Dear Allstaters,
in a repeated measures analysis, say, of a continuous outcome for a set of
patients observed at different time occasions, let's suppose we have a
time-varying binary predictor as "change of status" (i.e. transplant,
treatment, marital status, etc.) and let's assume, furthermore, that once
the patient has changed his/her status can't go back to the previous one,
then what's the natural meaning of the beta of a predictor like this?
Would be the following model specification (in the same fashion of
time-varying predictors for survival analysis) right?
yij=b0+b1*time+b2*status+eij
i=1 ... N subjects
j=1 ... ni time occasions
or would it be necessary to use one dummy for each time occasion so as to
specify at which follow-up time the change of status occurs?
Any suggestions?
Cheers,
Fabio
Fabio Pellegrini, Biostatistician
Laboratory of Clinical Epidemiology of Diabetes and Cancer,
Department of Clinical Pharmacology and Epidemiology
Consorzio Mario Negri Sud
via Nazionale
66030 Santa Maria Imbaro (Chieti), Italy
Phone: ++39 0872 570262
Fax: ++39 0872 578263
e-mail: <[log in to unmask]>
http://www.negrisud.it
|