i knew i should not have put that stuff about nnt's in - the point i was
trying to make was about more n of 1 (thanks trish) and in particular
honesty with patients about the need for 3 or 4 drugs to see which has the
best effect (least side effects).
Dr Martin Dawes
Director Centre for Evidence Based Medicine
University of Oxford
Nuffield Department of Medicine
John Radcliffe Hospital
OXFORD OX3 9DU
----- Original Message -----
From: Toby Lipman <[log in to unmask]>
To: Martin Dawes <[log in to unmask]>
Cc: Mike McDowall <[log in to unmask]>;
<[log in to unmask]>
Sent: Friday, June 23, 2000 12:55 PM
Subject: Re: Populations vs individuals (was Is it worth it?)
> In message <[log in to unmask]>, Martin Dawes
> <[log in to unmask]> writes
> >
> >I see a difference in the population approach of NNT's and NNH's to the
> >strategic apporach to care in the individual. If one has a selection of
> >alternative therapies why don't we try them. This would avoid the problem
of
> >"drug failure" which usually is interpreted as doctor failure when the
1st
> >drug is not succesful. I mean we talk about NNT's of 4 as great. yet that
> >means 3 individuals dont reach the arbitrary threshold.
> >
> It doesn't though does it? Not in real life. If you have a group of 4
> individuals and they are given a drug which produces an absolute risk
> reduction of 25% for a particular event, any number of them from 0 to 4
> could have the event prevented. The probability of any particular number
> having the event prevented will follow the binomial distribution.
>
> No doubt this explains why, even if the ARR is small, if a large number
> of small groups of patients are treated with a drug, then in some groups
> a much higher proportioin of patients will have successful outcomes than
> would be expected from population based data. So in the case of MS you
> are bound to have a few neurologists with, say, 30 or 40 MS patients,
> whose experience is that 70% or 80% have responded to beta interferon.
> These will then extrapolate falsely from their experience and become
> advocates of the drug's use. Of course the majority of neurologists will
> find that their success rate is not too far from the 10% or 15%
> predicted from RCTs and some will find it much lower. I guess it is also
> likely that patients whose treatment is successful might be more likely
> to continue follow up, which will bias the neurologist's opinion in
> favour of the treatment (does anyone know whether this has been
> researched?)
>
> Toby
> --
> Toby Lipman
> General practitioner, Newcastle upon Tyne
> Northern and Yorkshire research training fellow
>
> Tel 0191-2811060 (home), 0191-2437000 (surgery)
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|