Psychometric Evaluation of Patient-Reported Outcomes in Irritable Bowel Syndrome Randomized Controlled Trials: A Rome Foundation Report

Relationship between trichotomized baseline severity and responder status, defined using harmonized binary end point (aka, “adequate relief”). Data are provided for the overall, treatment, and placebo groups.

Relationship between trichotomized baseline severity and responder status, defined using 50% improvement from baseline severity. Data are provided for the overall, treatment, and placebo groups.

“Difference in difference” (DID) scores stratified by response status. The Figure depicts 2 series of data: 1 for adequate relief (AR) and 1 for 50% improvement in severity. The data are stratified by 23 variables, including bowel symptoms, HRQOL domains, and work productivity domains. Each bar reveals the mean DID scores for each variable between responders and nonresponders, stratified by responder definition. The higher the DID the better the discriminant validity. For example, using the AR response definition, the mean DID abdominal pain score between responder groups was 15. In contrast, using the 50% improvement definition, the mean DID score between groups was 22.5. Both bars exceed the threshold for “clinical significance” depicted by the dashed line at 5 points (ie, half standard deviation MCID definition).

Proportion of patients achieving an MCID for individual bowel symptoms: AR vs 50%. Panel A provides data in the IBS-C subpopulation, and panel B provides data in the IBD-D subpopulation. Each bar represents the results of an individual 2 × 2 table and depicts the difference in MCID achievement between responders and nonresponders. For example, 38% more IBS-C patients in the binary response group achieved an MCID in bloating vs those not achieving a binary response. In contrast, 52% more IBS-C patients in the 50% improvement group achieved an MCID in bloating vs those not achieving 50% improvement.