When trials use different outcome measures meta-analysis sometimes combine
these into an effect size. This is a measure of the effect in variations
from the mean rather than number of deaths, pain etc. The result is often
presented as x% on treatment y did better than the average result on
treatment z. I think that although this provides a statistically accurate
measure of the likelihood of an effect it is of very little clinical use,
because it is very hard to interpret what the result means. I was wondering
if list members thought effect sizes shouldn't be used or if there was a
case for them if the outcomes are relatively similar (e.g. two pain scales)
or if there were better ways of interpreting the results?
James Woodcock
________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|