Meta-analysis: the effect of dietary counseling for weight loss

Dietary and lifestyle modification efforts are the primary treatments for people who are obese or overweight. The effect of dietary counseling on long-term weight change is unclear.

To perform a meta-analysis of the effect of dietary counseling compared with usual care on body mass index (BMI) over time in adults.

Early studies (1980 through 1997) from a previously published systematic review; MEDLINE and the Cochrane Central Register of Controlled Trials from 1997 through July 2006.

English-language randomized, controlled trials (> or =16 weeks in duration) in overweight adults that reported the effect of dietary counseling on weight. The authors included only weight loss studies with a dietary component.

Single reviewers performed full data extraction; at least 1 additional reviewer reviewed the data.

Random-effects model meta-analyses of 46 trials of dietary counseling revealed a maximum net treatment effect of -1.9 (95% CI, -2.3 to -1.5) BMI units (approximately -6%) at 12 months. Meta-analysis of changes in weight over time (slopes) and meta-regression suggest a change of approximately -0.1 BMI unit per month from 3 to 12 months of active programs and a regain of approximately 0.02 to 0.03 BMI unit per month during subsequent maintenance phases. Different analyses suggested that calorie recommendations, frequency of support meetings, inclusion of exercise, and diabetes may be independent predictors of weight change.

The interventions, study samples, and weight changes were heterogeneous. Studies were generally of moderate to poor methodological quality. They had high rates of missing data and failed to explain these losses. The meta-analytic techniques could not fully account for these limitations.

Compared with usual care, dietary counseling interventions produce modest weight losses that diminish over time. In future studies, minimizing loss to follow-up and determining which factors result in more effective weight loss should be emphasized.