Twenty-six trials were included, with four on selective serotonin re-uptake inhibitors, three on lubiprostone, six on tricyclic antidepressants, eight on alosetron, and five on rifaximin. There were 8,595 participants (range 23 to 1,171). Follow-up ranged from 10 days to 10 weeks (where reported).
Constipation: There was no difference in dropouts due to adverse events between lubiprostone and placebo. The four trials of selective serotonin reuptake inhibitors did not supply enough data for a reliable meta-analysis of harm, but the drugs seemed to be safe, with 7.8% dropout from treatment and 9.2% dropout from placebo.
Diarrhoea: The pooled number needed to harm was 18.3 (95% CI 5.8 to 217.4) for tricyclic antidepressants, 19.4 (95% CI 8.5 to 90.1) for alosetron, and 8,971 for rifaximin. The relative risk was 2.71 (95% CI 1.14 to 6.44) with tricyclic antidepressants, 2.00 (95% CI 1.22 to 3.30) with alosetron, and 1.01 (95% CI 0.50 to 2.02) with rifaximin. Apart from alosetron (Ι²=71.8%), no statistical heterogeneity was observed in the pooled relative risks. No publication bias was found.
The number needed to treat to benefit one patient was 8.0 for tricyclic antidepressants, 7.5 for alosetron, and 10.6 for rifaximin. For tricyclic antidepressants 2.3 patients and for alosetron 2.6 patients benefited from treatment before one harm occurred. For rifaximin, 846 patients benefited before one harm. Adverse events were more common with tricyclic antidepressants and alosetron.