Home > DARE Reviews > Effectiveness of exercise on work...

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Database of Abstracts of Reviews of Effects (DARE): Quality-assessed Reviews [Internet]. York (UK): Centre for Reviews and Dissemination (UK); 1995-.

Database of Abstracts of Reviews of Effects (DARE): Quality-assessed Reviews [Internet].

Effectiveness of exercise on work disability in patients with non-acute non-specific low back pain: systematic review and meta-analysis of randomized controlled trials

P Oesch, J Kool, K Birger Hagen, and S Bachmann.

Review published: 2010.

CRD summary

This generally well-conducted review concluded that exercise interventions had a significant effect in the long-term on work disability in patients with non-acute non-specific low back pain, but that conclusions could not be made regarding different exercise types. The authors' conclusions reflected the evidence presented, but interpretation should take into consideration the limitations with the included trials.

Authors' objectives

To assess the effectiveness of exercise in reducing work disability in patients with non-acute non-specific low back pain.


A previous search undertaken in 2002 was updated to August 2008: MEDLINE (from 1966), EMBASE (from 1988), PEDro, Cochrane Library, NIOSHTIC-2 (US workplace health and safety resources database), and PsycINFO (previously PsycLIT from 1984). Search terms were not reported. Reference lists were manually searched.

Study selection

Randomised controlled trials (RCTs) that assessed the effects of exercise alone, or as part of a multidisciplinary treatment, in patients with non-acute non-specific low back pain (with a duration of at least four weeks), were eligible for inclusion. Pregnant women were excluded.

The primary outcome of interest was work disability, with at least 90% of participants available for the job market (i.e. were either employed, or unemployed and seeking work).

Included trials were of patients aged between 18 and 65 years (where reported); some were on sick leave, or their job situation was threatened, or they were out of work due to non-acute non-specific low back pain. Where reported, the duration of non-specific low back pain, or sick leave due to non-specific low back pain, ranged from three weeks to two years.

Trials compared two different exercise interventions, or compared exercise with usual care. Interventions were individually designed, or of standard design performed at home and/or under supervision. The interventions included the following types of exercise: strengthening, stretching, mobilisation or flexibility, aerobic, stabilisation, or a mixture of the previous. Interventions specifically designed to restore work-related physical capacity were classified as work-related, while some other interventions were classed as using a behavioural treatment approach (as defined in the review). Where reported, intervention durations ranged from three to 12 weeks. Intervention intensity also varied considerably between trials, with some interventions providing two one hour sessions per week and others providing 39 hours of treatment per week.

Work disability outcomes of interest were the number of participants on sick leave at follow-up and the number of participants returning to work or not returning to work at follow-up.

Two authors independently selected studies for inclusion, with any disagreements referred to a third reviewer.

Assessment of study quality

Two reviewers independently assessed trial validity using previously published criteria, including items on allocation concealment (selection bias), blinding of outcome assessor (detection bias), and intention-to-treat analysis (attrition bias). Trials meeting two or three criteria were categorised as high quality, while trials meeting one or none of the criteria were categorised as low quality. Disagreements were referred to a third reviewer.

Data extraction

Two reviewers independently extracted or calculated means and standard deviations on work-related outcomes which were converted into odds ratios (ORs) and 95% confidence intervals (CIs). Where trials compared two different exercise interventions with usual care, these were classed as two separate trials with the control group equally divided between the two intervention groups.

Methods of synthesis

Results were presented for all trials as a narrative synthesis, and where pooling was appropriate, a random-effects model was used to combine odds ratios and 95% confidence intervals by length of follow-up: short-term (closest to four weeks); intermediate-term (closest to six months); and long-term (closest to 12 months). Sensitivity analyses were conducted to assess the effect of trial quality on the findings.

Statistical heterogeneity was assessed using the I2 statistic. Meta-regression was conducted to identifty the effect of the following pre-specified trial details on the results: exercise design (individual care versus standard care), dose (high-dose versus low-dose exercise), delivery type (home-based exercises versus supervised exercises), administration within a cognitive behavioural approach (yes or no), and work context (yes or no).

Publication bias was assessed using funnel plots.

Results of the review

Twenty three RCTs (n=4,138 participants) were included in the review, with 20 included in the meta-analyses. Fourteen trials were of high quality (with eight reporting no selection, detection, or attrition bias) and 12 were of low quality (with five reporting all three types of bias). Follow-up ranged between four weeks and three years.

Qualitative synthesis showed contradictory findings in the short-term (seven RCTs), intermediate-term (five RCTs), and long-term (11 RCTs), as reported in the review.

Meta-analysis (13 RCTs, 17 comparisons) showed no statistically significant effects of exercise on work disability in the short-term or intermediate-term compared with usual care. In the long term, eight high quality trials/10 comparisons (n=1,992 participants) showed a statistically significant reduction in work disability in patients receiving exercise compared with usual care (OR 0.66, 95% CI 0.48 to 0.92; I2=72.8%); this remained statistically significant with the addition of two low quality trials/three comparisons (overall I2=67.4%). Meta-regression showed that the pre-specified trial characteristics did not statistically significantly affect work disability. There was evidence of publication bias in the short- and intermediate-term trials.

Comparison of different exercise interventions (13 RCTs comparing 15 exercise interventions): Meta-analysis showed that trials using higher dose exercise did not statistically significantly affect work disability (six high quality RCTs; I2=65.5%). Similarly, comparison of trials applying exercise with or without a behavioural treatment approach (three RCTs), and one RCT comparing a specifically designed work-related exercise versus a non specifically designed work-related exercise, showed no statistically significant effect on work disability. There was no evidence of publication bias.

Authors' conclusions

Exercise interventions had a significant effect in the long-term on work disability in patients with non-acute non-specific low back pain. Conclusions could not be made regarding different exercise types.

CRD commentary

The review question and supporting inclusion criteria were clearly stated. Relevant sources were searched for articles, although search terms were not reported and it was unclear whether a search for unpublished data was undertaken, which meant that potentially relevant data may have been missed. Publication bias was assessed and was found to be evident for some comparisons. The authors undertook each stage of the review in duplicate, reducing the potential for reviewer error and bias.

The authors assessed trials for quality using appropriate criteria; just over half were categorised as high quality. There was evidence of statistical heterogeneity for some comparisons, and there were differences in intervention methodologies. Also, it was difficult to determine whether there may have been clinical heterogeneity, as few patient characteristics were reported. However, the authors did go some way to investigate potential effects of heterogeneity using meta-regression, and they suggested that patients, social support and outcomes showed satisfactory homogeneity. The authors acknowledged the wide confidence intervals for short-term and intermediate-term trials, which may have affected the robustness of the findings.

This was a generally well-conducted piece of research and the authors' conclusions seem to reflect the evidence presented, but interpretation of their conclusions should take into account the above limitations with the included trials.

Implications of the review for practice and research

Practice: The authors did not state any implications for practice.

Research: The authors stated that further research is needed to assess the effects of individually designed home exercise interventions applied within a behavioural treatment approach and specifically aimed at restoring work-related physical capacity. Research should also focus on identifying an effective implementation process of exercise interventions aimed at early return to work.


Swiss National Research Foundation, Swiss National Research Program NRP 53, project number405340-111500/2.

Bibliographic details

Oesch P, Kool J, Birger Hagen K, Bachmann S. Effectiveness of exercise on work disability in patients with non-acute non-specific low back pain: systematic review and meta-analysis of randomized controlled trials. Journal of Rehabilitation Medicine 2010; 42(3): 193-205. [PubMed: 20411212]

Indexing Status

Subject indexing assigned by NLM


Disability Evaluation; Exercise Therapy /methods; Humans; Low Back Pain /diagnosis /psychology /rehabilitation; Outcome Assessment (Health Care); Randomized Controlled Trials as Topic; Rehabilitation, Vocational; Treatment Outcome; Work Capacity Evaluation



Database entry date


Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.

CRD has determined that this article meets the DARE scientific quality criteria for a systematic review.

Copyright © 2014 University of York.

PMID: 20411212

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...