475 The Fragility Index of the Randomized Trials for Irritable Bowel Syndrome With Constipation

Abstract

INTRODUCTION: In the age of frequentist analyses and standardized threshold P-values (<0.05); fragility index (FI), the minimum number of patients that needs to change from “non-responder” to “responder” for the P-value to become >0.05, has been suggested to aid our interpretation of the robustness of the “positive” large randomized control trials (RCTs) (Walsh, M. et al. JCE 2014). This number allows the reader to quickly know the number of patients that the study relies on for its significance, hence a higher index suggests a higher confidence in the results. Based on all available RCTs, we aimed to calculate the FI of medications that are approved by Food and Drug Association (FDA) for (Irritable bowel syndrome with constipation) IBS-C. METHODS: Data from 13 double-blind RCTs on linaclotide, plecanatide, lubiprostone and tegaserod were analyzed. FI was calculated by converting the number of “non-responders” in a dichotomized data to a “responders” to get a P-value > 0.05 by using the Fisher's exact test. We calculated the FI for the FDA endpoint if it was reported, otherwise the dichotomized endpoint evaluating its efficacy was used. A similar calculation was done for diarrhea as the most common side effect in RCTs on IBS-C in each respective study by converting those without diarrhea to those with diarrhea to calculate its FI. RESULTS: All 13 studies had a positive FI except for Johanson et al. studying lubiprostone since the study itself reported a nonsignificant outcome, as well as Novick at al. studying tegaserod (Table 1). The FI for diarrhea was positive for those in linaclotide and 2 studies for plecanatide and tegaserod but not in lubiprostone. The median (mean) FI for efficacy was 27 (28.5), 13 (17.7), 2 (2), and 6 (11) events and the median (mean) FI for diarrhea was 31.5 (29.3), 2 (3), 0 (0), and 8 (5.7) events for linaclotide, plecanatide, lubiprostone and tegaserod respectively. Linear regression showed a significant positive correlation between FI for efficacy and diarrhea with R2 = 0.54 (P = 0.004) (Figure 1). CONCLUSION: Median FI is appropriately positive for RCTs assessing IBS-C on medications currently approved by FDA. We recommend considering the addition of FI, as a simple and useful tool, in future systematic reviews to assist in appropriate and optimal decision making in patient care.