A nomogram for predicting sedation-related adverse events in elderly patients undergoing painless gastrointestinal endoscopy

Abstract Background Elderly patients undergoing painless gastrointestinal endoscopy are at increased risk for sedation-related adverse events (SRAEs) because of their greater physiological vulnerability and higher likelihood of comorbidities. Risk stratification before endoscopy may improve perioperative safety and individualize sedation and management. Objective This study aimed to develop and validate a model based on machine learning to predict risk of SRAEs in elderly undergoing painless gastrointestinal endoscopy. Methods Prospective data from 520 patients at least 60 years old who underwent painless gastrointestinal endoscopy between April 2023 and June 2024 at our medical center were randomly divided into a training set (n = 364) and validation set (n = 156). SRAEs were defined as intraoperative hypotension or hypoxemia, and independent predictors of SRAEs in the training set were identified through multivariate logistic regression. A nomogram to predict risk of SRAEs was developed using R software and tested against the validation set. Its performance was assessed in terms of receiver operating characteristic curves, calibration plots, and decision curve analysis. Results In the training set, hypotension occurred in 39.0% patients and hypoxemia in 33.5%, and independent predictors were older age, history of snoring, frailty, preexisting hypertension, chronic obstructive pulmonary disease, prolonged fasting before the procedure, and higher initial dose of etomidate-propofol. Conversely, regular physical activity was a protective factor. The nomogram built from the training set discriminated between people in the validation set who experienced SRAEs or not with an area under the curve of 0.95 (95%CI 0.92–0.99), it showed good calibration in the Hosmer–Lemeshow test (P = 0.405), and decision curve analysis demonstrated clinical utility across a wide range of threshold probabilities. Conclusion A predictive model based on readily available clinical variables can accurately estimate SRAE risk in elderly patients undergoing painless gastrointestinal endoscopy. The model may be useful for individualizing sedation and patient management. Trial registration Chinese Clinical Trial Registry ChiCTR2300069816. Registered on 23 March 2023.