I have the problem where i am trying to analyse a datset which has both clustered and recurrent events in it. It is important to include both and since some of the data is censored we have chosen PHREG to analyse this. I currently have the following:
ODS GRAPHICS ON;
proc phreg data=A plot(overlay)=survival COVM covs(aggregate);
baseline covariates=REGIMESA out=_null_;
ODS GRAPHICS OFF;
This accounts for ClinicID (so clustered events are included), but not for the variation of an individual that could have up to twelve Gaptime recordings within the study (so in the text above, recurrent events are not included).
I have read and re-read the SAS manual chapters on this and can deal with the two things individually, but not in the same analysis. Ultimately i want to compare the effects of two infection statuses (infected or not) on the gaptime factor of interest and produce a survival curve to illustrate that - but both individual (biological variation and clinic variation need to be included).
Any advice would be greatly appreciated. I am a complete novice with these method, so please be gentle!!
Proc mixed is not appropriate for this data and i have tried proc lifetest, but had a similar problem and could not get the graphical outputs i mention above.