2 Comments
User's avatar
oncologist's avatar

Great stuff! Have a question though. If the rate of early informative censoring is significant but equal between the arms, while the REASON for censoring is different (early dropout for toxicity in experimental and "disappointment" in control arm) would your method of the "censoring area" have a particular pattern?

Expand full comment
Timothée Olivier's avatar

Thanks, great question, the pattern does not indicate the reason. However you can identify areas where it's reasonable to make various assumptions (even though informative censoring can occur outside that "area")

Reasons matter a lot, I agree, you might be familiar with this work : https://www.sciencedirect.com/science/article/pii/S0959804924001187

Expand full comment