ABSTRACT
BACKGROUND: Patients with a new ostomy have high rates of unplanned healthcare utilization (UPHU). We used machine learning to assess which factors contributed the most to UPHU after ostomy construction. METHODS: We retrospectively studied new ostomy patients between 2018 and 2021 âat a single institution. The primary outcome was UPHU within 60 days of discharge. Factors that contributed the most to UPHU were assessed using a classification tree machine learning method. RESULTS: Among 318 patients, 30.8 â% of patients had an UPHU event. The classification tree identified diabetes mellitus as the most important factor associated with UPHU: 56 â% of diabetics had UPHU. Smoking history was the next most important factor: 77 â% of diabetics who smoked had UPHU. Patients who had diabetes, smoked, and had chronic kidney disease had the highest UPHU rate at 86 â%. DISCUSSION: Unplanned healthcare utilization after ostomy construction is highest among patients with diabetes, smoking history, and chronic kidney disease.