Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles.
All statistics is seldom pure and never simple- to paraphrase Oscar
Wilde. Another cynical 19th century figure, Mark Twain, implied that
statistics was no better than damned lies (1).
Number Needed to Treat would seem to be favoured by clinicans, as it
appears to give a human face to effect size. Unfortuately, Number Needed
to Treat is an ugly nightmare for statisticians.
The mathematical problem is that it is a reciprocal (of the absolute
risk difference). Thus the null hypothesis of no effect for Number Needed
to Treat is set at one divided by zero, which is infinity. The result is
grotesque complications of the confidence intervals (2), and it is thus
not always easy to fit Number Needed to Treat into a meta- analysis.
Translating Odds Ratios into Number Needed to Treat is impossible
without the patient's expected event rate and an algorithm/ table that is
very far from neat and clear (3).
Number Needed to Treat is very much an example of how medicine and
mathematics are incurably uncomfortable bedfellows. The border between
them is like a scar that never properly heals. Medicine and mathematics
are transformed and backtransformed into each other, to and fro, but the
results for the layman can be some sort of alien mess leading one away
from the kingdom of truth- into all sorts of manipulations/lies by lobby
groups and drug companies.
REFERENCES:
(1)NNT- seldom pure and never simple. Harry Hall. Rapid Response.
bmj.com, 27 Aug 2009
(2)Confidence Intervals and the Number Needed to Treat. D.G.Altman.
BMJ 1998;317:1309-12.
(3)A-Z of Medical Statistics. Filomena Pereira- Maxwell. Arnold.
1998. pgs. 49-50.
-As Oscar Wilde said about Truth. There are traps especially when
exposures are of different durations (Suissa, NEJM (2009) 361, 424-5.) I
find it (NNT) very useful as a proxy for the probability of benefit. Thus
an NNT of 80 means you have have only a 1 in 80 chance of benefit.
All Statistics- seldom pure and never simple
All statistics is seldom pure and never simple- to paraphrase Oscar
Wilde. Another cynical 19th century figure, Mark Twain, implied that
statistics was no better than damned lies (1).
Number Needed to Treat would seem to be favoured by clinicans, as it
appears to give a human face to effect size. Unfortuately, Number Needed
to Treat is an ugly nightmare for statisticians.
The mathematical problem is that it is a reciprocal (of the absolute
risk difference). Thus the null hypothesis of no effect for Number Needed
to Treat is set at one divided by zero, which is infinity. The result is
grotesque complications of the confidence intervals (2), and it is thus
not always easy to fit Number Needed to Treat into a meta- analysis.
Translating Odds Ratios into Number Needed to Treat is impossible
without the patient's expected event rate and an algorithm/ table that is
very far from neat and clear (3).
Number Needed to Treat is very much an example of how medicine and
mathematics are incurably uncomfortable bedfellows. The border between
them is like a scar that never properly heals. Medicine and mathematics
are transformed and backtransformed into each other, to and fro, but the
results for the layman can be some sort of alien mess leading one away
from the kingdom of truth- into all sorts of manipulations/lies by lobby
groups and drug companies.
REFERENCES:
(1)NNT- seldom pure and never simple. Harry Hall. Rapid Response.
bmj.com, 27 Aug 2009
(2)Confidence Intervals and the Number Needed to Treat. D.G.Altman.
BMJ 1998;317:1309-12.
(3)A-Z of Medical Statistics. Filomena Pereira- Maxwell. Arnold.
1998. pgs. 49-50.
Competing interests:
None declared
Competing interests: No competing interests