clustered survival analysis with interval censored data

[Peter_Y]

Hello:

  I recently encounter couple of datasets with specific structures and I wonder if anyone knows how to fit them with SAS. Suppose we give patients one of the two treatments and follow-up their next 3 hospital visits. The goal is to model time from baseline to first hospital visit and time between each follow-up visits. The 2 distinct features of this data are

1. Data is clustered. i.e patients within each hospital. Hospitals within provinces......

2. Patient will skip the visits. This is the part I get confused. If a patient skip visit 2, time between visit 1 and visit 2 (t12) is interval censored. Similaly, t23 is interval censored as well. But there is a constraint: t12 + t23 is fixed. How should I take into account the dependency in between-visit times ?

This kind of data seems to be very common in practice but I don't know the appropriate way to analyze it. Any help would be appreciated.

Thanks,

Peter