This sounds like a task for logistic regression using the confounder and
the risk factor. If you want to see if there's effect modification, use
the product of the risk factor and confounder. You may want to
categorize these variables.
Tony
Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Visintainer,
Paul
Sent: Monday, April 21, 2008 12:01 PM
To: [email protected]
Subject: RE: st: trend in ORs across ordered levels of a 3rd variable
Joseph,
Thanks for your input. But I don't think -epitab- addresses this
question. The output you provided gives the trend in ORs "adjusting"
for the confounder. What I wanted to know is whether we can detect a
linear pattern of the ORs over levels of the confounder (which, to me,
looks like a specific type of interaction)
Another example: suppose I want to know whether there is a difference
in the risk (odds) of death between males and females from trauma.
Suppose my third variable is level of consciousness (ordinal variable
measured at 4 levels). Say, my output shows that as level of
consciousness decreases, the OR for gender and death increases: (e.g.,
ORs at each level of consciousness: 1.0 at level 1, 1.5 at level 2, 1.9
at level 3, and 2.3 at level four), which suggests that men do worse at
lower levels of consciousness.
I suppose that one way to address this is to approach it as if
consciousness were a continuous variable, then look at the slopes for
consciousness in logit models run separately for men and women.
I can't think of any other approach.
-p
______________________________________
Paul F. Visintainer, PhD
School of Public Health
New York Medical College
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Joseph
Coveney
Sent: Saturday, April 19, 2008 3:21 AM
To: Statalist
Subject: Re: st: trend in ORs across ordered levels of a 3rd variable
Paul Visintainer wrote:
Is there an approach to analyzing the trend in odds ratios across the
ordered levels of a 3rd variable? For example,
Suppose I have the risk of obesity in high school students by gender
over three different grades:
Grade OR
10 1.5
11 1.9
12 2.2
There is a test of homogeneity to determine whether these ORs differ
across grade strata. Is there a test to determine whether the pattern
is linear across strata?
------------------------------------------------------------------------
--------
Are you looking for something other than -tabodds-?
Joseph Coveney
. webuse bdesop
. tabodds case alcohol [fweight = freq], or
------------------------------------------------------------------------
---
alcohol | Odds Ratio chi2 P>chi2 [95% Conf.
Interval]
-------------+----------------------------------------------------------
---
0-39 | 1.000000 . . .
.
40-79 | 3.565271 32.70 0.0000 2.237981
5.679744
80-119 | 7.802616 75.03 0.0000 4.497054
13.537932
120+ | 27.225705 160.41 0.0000 12.507808
59.262107
------------------------------------------------------------------------
---
Test of homogeneity (equal odds): chi2(3) = 158.79
Pr>chi2 = 0.0000
Score test for trend of odds: chi2(1) = 152.97
Pr>chi2 = 0.0000
*
* 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/
