- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Hello,
I'm seeking assistance with extracting odds ratios from the "Analysis of GEE Parameter Estimates" table in SAS. Specifically, I need to generate a table that presents each parameter alongside its corresponding log odds estimate, and lower and upper 95% confidence limits, converted into odds ratios.
While I'm aware that this could be done using an LS Means statement, I'm looking for an alternative method that could facilitate this process. Perhaps there's a way to utilize proc tabulate
or proc report
to create this table without manual calculation?
Any guidance on this would be greatly appreciated.
Thank you
Here's an example of the table (proc genmod output) that I would like to retrieve data from.
.
Accepted Solutions
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
So ... you are fitting GEE model with PROC GENMOD, ... fine to me.
(just be aware there's also a PROC GEE)
LSMEANS statement can contain all model effects you are interested in.
The syntax is :
-
LSMEANS <model-effects> </ options>;
See here for an example:
Home > Analytics > Stat Procs > How to calculate adjusted odds ratios and 95% CIs for GEE using proc genmod?
https://communities.sas.com/t5/Statistical-Procedures/How-to-calculate-adjusted-odds-ratios-and-95-C...
If you use "at" - option behind forward slash
-
-
lsmeans A / at means; lsmeans A / at x1=1.2; lsmeans A / at (x1 x2)=(1.2 0.3);
-
, then you need multiple obviously.
W.r.t. LSMESTIMATE statement ... Yes, you need one LSMESTIMATE statement per custom hypothesis test among the least squares means.
See here :
- PharmaSUG2011 -- Paper SP01
CONTRAST and ESTIMATE Statements Made Easy: The LSMESTIMATE Statement
Kathleen Kiernan, Randy Tobias, Phil Gibbs, and Jill Tao; SAS Institute Inc., Cary, NC
https://www.lexjansen.com/pharmasug/2011/SAS/PharmaSUG-2011-SAS-SP01.pdf - Using and Understanding LSMEANS and LSMESTIMATE
David J. Pasta, ICON, San Francisco, CA
https://www.lexjansen.com/wuss/2014/103_Final_Paper_PDF.pdf
(from WUSS conference ; Western Users of SAS® Software (WUSS))
Koen
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Are you using PROC GEE ??
I don't think an alternative method would make it easier.
It would make it more "manual" than LSMESTIMATE statement.
It's quite easy to do with :
- lsmeans statement
- lsmestimate statement
- oddsratio statement (maybe not available in PROC GEE)
Usage Note 42728: Producing odds ratios for logistic models in the GENMOD or GEE procedure
https://support.sas.com/kb/42/728.html
Usage Note 41516: Testing hypotheses and estimating odds ratios within and across specific logits in LINK=GLOGIT models
https://support.sas.com/kb/41/516.html
(The above usage note also applies when fitting the generalized logit model to repeated measures data using PROC GEE)
Koen
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Thank you for your response.
I'm using the Proc Genmod procedure.
Please correct me if I'm wrong but the LSMeans and LSMestimate need to be written for each parameter, correct? Is there a way to do this for all parameter included in the model without writing a LSMeans line for each parameter of interest?
Also, is there a resource you have discusses the placement of the values "1 -1 0" etc. for when you're working with a categorical variable that has more than two groups?
Thank you
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
So ... you are fitting GEE model with PROC GENMOD, ... fine to me.
(just be aware there's also a PROC GEE)
LSMEANS statement can contain all model effects you are interested in.
The syntax is :
-
LSMEANS <model-effects> </ options>;
See here for an example:
Home > Analytics > Stat Procs > How to calculate adjusted odds ratios and 95% CIs for GEE using proc genmod?
https://communities.sas.com/t5/Statistical-Procedures/How-to-calculate-adjusted-odds-ratios-and-95-C...
If you use "at" - option behind forward slash
-
-
lsmeans A / at means; lsmeans A / at x1=1.2; lsmeans A / at (x1 x2)=(1.2 0.3);
-
, then you need multiple obviously.
W.r.t. LSMESTIMATE statement ... Yes, you need one LSMESTIMATE statement per custom hypothesis test among the least squares means.
See here :
- PharmaSUG2011 -- Paper SP01
CONTRAST and ESTIMATE Statements Made Easy: The LSMESTIMATE Statement
Kathleen Kiernan, Randy Tobias, Phil Gibbs, and Jill Tao; SAS Institute Inc., Cary, NC
https://www.lexjansen.com/pharmasug/2011/SAS/PharmaSUG-2011-SAS-SP01.pdf - Using and Understanding LSMEANS and LSMESTIMATE
David J. Pasta, ICON, San Francisco, CA
https://www.lexjansen.com/wuss/2014/103_Final_Paper_PDF.pdf
(from WUSS conference ; Western Users of SAS® Software (WUSS))
Koen