S1166 Predictors of Improvement Among Patients With Disorders of Gut-Brain Interaction in a Virtual Multidisciplinary Care Model

Introduction: Disorders of gut-brain interaction (DGBI) such as irritable bowel syndrome (IBS) require multifaceted care including pharmacologic, dietary, and behavioral interventions. Traditional in-person care models can fall short in delivering comprehensive support. While digital delivery of gut-brain therapies is promising, real world data evaluating predictors of response across these modalities remains limited. Methods: We conducted a retrospective cohort study of adult patients (≥18 years) with confirmed DGBI diagnoses who enrolled in a virtual care program (Oshi Health) including services from GI providers, registered dietitians (RD), and behavioral health (BH). As part of the program, patients engaged with a gastroenterology advanced practice provider who designed an individualized care plan. We analyzed demographic and clinical characteristics, including baseline symptom burden, diagnosis, and care components received. Patient-reported outcomes on symptom severity and improvement were collected through in-program surveys. Multivariable logistic regression identified predictors of clinical improvement, defined as self-reported symptom improvement or resolution. Results: From January 2023 to May 2025, a total of 4,818 patients were evaluated; 3,485 (72%) engaged with multidisciplinary care. The mean age of the cohort was 43.3, most patients were women (70.5%), and IBS was the most common diagnosis (93.7% with any subtype). Overall, 81.2% of patients achieved symptom improvement over an average of 1.2 months. In unadjusted analysis, patients prescribed any neuromodulator had 1.73 greater odds of symptom improvement compared to those not prescribed a neuromodulator. This relationship was no longer significant after multivariable adjustment (odds ratio [OR] 1.56, CI 0.95-2.57). Among patients who received neuromodulators, higher numbers of GI, RD, and BH visits were associated with greater odds of improvement per visit (OR 1.4, 1.9, 2.2 respectively; all P < 0.05). Conclusion: This study characterizes the real-world application of a virtual care model and identifies patient-level predictors associated with clinical improvement and treatment response. These findings may inform tailored care pathways and help target which patients are most likely to benefit from specific interventions. Further analysis is planned to better delineate factors associated with overall clinical improvement, neuromodulator initiation, and treatment response patterns.