Discussion: View Thread

I have a clinical trial which has repeated measures taken at 3, 6, 9 and 12 months.  There are 3 arms in the trial but two of the arms are identical for the first 3 months.  They differ in what is being given subsequent to the 3 month time point.   The primary endpoint is continuous.  I'd like to use PROC MIXED to analyze but to constrain some of the parameter estimates so that the estimates of the two groups at the 3 month time point are identical.  Is this possible in PROC MIXED or do I have to use something like PROC NLMIXED? 

Thanks.   

-------------------------------------------
Roy Tamura
Associate Professor
University of South Florida
-------------------------------------------

Piece of cake...use lsmestimate to construct orthogonal effects of interest, which can be complex such as 50% weighted combined groups for early time points. Happy to help you code.

-------------------------------------------
Jason T. Machan
Director, Lifespan Biostatistics Core,
Lifespan Hospital System
Research Scientist, Biostatistics, Research
Rhode Island Hospital
Assistant Professor, Departments of Orthopaedics and Surgery
The Warren Alpert Medical School, Brown University
Director Biostatistics Externship, Adjunct Assistant Professor, Department of Psychology
University of Rhode Island
-------------------------------------------

Think carefully through what your goal is. Particularly because if this is to lead to a publishable paper,
the reviewers will ask whether those two groups are equivalent at three months.  You can only test that by
treating them as three separate groups all the way through.  So my recommendation as a consultant and
a reviewer is not to use this approach.  
Ray

-------------------------------------------
Raymond Hoffmann
Professor
Medical College of Wisconsin
-------------------------------------------



Original Message:
Sent: 09-06-2014 07:46
From: Jason Machan
Subject: Estimation Under Constraints in PROC MIXED

Piece of cake...use lsmestimate to construct orthogonal effects of interest, which can be complex such as 50% weighted combined groups for early time points. Happy to help you code.

-------------------------------------------
Jason T. Machan
Director, Lifespan Biostatistics Core,
Lifespan Hospital System
Research Scientist, Biostatistics, Research
Rhode Island Hospital
Assistant Professor, Departments of Orthopaedics and Surgery
The Warren Alpert Medical School, Brown University
Director Biostatistics Externship, Adjunct Assistant Professor, Department of Psychology
University of Rhode Island
-------------------------------------------

Yes...test...visualize...be cautious...but completely reasonable when appropriate.


-------------------------------------------
Jason T. Machan
Director, Lifespan Biostatistics Core,
Lifespan Hospital System
Research Scientist, Biostatistics, Research
Rhode Island Hospital
Assistant Professor, Departments of Orthopaedics and Surgery
The Warren Alpert Medical School, Brown University
Director Biostatistics Externship, Adjunct Assistant Professor, Department of Psychology
University of Rhode Island
-------------------------------------------



Original Message:
Sent: 9/6/2014 10:07:41 AM
From: rhoffmann@mcw.edu
Subject: RE: Estimation Under Constraints in PROC MIXED

Think carefully through what your goal is. Particularly because if this is to lead to a publishable paper, the reviewers will ask whether those two groups are equivalent at three months.  You can only test that by treating them as three separate groups all the way through.  So my recommendation as a consultant and a reviewer is not to use this approach.   Ray ------------------------------------------- Raymond Hoffmann Professor Medical College of Wisconsin -------------------------------------------
Original Message: Sent: 09-06-2014 07:46 From: Jason Machan Subject: Estimation Under Constraints in PROC MIXED Piece of cake...use lsmestimate to construct orthogonal effects of interest, which can be complex such as 50% weighted combined groups for early time points. Happy to help you code. ------------------------------------------- Jason T. Machan Director, Lifespan Biostatistics Core, Lifespan Hospital System Research Scientist, Biostatistics, Research Rhode Island Hospital Assistant Professor, Departments of Orthopaedics and Surgery The Warren Alpert Medical School, Brown University Director Biostatistics Externship, Adjunct Assistant Professor, Department of Psychology University of Rhode Island -------------------------------------------

I have not looked at Proc mixed for some time, but could you implement some hierarchical testing to ascertain whether the more complex model yields an incremental improvement in model fit? ------------------------------------------- David Mangen Owner Mangen Research Associates, Inc. -------------------------------------------
Original Message: Sent: 09-06-2014 15:20 From: Jason Machan Subject: Estimation Under Constraints in PROC MIXED Yes...test...visualize...be cautious...but completely reasonable when appropriate. ------------------------------------------- Jason T. Machan Director, Lifespan Biostatistics Core, Lifespan Hospital System Research Scientist, Biostatistics, Research Rhode Island Hospital Assistant Professor, Departments of Orthopaedics and Surgery The Warren Alpert Medical School, Brown University Director Biostatistics Externship, Adjunct Assistant Professor, Department of Psychology University of Rhode Island -------------------------------------------