chronic back pain.
One RCT with low risk of bias (Lambeek 2010) (134 workers) reported low quality evidence that intense physical conditioning with care as usual was more effective and clinically worthwhile than care as usual only for workers with chronic back pain in reducing duration of sickness absence, with a SMD of ‐4.42 (95% CI ‐5.06 to ‐3.79) at long‐term (12 months) follow‐up. The physical conditioning in this study was part of an integrated care protocol of three elements: integrated care management by a clinical occupational physician, workplace intervention and graded activity.
One RCT with high risk of bias (Bendix 1996) (74 workers) reported very low quality evidence that intense physical conditioning was more effective and clinically worthwhile than care as usual for workers with chronic back pain in reducing the proportion of workers off work in the short‐term (four months), with an OR of 0.16 (95% CI 0.05 to 0.49).
The pooled results of five RCTs (Bendix 1996; Corey 1996; Jensen 2001; Mitchell 1994; Skouen 2002) (1093 workers) showed moderate quality evidence that intense physical conditioning was more effective than care as usual for workers with chronic back pain in reducing the time to return‐to‐work at long‐term follow‐up. However, the size of the effect may not be clinically worthwhile with a SMD of ‐0.23 (95% CI ‐0.42 to ‐0.03). This comparison included workers who at baseline were on full or partial sick leave as well as workers who were at work but reported work disability ( Figure 4 ).
Forest plot of comparison: Intense PCP versus care as usual for workers with chronic back pain, outcome: Time to return‐to‐work at long‐term follow‐up
A sensitivity analysis of studies in which all workers were on full‐time sick leave due to chronic back pain at baseline (Corey 1996; Jensen 2001; Mitchell 1994) showed no difference in sickness absence duration between intense physical conditioning and exercise treatment at long‐term follow‐up (12 to 18 months), with a pooled SMD of ‐0.14 (95% CI ‐0.39 to 0.11).
The same effect was reported for another sensitivity analysis when only the two studies with low risk of bias (Jensen 2001; Skouen 2002) were included in this comparison, with a pooled SMD of ‐0.13 (95% CI ‐0.38 to 0.12).
Further subgroup analyses of differences in components of physical conditioning of these five studies did not change the results.
The pooled results of three RCTs (Bendix 1996; Jensen 2001; Skouen 2002) (297 workers) showed moderate quality evidence that intense physical conditioning was not more effective than care as usual for workers with chronic back pain in reducing sick leave time at two‐ to three‐year follow‐up, with a pooled SMD of ‐0.26 (95% CI ‐0.61 to 0.10) ( Figure 4 ).
Four studies measured the effects of intense physical conditioning versus exercise treatment for workers with chronic back pain on the mean number of days off work or return‐to‐work rate at various follow‐up times.
One RCT with high risk of bias (Roche 2007) (136 workers) reported very low quality evidence that intense physical conditioning was no more effective than active individual exercise therapy for workers with chronic back pain in reducing the proportion of workers off work, measured directly after the intervention programme of five weeks, with an OR of 1.00 (95% CI 0.37 to 2.70).
This study also reported on workers returning to any type of work (full time, partial or work adaptation) and again showed no difference in effect on return to any type of work.
The pooled results of two RCTs with a high risk of bias (Meyer 2005; Roche 2007) (114 workers) showed low quality evidence that there was no difference between intense physical conditioning and exercise treatment for workers with chronic back pain in the reduction of time to return‐to‐work at six‐month follow‐up, with a SMD of 0.19 (95 CI ‐0.63 to 0.24). The study by Meyer 2005 was actually a pilot study for a larger trial ( Figure 5 ).
Forest plot of comparison: Intense PCP versus exercise programme for workers with chronic back pain, outcome: Time to return‐to‐work at intermediate‐term follow‐up.
Three RCTs reported conflicting evidence on the effect of intense physical conditioning for workers with chronic back pain at long‐term follow‐up versus an exercise treatment. Due to high statistical heterogeneity, results could not be pooled (I² = 74%). Further analysis of components between the three physical conditioning programmes showed no differences. Two studies (Bendix 1997, high risk of bias; Roche 2007 low risk of bias) (182 workers) showed low quality evidence of a clinically worthwhile effect in favour of intense physical conditioning, with a SMD of ‐0.67 (95% CI ‐1.26 to ‐0.08). However, another RCT with high risk of bias from the same authors (Bendix 2000) (74 workers) showed very low quality evidence that there was no difference in effect, with a SMD of ‐0.04 (95% CI ‐0.50 to 0.41).
One RCT with high risk of bias (Bendix 1997) (52 workers) reported very low quality evidence that intense physical conditioning was more effective and clinically worthwhile than exercise therapy for workers with chronic back pain in reducing the time to return‐to‐work at very long‐term follow‐up (two years), with a SMD of ‐0.62 (95% CI ‐1.21 to ‐0.04).
Three studies reported on intense physical conditioning versus intense physical conditioning with CBT for workers with chronic back pain (Altmaier 1992; Jensen 2001; van den Hout 2003).
The pooled results of two RCTs with high and low risk of bias (Altmaier 1992; van den Hout 2003) (126 workers) showed low quality evidence that there was no difference in the reduction of the proportion of workers off work between intense physical conditioning and intense physical conditioning with CBT at the six‐month follow‐up, with a SMD of 0.26 (95% CI ‐0.50 to 1.03) ( Figure 6 ).
Forest plot of comparison: Intense PCP versus intense PCP with CBT for workers with chronic back pain, outcome: Time to return‐to‐work at long‐term follow‐up.
Altmaier 1992 also reported on the effectiveness of intense physical conditioning that included CBT for workers returning to full‐time work and for those who had made constructive efforts toward re‐employment. They concluded that this was not more effective in reducing the proportion of workers who were off work when compared with intense physical conditioning without CBT.
The pooled results of two RCTs with low risk of bias (Jensen 2001; van den Hout 2003) (193 workers) showed moderate quality evidence that there was no difference in the reduction of sickness absence between intense physical conditioning without CBT and intense physical conditioning with CBT at one‐year follow‐up, with a pooled SMD of 0.05 (95% CI ‐0.30 to 0.40) ( Figure 6 ).
One RCT with low risk of bias (Jensen 2001) (117 workers) reported low quality evidence that there was no difference in the reduction of time off work between intense physical conditioning without CBT and an intense physical conditioning programme with CBT at the 36‐month follow‐up, with a SMD of 0.19 (95% CI ‐0.18 to 0.56).
Two studies measured the effect of intense physical conditioning versus CBT for workers with chronic back pain (Bendix 1997; Jensen 2001). CBT in these studies was delivered by psychologists. When studies were pooled, we found high statistical heterogeneity that may have been caused by differences in components of the physical conditioning. Further analysis showed that Jensen 2001 included a workplace visit and explicit return‐to‐work advice in their programme. However, these two components were also available for the control situation and therefore we have not been able to explain the high statistical heterogeneity.
One RCT with high risk of bias (Bendix 1997) (57 workers) reported very low quality evidence of a clinically worthwhile effect in reducing the number of workers off work in favour of intense physical conditioning compared with CBT after one and two years follow‐up, with an OR of 0.04 (95% CI 0.01 to 0.17) and an OR of 0.14 (95% CI 0.05 to 0.43) respectively.
Another RCT with low risk of bias (Jensen 2001) (103 workers) reported low quality evidence of a clinically worthwhile effect in reducing time off work in favour of intense physical conditioning compared with CBT after one and a half years follow‐up, with a SMD of ‐0.46 (95% CI ‐0.85 to ‐0.07), but this effect was gone after three years with a SMD of ‐0.13 (95% CI ‐0.51 to 0.26).
One RCT with low risk of bias (Skouen 2002) (109 workers) reported low quality evidence that there was no difference between intense physical conditioning and light physical conditioning in reducing the time to return‐to‐work for workers with chronic back pain at the one‐ and two‐year follow‐up, with a SMD of 0.10 (95% CI ‐0.28 to 0.48) and SMD of 0.20 (95% CI ‐0.18 to 0.58) respectively.
We included 25 studies in this updated review that evaluated the effectiveness of physical conditioning as part of a return to work strategy for workers with back pain. Based on these studies, we analysed 17 comparisons and were able to pool data in a few cases. There was high statistical heterogeneity and conflicting results about the effect of intense physical conditioning versus care as usual for workers with subacute back pain. Further exploration of the components of the physical conditioning programmes using subgroup analysis pointed towards a positive effect of a workplace visit included in the programme. However, these results were based on indirect evidence only and we should therefore be cautious to reach any conclusions. Moreover, high statistical heterogeneity and conflicting results were found for the effect of intense physical conditioning versus exercise treatment, which could not be explained by differences in components or style of delivery of the programmes. Our review showed the following results.
(1) There is low quality evidence that for workers with acute back pain both light and intense physical conditioning may have little or no impact on sickness absence duration compared with care as usual.
(2) There is low quality evidence that for workers with subacute back pain light physical conditioning may have little or no impact on sickness absence duration compared with care as usual. In addition, a multidisciplinary intervention supplementing a brief clinical intervention does not have added value in terms of return to work compared to a brief clinical intervention only.
(3) The evidence on the effectiveness of intense physical conditioning with care as usual versus care as usual only in workers with subacute back pain is conflicting. Further subgroup analysis shows that if the intervention is executed at the workplace, or includes a workplace visit, it may reduce sickness absence duration at intermediate and long‐term follow‐up based on low quality evidence.There is moderate quality evidence for this effect at the very long‐term follow‐up.
(4) Based on moderate quality evidence intense physical conditioning probably reduces sickness absence duration for workers with subacute back pain at long‐term follow‐up compared with a multidisciplinary exercise treatment. However, based on low quality evidence there may be no or little difference in effect at intermediate follow‐up, but physical conditioning may again reduce sickness absence duration at short‐term follow‐up.
(5) Based on low quality evidence, intense physical conditioning may have little or no effect in reducing sickness absence duration compared with a cognitive intervention for workers with subacute back pain at short‐term follow‐up.
(6) Based on moderate quality evidence, intense physical conditioning probably slightly reduces duration of sickness absence compared with care as usual for workers with chronic back pain, but only at long‐term follow‐up. Based on low quality evidence intense physical conditioning together with care as usual may reduce sickness absence duration compared with care as usual, but only at long‐term follow‐up.
(7) The evidence on the effectiveness of intense physical conditioning programmes versus exercise therapy for workers with chronic back pain at long‐term follow‐up is conflicting and missing at other follow‐up times.
(8) There is low quality evidence that intense physical conditioning without a workplace component may have little or no effect on sickness absence duration compared with light physical conditioning for workers with subacute and chronic back pain.
(9) There is very low to moderate quality evidence that CBT is probably not superior to physical conditioning as an alternative or in addition to physical conditioning.
The objective of this updated review was to compare the effectiveness of work‐related physical conditioning in reducing time lost from work or improving work status for workers with back pain. Return to work and improving work status are the most important and relevant outcome measures in occupational health care. Other outcome measures such as functional status, psychological well‐being and satisfaction with the treatment are also important, however we chose to focus only on sickness absence and work status in an attempt to simplify this complex study topic.
Due to the variety of interventions, comparisons and follow‐up times, there were no studies available for some comparisons and only single studies for others. In this update we divided comparisons between physical conditioning in addition to care as usual and physical conditioning only. This resulted in 17 comparisons with physical conditioning. The comparisons with light physical conditioning programmes only included single studies. Because most included studies had small sample sizes, the results of the light physical conditioning programmes were not significantly effective, but had 95% confidence intervals that included the possibility of a clinically worthwhile effect. Therefore the effects of light physical conditioning programmes are still uncertain.
Another issue is the high statistical heterogeneity in the comparison of intense physical conditioning versus usual care. We ascribed this to differences in components of the physical conditioning programmes. We reported further subgroup analyses in an attempt to explain differences in the effect of physical conditioning programmes. Workplace involvement in the intervention may explain the positive results reported for intense physical conditioning versus usual care for workers with subacute back pain. However, no new studies were found to confirm added value for physical conditioning versus exercise therapy for chronic back pain. For these types of workers the evidence remains conflicting at long‐term follow‐up.
This updated review has good external validity because the participants in the trials resemble workers in occupational health practice and the interventions are considered feasible to carry out in daily medical practice. However, there may be differences in social benefit systems between the countries in which the studies were carried out, which may have an additional effect on the return‐to‐work outcome.
Sixteen of the 25 studies that were included were assessed as having a low risk of bias. Eight studies had serious flaws due to unclear or no allocation concealment. Sparse data due to the small number of participants in most studies lowered the quality of the evidence further. The three newly included studies were all of high quality (Bethge 2011; Jensen 2011; Lambeek 2010). Due to the new division of comparisons in this updated version of the review there were 17 comparisons for the included studies. This complicated further pooling of studies and therefore the conclusion of the previous updated review (Schaafsma 2010) is still accurate.
The effectiveness of interventions with involvement of the workplace is supported by current literature (Anema 2007; van Oostrom 2009). It is argued that workplace‐based interventions can change both workers' and supervisors' perceptions about the workers' return‐to‐work capabilities and opportunities (Anema 2007). In our review, a subgroup analysis of interventions that were done at the workplace or included a workplace visit seems to point towards a positive effect when a workplace intervention is included in an intense physical conditioning programme. The positive effect of workplace involvement is also substantiated with the newly included high quality study of Lambeek 2010 for chronic back pain. In this study the physical conditioning was part of integrated care management with particular workplace involvement. What part of this integrated care management was actually (most) effective is unclear.
In the meta‐regression analysis based on the studies of the previous review (Schaafsma 2011), involvement of the workplace provided no explanation for any difference in effect of physical conditioning as part of a return to work strategy. This meta‐regression analysis will therefore need to be updated with these three new studies. In this updated review we found that for subacute back pain intense physical conditioning is more effective than exercise therapy at long‐term follow‐up. The difference between physical conditioning and exercise lies in the explicit objective of enabling the individual to return to work in the conditioning programmes (Schaafsma 2011). However, for workers with both subacute and chronic back pain the results of included studies did not clearly and consistently support physical conditioning over exercise therapy at different follow‐up times. The additional effect of the focus on return to work within physical conditioning compared to exercise therefore remains unclear. Furthermore, the study of Jensen 2011 also reported that no additional effect could be measured for a multidisciplinary intervention on top of a clinical intervention in terms of return to work. Therefore, also for subacute back pain the addtional effect of a return to work focus remains unclear.
For workers with chronic back pain, the combined results did show there is a small positive effect of intense physical conditioning in the long term compared with usual care. However, this combined result may not be clinically worthwhile according to our definition. The meta‐analysis consisted of five studies (Bendix 1996; Corey 1996; Mitchell 1994; Jensen 2001; Skouen 2002). The study by Lambeek 2010 differed from these five as they compared physical conditioning in addition to care as usual. However, it can be argued that the participants in this study did not receive any additional care as usual. Besides, it is not always clear what care as usual is. It might still be that alternative interventions use fewer resources and thus are less costly but equally effective. For example, there was one small study that showed no difference in return to work between physical conditioning programmes and a cognitive intervention that consisted of only two consultations. These consultations focused on explaining pain mechanisms, advice on how to stay active and take responsibility, and were delivered by a physical medicine specialist. More research is needed to clarify if such a 'light' intervention is actually enough to help workers with subacute back pain return to work.
The original review also highlighted the need for effective physical conditioning programmes to have a cognitive behavioural approach. In the previously updated review, we defined the cognitive behavioural approach (CBA) as an intervention component that encouraged workers with back pain to focus on functional gains rather than pain, as defined by Fordyce 1976. A CBA was included in 16 out of the 25 studies, mostly as a graded activity approach. A subgroup analysis of studies with a CBA included in their intervention could not be done because CBA was either included in all studies or in none of the meta‐analyses reported.
We found that the effectiveness of physical conditioning is not consistent at different times of follow‐up. Intuitively one would expect that the effect of physical conditioning is largest at the end of the intervention programme, with the effect then gradually wearing off. However, for intense physical conditioning with workplace intervention versus care as usual, there is a non‐significant outcome at short‐term follow‐up but significant outcomes at longer‐term follow‐up. This phenomenon, that the intervention is only effective at long‐term follow‐up, was also seen in some single studies. It might be that this is not due to the mechanism of the intervention but is the result of publication bias by which only positive long‐term results are published and not the negative ones.
We did not find that other aspects of the intervention. such as involvement of an occupational health physician, specific return‐to‐work advice, inclusion of ergonomic advice or specific occupational training, influenced time lost from work.
Other possible explanations for the high statistical heterogeneity of included studies may be the differences in healthcare or social security systems between countries. Also, different type of workers, for example blue collar versus white collar workers between studies or the type of industry branch, may have an impact on sickness absence. However, we did not have the opportunity to extensively study this.
In this updated review, we pooled data whenever possible. For this purpose we had to recalculate data from the original studies for example odds ratios and mean differences into standard mean differences (SMDs). These recalculations sometimes resulted in different conclusions than those reported by the authors. Differences could be attributed to recalculation of workability into sickness absence and recalculation to standard deviations from studies when only P values or 95% CIs were provided in the original literature. The advantage of using SMD was that it increased the number of possible meta‐analyses.
Results of this updated Cochrane review are supported by previously published systematic reviews, even though their scope and methods are different. However, there are no existing reviews that focus solely on return to work with specific exercise interventions that have an explicit relation with work. In an older review by Ostelo 2000, it was found that CBT has no added value compared with waiting list controls on short‐term pain relief for chronic low‐back pain and no difference in effect was found between CBT and exercise. This finding supports the current review as the results of intense physical conditioning in terms of return to work are better than CBT in the long term but show no difference in effect in the very long term.
Our results are also in line with another more recent systematic review on specific functional restoration programmes that concluded that return‐to‐work improved when programmes incorporated the provision of suitable modified duties (Poiraudeau 2007). In this review, we found that physical conditioning programmes that particularly included workplace involvement reduced sickness absence for workers with subacute or chronic back pain compared to care as usual.
A Cochrane review on the effectiveness of back schools reported that there is moderate evidence suggesting that back schools for chronic low‐back pain in an occupational setting are more effective than other treatments, placebo, or waiting lists on pain, functional status and return to work during short‐ and intermediate‐term follow‐up (Heymans 2005). In the current updated review, back school interventions were only considered as physical conditioning programmes if there was a relationship with the work situation. We confirm that for workers with chronic back pain, intense physical conditioning in addition to care as usual is significantly more effective in reducing sick leave time compared with care as usual only. However, this result is based on one study only. In addition, physical conditioning only compared to care as usual may only have a small effect, which may not be clinically worthwhile in the long term, based on five studies.
A Cochrane review on the effectiveness of exercise therapy showed positive results for improving pain and function at short‐term follow‐up. Further, this review found evidence for effectiveness of graded activity exercise programmes in subacute low‐back pain in occupational settings, although the evidence for other types of exercise therapy in other populations is inconsistent (Hayden 2005). Since these conclusions are based on the same studies, these findings are not surprising.
For workers with acute and subacute back pain, there was only low quality evidence regarding the effects of physical conditioning on sickness absence duration. Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Implications for practice are therefore uncertain.
For workers with chronic back pain, there is moderate quality evidence that intense physical conditioning probably reduces sickness absence duration slightly compared to care as usual. The effect may not be clinically relevant, although further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Implications for practice are therefore uncertain.
More than half of the included studies showed a low risk of bias, especially for the most recent studies, which implies that high quality studies are feasible in this area. More studies are needed that compare the effect of physical conditioning as part of coordinated and tailored work rehabilitation programmes in addition to usual care versus usual care only for workers with back pain. More research is also needed to understand the mechanism behind physical conditioning and return to work at different times of follow‐up and for various durations of back pain. Qualitative research would be an appropriate tool to reveal ideas and attitudes of workers that could be important factors in return to work. We recommend that work outcomes of new research are reported as the mean number of full‐time and part‐time sick days and analysed as censored events with survival analysis.
: Jan 5, 2005. Would be interesting to see the comparison between CBT and CBT‐exercise programme to see if there was additive benefit of exercise.
: Jan 5, 2005. A response was sent directly to Dr Hardy, in which he was referred to the updated review on 'Behavioural treatment for chronic low‐back pain', published in The Cochrane Library 2005 (1), for this comparison.
| Date | Event | Description |
|---|---|---|
| 5 June 2013 | New citation required but conclusions have not changed | Conclusions have not changed. |
| 5 June 2013 | New search has been performed | 3 new trials were incorporated into this update. |
Protocol first published: Issue 4, 1999
Review first published: Issue 1, 2003
| Date | Event | Description |
|---|---|---|
| 19 January 2011 | Amended | Contact details updated. |
| 17 February 2010 | Amended | typos corrected in abstract and plain language summary |
| 9 June 2009 | New citation required and conclusions have changed | New GRADE approach to assessment of the quality of the evidence led to change in conclusions. |
| 30 June 2008 | New search has been performed | inclusion/exclusion criteria, methodology and outcome measures revised. Based on changes, literature search updated. New criteria led to exclusion of 12 studies from the previous review. Revised and updated search led to inclusion of 15 new studies. |
| 22 May 2008 | Amended | Converted to new review format. |
| 23 May 2003 | Amended | Issue 3, 2003 The authors of this review have made one modification since Issue 2, 2003. They removed the sensitivity analysis and subsequent tables (01/02) as they discovered an error with the data entered. Accordingly, they have also modified relevant text and removed one reference (Altman 2001). |
| 30 April 2003 | New citation required and conclusions have changed | Substantive amendment |
Rachel Couban from the Cochrane Back Review Group who updated and refined the literature searches.
Search strategy for MEDLINE (Ovid):
1 randomized controlled trial.pt.
2 controlled clinical trial.pt.
5 drug therapy.fs.
10 (animals not (humans and animals)).sh.
13 exp Back Pain/
15 exp Low Back Pain/
16 (lumbar adj pain).ti,ab.
20 sciatic neuropathy/
24 work conditioning.mp.
25 work hardening.mp.
26 functional restoration.mp.
27 exercise$.mp. or exp Exercise/
28 gym$ prog$.mp. [mp=title, abstract, original title, name of substance word, subject heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier]
29 disability evaluation.mp. or exp Disability Evaluation/
31 exp Work Capacity Evaluation/
34 11 and 23 and 33
35 limit 34 to yr="2008 ‐ 2012"
36 limit 34 to ed=20080601‐20120320
There is a low risk of selection bias if the investigators describe a random component in the sequence generation process such as: referring to a random number table, using a computer random number generator, coin tossing, shuffling cards or envelopes, throwing dice, drawing of lots, minimisation (minimisation may be implemented without a random element, and this is considered to be equivalent to being random).
There is a high risk of selection bias if the investigators describe a non‐random component in the sequence generation process, such as: sequence generated by odd or even date of birth, date (or day) of admission, hospital or clinic record number; or allocation by judgement of the clinician, preference of the participant, results of a laboratory test or a series of tests, or availability of the intervention.
There is a low risk of selection bias if the participants and investigators enrolling participants could not foresee assignment because one of the following, or an equivalent method, was used to conceal allocation: central allocation (including telephone, web‐based and pharmacy‐controlled randomisation); sequentially numbered drug containers of identical appearance; or sequentially numbered, opaque, sealed envelopes.
There is a high risk of bias if participants or investigators enrolling participants could possibly foresee assignments and thus introduce selection bias, such as allocation based on: using an open random allocation schedule (e.g. a list of random numbers); assignment envelopes were used without appropriate safeguards (e.g. if envelopes were unsealed or non‐opaque or not sequentially numbered); alternation or rotation; date of birth; case record number; or other explicitly unconcealed procedures.
There is a low risk of performance bias if blinding of participants was ensured and it was unlikely that the blinding could have been broken; or if there was no blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding.
There is a low risk of performance bias if blinding of personnel was ensured and it was unlikely that the blinding could have been broken; or if there was no blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding.
There is low risk of detection bias if the blinding of the outcome assessment was ensured and it was unlikely that the blinding could have been broken; or if there was no blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding, or:
for patient‐reported outcomes in which the patient was the outcome assessor (e.g. pain, disability): there is a low risk of bias for outcome assessors if there is a low risk of bias for participant blinding (Boutron 2005);
for outcome criteria that are clinical or therapeutic events that will be determined by the interaction between patients and care providers (e.g. co‐interventions, length of hospitalisation, treatment failure), in which the care provider is the outcome assessor: there is a low risk of bias for outcome assessors if there is a low risk of bias for care providers (Boutron 2005);
for outcome criteria that are assessed from data from medical forms: there is a low risk of bias if the treatment or adverse effects of the treatment could not be noticed in the extracted data (Boutron 2005).
There is a low risk of attrition bias if there were no missing outcome data; reasons for missing outcome data were unlikely to be related to the true outcome (for survival data, censoring unlikely to be introducing bias); missing outcome data were balanced in numbers, with similar reasons for missing data across groups; for dichotomous outcome data, the proportion of missing outcomes compared with the observed event risk was not enough to have a clinically relevant impact on the intervention effect estimate; for continuous outcome data, the plausible effect size (difference in means or standardised difference in means) among missing outcomes was not enough to have a clinically relevant impact on observed effect size, or missing data were imputed using appropriate methods (if drop‐outs are very large, imputation using even 'acceptable' methods may still suggest a high risk of bias) (van Tulder 2003). The percentage of withdrawals and dropouts should not exceed 20% for short‐term follow‐up and 30% for long‐term follow‐up and should not lead to substantial bias (these percentages are commonly used but arbitrary, not supported by literature) (van Tulder 2003).
There is low risk of reporting bias if the study protocol is available and all of the study's pre‐specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre‐specified way, or if the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre‐specified (convincing text of this nature may be uncommon).
There is a high risk of reporting bias if not all of the study's pre‐specified primary outcomes have been reported; one or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not pre‐specified; one or more reported primary outcomes were not pre‐specified (unless clear justification for their reporting is provided, such as an unexpected adverse effect); one or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta‐analysis; the study report fails to include results for a key outcome that would be expected to have been reported for such a study.
Bias due to dissimilarity at baseline for the most important prognostic indicators.
There is low risk of bias if groups are similar at baseline for demographic factors, value of main outcome measure(s), and important prognostic factors (examples in the field of back and neck pain are duration and severity of complaints, vocational status, percentage of patients with neurological symptoms) (van Tulder 2003).
There is low risk of bias if there were no co‐interventions or they were similar between the index and control groups (van Tulder 2003).
There is low risk of bias if compliance with the interventions was acceptable, based on the reported intensity or dosage, duration, number and frequency for both the index and control intervention(s). For single‐session interventions (e.g. surgery), this item is irrelevant (van Tulder 2003).
There is low risk of bias if all randomised patients were reported and analysed in the group to which they were allocated by randomisation.
There is low risk of bias if all important outcome assessments for all intervention groups were measured at the same time (van Tulder 2003).
There is a low risk of bias if the study appears to be free of other sources of bias not addressed elsewhere (e.g. study funding).
The GRADE criteria were operationalised in the following way.
Limitations of the study refers to the risk of bias assessment of studies. Studies with more than 5 points on the risk of bias assessment were regarded as studies with a low risk of bias. If 75% or more of the studies scored above 5, this item was scored as: no limitations. If between 50% and 75% of the studies scored above 5, this was scored as: serious limitations. If less that 50% of the studies scored above 5: very serious limitations.
Consistency refers to the similarity of estimates of treatment effects for the outcome across studies. Study results were considered consistent if direction, effect size and statistical significance were sufficiently similar to lead to the same conclusions. Consistency in direction was defined as 75% or more of the studies showing either benefit or no effect of the workplace intervention. In the case of a benefit, consistency in effect size was defined as 75% or more of the studies showing a clinically important or unimportant effect. Minimal clinically relevant differences were derived from the original review as mentioned above in the quantitative analysis. Consistency in statistical significance was defined by the Chi 2 test for heterogeneity.
Directness (generalisability) refers to the extent to which the workers, interventions, and outcomes in the studies were comparable to those defined in the inclusion criteria of the review. If there was uncertainty about generalisablity of the results, or if the results were more applicable to a specific population than a general population on work disability, serious or very serious limitations were assigned.
Precision of the evidence refers to the confidence in the results. It takes into account the number of studies, patients, and events; and width of the CIs for each outcome. Data were interpreted to be imprecise as multiple studies were combined in a meta‐analysis but the CI (confidence interval) was sufficiently wide that the estimate could either support or refute the effectiveness of the workplace intervention. In the case of imprecise data serious limitations were assigned. Serious limitations could also be assigned if data were judged to be sparse, that is if only one study was available for an outcome, or fewer than 75% of the studies presented data that could be included in the meta‐analysis.
Publication bias refers to the probability of selective publication of studies and outcomes.The overall quality of the evidence for each outcome was the result of the combination of the assessments in all domains.
New search for studies and content updated (no change to conclusions)
