Development and validation of a nomogram for predicting postoperative delirium in elderly patients with hip fracture based on preoperative biochemical indicators

Background Postoperative delirium (POD) is a common and serious complication in elderly patients undergoing hip fracture surgery, associated with prolonged hospital stay, functional decline, and increased mortality. This study aimed to develop and validate a practical nomogram for predicting POD using preoperative biochemical parameters and clinical characteristics. Methods A retrospective cohort of 473 elderly patients (≥65 years) who underwent hip fracture surgery between January 2020 and September 2025 was analyzed. Exclusion criteria included: preoperative dementia; severe renal insufficiency (serum creatinine >177μmol/L in males or >133μmol/L in females); severe liver dysfunction (Child-Pugh class C); pathological fractures (e.g., primary or metastatic tumors); severe communication barriers; and missing preoperative biochemical data. POD was diagnosed using a validated two-step chart review method. Independent risk factors were identified via univariate and multivariate logistic regression with class weight adjustment. Feature selection was performed using Recursive Feature Elimination with Cross-Validation (RFECV) based on Random Forest. The nomogram was constructed and evaluated using ROC curves, calibration curves, and decision curve analysis (DCA). Results The incidence of POD was 16.07% (76/473). Independent predictors included age, preoperative ALT, AST, HCT, LYM%, and Na. RFECV selected 6 optimal features. The nomogram demonstrated good discrimination with an AUC of 0.764 (95% CI:0.645-0.865) on the test set and mean AUC of 0.745 ± 0.040 in 5-fold cross-validation. Calibration was acceptable (Brier score = 0.1765), though some deviation existed at higher predicted probabilities. DCA showed net clinical benefit across a wide range of threshold probabilities (approximately 0.0–0.2 and >0.7), outperforming treat-all or treat-none strategies in relevant ranges. Conclusions We developed a user-friendly nomogram incorporating six easily obtainable preoperative variables for predicting POD risk in elderly hip fracture patients. This tool may facilitate early risk stratification and targeted preventive interventions.