EHC model

From: alison@c255.ucsf.EDU (ABoeckmann) Subject: EHC model Date: Tue, 13 Oct 1998 09:50:57 -0700 (PDT) Last week I sent some suggestions for modelling EHC. The code in $PK looked like this: IF (NEWIND.LT.2) THEN EHYN=0 OLDTIM=0 ENDIF IF (OLDTIM.EQ.ALAG5) EHYN=1 IF (OLDTIM.EQ.ALAG6) EHYN=0 OLDTIM=DOSTIM This works with population data. With population data, NEWIND is 2 except for the first event record of each individual record. However, it does not work with single-subject data. It might be imagined that NEWIND is 2 with all event records of a single-subject data set except with the first event record. However, remembering that (with either population or single-subject data) NEWIND is 1 with the first event record of each individual record, and that an individual record is any contiguous set of data records with a common ID value, NEWIND can be 1 with a number of event records of the data set, and not just with the first event record. OLDTIM will be set to 0 with every record after the first observation, and EHYN will never be set to 1. The test with single-subject data should be: IF (NEWIND.LT.1) THEN which tests for the first event record of the data set. Now the EHYN flag is set properly. It is unfortunate that the code I wrote, using NEWIND, must be different for population vs. single-subject data. It would be better to test some item in the data record. With some data sets, one can use TIME. The test is: IF (TIME.EQ.0) THEN etc. This test works well for both population and single-subject data if each subject's data consist of dose events at TIME=0 followed only by observation events. With other data sets, more complicated code is needed.