Identification of Risk Factors for Postoperative Delirium and Targeted Preventive Strategies in Elderly Patients in the Emergency Surgical ICU.

AIM: Elderly patients admitted to the intensive care unit (ICU) after emergency surgery represent a uniquely high-risk population for postoperative delirium (POD). However, the key risk factors specific to this group have not been comprehensively integrated, and a corresponding clinical prediction tool is lacking. This study aimed to elucidate the clinical features, identify independent risk factors for POD, and develop and validate a prediction model tailored for elderly patients in the emergency surgical ICU. METHODS: This retrospective study included 496 elderly patients (≥60 years) admitted post-emergency surgery. Delirium was assessed using the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). Patients were randomly divided in a 7:3 ratio into training (n = 347) and validation (n = 149) sets. Logistic regression identified independent predictors, which were used to construct a nomogram. Model performance was evaluated via receiver operating characteristic (ROC) curves, calibration, and decision curve analysis (DCA). RESULTS: POD incidence was 33.1%, predominantly the hypoactive subtype (64.6%), with peak onset on postoperative day 2 (57.9%). Five independent risk factors were identified: higher Sequential Organ Failure Assessment (SOFA) score (Odds Ratio [OR] = 3.818), elevated Procalcitonin (PCT) (OR = 1.594), longer mechanical ventilation (OR = 1.283), intraoperative hypotension (OR = 2.666), and physical restraint use (OR = 3.003). A nomogram was developed using these variables, which demonstrated excellent discriminative performance, with area under the curves (AUCs) of 0.937 and 0.913, in the training and validation datasets, respectively. DCA confirmed the clinical utility of our proposed model. CONCLUSIONS: This study identified five key and modifiable risk factors for POD in elderly emergency surgical ICU patients and successfully developed and validated a high-performance nomogram prediction tool. This model aids in the early identification of high-risk patients, providing a basis for implementing targeted bundled prevention strategies, which holds promise for improving clinical outcomes in this vulnerable population.