There is no simple yes or no answer to the question of whether
observational studies are as reliable as RCTs. Examples can be found
where they have given similar results. In other cases, such as the
recent HRT example as well as antioxidants in the diet, they have not.
The problem is that one doesn't know which situation one is in until the
RCTs are done.
In some situations, in which all the important factors and variables are
known, proper matching or data adjustments should be possible. But many
situations are too complicated and poorly understood to know all the
important factors and variables. Again, how does one know which
situation applies?
If observational data is all there is, then one has to use it, but with
a grain of salt. But if RCTs are possible, then observational data is
no substitute for large, generalizable RCTs. Even if they give the same
result in the end, how does one know that until the RCT is done?
David L. Doggett, Ph.D.
Senior Medical Research Analyst
Health Technology Assessment and Information Service
ECRI, a non-profit health services research organization
5200 Butler Pike
Plymouth Meeting, PA, USA 19462
Phone: (610) 825-6000 x5509, FAX (610) 825-6834
e-mail [log in to unmask]
|