25gmds079 10.3205/25gmds079 urn:nbn:de:0183-25gmds0798 Meeting Abstract Alidan Alidan Duoerkongjiang D

Universitätsklinikum Hamburg-Eppendorf, Hamburg, Germany

author Ozga Ozga Ann-Kathrin AK

Institut für Medizinische Biometrie und Epidemiologie, Universitätsklinikum Hamburg-Eppendorf, Hamburg, Germany

author German Medical Science GMS Publishing House

Düsseldorf

610 20251103 engl This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). M0631 079 Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie 70. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS) V: Medical Biometry 2: Analysemethoden Jena 20250907 20250911 Abstr. 33 TextIntroduction: In clinical trials, time-to-event analyses are essential for evaluating treatment efficacy and understanding patient outcomes. Traditional methods such as the Cox proportional hazards model or Kaplan-Meier estimation typically focus on time-to-first events. However, this approach may overlook recurrent and competing events, which are common in chronic disease settings and can provide critical insights into disease progression and treatment impact.Methods: We present a systematic overview of statistical methods for multiple time-to-event outcomes, including recurrent and terminal events. These methods include the weighted hazard ratio by Rauch, Wei-Lachin approach, Andersen-Gill (AG) model, Wei-Lin-Weissfeld model, Prentice-Williams-Peterson (PWP) models, joint frailty models, Ghosh and Lin’s approach, and prioritized composite outcome analyses (e.g., win ratio, win odds). We applied these methods to data from a randomized, controlled cardiovascular trial.Results: Each method yielded consistent but nuanced differences in estimating treatment effects. For instance, prioritized outcomes (e.g., win ratio) highlighted early benefits in non-fatal events, while joint models better captured long-term competing risks. The AG and PWP models offered different event-order interpretations. These findings demonstrate that model choice significantly influences interpretation, especially regarding the clinical prioritization of events.Discussion: Our results underscore that no single model is universally optimal. Instead, each method answers a slightly different clinical question. We discuss practical implementation considerations, the interpretability of effect measures, and model assumptions in the context of RCTs.Conclusion: Moving beyond the time-to-first-event paradigm allows for richer, more clinically relevant analyses. This work aims to support applied researchers in selecting and interpreting appropriate methods for multiple-event survival data in randomized trials.The authors declare that they have no competing interests.The authors declare that a positive ethics committee vote has been obtained. 0 0 0 0