Development and Validation of a Clinical Prediction Model for Elderly Patients with Preoperative mild cognitive impairment: A Prospective Cohort Study

Abstract

Abstract Background Mild cognitive impairment (MCI) in elderly patients undergoing surgery is neglected easily by clinicians and families. Preoperative patients with MCI are more likely to suffer from postoperative cognitive dysfunction and postoperative delirium, so an effective MCI prediction method has important implications for ameliorating perioperative cognitive function. Objective This study is designed to construct a predictive model to provide a novel approach for preoperative MCI diagnosis in geriatric patients. Methods Patients over 65 years old who underwent elective surgery with general anesthesia were screened. Patients were randomly divided into training cohort (n = 258) and test cohort (n = 49) by the ratio of 8:2, and baseline demographic variables and characteristics of the patients in the different cohort were compared. The least absolute shrinkage and selection operator (LASSO) regression was used to identify risk factors in the training cohort. A nomogram was constructed based on the logistic regression. Receiver operating characteristic (ROC) curves and calibration charts were drawn in the training cohort and test cohort respectively to evaluate the diagnostic value of the prediction model. The decision curve analysis (DCA) was used to value the clinical utility of the prediction model. Results In this study, a total of 307 elderly surgical patients were enrolled, including 137 patients with MCI and 170 patients with normal cognitive function. Multivariate analysis showed that history of more than two operations, higher urea nitrogen, lack of education, body mass index (BMI) < 24kg/m2 and lower albumin/globulin ratio were the independent risk factors for preoperative MCI. The C statistic of the prediction model in the training cohort and test cohort was 0.754 (95%CI, 0.695–0.812) and 0.708 (95%CI, 0.559–0.856) respectively. The threshold probability of the net benefit ranged from 45–81% in the DCA. Conclusions The independent risk factors for preoperative MCI in elderly patients were two or more operations, higher blood urea nitrogen level, shorter years of education, BMI < 24kg/m2, and lower albumin/globulin ratio. The predictive model has a certain diagnostic value for preoperative MCI in elderly patients, and provides a novel method for anesthetists to evaluate preoperative cognitive function in elderly patients.