The Prediction of Postoperative Delirium Using the Preoperative Assessments of Frailty and Cognitive Impairment in Aged Patients

Purpose: Frailty and cognitive impairment are closely associated with postoperative delirium. The purpose of this study was to compare the ability of screening tools assessing preoperative frailty and cognitive impairment to predict Postoperative delirium (POD) and the association with prevalence of postoperative complications, Intensive Care Unit (ICU) admission, and the hospital length of stay. Patients and Methods: Two hundred and ninety-nine patients aged ≥60 years presenting for elective major thoracic or abdominal surgery were divided into preoperative frailty and no frailty groups or cognitive impairment and no cognitive impairment groups. The primary outcome was the incidence of postoperative delirium. The secondary outcomes included postoperative complications, ICU admission, and hospital lengths of stay. Results: Frailty (25.6% VS 14.6%, P = 0.017) and cognitive impairment (32.7% VS 13.4%, P < 0.001) were associated with POD. However, the area under the receiver operating characteristic curve (AUC-ROC) between frailty (0.657 [95% CI 0.60-0.71]) and cognitive impairment (0.661 [95% CI 0.60-0.71]) for POD was not different (P = 0.9) and both lower than the integrated predictive model of age, body mass index (BMI), American Society of Anesthesiologists (ASA) status, duration of surgery, morphine equivalent, surgical risk, frailty and cognitive impairment (0.814 [95% CI 0.77-0.86], P < 0.0001, P < 0.0001). Besides, frailty (15.6% vs 6.3%, P = 0.010) and cognitive impairment (16.3% vs 8.0%, P = 0.029) was associated with the incidence of postoperative complications. Conclusion: Preoperative frailty and cognitive impairment were associated with POD. However, preoperative frailty or cognitive impairment by themselves were comparably poor predictors of POD. A comprehensive predictive model including age, BMI, ASA status, duration of surgery, morphine equivalent, surgical risk, frailty and cognitive impairment was more useful to predict POD.