Dear Andrew,
The difference between Peto's OR and normal OR is probably two-fold:
1). OR = p1(1-p0)/p0(1-p1)and Peto's OR = (O-E)/v;
2). SE(lnOR) = 1/a + 1/b + 1/c + 1/d (Woolf's method, for example) and
Peto's SE(lnOR) = square root of 1/v.
The Peto's method was developed from Mantel-Haenzel's method (1959) and was
firstly applied by Yusuf and Peto (1985). It is a likelihood estimate of OR
by calculating the difference between observed events(O) and expected events
(E), then divided by the variance (v) of this difference.
If you have to chose one of them for your analysis, Peto's method is
recommended, particularly when the sample size is small, as in such
condition the Peto's method would produce:
(a) a better estimate for the relative risk;
(b) a comparatively narrow confidence interval.
Hope this help
Weiya Zhang
***********************************************************
Dr W Y Zhang
Centre for Evidence-Based Pharmacotherapy
Pharmaceutical Sciences
Aston University
Aston Triangle
Birmingham B4 7ET
UK
Tel: +44 121 359 3611 ext 5535
Fax: +44 121 359 0733
E-mail: [log in to unmask]
Web: http://www.aston.ac.uk/pharmacy/cebp/
************************************************************
----- Original Message -----
From: Andrew Jull <[log in to unmask]>
To: <[log in to unmask]>
Sent: 10 December 1999 02:51
Subject: Peto ORs
> Dear all
>
> can anybody inform me what the difference between a Peto OR and normal OR
is
> and why one might use one over the other.
>
> Andrew Jull
> Clinical Nurse Consultant
> Auckland Hospital
> Private Bag 92024
> Auckland
> NEW ZEALAND
> Phone: +64 9 3797440
> Fax:+64 9 3072818 (external)
> 7718 (internal)
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|