Re: comparing theta's

From: orjan.nordle@activebiotech.com Date: Mon, 29 Mar 1999 9:59:36 +0200 Subject: Re: comparing theta's Peter, Pascal and Lewis answered the question if there were a significant difference between the three thetas. However, your question was if the thetas were the same, i.e. an equivalence question. I haven't seen so much about equivalence together with NONMEM but I think that it may be useful, especially in the toxicokinetic studies. If we define equivalence between two thetas as "a difference less than 20%" then we have to reverse the hypotheses: H0: d<0.8 or d>1.25 H1: .8<d<1.25 where TH(2)=TH(1)*d or d=TH(2)/TH(1) If we can reject H0 then H1 is accepted and can conclude equivalence. How can we do that in NONMEM? My suggestion is to replace TH(2) with TH(1)*.8 in one run and TH(1)*1.25 in one run. If the objective function from both runs is more then three??? units larger than the value achieved when both TH(1) and TH(2) are estimated, then H0 is rejected and H1 accepted. Equivalence are proved! NONMEM users: is this an acceptable method? Orjan Nordle Active Biotech Research AB, Lund, Sweden