Dear all
I fear mailing to this list. You are all so bloomin clever! However, I
have a problem which i am sure that somebody could sort me out with.
Simply, can I estimate PAR% from survey data?Specifically, can I
estimate the PAR% of obesity for diabetes in a survey population
(using the odds ratios from a logistic regression model).
I have seen this done in a number of papers and have not liked it. The
method, as described in Armitage and Berry and the Epidemiology text
by Hennekens refers to cohort and case control studies, with Relative
Risks not Odds Ratios.
If it is ligitimate to use ORs as proxies for RRs, I would like to do
so. In this case, I am obviously dealing with a more *prevalent
conditions* (type 2 diabetes and hypertension) rather than incident
ones (food poisoning or fractured femur).
Please be gentle with me, if this is a really stupid question.
Regards
Terry
*********************************************
Dr Terry Aspray
Wellcome Laboratories
Royal Victoria Infirmary
Newcastle upon Tyne
NE1 4LP
Great Britain
tel +44 (0)191 222 5407 fax +44 (0)191 261 1763
*********************************************
Dr Terry Aspray
Wellcome Laboratories
Royal Victoria Infirmary
Newcastle upon Tyne
NE1 4LP
Great Britain
tel +44 (0)191 222 5407 fax +44 (0)191 261 1763
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|