I am planning the analysis of a randomized, 2-arm, parallel-group treatment

study with serial measurements for the the continuous outcome variable of

interest. Let's assume that we have treatment and 2 other fixed factors, and

that there will be a (pre-randomization) baseline assessment (Visit 0) and

10 follow-up assessments at monthly intervals (Visits 1 - 10) for the

dependent variable.

Normally, I would specify a MMRM (using the Mixed procedure) that includes

the post-baseline measurements of the outcome of interest as the dependent

variable (i. e., I would include a Time variable with visits 1 through 10)

and use the baseline value of the dependent variable as a covariate. There

may, however, be randomized subjects who will drop out before the first

post-baseline assessment of the dependent variable. For regulatory reasons,

I will nevertheless have to include these patients into the model at least

in a sensitivity analysis.

My idea was now to omit the baseline value of the dependent variable as a

covariate, and to let the trajectories start at Visit 0 instead of at Visit

1. This would, however, imply that the treatment effect still needs to be

modeled starting at Visit 1, i. e., without the treatment group difference

at baseline (vVsit 0).

How do I specify this in SPSS Mixed?

Thank you very much,

Andreas

--

Sent from:

http://spssx-discussion.1045642.n5.nabble.com/=====================

To manage your subscription to SPSSX-L, send a message to

[hidden email] (not to SPSSX-L), with no body text except the

command. To leave the list, send the command

SIGNOFF SPSSX-L

For a list of commands to manage subscriptions, send the command

INFO REFCARD