Dear List Members:
Can any one explain me why I get different results dealing with different
ways of NNH calculations?
>From the following table:
End point
present absent
treated 20 989
control 13 1013
Using CATmaker, I got:
NNH: -145 (CI95%: -55 to infinite)
Using Arcus Biomedical:
NNH: -140 (CI95%: -52 to infinite)
By hand calculation from Absolute Risk Reduction (1/ARR):
CI95% for the differences between two proportions were calculated following
Gardner & Altman, Statistics with confidence, chapter 4, page 30.
NNH: -140 (CI95%: -52 to -259)
By hand calculation using RR (1/(CER x RRR)) as suggested by Chatellier et
al, BMJ 1996;312:426-9. RR and its CI95% were calculated using EPI-INFO vers
6.03
NNH: -141 (CI95%: -37 to -358)
As you can see two procedures yield an infinite limit and the other two
yield an entire number?. Also, the two hand calculation yield different
limitis?.
I understand that an infinite limit said that probably that adverse event
never happen. But, this is different to say that could happen each x
patients treated, irrespective how big this number is.
Thanks a lot in advance for your help.
Roberto Lede
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|