Bioequivalence and Bioavailability Forum

Iassou ElMag!

❝ Could you please tell me how can I find the intra-subject variability from Winnonlin? I would like you to tell me the exact steps I have to follow in order to have intra-subject variability result in the Winnonlin output.

In the current versions (WinNonlin5.3 and Phoenix 6.1) for a 2×2 cross-over on log-transformed data you get it in the "Final Variance Parameters"-tab of the results worksheet. Run the example data-set "Data22.PWO"; you should get:
Dependent   Units Parameter              Estimate
Ln(AUClast)       Var(Sequence*Subject)  0.097923
Ln(AUClast)       Var(Residual)          0.053277
Ln(AUClast)       Intersubject CV        0.320745
Ln(AUClast)       Intrasubject CV        0.233928  (≈23.4%)

If you have a higher-order design (e.g., two tests and one reference) you have to calculate it manually from the residual variance. I had some discussions with Pharsight last May; maybe it will be included in the next release of Phoenix (6.2, expected end of March).^*)
CVintra=sqrt(exp(Var(Residual))-1)

If you have a replicate design (at least the reference administered twice) in the first window of the BE wizard select 'Type of Bioeqivalence' [•] Population/Individual (not Average!). In the result worksheet (example data-set "Seq2Per4.pwo") you should get the within-subject sigma as:
SigmaWR 0.3566
Calculate CVWR=sqrt(exp(SigmaWR2)-1)=0.36825
Don't know how to calculate CV_WT; consider asking at Pharsight's Extranet.

---

^*) Still not available in v6.2...