Abstract
BACKGROUND: Unplanned extubation (UE) after endoscopic retrograde cholangiopancreatography plus endoscopic nasobiliary drainage (ENBD) increases patient morbidity and prolongs hospitalization duration. AIM: To construct a risk prediction model for UE in patients undergoing ENBD to provide evidence for clinical nursing. METHODS: A multicenter retrospective study was conducted, collecting data from 981 patients undergoing ENBD from three hospitals in Chongqing from January 2018 to June 2024, randomly allocated to modeling and validation groups in a 7:3 ratio. Logistic regression analysis was used to screen independent risk factors, construct prediction models, and draw nomograms. RESULTS: The overall incidence of UE was 6.12% (60/981). The majority (70.00%) of extubations occurred within 24-72 h postoperatively. Multivariate logistic regression analysis identified age ≥ 61 years [odds ratio (OR) = 2.341, 95% confidence interval (CI): 1.28-4.27], smoking history (OR = 2.876, 95%CI: 1.54-5.37), prolonged fasting time (OR = 1.124, 95%CI: 1.05-1.20), prolonged catheter duration (OR = 1.286, 95%CI: 1.09-1.52), and consciousness changes (OR = 3.152, 95%CI: 1.69-5.89) were independent risk factors while serum albumin was a protective factor (OR = 0.912, 95%CI: 0.87-0.95). The model receiver operating characteristic area under curve was 0.881 with accuracy of 80.36%, sensitivity of 83.59%, and specificity of 74.88%. A nomogram total score ≥ 199 points corresponded to a high-risk threshold. CONCLUSION: The six-factor risk prediction model had good discrimination and accuracy, which can provide clinical nursing staff with scientific evidence to identify patients at high risk and help reduce the incidence of UE.