Background: Postoperative delirium is a common complication that is distressing. This study aimed to demonstrate a prediction model for delirium. Methods: Among 203,374undergoing non-cardiac surgery between January 2011 and June 2019 at Samsung Medical Center, 2,865 (1.4%) were diagnosed with postoperative delirium. After comparing performances of machine learning algorithms, we chose variables for a prediction model based on an extreme gradient boosting algorithm. Using the top five variables, we generated a prediction model for delirium and conducted an external validation. The Kaplan–Meier and Cox survival analyses were used to analyse the difference of delirium occurrence in patients classified as a prediction model. Results: The top five variables selected for the postoperative delirium prediction model were age, operation duration, physical status classification, male sex, and surgical risk. An optimal probability threshold in this model was estimated to be 0.02. The area under the receiver operating characteristic (AUROC) curve was 0.870 with a 95% confidence interval of 0.855–0.885, and the sensitivity and specificity of the model were 0.76 and 0.84, respectively. In an external validation, the AUROC was 0.867 (0.845–0.877). In the survival analysis, delirium occurred more frequently in the group of patients predicted as delirium using an internal validation dataset (p < 0.001). Conclusion: Based on machine learning techniques, we analyzed a prediction model of delirium in patients who underwent non-cardiac surgery. Screening for delirium based on the prediction model could improve postoperative care. The working model is provided online and is available for further verification among other populations. Trial registration: KCT 0006363.

Background: Hyperglycemia has shown a negative association with cognitive dysfunction. We analyzed patients with high preoperative blood glucose level and hemoglobin A1c (HbA1c) level to determine the prevalence of postoperative delirium.Methods: We reviewed a database of 23,532 patients with diabetes who underwent non-cardiac surgery. Acute hyperglycemia was defined as fasting blood glucose > 140 mg/dl or random glucose > 180 mg/dl within 24 h before surgery. Chronic hyperglycemia was defined as HbA1c level above 6.5% within three months before surgery. The incidence of delirium was compared according to the presence of acute and chronic hyperglycemia.Results: Of the 23,532 diabetic patients, 21,585 had available preoperative blood glucose level within 24 h before surgery, and 18,452 patients reported levels indicating acute hyperglycemia. Of the 8,927 patients with available HbA1c level within three months before surgery, 5,522 had levels indicating chronic hyperglycemia. After adjustment with inverse probability weighting, acute hyperglycemia was related to higher incidence of delirium (hazard ratio: 1.33, 95% CI [1.10,1.62], P = 0.004 for delirium) compared with controls without acute hyperglycemia. On the other hand, chronic hyperglycemia did not correlate with postoperative delirium.Conclusions: Preoperative acute hyperglycemia was associated with postoperative delirium, whereas chronic hyperglycemia was not significantly associated with postoperative delirium. Irrespective of chronic hyperglycemia, acute glycemic control in surgical patients could be crucial for preventing postoperative delirium.

This study aimed to evaluate the association between intraoperative blood loss and myocardial injury after non-cardiac surgery (MINS), which is a severe and common postoperative complication. We compared the incidence of MINS based on significant intraoperative bleeding, defined as an absolute hemoglobin level < 7 g/dL, a relative hemoglobin level less than 50% of the preoperative measurement, or need for packed red cell transfusion. We also estimated a threshold for intraoperative hemoglobin level associated with MINS. We stratified a total of 15,926 non-cardiac surgical patients with intraoperative hemoglobin and postoperative cardiac troponin (cTn) measurements according to the occurrence of significant intraoperative bleeding; 13,416 (84.2%) had no significant bleeding while 2,510 (15.8%) did have significant bleeding. After an adjustment with inverse probability weighting, the incidence of MINS was higher in the significant bleeding group (35.2% vs. 16.4%; odds ratio, 1.58; 95% confidence interval, 1.43–1.75; p < 0.001). The threshold of intraoperative hemoglobin associated with MINS was estimated to be 9.9 g/dL with an area under the curve of 0.643. Intraoperative blood loss appeared to be associated with MINS. Further studies are needed to confirm these findings. The cohort was registered before patient enrollment at https://cris.nih.go.kr (KCT0004244).

This study aimed to investigate the association between glucose dysregulation and delirium after non-cardiac surgery. Among a total of 203,787 patients who underwent non-cardiac surgery between January 2011 and June 2019 at our institution, we selected 61,805 with available preoperative blood glucose levels within 24 h before surgery. Patients experiencing glucose dysregulation were divided into three groups: hyperglycemia, hypoglycemia, and both. We compared the incidence of postoperative delirium within 30 days after surgery between exposed and unexposed patients according to the type of glucose dysregulation. The overall incidence of hyperglycemia, hypoglycemia, and both was 5851 (9.5%), 1452 (2.3%), and 145 (0.2%), respectively. The rate of delirium per 100 person-months of the exposed group was higher than that of the unexposed group in all types of glucose dysregulation. After adjustment, the hazard ratios of glucose dysregulation in the development of delirium were 1.35 (95% CI, 1.18–1.56) in hyperglycemia, 1.36 (95% CI, 1.06–1.75) in hypoglycemia, and 3.14 (95% CI, 1.27–7.77) in both. The subgroup analysis showed that exposure to hypoglycemia or both to hypo- and hyperglycemia was not associated with delirium in diabetic patients, but hyperglycemia was consistently associated with postoperative delirium regardless of the presence of diabetes. Preoperative glucose dysregulation was associated with increased risk of delirium after non-cardiac surgery. Our findings may be helpful for preventing postoperative delirium, and further investigations are required to verify the association and mechanisms for the effect we observed.

The cardioprotective effects of volatile anesthetics versus total intravenous anesthesia (TIVA) are controversial, especially in patients undergoing non-cardiac surgery. Using current generation high-sensitivity cardiac troponin (hs-cTn), we aimed to evaluate the effect of anesthetics on the occurrence of myocardial injury after non-cardiac surgery (MINS). From February 2010 to December 2016, 3555 patients without preoperative hs-cTn elevation underwent non-cardiac surgery under general anesthesia. Patients were grouped according to anesthetic agent; 659 patients were classified into a propofol-remifentanil total intravenous anesthesia (TIVA) group, and 2896 patients were classified into a volatile group. To balance the use of remifentanil between groups, a balanced group (n = 1622) was generated with patients who received remifentanil infusion in the volatile group, and two separate comparisons were performed (TIVA vs. volatile and TIVA vs. balanced). The primary outcome was occurrence of MINS, defined as rise of hs-cTn I ≥ 0.04 ng/mL within postoperative 48 hours. The secondary outcomes were 30-day mortality, postoperative acute kidney injury (AKI), and adverse events during hospital stay (mortality, type I myocardial infarction (MI), and new-onset arrhythmia). In propensity-matched analyses, the occurrence of MINS was lower in the TIVA group compared to the volatile group (OR 0.642; 95% CI 0.450–0.914; p = 0.014). However, after balancing the use of remifentanil, there was no difference between groups in the risk of MINS (OR 0.832; 95% CI 0.554–1.251; p-value = 0.377). There were no significant associations between the two groups in type 1 MI, new-onset atrial fibrillation, in-hospital and 30-day mortality before and after balancing the use of remifentanil. However, the incidence of postoperative AKI was lower in the TIVA group (OR 0.362; 95% CI 0.194–0.675; p-value = 0.001). After balancing the use of remifentanil, volatile anesthesia and TIVA showed comparable effects on MINS in patients undergoing non-cardiac surgery without preoperative myocardial injury. Further studies are needed on the benefit of remifentanil infusion.

Background: Machine-learning techniques are useful for creating prediction models in clinical practice. This study aimed to construct a prediction model of postoperative 30-day mortality based on an automatically extracted electronic preoperative evaluation sheet. Methods: We used data from 276,341 consecutive adult patients who underwent non-cardiac surgery between January 2011 and December 2020 at a tertiary center for model development and internal validation, and another dataset from 63,384 patients between January 2011 and October 2021 at another center for external validation. Postoperative 30-day mortality was 0.16%. We developed an extreme gradient boosting (XGB) prediction model using only variables from preoperative evaluation sheets. Results: The model yielded an area under the curve of 0.960 and an area under the precision and recall curve of 0.216, which were 0.932 and 0.122, respectively, in the external validation set. The optimal threshold calculated by Youden’s J statistic had a sensitivity of 0.885 and specificity of 0.914. In an additional analysis with balanced distribution, the model showed a similar predictive value. Conclusion: We presented a machine-learning prediction model for 30-day mortality after non-cardiac surgery using preoperative variables automatically extracted from electronic medical records and validated the model in a multi-center setting. Our model may help clinicians predict postoperative outcomes.