Dear allstaters,
in a clinical trial with a quality of life measurement (SF-36) as the
primary endpoint with an expected death rate of 5-10%, how to analyse the
ITT population? How to include the missing values due to death? Consider
them as "failure" - the worst possible quality of life score? Actually, I
planned to analyse the score as normally distributed. (Maybe I should not
do so? - But otherwise I must plan a RCT with clusters with non-normal
maybe even with a nonparametric endpoint?)
What would YOU do?
Best regards,
Steffen
---
Dr. Steffen Witte
Abteilung Medizinische Biometrie & SDGC
INF 305, D-69120 Heidelberg
T: +49-6221-56-4371, F: +49-6221-56-4195
mailto:[log in to unmask]
www.biometrie.uni-heidelberg.de
|