The authors concluded that interventions to promote walking in groups increased physical activity in adults. The review methods had some limitations, but the authors' conclusions reflect the evidence presented and appear to be reliable.

To evaluate the efficacy of promoting walking in groups, for increasing physical activity in adults.

Seven databases, including MEDLINE, PsycINFO, and CINAHL, were searched, for English-language publications from January, 1980 to March, 2012. Search terms were reported. Forward and backward citation searches were made from included articles, and the reference lists of relevant reviews were handsearched. The authors of relevant studies were contacted to identify any unpublished data.

Eligible were randomised controlled trials, non-randomised controlled trials, and pre-post studies that investigated the efficacy of walking-group interventions in promoting physical activity in adults (aged over 18 years). Studies had to use objective or self-reported physical activity measures, and focus on the frequency, duration or distance of physical activity. Interventions had to be those where people walked collectively in groups; where they walked with or without a leader; or where people could choose a partner to walk with. Many exclusion criteria were reported in the paper.

Most of the included studies were performed in the USA; the others were performed in the UK (one study), Canada, Australia, Japan, or China. The mean age of participants ranged from 44 to 88 years; a few were classed as being at risk, or had a diagnosis of chronic disease, such as cardiovascular disease or mental decline. Just over half of the interventions had just one component (walking in groups); the others included other physical activity, or a supportive programme of patient education on chronic diseases or other lifestyle behaviours. Most interventions were community based, and were delivered by researchers, lay people, professionals, or trained graduate students, in training sessions, discussion groups, or one-to-one visits, or using educational materials. The studies varied in the intensity and duration of their interventions, and their measures of physical activity.

One reviewer selected studies for inclusion in the review.

Study quality was assessed using criteria from the Cochrane Consumers and Communication Review Group, covering randomisation, allocation, blinding, baseline comparability, follow-up, valid measures, ethical approval, and informed consent. Studies with a score of 4 or more (meeting at least half of the criteria) were classified as high quality, and studies with scores of between 0 and 3 were classified as being low quality.

The authors did not state how many reviewers assessed quality.

Data for the outcome of physical activity duration, frequency, or distance, were extracted to calculate mean differences, with 95% confidence intervals.

Two reviewers independently extracted the data; any discrepancies were resolved by discussion.

Mean differences and 95% confidence intervals, from individual studies, were pooled using a random-effects model. The results were reported as standardised mean differences. Heterogeneity between studies was assessed using the Q test.

To assess the risk of publication bias, the fail-safe N was calculated, specifying the number of studies with a zero effect that were needed to change a statistically significant result to a non-significant one. A sensitivity analysis was performed by study quality, and moderator analyses were performed for study characteristics (reported in paper). Pairwise z tests were performed to find statistically significant differences in effect sizes between subgroups of studies, varying by certain characteristics, such as participant gender or age.

Nineteen studies were included in the review and meta-analysis (4,572 participants). Nine were randomised controlled trials (2,230 participants), six were non-randomised controlled trials (1,193 participants), and four were pre-post studies (1,149 participants). Twelve of the 19 studies were rated high quality, and the other seven were rated low quality. Thirteen potentially relevant studies were excluded because missing data could not be obtained from the study authors.

Interventions to promote walking in groups were found to be moderately efficacious in increasing physical activity (SMD 0.52, 95% CI 0.32 to 0.71; 19 studies). The fail-safe N was large (753 studies), indicating a low risk of publication bias. Sensitivity analysis revealed that low-quality studies tended to produce larger effect sizes (SMD 0.59, 95% CI 0.28 to 0.91; seven studies) than high-quality studies (SMD 0.49, 95% CI 0.24 to 0.75; 12 studies). This difference between low- and high-quality studies was statistically significant (z=1.55; 19 studies; p=0.05).

The heterogeneity between the studies was high; only 23.21% of the variance could be explained by sampling error (no further information provided). Moderator analyses showed that studies reporting outcomes over six months had significantly larger effect sizes than those reporting outcomes up to six months. Similar findings were reported for studies targeting both genders versus women only, and studies targeting older adults versus younger adults. No significant differences in effect size were found between interventions delivered by professionals, and those delivered by lay people.

Interventions to promote walking in groups increased physical activity in adults.

The review question was clear and supported by reproducible inclusion criteria. Relevant databases were searched and some effort was made to locate unpublished data, but the restriction to studies in English may mean that relevant studies were missed. An attempt was made to minimise reviewer error and bias in data extraction, but not in study selection, and it was unclear for quality assessment.

The quality of most of the included studies was high because they met at least half of the criteria, but the Cochrane tool was not designed to produce a score for quality, and not all of the criteria were applicable to non-randomised trials and pre-post studies. So, the quality results may be misleading, particularly for studies that were not randomised controlled trials. Study details were presented clearly, but no information was given on the control groups. Given this, and the heterogeneity between studies, it is possible that the statistical methods of synthesis may not have been appropriate. Nevertheless, all studies showed a positive effect, with their intervention, and the overall conclusion reflects this. The authors stated that the review findings should be treated cautiously for different demographic groups, because most participants were white, middle-aged women.

The review methods had some limitations, but the authors' conclusions were in accordance with the evidence and appear to be reliable.

Practice: The authors stated that walking interventions that provide social support for behaviour change could improve maintenance of behaviour change. Consideration of participants' self-efficacy and outcome expectations might improve long-term outcomes. Tailored interventions should be provided for men, and lay people should be trained to effectively promote walking in groups.

Research: The authors stated that randomised controlled trials with objective, long-term measures of behavioural outcomes were required. Pragmatic research, with rigorous designs and reliable, long-term outcome measures, was needed to investigate how behaviour change can be achieved and maintained in the real world. Such research should investigate the intervention components that were important for short- and long-term behaviour change, and test the cost-effectiveness of the interventions.

Funded by Coventry University, UK.

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.