You have done nothing wrong except for interpreting the inverse of the upper
confidence limit of the ARR as the upper limit of the NNT, whereas it is the
lower limit (that what happens when you take an inverse function).
For a visual picture of what is happening try plotting the graph of NNT = 1/ARR
(i.e. y=1/x) and see what happens when the ARR becomes negative !
John Aldous
Consultant in Public Health Medicine
Hillingdon Health Authority
e-mail: [log in to unmask]
____________________Reply Separator____________________
Subject: NNT confidence intervals [MsgID:3CA2]
Author: [log in to unmask]
Date: 09/04/99 17:28
I am hoping for a little statistical guidance with the 95% CI around an
NNT.
Here are the specifics I am working with: The CER is 0.8, the EER is
0.585, so the ARR is 0.215. The CI around the ARR is -0.00256 to 0.433.
So far this makes sense to me. Now, the NNT = 4.65, but taking 1/ARR for
the confidence limits, I get a 95% CI on my NNT of -390 to 2.3. This
doesn't even include my value of 4.65, and doesn't intuitively make sense.
What am I doing wrong?
Julie Brown, MD
Pediatric Emergency Medicine Fellow, box CH-04
Children's Hospital and Regional Medical Center
Seattle, WA 98105
e-mail: [log in to unmask]
fax: 206 527 3892
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|