Dear Colleagues,
we are involved in an equivalence trial in a clinical research
project and have to analyse the data. The primary endpoint is the
amount of an analgesic drug, measured in [mg]. The amount is the sum
of the taken analgesic drug in one day. Sometimes not in the whole
range of time the taken drugs could be captured - some information is
missing, whereat the occurence is probably at random. The measured
sum of analgesics is therefore just a lower bound of that what was
taken. The question is, how to handle these missing data in the
equivalence trial. Does it make sense to use imputation techniques,
and if yes, which method will be a correct one (for example
imputation of the means)? We would be grateful if you could give us
some hints or proposals for solution.
Best wishes
Andreas Deckert
Dipl.-Inform. Med. Andreas Deckert
Institut fuer Medizinische Biometrie und Informatik
Universitaet Heidelberg
INF 305
69120 Heidelberg
TEL: +49 (0) 6221 56 4149
FAX: +49 (0) 6221 56 4195
email: [log in to unmask]
http://www.biometrie.uni-heidelberg.de
|