The authors concluded that regular exercise training may improve functional performance in older adults with differing levels of ability, but more research is required. In view of the poor quality and clinical differences between the primary studies, inconsistency of results and lack of information about effect sizes, these conclusions should be interpreted with some caution.

To evaluate the effects of exercise training on physical performance in frail older people.

PubMed, EMBASE and the Cochrane Central Register of Controlled Trials were searched from 1995 to August 2007. Search terms were reported. The search was restricted to full text articles in English.

Randomised controlled trials (RCTs) of exercise programmes for frail older adults were eligible for inclusion. Included trials had to report performance-based measures of physical function and include a comparison group receiving no exercise or light-intensity activity. Trials were required to describe the study population as ‘frail” in the abstract or title.

Participants in most of the included trials were community-dwelling older people, but some trials included participants from care facilities, such as sheltered housing and nursing homes. The mean participant age ranged from 77 to 88 years, and most were women. Criteria for frailty (where reported) varied widely, including specific functional measures, a history of falls or fracture and a diagnosis of heart failure.

The exercise protocols used in the included trials varied widely. The interventions included resistance training, Tai Chi and multi-component training (resistance training with endurance, flexibility and balance exercises). In most cases interventions were administered to groups in an institutional setting, with or without some home sessions. The usual frequency was two or three times a week. Control conditions included no intervention, or activities such as lower intensity exercises at home, social activities and lifestyle advice. A wide range of performance measures was used, including performance-based summary scores (e.g. the modified Physical Performance Test), distance walked in six minutes, gait speed and the timed Get-Up-and-Go test. Adverse events were not systematically reported. The duration of the intervention ranged from ten weeks to 28 months. Mean attendance rate varied: the lowest reported was 42% over one year follow-up, but methods of measuring adherence varied.

One reviewer scanned all articles retrieved and excluded those completely out of scope; two other reviewers independently selected studies for inclusion. Disagreements were resolved by discussion.

The Verhagen scale was used to score trials for the following components of validity: randomisation method; allocation concealment; group equivalence at baseline; specification of inclusion criteria; blinding (of outcome assessor, exercise trainer and participants); reporting of effect estimates and measures of variability; and use of intention-to-treat analysis. A maximum of 9 points were awarded, with studies scoring 5 or more points deemed to be high quality.

Three reviewers independently assessed study validity. Disagreements were resolved by discussion.

Data were extracted on the results for each functional outcome in the included trials. Statistically significant between-group differences were highlighted in a table.

Three reviewers independently extracted the data. Disagreements were resolved by discussion.

Trials were combined in a narrative synthesis, organised by type of intervention.

Twenty RCTs were included in the review (n=2,515). Sixteen RCTs were rated as high quality. Quality scores ranged from 2 to 7 points. Outcome assessment was blinded in 14 RCTs, participants were blinded in five RCTs and the trainer was blinded in one RCT. In seven RCTs, groups were not similar at baseline. Two RCTs did not specify inclusion criteria. Sixteen RCTs used intention-to-treat analysis (for participants with complete data) or imputed missing data. Only three RCTs reported point estimates and 95% confidence intervals for differences between the groups. The mean drop-out rate was 18% (range 4 to 32%). Most trials failed to report their methods in sufficient detail.

For effectiveness of physical exercise versus controls, there was a statistically significant improvement in the intervention group in: three out of four RCTs using a performance-based summary score; nine of 17 RCTs reporting the distance walked in six minutes or gait speed; and two of nine RCTs reporting the timed Get-Up-and-Go test. Six RCTs, all rated high quality, found no significant benefit from the intervention on any performance measure; five of these were performed in a highly frail population.

Regular exercise training may improve functional performance in older adults with differing levels of ability, but more research is required.

The objectives and inclusion criteria of the review were clear. Relevant sources were searched, but the restriction to trials using the specific term ‘frail’ means that relevant trials using other terminology may have been missed (as the authors acknowledged). Steps were taken to minimise the risk of reviewer bias and error by having more than one reviewer independently make decisions on study selection, validity assessment and data extraction, but the initial stages of study selection involved only one reviewer. The decision to combine trials in a narrative synthesis appeared appropriate, but no indication of effect sizes was provided. Also, there was inconsistency between the text and tables as to whether significant effects were measured between or within groups. Heterogeneity between the trials was well addressed in the text, as were other potential sources of bias. In view of the poor quality and clinical differences between the primary studies, inconsistency of results and lack of information about effect sizes, the authors' conclusions should be interpreted with some caution.

Practice: The authors stated that professionals working with older people need to be aware of functional abilities of this population, to ensure that exercise programmes offered are sufficiently challenging. Programmes should be enjoyable and manageable, to improve adherence.

Research: The authors stated that larger and more rigorous studies are required to determine the optimum type, intensity, duration and frequency of exercise for older adults. Blinding of participants, programme delivery and outcomes assessment is recommended. Researchers should agree on a clear definition of frailty, a standard method for assessing outcomes (e.g. the Physical Performance Test) and should systematically report adverse events.

Department of Public and Occupational Health; The EMGO Institute; TNO-VU University Medical Center, Amsterdam, Research Center Physical Activity, Work and Health, Body@Work; VU University Medical Center, Amsterdam.

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.