J Anesth. 2026 May 6. doi: 10.1007/s00540-026-03758-0. Online ahead of print.
ABSTRACT
PURPOSE: To determine the prevalence of disability-free survival (DFS) five years after elective non-cardiac surgery in older adults, and to identify preoperative factors associated with DFS using conventional statistical analyses and machine learning methods.
METHODS: In this prospective cohort study conducted at a single tertiary hospital in Japan, 2878 patients aged ≥55 years who underwent elective non-cardiac surgery under general anesthesia between 2016 and 2018 were enrolled and followed for 5 years. DFS was defined as survival without significant functional disability, assessed using the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0). Multivariable logistic regression and machine learning models were used to identify preoperative predictive factors.
RESULTS: At 5 years after surgery, 80.6% (n = 2321) of the patients achieved DFS. Factors significantly associated with non-DFS included older age (odds ratio [OR]: 1.76), symptomatic cerebrovascular disease (OR: 1.94), low serum albumin (OR: 0.70), poor nutritional status (OR: 0.80), and malignancy (OR: 1.53). The machine learning ensemble model, incorporating support vector machine, neural network, and XGBoost algorithms, achieved a balanced accuracy of 0.69 and an area under the receiver operating characteristic curve (AUC) of 0.76.
CONCLUSION: Multiple preoperative factors were independently associated with long-term DFS after surgery. Machine learning methods demonstrated moderate predictive accuracy, indicating potential clinical utility pending further model refinement.
PMID:42089911 | DOI:10.1007/s00540-026-03758-0