Dear statalisters,
I have a longitudinal (repeated-measures) data collected on patients
with heart failure, starting at the index hospitalization and at their
follow-up. Patients are followed until death or drop-out.
The survival endpoint here is total mortality.
At each visit (index hospitalization, office visits or
rehospitalizations) together with traditional variables, a new
continuous biomarker was collected.
Thus, I am interested in modeling simultaneously the longitudinal and
the time-to-event part of the data using shared parameter models as
described by Tsiatis & Davidian (2004). Here the main goal is to
evaluate
the prognostic value (on mortality) of this biomarker while accounting
for the effect of a other time-dependent covariates (continuous and
discrete) which are assumed being measured with error; secondarily,
to correct for nonrandom dropout (informative censoring) in the
analysis of longitudinal outcomes.
I did explore stcrreg with time-dependent covariates (the pneumonia
example), but it seems it doesn't take into account the covariates'
history until the time when the competing event occurred.
I appreciate if anyone has written a stata code suitable for such type
of analysis.
1) Tsiatis A, Davidian M (2004). \Joint Modeling of Longitudinal and
Time-to-Event Data: An Overview." Statistica Sinica, 14, 809-834.
Thank you in advance,
Eduardo
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/