Mo1131 Development and Validation of Prism: A Patient Reported Outcome Measure for GERD Patients Who Are Partial or Non-Responders to PPI Therapy

Background: Most patients with gastro-esophageal reflux disease (GERD) experience relief following treatment with proton pump inhibitors (PPI). As many as 17-44% of patients, however, exhibit partial or non-response to PPI therapy. Self-report of symptoms is fundamental to diagnosing and monitoring treatment response in these patients, yet most extant GERD PRO instruments fail to meet development best-practices as described by the FDA. We therefore sought to develop and validate a PRO instrument for clinical trials involving patients with GERD who are PPI partial-responders. Methods: We conducted a systematic literature review, held patient focus groups, convened an expert panel, and conducted cognitive interviews in order to establish content validity. Eligible participants took PPI therapy for at least 8 weeks, had undergone an upper endoscopy revealing no more than L.A. Classification Grade A esophagitis, and scored at least 8 points on the GerdQ. We used ATLAS.TI software to code focus group transcripts, generating frequencies at the domain, subdomain, and primary code levels. Results informed the construction of a conceptual framework, and guided development of 26 draft items. These items and subsequent revisions were reviewed by two expert panels and debriefed with cognitive interview participants. These efforts resulted in PRISM, a 21-item instrument that underwent psychometric evaluation during a Phase IIB drug trial (Protocol SPD557-206). Results: A diverse sample participated in focus groups (N=38; mean age=50.8; 55%M), saturating the codebook and supporting a preliminary conceptual framework. Following drafting of items, a new sample of patients participated in cognitive interviews for purposes of item revision (N=20; Age=52.6; 55%M). During the trial, an exploratory item assessment (n=104) revealed that all but one of the PRISM items met expectations for internal and external indices. A confirmatory factor analysis (n=220) resulted in a 4-factor model displaying the highest goodness-of-fit scores. All domains had a high inter-item correlation (Cronbach's α>0.8). Test-retest reliability was excellent with intraclass correlation also exceeding 0.8. Evidence for convergent validity was strong, with highly significant (p<.01) correlations between average weekly PRISM scores and severity anchors, and significant (p<.05) correlations with RDQ sub-scales. Three of 4 PRISM sub-scale scores were highly responsive to changes in both self-reported severity and symptom relief. Cumulative distribution functions revealed significant differences between responders and non-responders. Conclusion: PRISM demonstrates strong validity and psychometric properties in a clinical trial setting. Developed in line with FDA guidance on PROs, PRISM represents an new outcome measure for GERD patients with partial or nonresponse to PPI therapy.