Thirty-two RCTs were included in the review (n=1,482): 30 RCTs in adults and two RCTS in children. The number of patients in each trial ranged from 14 to 125. There was no evidence of publication bias.
Primary outcomes: Overall, compared with placebo, pharmacological interventions were associated with greater weight loss (WMD -1.99kg, 95% CI -2.77 to -1.20, I2=86%; 29 RCTS, n=1,013). Compared with placebo, metformin had the greatest weight loss (WMD -2.94kg, 95% CI -4.89 to -0.99, I2=91%; seven RCTs, n=334), followed by d-fenfluramine (WMD -2.60kg, 95% CI -5.14 to -0.06; one RCT, n=16), sibutramine (WMD -2.56kg, 95% CI -3.91 to -1.22; I2=40%; two RCTs, n=55), topiramate (WMD -2.52kg, 95% CI -4.87 to -0.16; I2=75%; two RCTs, n=133) and reboxetine (WMD -1.90kg, 95% CI -3.07 to -0.72; I2=0%; two RCTs, n=79). Compared with placebo, there were no significant differences in weight loss with amantadine, dextroamphetamine, famotidine, fluoxetine, nizatidine, orlistat, fluvoxamine, metformin plus sibutramine and rosiglitazone.
Secondary outcomes: There was no significant difference between pharmacological agents compared with placebo for nausea, all-cause discontinuation and psychiatric symptoms. The findings were conflicting for the effects of different drugs on carbohydrate metabolism and blood lipids (findings reported in the review).
Sensitivity analysis: There were few differences in outcomes when analyses were conducted for type of trial, trial duration, hospitalisation status and lifetime antipsychotic treatment duration; the level of statistical heterogeneity remained generally high.