Development and Validation of a Prediction Model for Enteral Feeding Intolerance in Critical Ill Patients: A Retrospective Cohort Study

To construct and validate a prediction model for enteral feeding intolerance in critically ill patients during the first 7 days of enteral feeding. A retrospective cohort study. We reviewed the medical records of two intensive care units from January 2015 to August 2023, to develop a prediction model by univariate analysis and logistic regression analysis. Model's performance was evaluated through discrimination, calibration and decision curve analysis. This study involved a total of 471 patients, with an enteral feeding intolerance incidence rate of 35.7%. The prediction model comprised six variables, namely neurological disease, chronic gastrointestinal disease, Acute Physiological and Chronic Health Assessment II score, sedatives, acid suppressants and serum albumin. The model showed robust discrimination, calibration and clinical net benefit, indicating significant potential for practical application with readily available variables. The model demonstrated strong predictive performance in assessing the risk of enteral feeding intolerance during the early stage of nutrition initiation. Enhancing clinicians' capacity to reduce the incidence of enteral feeding intolerance and improve patient outcomes. The prediction model shows a good capacity to discriminate critically ill patients at risk of enteral feeding intolerance, is helpful to provide personalised care. TRIPOD + AI checklist. No patient or public contribution. https://www.chictr.org.cn/ ChiCTR2400090757.