A clinical prediction model for postoperative delirium following surgical treatment of femoral fractures in older adult patients: development and validation study.

OBJECTIVE

Postoperative delirium (POD) represents a significant postoperative complication in older adult patients undergoing femoral fracture surgery, associated with substantial mortality, extended hospitalization, and heightened healthcare expenditure. This study aimed to develop and validate a clinical model to predict POD in older adult patients with femoral fractures, facilitating early risk identification and targeted interventions.

METHODS

We retrospectively analyzed patients aged ≥ 65 years who underwent femoral surgery at the First Affiliated Hospital of Dalian Medical University from June 2017 to June 2023. Demographic, laboratory, and perioperative data were collected, with POD as the primary outcome. After multivariable logistic regression identified independent predictors, which were incorporated into a nomogram. The cohort was divided into training (70%) and validation (30%) sets. Model performance was assessed by area under the ROC curve (AUC), Hosmer-Lemeshow(HL) test, calibration plots, and decision curve analysis (DCA). The multivariable analysis identified only three independent predictors of postoperative delirium: tranexamic acid use, preoperative glucose levels, and preoperative lymphocyte counts.

RESULTS

Among 2,029 patients, POD incidence was 23.8%. Multiple factors were associated with POD in univariable analysis, including history of cerebral infarction, laboratory abnormalities, and perioperative factors. However, multivariable analysis identified only three independent predictors. Notably, tranexamic acid use (adjusted OR = 0.32, 95% CI: 0.24-0.44) and higher preoperative lymphocyte counts (adjusted OR = 0.60, 95% CI: 0.48-0.75) were protective, while elevated preoperative blood glucose increased risk (adjusted OR = 1.41, 95% CI: 1.35-1.49). The model showed strong discrimination (AUC 0.85 training, 0.80 validation) and good calibration (test HL P = 0.385). DCA indicated significant net clinical benefit within probability thresholds of 0.2-0.6. Sensitivity and specificity at the optimal cut-off point (0.257) were 82.4% (95% CI: 80.0%-85.0%) and 75.0% (95% CI: 71.0%-80.0%), respectively.

CONCLUSION

This validated model, integrating multidimensional perioperative factors, enables individualized risk stratification for POD in older adult femoral fracture patients. Key predictors include tranexamic acid use, preoperative glucose, and preoperative lymphocyte counts, supporting early identification and preventive strategies.