Influencing Factors and Prediction Model of Postoperative Delirium for Elderly Femoral Neck Fracture Patients Under Remimazolam Intrathecal Anesthesia.

Aim: This study aims to explore the factors affecting postoperative delirium (POD) in elderly femoral neck fracture (FNF) patients under intrathecal anesthesia with remimazolam and to construct a prediction model.

Methods: A retrospective study was conducted on 140 eligible patients who were diagnosed with FNFs in our hospital from January 2022 to October 2024 and underwent surgical treatment under intrathecal anesthesia with remimazolam. All individuals were grouped into a delirium group and a non-delirium group according to the occurrence of POD. The included patients were randomly stratified into a training set and a validation set at a ratio of 7:3. Clinical data were collected, and influencing factors were screened using various statistical methods. A nomogram model was constructed to evaluate its performance.

Results: Apparent differences between the groups were identified in glucose, lactate level, oxygen partial pressure (PO₂), Glasgow Coma Scale (GCS) result, and alcohol consumption. The key factors used in constructing the prediction model included glucose (odds ratio [OR] = 1.011, 95% confidence interval [CI]: 0.980-1.024, p = 0.056), lactate (OR = 1.726, 95% CI: 1.252-2.660, p = 0.003), PO₂ (OR = 0.988, 95% CI: 0.977-0.996, p = 0.007), GCS score (OR = 0.346, 95% CI: 0.184-0.550, p < 0.001), and alcohol consumption (OR = 3.140, 95% CI: 0.904-12.310, p = 0.081). The model exhibited favorable discriminative ability, with the training set having an area under the curve (AUC) of 0.882 (95% CI: 0.784-0.952) and the validation set an AUC of 0.941 (95% CI: 0.877-1.000). The calibration curve revealed a strong concordance between the predicted and actual values, suggesting good accuracy. Based on the decision curve analysis (DCA) curve, the model exhibited a remarkable capacity in predicting POD in both the training and the validation sets.

Conclusions: The model constructed on the basis of multiple key influencing factors identified in this study was proved to be helpful for the early recognition of high-risk patients for POD, laying a foundation for personalized perioperative management to lessen the incidence of delirium and ameliorate the prognosis. However, this investigation has limitations, requiring optimization and improvement in the future.