Bioequivalence and Bioavailability Forum

Dear both,

seems that you are having fun!

❝ ❝ In my gut feeling there must be some! Remenber our discussion that Kinetica is Number three?

Since Kinetica is widely-used in the Middle East I’m in contact with regulators from Saudi Arabia and Oman, who are exploring the issue since last October. They confirmed the defect imme­di­a­tely. From the information given in the “Thermo Scientific Informatics Support Resource Center” (see this post) I seriously suspected that only the crossover defect was cor­rected (our second paper was not mentioned). Ten days ago I got a response from statisticians of the SFDA (THX!). They com­pared our re­fe­rence data sets in Phoenix 6.3 to the new “Service Release 1” of Kinetica 5.1 (of Dec 24, 2014).

Bingo!
Imbalanced crossovers are OK (PE and 90% CI).
Of course, the test for se­quence effects in cross­overs (here) and the CVs (there) is as flawed as it ever was.
Parallel designs with un­equal group sizes are still wrongly calculated (exactly the “old” results…). ♻

❝ You can look up in ICH E6 that software must perform for the intended purpose (clause 5.5.3.)

Don’t forget Annex III to Procedure for Conducting GCP Inspections requested by the EMEA: Computer Systems (2007), which is a one-pager leading to the very exhaustive PIC/S Guidance: Good Practices for Computerised Systems in Regulated “GXP” Environments (2007)