Data-Driven Phenotyping Reveals Nonuniform Association Between Age and Mortality After Aortic Surgery: Retrospective Cohort Study of UK Biobank Data.

Background: Life expectancy and age are frequently considered factors to assess perioperative and postoperative mortality risks in patients affected by aortic pathologies, which can affect the decision whether to suggest invasive treatment. Objective: This study aims to investigate the association between age and all-cause mortality after invasive aortic treatment. Methods: Unsupervised clustering (k-means) using data from the UK Biobank was conducted for patients with aortic pathologies (International Classification of Diseases, Tenth Revision [ICD-10] group I71) receiving endovascular or open surgical treatment. Clustering variables encompassed demographic and clinical parameters. Survival analyses (postoperative survival time in days to all-cause death) between clusters and cluster-derived age groups were conducted. Results: The study included 1801 individuals undergoing surgical or endovascular repair for aortic aneurysms. Unsupervised cluster analysis identified distinct groups primarily based on age, both in models using 2 or 3 clusters. Clusters with older patients at surgery exhibited lower postoperative survival, with perioperative mortality disproportionately affecting these groups. While age was significantly associated with postoperative mortality overall (hazard ratio 1.07, 95% CI 1.05-1.08), this association diminished in older clusters after excluding perioperative deaths, a trend confirmed in analyses adjusted for relevant confounders. Conclusions: Unsupervised cluster analysis revealed age as the primary factor distinguishing patient groups undergoing invasive treatment for aortic pathologies. However, age at surgery appears to have different consequences in certain age brackets, indicating a complex nonuniform relationship.