Hi friends,
How to treat protocol violators, i.e. subjects being accidentally treated
in another group than assigned randomly or by specified characteristics.
Mainly three possible handlings have been suggested:
1. leave them in the original ITT group, at least for efficacy, not for safety.
More such occasions, if happening randomly, may average out.
2. put them in the actual AST group, if by accident and no further problems.
3. remove them from the analysis, at least for efficacy.
Bias may occur with any of these "solutions".
These suggestions will be summarized / tabulated by responder below:
Responder ? ITT ? AST ? exclude ? remarks ?
??????????????????????????????????????????????????????????????????????????
Robert Newcombe ? generally ? if random ? ? power ?
? ? mix-up ? ? reduction ?
??????????????????????????????????????????????????????????????????????????
Laura Gray ? ITT analysis? PP analysis ? ? as choosen ?
??????????????????????????????????????????????????????????????????????????
John Whittington ? strict, ? common sense? ? analyze both?
? conservative? most ? ? ways, prescr?
? approach ? realistic ? ? vs taken ?
??????????????????????????????????????????????????????????????????????????
[known to me] ? if minor ? if minor ? if major ? as sponsor ?
??????????????????????????????????????????????????????????????????????????
Martin Bland ? justified ? ? ? comparable ?
??????????????????????????????????????????????????????????????????????????
Jochen Weber ? efficacy ? safety ? ? intention ?
??????????????????????????????????????????????????????????????????????????
John Hughes ? efficacy ? safety ? ? ?
??????????????????????????????????????????????????????????????????????????
Graham Wetherill ? practical, ? unrelated ? no treatment? in protocol ?
? conservative? to group; ? incomplete, ? and/or SAP; ?
? if not ? sensitivity,? withdrawn ? consider if ?
? foreseen ? safety ? ? more often ?
??????????????????????????????????????????????????????????????????????????
Sincere thanks to all of you for your worthwhile contributions!
For those of you who are interested in it I have made all individual
contributions available in a single text file, which you can view at:
http:/home.hccnet.nl/jim.groeneveld/protocol/deviations.txt
In our case we have randomized sequences (groups) in different categories.
Subjects (pts) belong to these categories dependent on specifications.
A few subjects were laying on the borderlines between categories.
They have been classified in one category, while later one it appeared
that they actually would belong in another one. In their first category
they now do not meet the inclusion criteria and should best be excluded.
They can not be moved to another category because treatments differ between
categories. So we have decided to exclude these subjects, at least for efficacy.
Regards - Jim.
--
. . . . . . . . . . . . . . . . . . . . . .
Jim Groeneveld, MSc.
Biostatistician
Science Team
Vitatron B.V.
Meander 1051
6825 MJ Arnhem
P.O.Box 5227
6802 EE Arnhem
Tel: +31/0 26 376 7365
Fax: +31/0 26 376 7305
[log in to unmask]
www.vitatron.com
[usual disclaimer]
|