From Fri May 26 16:23:21 1995
Subject: Data from multiple hospitalizations

I am analyzing a set of observational data in children treated with an antibiotic. In some cases, a child may have measurements from multiple hospitalizations over several years (the antibiotic can only be given in a hospital setting). My current approach is to run a cumulative clock for each hospitalization, but then reset the clock to 0 at the start of each subsequent hospitalization (EVID = 4 on the $INPUT). Does this sound reasonable? Has anyone used another approach to handle data like this? Thanks in advance for your help.

Denise Roe