This review concluded that stakeholder participation and workplace modification were effective and cost-effective at helping employees with musculoskeletal conditions on sick leave return to work. The authors' conclusions reflect the evidence presented and appear to be broadly reliable.

To determine whether interventions involving the workplace are more effective and cost-effective than those which do not have workplace involvement at helping employees on sick leave return to work.

MEDLINE, CINAHL, PsycINFO, AMED, Applied Social Sciences Index and Abstracts (ASSIS), British Nursing Index, Business Source Premier, the Cochrane Library, Current Contents, International Bibliography of the Social Sciences, Sociological Abstracts and the Science, and Social Citation Index were searched from 1990 for English language studies. Health Management Information Consortium, National Research Register and SIGLE were searched for grey literature. NHS EED, OHE HEED, and EconLit were searched from 1990 for relevant studies on economic evaluation. Search terms were reported. Reference lists of retrieved papers and reviews were reviewed. Experts were contacted for additional studies.

To be eligible for inclusion in the review, a study had to include employees on sick leave of at least two working weeks at the time of the intervention. The intervention had to involve the workplace. The control intervention should not include any involvement of the workplace. Eligible studies had to be controlled longitudinal studies and/or cost-benefit/cost-effectiveness analyses of such studies. The workplace element could form the entire or a component of the intervention. Workplace interventions had to involve interventions taking place (in full or in part) at the workplace of the employee, or involve delivery of the intervention by direct contact with the employer or a representative (employee's supervisor or employer's occupational health services). Eligible studies were required to report data on return to work.

Studies were excluded if the workplace element of the intervention consisted of only education or advice concerning ergonomics or the workplace without a work site visit or contact with the workplace or employer. Also excluded were studies of self-employed people, studies with mixed populations on sick leave and in-work, and studies with mixed populations of employed and unemployed participants that did not report outcomes separately for the participants in formal employment,

Most of the included studies recruited participants who were on sick leave because of back pain. The mean age, length of sickness absence, and type of intervention varied across studies. Included studies were conducted in Europe and Canada.

Ninety percent of the potentially relevant citations and abstracts were screened by a single individual. The retrieved full articles were assessed by one reviewer and cross-checked by a second; queries or disagreements were resolved with a third reviewer.

Quality of the of randomised controlled trials (RCTs) was assessed on randomisation, recruitment, blinding and types of analyses using CRD quality assessment criteria. The non-randomised studies were assessed for quality using the Downs and Black checklist. Economic evaluations were assessed with the checklists developed by Drummond and by Eddy.

Two reviewers independently assessed quality of the included studies.

Summary statistics in the form of rate and time of return to work for treatment and control groups were extracted.

One reviewer extracted the data which were double-checked by a second reviewer.

The studies were pooled in a narrative synthesis, grouped by the principal component of the intervention.

Nine studies, eight RCTs and one non-RTC (n=3,134 participants) reported in ten articles met the inclusion criteria. Most of the studies were good or moderate quality. Follow-up was one year or more in most studies; 13 weeks was the shortest follow-up.

Intervention involving employees, health practitioners and employers working collaboratively in implementing work modifications showed significant improvement in rates of return to work compared with usual care, or other interventions that did not involve the workplace (six studies). Of the three remaining studies, one study found some improvement in the rate of return to work, but this was not statistically significant; two studies did not find any difference between workplace interventions and control.

Three of four studies that evaluated time to return to work found that the intervention group required significantly shorter time to return to work than the control group or the functional restoration therapy group that received exercise interventions.

Two moderate quality economic evaluations of RCTs were included. Both indicated that interventions with a workplace component appeared to be more cost-effective than those without workplace interventions. One study estimated that the cost per return to work-day gained was £17 and that these interventions dominated usual care. In the second study, cost-benefit and cost-effectiveness analyses found that the cost per day on all full benefit saved for the insurance provider by the intervention was £46 and the ratio of benefits to cost for the intervention was found to be 7.7.

Stakeholder participation and work modification were more effective and cost-effective at helping employees with musculoskeletal conditions on sick leave return to work compared with other non-collaborative workplace interventions, including exercise or usual care.

This review addressed a well-defined question in terms of participants, interventions, outcomes, and study design. The authors searched numerous relevant databases, but the restriction of the review to studies published in English meant that the possibility of language and publication biases could not be ruled out. To minimise bias and errors during the review process, two or more reviewers were involved in selecting studies, extracting data, and assessing quality of the included studies.

Relevant quality assessment criteria were applied; the results indicated that most of the included studies were considered as moderate or good quality. The characteristics of the individual trials were presented. The choice of narrative synthesis appeared to be appropriate given that there was significant heterogeneity between the included studies.

The authors' conclusions reflect the evidence presented and appear to be broadly reliable.

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

Research: The authors stated that larger controlled trials of workplace interventions that consider other populations who experience long-term sickness absence, especially those with stress and mental health issues are warranted.

This review drew on data for a series of reviews funded by the UK National Institute for Health Clinical excellence (NICE).

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.