Development of a nomogram for predicting the probability of postoperative delirium in patients undergoing free flap reconstruction for head and neck cancer

Purpose To develop nomogram for prediction of postoperative delirium (POD) in patients undergoing ablative and reconstruction surgery for head and neck cancer. Methods Total 341 patients were retrospectively analyzed, and clinical variables in preoperative, intraoperative and postoperative periods were compared between delirium group (n = 89) and non-delirium group (n = 252). Multivariate logistic regression, receiver operating characteristics curve, and area under the curve (AUC) were used to generate and test a nomogram, which performance was evaluated by 10-fold cross validation (CV) procedure. Results In univariate and multivariate analysis, age, history of psychiatric disorder, marital status, preoperative numeric rating scale for pain, ASA classification, and ICU stay period were identified as significant risk factors. Using these factors, nomogram for predicting the POD was developed and it showed sensitivity of 61.8%, specificity of 75.4%, PPV of 47.0%, and NPV of 84.8% (Youden's index of 0.372). In 10-fold cross validation set, corresponding values were 44.9%, 84.1%, 50.0% and 81.2% (Youden's index of 0.337). AUC was comparable between two sets (0.7407 and 0.6898). Conclusions Proposed nomogram showed fair discriminative power for POD risk in head and neck cancer patients undergoing major surgery.