RESEARCH METHODOLOGY POSTER PRESENTATION ABSTRACTS
Please note: All posters will be displayed in
Salons 10-12.
THURSDAY, 5:30-6:15 PM
Aickin M.
Randomization in Early-Phase CAM Trials.
Program in Integrative Medicine, and Department
of Family & Community Medicine, University of Arizona maickin@earthlink.net
Following the very strong advice of allopathic
researchers, CAM/IM researchers have tended to randomize study participants
to treatment groups, even in rather early-phase, very small studies.
A number of reasons will be detailed why this is a weak research strategy.
Most of the arguments favoring randomization are misleading or wrong.
In particular:
1. Randomization does not balance the treatment
groups in small studies, or studies with multiple analysis groups, when
there are a reasonable number of prognostic factors.
2. Randomization is not the only method of balancing unmeasured variables
in large trials; in fact it is one of the least efficient ways of doing
this.
3. It is not appropriate to ignore prognostic factors in the analysis
of a randomized trial, since this creates bias.
4. The fundamental statistical justification for randomization (that
it validates the analysis) is based on several misconceptions.
5. It is not difficult to employ balancing methods that improve on randomization,
and usually quite considerably so..
Some of these facts have been known since Taves'
seminal article published thirty years ago, and others emerge from recent
research. The overall conclusion is that editors, referees, grant reviewers,
and researchers themselves should seriously revisit the issue of randomization
in small studies, and consider whether the above facts do not suggest
that it is out of place. Several alternatives to randomization will
be indicated and resources identified.
Back