A Novel Preoperative Electroencephalogram-Derived Index to Predict Early Postoperative Delirium in Elderly Patients After Hip Fracture Surgeries: Development of a Prediction Model.

Introduction: It is appealing to accurately predict postoperative delirium (POD) before surgeries. In this study, it was hypothesized that a novel electroencephalogram-derived index, the delirium index (DELi), could extract latent information regarding the predisposing factors of POD preoperatively. This study was aimed at developing a concise model that incorporated this DELi score to predict the early POD of elderly patients after hip surgeries. Materials and Methods: Elderly patients scheduled for elective hip fracture surgeries were prospectively enrolled in a tertiary care hospital from November 2020 to June 2022. DELi scores and patient characteristics (age, sex, types of fracture and surgery, the time interval between fracture and surgery, cognitive function assessed using the Montreal Cognitive Assessment (MoCA), and frailty status assessed using the FRAIL scale) were collected preoperatively as candidate predictors. POD diagnosed using the confusion assessment method (CAM) was the outcome. Least absolute shrinkage and selection operator (LASSO) regression analysis was used to select predictors. Then, these predictors were entered into a backward logistical regression analysis to develop a prediction model. Discrimination, calibration and clinical utility were validated using the bootstrapping method. Results: All data (144 qualified patients of 170) were used for development. POD was observed in 71 patients (49.3%). Preoperative DELi scores predicted early POD (the area under the curve (AUC) = .786, 95% confidence interval (CI): .712, .860, in internal validation). A nomogram with MoCA, FRAIL scale and DELi score was constructed with excellent discrimination (AUC = .920, 95% CI: .876, .963, in internal validation), accredited calibration ( Conclusions: Preoperative DELi scores predicted the early POD of elderly patients after hip surgeries. A concise prediction model was developed and demonstrated excellent discrimination.