the way we routinely generate NNTs for continuous data is:
1. decide what constitutes a clinically (or financially) important change
(from the average length-of-stay) in the continuous measure (from 5 days
down to 4.5 days?;down to 4 days?) from the viewpoint of the folks who
will benefit from the evidence (in your example, sounds like both admin
folks and maybe patients [who would want to be sure they weren't hurt by
earlier discharge].
2. that becomes the break-point, and you determine the rates at which
control (CER, say 20%) and experimental (EER, say 55%) patients get
discharged within x days.
3. then EER-CER = an absolute increase (ARI=55% - 20% = 35%) in early
discharge , and the NNT to get one more patient discharged early is 1/ARI
= 1/35% = 3
5. in similar fashion (especially when different consumers of your study
have different ideas about what's a clinically/financially important
change in length of stay), you could generate NNTs for getting one more
patient discharged in 3 days, 2 days, 4.5 days, etc. (but the usual
cautions about data-dependent analyses apply).
hope this helps
cheers,
dls
............................................................................
Prof David L. Sackett
Director, NHS R&D Centre for Evidence-Based Medicine
Consultant in Medicine Editor, Evidence-Based Medicine
Nuffield Department of Medicine, University of Oxford
Level 5, John Radcliffe Hospital, Oxford OX3 9DU, England
Phone: +44-(0)1865-221320 Fax: +44-(0)1865 222901
Email: [log in to unmask] WWW: http://cebm.jr2.ox.ac.uk
As of 1 June 1999: The Kilgore Trout Research
and Conference Centre at Irish Lake,
RR #1, Markdale, Ontario, Canada N0C 1H0
............................................................................
On Thu, 1 Apr 1999, Robert Davis wrote:
> Can one of the participants on this list point us to a good discussion
> on the proper 'framing' of NNT in the context of a continuous outcome
> measure.
>
> For example, a rather bright fellow has posed this question to me:
> Treatment "X" shortens the hospital stay for disease "Z" by 0.50
> hospital days. Is there a relevant way to present this that is roughly
> equivalent to a NNT?
>
> I thought, off hand, that this would be somewhat equivalent to
> suggesting that 2 children would have to be treated to 'save' one
> hospital day. Fundamentally this appears overly simplistic, but I would
> be interested in hearing how others might have dealt with this before.
>
> Thanks in advance,
>
> Robert L. Davis, MD, MPH
> Pediatric Evidence Based Program
> Department of Pediatrics
> University of Washington
> Seattle, WA , USA
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|