This review of psycho-behavioural interventions for preventing obesity in multi-ethnic adults in the USA concluded that future programs should include individual sessions, family involvement and problem-solving, as well as lifestyle changes. The review process appeared to be well conducted, but the methods of pooling data and lack of quality assessment cast doubt on the reliability of the authors' conclusions.

To establish which components of psycho-behavioural interventions are more effective for reducing body weight, body mass index (BMI) or body fat percentage in multi-ethnic and minority adults.

MEDLINE, ERIC (Education Resources Information Center), SPORTDiscus, CINAHL Plus, Academic Search Premier, PsycARTICLES and Health Source Nursing/Academic Edition were searched alongside reference lists. Search terms were reported. Only articles published in English between January 1980 and July 2006 were included. Consultation with experts was also undertaken.

Studies of interventions with a psycho-behavioural component, including a control group, where at least 50% (or 25) of participants were from a minority group, conducted in the USA, were eligible for inclusion. Eligible outcomes were body weight, BMI or percentage body fat. Studies of pharmacological interventions only were excluded.

Most of the included studies were based in a clinic, the rest were based in the community. Interventions included exercise (walking, jogging, aerobics), changes to diet (nutritional advice, calorie restriction, reducing fat, and increasing fruit and vegetable intake), and medication. Included studies apparently compared active interventions assessing these components individually or in combinations of two or three components. The proportion of minority adults ranged from 18 to 100%, with ages ranging from 18 to 79 years. Almost half of the included studies were of women only and the rest (apart from one) were of mixed sex. The median duration of treatment was 12 months (range 2.75 to 58 months).

Studies were selected by two reviewers independently with disagreements resolved by consensus.

The authors did not state that they assessed validity.

The mean weight at baseline and at the end of the intervention were extracted and used to calculate a standardised effect size (Cohen's d), with 90% confidence intervals (CI), for each intervention group within each study. If actual weights were not reported, effect sizes at six months were extracted.

Data were extracted by two reviewers independently.

The mean or median of all the effect sizes appears to have been calculated, further details were not reported. Results were presented grouped by the number of components involved and the type of intervention.

Twenty four studies (n=13,326 participants with 5,355 minority adults) were included in the review. The mean weight loss was 3.5kg (range 0.4 to 8.6kg).

Effect sizes over all studies ranged from -0.07 to 1.01 and the median was 0.28. Studies evaluating only one component showed a low mean effect size of 0.08 (90% CI -0.04 to 0.35, five studies). The mean effect size for two component interventions was 0.22 (90% CI 0.05 to 0.40, 13 studies) and for three component interventions was 0.52 (90% CI 0.39 to 0.65, six studies).

Interventions aimed at altering lifestyles through changes to diet and exercise had a mean effect size of 0.35 (90% CI 0.19 to 0.50, 19 studies), whereas non-lifestyle interventions had a mean effect size of 0.09 (90% CI -0.03 to 0.35, five studies).

Interventions conducted in individual sessions had a mean effect size of 0.40 (90% CI 0.24 to 0.56, 15 studies) compared with a mean effect size of 0.08 (90% CI -0.04 to 0.30, nine studies) for group interventions.

Other results were reported in the paper.

Future interventions for preventing obesity which target multi-ethnic and minority adults might benefit from incorporating individual sessions, family involvement and problem solving strategies into multi-component interventions designed to address lifestyle changes.

This review had a clearly stated question and specified broad study inclusion criteria. The search covered a number of relevant databases and only papers published in English were included but, as the aim of the project was to look at interventions conducted in the USA, language bias was not a problem. Study selection and data extraction was performed by two reviewers independently, which reduced the possibility of error and bias. The authors did not assess study quality, or report any information about the designs of the included studies.

The results were based on change data from studies with multiple treatment arms and the methods of meta-analysis were unclear. It appeared that the results were the mean of a number of effect sizes without any weighting, which is not a valid method of pooling results. The inclusion criteria stated that all studies needed to be comparative, but it seemed that results comparing interventions within a study had not been presented.

Based on the presentation of data, the results from these studies and any meta-analysis may not be reliable.

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

Research: The authors stated that more studies where the people delivering the intervention are of the same race/ethnicity as the participants are needed. Future studies should also provide results by ethnic group and report weight changes before and after the intervention together with their standard deviations.

Not stated

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.