Prediction model construction for postanesthesia care unit retention in patients undergoing laparoscopic cholecystectomy under general anesthesia

Objective To analyze and construct a predictive model for postanesthesia care unit (PACU) retention in patients undergoing laparoscopic cholecystectomy (LC) under general anesthesia, providing reference for the efficient operation of PACU. Methods Patients who underwent general anesthesia LC at Mianzhu People's Hospital from January 2023 to November 2024 were selected as the research subjects. Clinical data of patients before and during operation, and during PACU were collected. Patients were divided into PACU retention group and non-retention group according to whether the PACU stay was longer than 1 hour. Lasso regression and Logistic regression were used to select and construct predictive model and then nomogram were constructed. Using Bootstrap method to resample 1,000 times for internal validation. The discrimination, calibration, clinical applicability, and effectiveness of the prediction model were evaluated using the receiver operating characteristic curve and its area under the curve (AUC), Hosmer-Lemeshow goodness of fit test, calibration curve, clinical decision curve, and clinical impact curve, respectively. Results A totol of 600 patients were included and 104 experienced PACU retention. Through Lasso regression and Logistic regression, K ion concentration, tracheal extubation time, agitation during awakening, and nausea and vomiting during PACU were identified as influence factors for PACU retention. The AUC of the prediction model was 0.803[95%CI(0.755, 0.850)]. The Hosmer-Lemeshow goodness of fit test showed no statistically significant difference (χ2=1.660, P=0.572). Based on Bootstrap resampling 1,000 times, the results showed AUC was 0.803[95%CI(0.756, 0.853)]. The calibration curve showed a basic fit between the predicted probability and the actual occurrence probability. The clinical decision curve showed that within the threshold for predicting probability range of 0.00~0.68, the clinical net benefit of intervening based on the predicted probability of the model was higher than that of not intervening in all individuals and intervening in all individuals. The clinical impact curve showed that when the threshold for predicting probability was greater than 0.30, there was an overlap match between the predicted PACU retention rate and the actual PACU retention rate per 1,000 people. Conclusion K ion concentration, tracheal extubation time, agitation during awakening, and nausea and vomiting during PACU are influencing factors for PACU retention of LC patients. The predictive model constructed based on these factors has good discrimination, calibration, clinical applicability, and effectiveness, and can provide reference for identifying high-risk PACU retention patients after LC surgery.