Peter Ellis wrote, in answering Andrew Jull's question:
>...The number needed to treat should remain reasonably constant (if
>patients have similar prognosis etc ). Attaching a utility to the NNH
>based on an individual's perception of that harm may changes the value
>of the action thershold which you describe.
This sounds familiar and Beyesan (and perhaps dating back a few years? eg:
Schwartz WB, Gorry GA, Kassirer JP, et al. Decision Analysis and Clinical
Judgement. AM J MED 1973;55:459-472; Pauker SG, Kassirer JP. The Threshold
Approach to Clinical Decision Making. N ENGL J MED 1980;302(20):1109-1117;
Kassirer JP, Moskowitz AJ, Lau J, et al. Decision Analysis: A Progress
Report. ANN INTERN MED 1987;106:275-91). A matter of balancing benefit
(measured as NNT) against risk (NNH), ideally weighted by perceived
utilities, rather than using NNT under certain circumstances and NNh under
others.
Ben Djulbegovic wrote:
>...The key is of course to use the same units for NNT and NNH. You
>shouldn't combined survival/mortality data with morbidity data or with
>preferences...
I don't understand why these different types of data shouldn't be combined
in comparing benefit (eg: NNT for surviving longer) against risk (eg: NNH
for serious deterioration in quality of hopefully longer life).
David Birnbaum, PhD, MPH
Clinical Assistant Professor
Dept. of Health Care & Epidemiology
University of British Columbia, Canada
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|