Analysis of Risk Factors for Delayed Discharge from the Post-Anesthesia Care Unit in Patients Undergoing General Anesthesia and Establishment and Validation of a Predictive Model: A Retrospective Case-Control Study

Background : Delayed discharge from the post-anesthesia care unit (PACU) after general anesthesia is a common complication in clinical anesthesia, resulting from the combined effect of multiple risk factors. It compromises the quality of postoperative recovery while diminishing the efficiency of perioperative turnover. Our study attempts to determine the risk factors for delayed PACU discharge and to create and validate a nomogram predictive model. Methods : A total of 746 patients with delayed PACU discharge after general anesthesia were enrolled. Using a 1:1 matching design (consistent gender and age ±2 years), 746 eligible patients without delayed discharge were selected as controls. Both the delayed and non-delayed discharge groups were split into a training set (n=1046) and a test set (n=446) at a 7:3 ratio. Logistic regression analysis was performed in the training set to develop a risk prediction model, which was then validated in the test set. The discriminative ability, model calibration, and clinical utility were assessed via receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA), correspondingly.The goodness-of-fit for the calibration curves was determined using the Hosmer-Lemeshow (HL) test. Results : A predictive nomogram model was developed using eight significant variables identified through multivariate logistic regression analysis: Body Mass Index (BMI), rocuronium dosage, Fasting blood glucose (FBG) , blood loss, PACU agitation, PACU pain, PACU hypoxemia, and PACU blood gas analysis. For the training set, the area under the ROC curve (AUC) was 0.888 (95% confidence interval [CI]: 0.868–0.908), and the corresponding value in the test set was 0.887 (95% CI: 0.856–0.918). Calibration curves indicated a high degree of agreement between predicted probabilities and actual probabilities. The P-values of the HL test in the training set and test set were 0.53 and 0.15, respectively, indicating good goodness-of-fit. DCA demonstrated that when the predicted probability exceeded 10%, using this model to predict delayed PACU discharge and implement intervention measures would yield greater benefits. Conclusion : This study created and validated a predictive model to estimate the likelihood of delayed PACU discharge in patients following general anesthesia.