From: "Zheng_Jenny" Zheng_Jenny@allergan.com
Subject: [NMusers] Outlier in subject level
Date: Mon, 2 Oct 2006 08:38:08 -0700
Dear NM user,
There has been very interesting discussion on outlier identification/exclusion
at observation level.
What is the general rule to identify/exclude outlier in subject level? For example, I
use FOCEI to fit the pooled data from phase 1 and phase 2. The minimization was successful.
However, the box plot and histogram of the individual CL and V suggested that two subjects
clearly have very different values from the population (typical CL is about 47 L/h but the
CL in these two subjects were 250 L/h and 350 L/h. Should these two subjects be included
in the analysis? I keep the two in the covariate analysis since I do not have good
justification on excluding them. Any comments? Thanks,
Jenny
_______________________________________________________