Dear Michael
Paul is correct
As an alternative, you might email the authors to determine if:
1. They calculated the LR and just didn't report it
2. They might be willing to share enough data to allow you to do it
yourself
Bill
William D. Grant, Ed.D.
Associate Dean, Graduate Medical Education
Executive Vice Chair, Research Professor, Family Medicine
Director, Center for Evidence Based Practice
Department of Familiy Medicine
SUNY Upstate Medical University
475 Irving Ave #200
Syracuse, NY 13210
315-464-6997 (p) 315-464-6982 (f)
STATEMENT OF CONFIDENTIALITY
The information contained in this electronic message and any attachments to
the message are intended for the exclusive use of the addressee(s) and may
contain confidential or privileged information that is protected by law. If
you are not the intended recipient, please notify the sender at (315)
464-6997 or at [log in to unmask], and destroy all copies of this message
and any attachments. This communication may not be forwarded without the
express written permission of the original author.
>>> Paul Glasziou <[log in to unmask]> 5/25/2004
10:34:47 AM >>>
Dear Michael,
No - generally you can't get to LRs from the OR without additional info.
The relationship is:
OR = LR+ve / LR-ve
or
log OR = Log LR+ve - Log LR-ve
So the OR is "split" between the two LRs, but without more info you can't
know where the split has been made,
Cheers
Paul Glasziou
At 25/05/2004, Michael Gross wrote:
>Danesh et al published their analysis of the association of C-reactive
>protein and C.A.D. (NEJM 2004: 350;14 pp1387-97) and found it to be
>less strong that a previous series of studies and subsequent published
>quidelines. They suggest that CRP may not be a useful measurement even
>in intermediate risk people. I hoped to calculate the likelihood ratio
>from their charts so that I could apply it to the standard Framingham
>risk tables to obtain post-test probability and see for myself.
>However, their statistical analysis is only expressed in odds ratios
>after adjustments for coronary risk factors, etc. Is there any way to
>convert these OR to LR? Thanks....
> Michael Gross, MD, MPH
> Lycoming Medical Associates
> Hughesville, PA, USA
Paul Glasziou
Department of Primary Health Care &
Director, Centre for Evidence-Based Practice, Oxford
ph: 44-1865-227055 www.cebm.net
|