Prev: Which SAS product am I running?
Next: calculating an exact standardized mortality ratio (SMR)
From: D T on 16 Oct 2009 09:46
I am running a logistic regression with provider
intervention as dependent variable=2C and a range of demographic variables =
as independent
variables. I use domain or subpopulation analysis so that I get separate
analyses for persons with normal weight=2C and persons with above normal we=
ight. I
am wondering about my odds ratio output=2C for example for the =93above nor=
mal
weight group=94=97why would I see the normal/underweight BMI listed with an=
OR
assigned while no one with normal/underweight BMI is in the subpopulation (=
persons
with above normal weight)? I can ascertain that by a simple proc freq (proc=
freq=3B tables bmicats=3B where weightgrp=3D2=3B run=3B). It does not seem=
to have to do with the format I am using either because I
double-checked by assigning a format that would not have a category for
normal/underweight BMI.

Can anyone explain?

=20

Here is my code (simplified):

=20

PROC SURVEYLOGISTIC
data=3Dpatient=3B=20

domain weightgrp=3B

class bmicats (param=3Dref ref=3D'BP above:
<10/5 mm Hg') agegrp (param=3Dref ref=3Dfirst) racegrp (param=3Dref ref=3D'=
White') gender
(param=3Dref ref=3D=92Male=92)=20

condit (param=3Dref ref=3Dlast)/order=3Dinternal=3B

CLUSTER prct =3B=20

MODEL=20
intervt (event=3D'Yes') =3D bmicats agegrp racegrp gender condit /parmlabel=
rsquare
ridging=3Dabsolute=3B=20

strata strata/list=3B

WEIGHT wgt=3B=20

format gender $gender. agegrp agegrp.
racegrp racegrp. bmicats bmicats. condit condit. weightgrp wt. =3B =20

run=3B

Thanks!
D.T.
=20
_________________________________________________________________
Hotmail: Trusted email with Microsoft=92s powerful SPAM protection.
http://clk.atdmt.com/GBL/go/177141664/direct/01/=
 | 
Pages: 1
Prev: Which SAS product am I running?
Next: calculating an exact standardized mortality ratio (SMR)