B1148
Title: Considerations in defining estimands for clinical trials of complex disease processes
Authors: Richard Cook - University of Waterloo (Canada) [presenting]
Alexandra Buhler - University of Waterloo (Canada)
Jerald Lawless - University of Waterloo (Canada)
Abstract: Assessing new interventions in clinical trials can be challenging when disease processes are complex, and no single outcome is sufficient to characterize the response to treatment. Examples arise when several, possibly competing, events can occur during follow-up or when intercurrent events arise, which makes simple estimands more difficult to interpret. The importance of supportive secondary analyses is highlighted for gaining insights into treatment effects in such settings. It stresses that causal inferences have relevance to the real-world use of treatments. To conclude, some cautionary remarks about artificial censoring and the use of inverse probability weighting and some recommendations are provided.
