Isla Roberts writes:
> has anyone carried out an epidemiology of RCT's on any
>subject?
I'm not sure what you mean by epidemiology of RCTs (randomized control
trials) but a recent issue of the British Medical Journal (October 31) had a
whole series of articles reviewing the RCT. One that I chose for
presentation at a journal club is
"Content and quality of 2000 controlled trials in schizophrenia over 50
years" Ben Thornley and Clive Adams BMJ 1998; 317: 1181-1184.
I think a good subtitle for it would be "You don't have to be crazy to
publish research about schizophrenia, but it helps."
The authors point out the grossly inadequate sample sizes in most of these
studies. The average study had only 65 patients total, and only 3% of the
trials had adequate power according to the authors' criteria.
Even worse, there was absolutely no consistency in how the papers measured
outcomes. There were 640 different instruments used, and 369 were used only
once. I feel very sorry for anyone who has to do a meta-analysis on this
mess.
The quality of the reporting was terrible. Only 4% of the studies provided
sufficient information about the methods of allocation, and only 22%
provided explicit information about blinding.
With these sorts of results, I wonder how successful Evidence Based Medicine
can be.
For what it's worth, this is not a surprising result. What these authors
describe about their field is also true for many other research fields.
Every review I have ever read about the quality of published research has
been rather depressing.
Steve Simon, [log in to unmask], Standard Disclaimer.
Ask Professor Mean: http://www.cmh.edu/stats/profmean.htm
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|