Twenty-one RCTs (13,759 participants) were included in the review. Sample size ranged from 46 to 3,277 participants (mean sample size 655). Jadad quality assessment scores ranged from 2 to 6 points. Details of the quality assessment were not reported. Duration of follow-up ranged from four to 48 months (mean 14.7 months).
Weight loss: Sixteen trials reported weight loss as an outcome. Orlistat (MD -2.39kg, 95% CI -3.34 to -1.45kg; six trials), sibutramine (MD -3.73kg, 95% CI -6.00 to -1.46kg; seven trials), and rimonabant (MD -3.66kg, 95% CI -4.17 to -3.15kg; three trials) were all associated with significant weight loss in the treatment group compared to the placebo group. The pooled total mean difference for all three drugs was -3.13kg (95% CI -4.00 to -2.26kg; 16 trials), which indicated that anti-obesity drugs generally lead to significant weight loss when compared with placebo. Heterogeneity was observed in the orlistat (Ι²=69.6%), sibutramine (Ι²=91.1%), and overall (Ι²=88.2%) analyses.
Cardiovascular risk factors
Total cholesterol: The pooled total across drugs indicated a significant reduction of total cholesterol in the treatment compared to the placebo group (MD -0.12mmol/L, 95% CI -0.22 to 0.03; 12 trials). However, only orlistat was associated with a significant reduction (six trials).
Low-density lipoprotein: The pooled total across drugs indicated a significant reduction of low-density lipoprotein in the treatment compared to the placebo group (MD -0.10mmol/L, 95% CI -0.18 to -0.01; 12 trials). Only orlistat was associated with a significant reduction (six trials).
High-density lipoprotein: None of the three drugs individually nor the pooled analysis across drugs revealed a significant difference in change in high-density lipoprotein between groups (12 trials).
Triglycerides: The pooled total across drugs did not indicate a significant difference in change in triglycerides between the treatment and the placebo group (11 trials). When investigated separately, sibutramine showed a significant effect on triglyceride reduction (three trials).
Fasting glucose: The pooled total across drugs indicated a significant reduction of fasting glucose in the treatment compared to the placebo group (MD -0.08mmol/L, 95% CI -0.13 to -0.04; nine trials). Only orlistat was associated with a significant reduction (four trials).
Systolic blood pressure: The pooled analysis (14 trials) reported no significant differences in changes in systolic blood pressure between groups. However, both orlistat (four trials) and rimonabant (three trials) were associated with significant reductions.
Diastolic blood pressure: The pooled analysis (14 trials) reported no significant differences between groups in diastolic blood pressure. However, both orlistat (four trials) and rimonabant (three trials) were associated with significant reductions.
Adverse events: A range of adverse events outcomes were assessed, including hypertension, tachycardia, gastrointestinal adverse events, infections/infestations, dry mouth, headache, rash, abdominal pain, upper respiratory tract infection, influenza/influenza symptoms, and pharyngitis. Overall, use of any anti-obesity drug increased the risk of developing any adverse event compared with placebo (RR 1.08, 95% CI 1.02 to 1.14; 11 trials).
Ι² values for heterogeneity were not reported consistently for cardiovascular risk factor and adverse events analyses.
The heterogeneity observed in the analyses for weight loss, high-density lipoprotein, and triglycerides was explored. Sequential removal of one trial at a time from each of the analyses did not alter the authors' conclusions. Subgroup analyses for the three outcomes affected confirmed the presence of heterogeneity between the included trials.