Twenty-nine double-blind RCTs were included (n=17,818 patients). Twenty-seven trials scored 4 or 5 on the Jadad quality scale.
5-HT3 antagonists (11 trials; 7,216 patients): Eight studies evaluated alosetron (n=4,987 patients) and three evaluated cilansetron (n=2,229 patients). 5-HT3 antagonists were associated with a statistically significant decrease in the proportion of patients with persistent symptoms compared with placebo (RR 0.78, 95% CI 0.71 to 0.86; NNT 7, 95% CI 5 to 11). Significant heterogeneity was found (I2=80%). There was no evidence of publication bias. The association remained significant with both alosetron (RR 0.79, 95% CI 0.69 to 0.90; I2=85%) and cilansetron (RR 0.75, 95% CI 0.69 to 0.82; I2=30%).
Adverse events were significantly increased with alosetron compared with placebo (64% versus 55%; RR 1.19, 95% CI 1.09 to 1.30; NNH 10, 95% CI 7 to 15; seven studies). There was no significant difference in serious adverse events between alosetron and placebo. Four patients developed ischaemic colitis. None of the cilansetron studies reported adverse events.
5-HT4 agonists (11 trials of tegaserod, 9,242 patients): 5-HT4 agonists were associated with a statistically significant decrease in the proportion of patients with persistent symptoms compared with placebo (55% versus 63.5%; RR 0.85, 95% CI 0.80 to 0.90; NNT 10, 95% CI 8 to 14). Significant heterogeneity was found (I2=57%). There was evidence of publication bias (asymmetrical funnel plot and Egger’s test p=0.04). There was no significant difference in adverse events between tegaserod and placebo. Two cardiovascular events were reported with tegaserod.
Mixed 5-HT3 antagonists/5-HT4 agonists (seven trials, 1,043 patients, including four trials of cisapride with 317 patients): There was no significant difference between mixed 5-HT3 antagonists/ 5-HT4 agonists and placebo in the proportion of patients with persisting IBS symptoms (64% versus 57%; I2=59%) for cisapride and renzapride whether combined or compared separately with placebo. There was no evidence of publication bias. Renzapride was associated with a statistically significant increase in diarrhoea (RR 2.17, 95% CI 1.26 to 3.74; three trials). There was no significant difference in adverse events between cisapride and placebo (three studies).
Results of other sub-group analyses were also reported.