I am trying to do the partitioned survival plot for a Q-TwisT analysis. I am not sure what I should use as censoring when calculating the Toxicity state as the survival curve is the time that the patients suffered adverse events before relapse. It is not the common time to event, it is how long a patient does not feel right because of the treatment. What I saw in another project is that the censoring they used is the same censoring that progression free survival curve, but we would be ignoring the adverse events of patients with adverse events without progression of the disease. Should the censoring be whether the patient had or not AE before progression or before the end of follow-up?.