Title: GFDsurv: A flexible toolbox to analyse nonproportional hazards in factorial survival designs
Authors: Marc Ditzhaus - Otto-von-Guericke University Magdeburg (Germany) [presenting]
Dennis Dobler - Vrije Universiteit Amsterdam (Netherlands)
Arnold Janssen - Heinrich-Heine University Duesseldorf (Germany)
Markus Pauly - Technical University of Dortmund (Germany)
Abstract: While the log-rank test and hazard ratios were the gold standard in time-to-event analysis for a long time, there is a recent trend towards alternative methods not relying on the proportional hazard assumption. The reason for this change is violations of the proportional hazard assumption frequently observed in real data, among others, in oncology. To tackle this, we developed different strategies to handle non-proportional hazards in factorial designs and implemented them in the R-package GFDsurv including a user-friendly shiny app. These strategies cover a nonparametric approach and procedures based on novel estimands, such as concordance probabilities, survival medians and, in the near future, restricted mean survival times. We present the methodology behind this flexible toolbox including appropriate resampling strategies for a better small sample size performance. Moreover, its application is illustrated by analysing a recent study on asthma, for which the assumption of proportional hazards is not justifiable.