Roger
You might find the answer to one of your questions in the graphs on this
page:
www.numberwatch.co.uk/RR&OR.htm
The graphs show how RR and OR vary with the event rates in control and
experimental groups (these rates are called p and q in the graphs). When
you compare the RR and OR graphs you need to take into account the fact
that they have very different scales for their vertical axes.
Michael Power
NHS CKS (Clinical Knowledge Summaries) Service
cks.library.nhs.uk
On Mon, 18 Feb 2008 15:24:38 -0300, Roger Keller <[log in to unmask]>
wrote:
>Hello,
> There is large literature comparing and discussing the use of odds
ratio and prevalence ratio. It may seem quite on old issue today, but I
still have some unclear points to my mind. If someone could help I would
be grateful. Among many other questions, I ask:
>
> - when may a common cross-sectional study (survey) be regard as a fixed
cohort? Is this an easy assumption? If it is a dynamic population, and the
effect measure is then the rate ratio (not the risk ratio), why does the
rare disease assumption need to be met?
>
> - odds ration overestimates the true risk ratio when the backgroud
proportion of disease is high. However, after adjustment by sereval
covariates, the baseline prevalence many times became very low. When
shoud the proportion of disease be rare, before or after adjusment?
>
> Thanks for any comment.
> Roger.
>
>
>___________________________________________
>Roger Keller Celeste, CD, MSc, Guest Researcher
>Centre for Health Equity Studies - CHESS
>Stockholm University/Karolinska Institute
>Sveavägen 160, Sveaplan
>SE-106 91 STOCKHOLM
>
>---------------------------------
>Abra sua conta no Yahoo! Mail, o único sem limite de espaço para
armazenamento!
|