Nine RCTs were included (n=9,635 patients). All trials had a low risk of bias. All but one RCT reported adequate sequence generation and allocation concealment, all were blinded, and none reported incomplete outcome data or selective outcome reporting. Follow-up duration ranged from three to 24 months.
The use of rimonabant as a dose of 5mg was not associated with a greater incidence of adverse events or increased discontinuation rates. Rimonabant 20mg was associated with an increased risk of adverse events (RR 1.35, 95% CI 1.17 to 1.56), discontinuation (RR 1.79, 95% CI 1.35 to 2.38), psychiatric adverse events (RR 2.35, 95% CI 1.66 to 3.34), and discontinuation due to nervous system adverse events (RR 2.35, 95% CI 1.49 to 3.70). The number-needed-to-harm at a dose of 20mg was 22 for any adverse event, 18 for discontinuation due to adverse events, and 30 for discontinuation due to psychiatric adverse events.
There was evidence of publication bias for the outcomes serious adverse events, discontinuations for adverse events, and psychiatric adverse events.