A Systematic Review of Efficacy in Irritable Bowel Syndrome: How the Number Needed to Treat May be a Misleading Metric.

Abstract

Using the number needed to treat (NNT) to compare pharmacotherapies for irritable bowel syndrome (IBS) is complicated by evolving trial endpoints. Standardizing NNT estimates across these endpoints, including future development of a conversion framework linking historical subjective measures to contemporary Food and Drug Administration (FDA) response criteria, may improve the validity of therapeutic efficacy comparisons. PubMed, Web of Science, and Cochrane databases were searched through October 2024 to identify placebo-controlled clinical trials investigating IBS pharmacotherapies that reported both the subjective adequate relief and contemporary FDA endpoints. Calculated NNTs were compared between different endpoints for the same trials and across the same endpoints for different indications using a random-effects model. Among the 15 included trials, the FDA endpoint NNT was on average 2 units higher than the subjective adequate global relief endpoint ( P < 0.01) and no significant NNT differences were noted between tricyclics and other therapies ( P = 0.41). When using the subjective relief endpoint to enable the inclusion of tricyclics, the NNTs for drugs for constipation-predominant IBS, diarrhea-predominant IBS, and IBS global symptoms were 5 ( P < 0.0001), 8 ( P < 0.0001), and 8 ( P = 0.001), respectively. When standardized endpoints are used, the NNTs for included IBS pharmacotherapies are similar across indications. Failure to account for endpoint heterogeneity in clinical trials for IBS may distort comparative effectiveness conclusions, undermining clinical decision-making.