From: Sriram Krishnaswami <>
Subject: Inter occasion variability
Date: Sat, 26 Feb 2000 13:18:17 -0500

Dear NMusers

I am trying to PK/PD model a rich data set comprising 14 subjects. I basically have the PK profile, the baseline PD profile (endogenous cortisol) over 24 hours (because cortisol exhibits circadian rhythm) and the drug induced PD (Suppressed cortisol levels) profile over 24 hours. I am currently doing the two-stage procedure of fitting each individual using a well established PK/PD model and finally hope to use means and SDs to make some sense out of them.

But, I find that in some cases (actually half of the patients), the drug induced levels of cortisol are higher than the baseline levels. The study was part of a lenghty crossover design with patients taking placebo over 5 days followed by 3 weeks washout and then followed by the drug treatment over 5 days. We are wondering if this is due to inter occasion variability.

Would this be a candidate to evaluate inter-occasion variability? If so , please suggest a way to do this. If my understanding is right, one usually does inter-occasion variability if the same conditions are repeated on a different occasion. In our case, in one occasion we have the placebo effect and in the other occasion we have the drug effect.

Thanks in advance






From: "Stephen Duffull" <>
Subject: RE: Inter occasion variability
Date: Mon, 28 Feb 2000 09:08:05 +1000

Hi Sriram

As I see your problem you could only add IOV to parameters that are common between the placebo and active dose model (if I have read your description properly). For example at the begining of each study period you will presumably model the concentration of cortisol (ie E0) which could then have IOV on it to account for random variability in the baseline cortisol conc. Hempel et al [Clin Pharmacol Ther 1998;64:622-35] did something similar with 24 hour BP control.


Stephen Duffull
School of Pharmacy
University of Queensland
Brisbane, QLD 4072
Ph +61 7 3365 8808
Fax +61 7 3365 1688