The fragility of randomized placebo-controlled trials for irritable bowel syndrome.

Abstract

The fragility index (FI) represents the number of participants whose status in a trial would have to change from a non-event (not experiencing the primary endpoint) to an event (experiencing the primary endpoint) in order to turn a statistically significant result into a non-significant result. We sought to evaluate the fragility indices of irritable bowel syndrome [IBS-mixed (IBS-M), IBS-constipation (IBS-C), & IBS-diarrhea (IBS-D)] trials. Irritable bowel syndrome trials published in high-impact journals were identified from Medline. Trials had to be in adults, randomized, parallel-armed, with at least one statistically significant binary outcome, and an achieved primary endpoint of therapeutic efficacy. FI and correlation coefficients were calculated, and regression modeling used to identify predictors of a high FI. Twelve trials were analyzed with a median FI of 6 (range: 0-38). Median sample size in all trials was 366 (range: 44-856). Trial publication year (p=0.71), journal impact factor (p=0.52), duration of study (p=0.12), and number need to treat [NNT] (p=0.29) were not predictive of a high FI. While a lower p-value correlated with a higher FI (p=0.039), no correlation was noted between FI and impact factor (R=-0.20, p=0.52), trial publication year (R=0.12, p=0.71), duration of trial (R=-0.46, p=0.13), NNT (R=-0.34, p=0.29), and sample size (R=0.23, p=0.5). The highest FI was in a Ramosetron trial (FI=30) for IBS-D. A median of six participants is needed to nullify results in the included IBS trials suggesting how easily statistical significance based on a threshold p-value may be overturned.