At 13.41 21/09/2003 +0100, Ian Bowns wrote:
>I suspect you could make a start to explore it. Is it possible to
>construct a trial (probably cluster randomised) where some patients are
>treated for a common condition, where patient variability is
>acknowledged. Some could be treated "by the book" so to speak, others
>by the experience of the clinician. It is increasingly possible to know
Assuming that "the book" is right, the results would be affected by i)As
often the doctors in the experience group would do the same as is in the
book (although for other reasons ), ii) How much better is the "book"
solution as compared to possible experience-based alternatives, and iii)
How often doctors in the "book" group fail to do what is in the book
(either by mistake, lack of time or resources, or because different
patient's preferences)
An easy (even trivial) way of doing this it is to compare the rate of
antibiotic-resistant infections in surgical or haematological wards with or
without a policy restricting the choice of wide spectrum antibiotics. But
then, what would be the use of such a trial?
cheers,
Piersante Sestini
|