Abstract
BACKGROUND: This investigation aimed to establish pivotal determinants of postoperative delirium (POD) following radical hysterectomy for cervical carcinoma (CC) and formulate an individualized risk stratification tool. METHODS: We conducted a retrospective cohort study encompassing 253 geriatric patients undergoing radical hysterectomy for CC between 2021 and 2025. We systematically evaluated potential predictors using a two-phase regression model: first through univariate analysis (P < 0.1), followed by multivariate logistic regression (P < 0.05) to identify independent predictors of POD. Key clinical, demographic, and laboratory variables were included in the analysis. The incidence of POD was assessed using the Confusion Assessment Method (CAM) during the 7-day perioperative period. The predictive nomogram was developed using R and was rigorously validated through both internal cohort validation and external validation. ROC, calibration, and decision curve analyses were used to assess the nomogram's predictive performance. RESULTS: The POD incidence reached 16.2% (n = 41) during the 7-day postoperative surveillance. Multivariable analysis delineated five independent predictors: advanced age (OR = 1.12, P = 0.031), depressed albumin-fibrinogen ratio (AFR; OR = 0.69, P = 0.029), elevated neutrophil-lymphocyte ratio (NLR; OR = 3.51, P = 0.001), Controlling Nutritional Status (CONUT) score (OR = 1.81, P = 0.003), and Geriatric Nutritional Risk Index (GNRI; OR = 0.94, P = 0.001). The constructed nomogram exhibited robust discriminative capacity, achieving area under curve (AUC) values of 0.821 and 0.966 in internal and external validations, respectively. CONCLUSIONS: This research introduced an effective nomogram prediction model for predicting POD after radical hysterectomy for CC, providing a straightforward and visual method for individualized risk assessment.