I guess they refer to the proper use of prevalence ratio and odds ratio as
estimates of the risk ratio (RR) in cross-sectional studies. With
cross-sectional data cs will result in a prevalence ratio, while cc will
result in an odds ratio. Whether you use cs or cc will depend on the nature
of your outcome (it's your call, not Stata's). For outcome with a high
prevalence, the odds ratio will overestimate the effect of the exposure, and
the prevalence ratio may be more appropriate. However, the prevalence ratio
will tend to underestimate the risk ratio.
Let P=Prevalence
Pe=prevalence in exposed
Po=Prevalence in no exposed
De=Average duration of disease in exposed
Do=Average duration of disease in non-exposed
I=Incidence (risk)
PR=Prevalence ratio
P = (I x D) x (1-P)
PR = Pe / Po
PR = (Ie / Io) x (De / Do) x [(1-Pe) / (1-Po)]
= IRR x (De / Do) x [(1-Pe) / (1-Po)]
If (De / Do) ≠ 1 then you will have survival bias
[(1-Pe) / (1-Po)] is the "Prevalence complement bias"
Since in most cases Pe > Po, then the "prevalence complement bias
will be less than 1, and the PR will underestimate the RR. The magnitud of
the bias will also depend on the magnitud of the true PR. The larger the
true PR, the greater the bias.
Leonelo Bautista
University of Wisconsin, Madison
Population Health Sciences
-----Original Message-----
From: [email protected] [mailto:owner-statalist@hsphsun2.
harvard.edu] On Behalf Of Ward Hagar
Sent: Friday, April 23, 2004 10:32 AM
To: [email protected]
Subject: st: cc vs cs
The reference manual and online help documentation states that cs is used
for cohort studes and, "in some cases, cross-sectional studies." The cc test
is for case-control and cross sectional studies.
I looked throughout the reference manual, but could not determine what the
"some cases" were for usnig "cs" over "cc" for cross-sectional data.
Help!
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
