Thirteen RCTs were included; five were of cognitive-behavioural therapy, and eight were of psycho-education or other interventions. Four trials looked at interventions for individuals, the other nine were for groups. Four trials included physical activity, and three included a diet. The trials were at an unclear risk of bias; randomisation was poorly described and allocation concealment rarely mentioned. Incomplete outcome data was a common problem, and in one trial participants were paid to take part.
A significant difference in favour of the intervention was noted, with a reduction in mean BMI of -0.98kg/m² (95% CI -1.31 to -0.65) across all 13 trials. There were no significant differences in drop-out rates between intervention and control groups.
Subgroup analyses suggested that weight gain prevention, intervention for individuals, psycho-education, and including diet or physical activity, were most effective.
Statistical heterogeneity was not significant (30%), but there were numerous sources of clinical heterogeneity: initial weight, length of illness, diagnosis, concurrent drug therapy, objective of the trial, and type and duration of intervention.