Competing events define who remains observable as high risk
Death before the study outcome, additional treatment, and loss of eligibility for resection follow-up can each remove a patient from the population still able to experience the event of interest. In the article indexed as PMID 42407332, the evidence needed to interpret an anomaly-detection label therefore begins with the endpoint and the competing events, not with the label alone. If a competing event is treated as ordinary censoring, the resulting risk set represents patients who remain event-free and competing-event-free; that denominator changes over time and may answer a narrower clinical question than intended. A cause-specific hazard could address the instantaneous event rate among those still in that risk set, whereas cumulative incidence would address the observed probability of the event while acknowledging competing outcomes. Those estimands are not interchangeable, and the publication title does not reveal which was targeted or how changing care pathways were handled. Please report the event definition, each competing event, the estimand, and whether anomaly status was evaluated against cause-specific risk, cumulative incidence, or another time-dependent quantity.
