Log in
Register|
jag009 ★★★ NJ, 2013-05-17 00:15 (4378 d 01:08 ago) Posting: # 10595 Views: 7,688 |
|
|
Hi everyone, I am back with some more interesting data from a 3-period 2-treatment SD partial replicate study. Out of academic curiosity, I analyzed the data (T vs 2xR) using the HVD SAS code from FDA's progesterone guidance to compute both RASBE (Proc GLM) and ABE (Proc Mixed) for 3 parameters. Here are the results for ln(AUCt). I also (back) computed the intra-subject CV ABE's 90% CI using PowerTost (Thanks Helmut  ).RASBE (Within sub SD of R > 0.294) T/R Ratio 95% Upper Bnd Within-Sub SD of R Within-Sub var of RABE T/R Ratio% 90% CI Intra-sub CV%Anyone smells something fishy here? sWR is greater than the RASBE criteria and yet intra-CV from ABE is less than 30% Here is the data. (Sequence 1 = ABB, 2=BAB, 3=BBA) Order = Subject, Sequence, Period, Treatment, ln(AUCt) 1 BAB 2 A 8.526364803Thanks John Edit: Category changed. [Helmut] |
|
Helmut ★★★   Vienna, Austria, 2013-05-17 04:03 (4377 d 21:20 ago) @ jag009 Posting: # 10596 Views: 6,180 |
|
|
Hi John! ❝ RASBE (Within sub SD of R > 0.294) PHX6.3 ![[image]](img/uploaded/image28.png) Note that from ilat you not only get the PE but also the 90% CI:102.11–119.55from s²WR → CVWR 39.24%❝ ABE ❝ ❝ PE and CI. I get a s²WR of 0.139466 (PHX’ terminology: Var(Period*Formulation*Subject)_21) → CVWR 38.69%Have a look at s²WT. Do you believe in this value? I don’t. Stupid design. ❝ I also (back) computed the intra-subject CV ABE's 90% CI using PowerTost 100*CVfromCI(lower=1.0184, upper=1.1914, n=52, design="2x3x3", robust=TRUE)100*CVfromCI(lower=1.0211, upper=1.1955, n=52, design="2x3x3") ❝ Anyone smells something fishy here? Strange. From the ABE model I get a CVWR which is pretty close to the one from RSABE. I think we have to wait for Detlew returning from his vacation in three weeks. BTW, update PowerTOST to the latest version 1.1-03 (published 2013-05-03). See the NEWS. — Dif-tor heh smusma 🖖🏼 Довге життя Україна! ![[image]](https://static.bebac.at/pics/Blue_and_yellow_ribbon_UA.png) Helmut Schütz ![[image]](https://static.bebac.at/img/CC by.png) The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
d_labes ★★★ Berlin, Germany, 2013-05-20 19:28 (4374 d 05:55 ago) @ Helmut Posting: # 10606 Views: 5,295 |
|
|
Boys! From my vacation: What is your problem ?CVfromCI() gives you the CV for the difference T vs. R.This is some pooled value of the intra-subject variances of T and R 1). So don't expect to get a value comparable to s2wR! 1) R.J. McNally Tests for Individual and Population Bioequivalence Using 3-Period Crossover Designs online here. gives in the context of appropriate intra-subject contrasts (aka progesterone guidance): s2I=(s2D + s2wT + s2wR/2)where s2I is the variance of the difference T-R used for calculating the 90% CI, s2D is the subject-by-formulation interaction, s2wT and s2wR are the intra-subject variabilities of Test or Reference, respectively. Since you didn't provide me s2wT try it by your own assuming s2D=0 which in many cases could be reasonably assumed for the partial replicate design. Remember our discussion on specifying compound symmetry in the FDA Proc Mixed code. — Regards, Detlew |
|
Helmut ★★★ Vienna, Austria, 2013-05-20 20:17 (4374 d 05:06 ago) @ d_labes Posting: # 10607 Views: 5,283 |
|
|
Daddy! ❝ From my vacation: Do I smell some signs and symptoms of forum-addiction? ❝ ❝ This is some pooled value of the intra-subject variances of T and R 1). Yep. ❝ So don't expect to get a value comparable to s2wR! OK, OK. ❝ ❝ where s2I is the variance of the difference T-R used for calculating the 90% CI, s2D is the subject-by-formulation interaction, s2wT and s2wR are the intra-subject variabilities of Test or Reference, respectively. ❝ ❝ Since you didn't provide me s2wT try it by your own assuming s2D=0 which in many cases could be reasonably assumed for the partial replicate design. ❝ Remember our discussion on specifying compound symmetry in the FDA Proc Mixed code. Yessir. PHX gives me a standard error of the difference (FDA’s ABE code) of 0.0468069; s²WR 0.139466 and s²WT 0.0153728 [sic]. Not negligible S×F of 0.166035. As we know SAS will spit out different values for s²WT and the S×F. Stupid enough with John’s data no warning. If I ignore the S×F I get 29.805% for the pooled CVW – close to PowerTOST.For completeness results of different parameterizations of the variance structure in PHX: s²WR s²WT (= FDA’s)Compound Symmetry 0.139466 0.0833300Enjoy your vacation!  — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
jag009 ★★★ NJ, 2013-06-04 18:04 (4359 d 07:19 ago) @ d_labes Posting: # 10726 Views: 5,056 |
|
|
Hi Detlew, Let me irritate you a bit this morning   ❝ Boys! ❝ ❝ From my vacation: What is your problem ❝ ❝ ❝ This is some pooled value of the intra-subject variances of T and R 1). ❝ So don't expect to get a value comparable to s2wR! If the Intra-CV derived from the pool'd variance is higher/lower than the intra-CV derived from the S2wR, then could one suggest that the difference is attributed to the test formulation (meaning test has larger or smaller variability)? See some interesting data below. I ran Proc Mixed (FDA progesterone HVD SAS code) to obtain the reference residual variance and computed the corresponding intra-cv. I then used R to back-calculate the intra-cv from the 90% CI generated by Proc Mixed (first column data). Formulation 1 Thanks John |
Ing. Helmut Schütz