All Times EDT
Keywords: estimands, censoring, time to event, noumenal, phenomenal, intercurrent event
In oncology clinical trials with time-to-event endpoints, censoring rules have traditionally been defined and applied following standard approaches based on longstanding regulatory guidelines such as the 2007 FDA Cancer Endpoint Guidance. The estimands framework’s mandate to precisely define the clinical question of interest, with attention to the handling of intercurrent events, requires a careful discussion and precise definition of how censoring rules are applied, and alignment with study design and conduct. We propose distinguishing two concepts. Noumenal censoring approaches addresses intercurrent events traditionally handled by censoring which influence the clinical question itself, i.e. define the pre-specified estimand. Phenomenal censoring addresses events that the estimand framework characterizes as missing data, such as administrative censoring (i.e. defines missing data handling in the estimate). Strategies for dealing with the most relevant intercurrent events are discussed in the context of this framework as well as the impact on practical consideration like study design and data collection. These considerations also apply to defining relevant and interpretable sensitivity and supplementary analyses. Strategies and options to address design or context limitations are discussed. In some cases, long-established methods may need reconsideration. The authors recommend defining trial-specific censoring rules as well as considering when noumenal and phenomenal approaches are to be used.