In article <[log in to unmask]>, James D Woodcock
<[log in to unmask]> writes
>Dear Julia,
> The best method depends on the question you want to answer.
>Pragmatic treatment, in which case all 160 should be analysed, or only
>treating biologically confirmed cases the 111. I find that many papers
>claim to use ITT but in fact use a whole host of reasons for excluding
>people from analysis, some legitimate some less so. Part of the problem is
>that researchers are told ITT is good and so claim to use ITT analysis.
>But in this case if they wanted a trial in biologically confirmed cases
>then they should have randomised people after biological confirmation of
>pathogen.
>
>James Woodcock
>
I agree with most of what you've said, but... Treatment would probably
be started before bacteriological confirmation in order to be effective,
so the former course of action (randomise, and probably analyse, the
160) is probably most appropriate. After all, the clinician using the
results would also be treating based upon suspicion and not confirmed
cases.
--
Ian Bowns
Senior Research Fellow
Health Policy and Management
School of Health and Related Research (ScHARR)
University of Sheffield
Regent Court
30 Regent Street
Sheffield
S1 4DA
Tel; +44 (0)114 2220742
Fax; +44 (0)114 2220798
E-mail; [log in to unmask] (work)
E-mail [log in to unmask] (home)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|