Log in
Register|
AngusMcLean ★★ USA, 2014-03-02 19:01 (4068 d 19:06 ago) Posting: # 12546 Views: 10,549 |
|
|
In the Bioequivalence output in Phoenix WinNonlin "Partial tests" are given and so are "sequential tests" in tabs. The tests are described in the Users manual (see link) and below material from the manual. I am wondering which tab one would use in for regulatory report or would one submit both? There are very small differences in the numbers for a study I am working on (one of the subjects did not show up for the last period), whereas for another example of a 3 way crossover study I have (all subjects completed) for that example the partial and sequential data is exactly the same. My current is unbalanced on account of losing a subject. http://screencast.com/t/lNW0Hiy8KG from the manual: Sequential Tests worksheet Angus Edit:Category changed. [Helmut] |
|
ElMaestro ★★★ Denmark, 2014-03-02 19:28 (4068 d 18:39 ago) @ AngusMcLean Posting: # 12548 Views: 9,115 |
|
|
Hello Angus, ❝ In the Bio equivalence output in Phoenix WinNonlin "Partial tests" are given and so are "sequential tests" in tabs. The tests are described in the Users manual (see link) and below material from the manual. I am wondering which tab one would use in for regulatory report or would one submit both? Welcome to the strange world of type III sums of squares (your partial tests). Invented by SAS, not widely agreed but widely used implicitly since this is what SAS just does. The sequential tests are called type I. The residual and treatment effect differences will be the same for both a type I and type III anova, since both are based on the same model. It is the model that determines the residual, not the anova. The only difference will be the SS and MS values, and hence F-tests and p-values, for the fixed factors. You will only see a difference in imbalanced datasets (e.g. when the number of subjects in RT and TR differ). Since the residual and the treatment effect differences are the same for type I and type III, the confidence intervals will be invariant and you can submit one of them or both as you please. It is quite common to submit both (I speculate this is due to SAS habits?). I have no idea why Pharsight thinks any user needs some info relating to QR factorisation when trying to figure out what type III tests are. The type III tests are done by comparing the grand residual to the residual of a model without the factor for which we are testing. That's all there is to it (although sequence is a little tricky due to the crossover but that's a story for another day). — Pass or fail! ElMaestro |
|
AngusMcLean ★★ USA, 2014-03-02 21:40 (4068 d 16:27 ago) @ ElMaestro Posting: # 12549 Views: 9,000 |
|
|
Thank for your remarks: it seems that you have a good knowledge of this bizarre field. SAS is of course a major program here; they are 600 miles from here in North Carolina. I had a small version of the program a while ago. Since the study is not a balanced study, on account of a dropout subject, I will then submit both. Why? The fear is of the FDA asking for additional information; if you only file one partial or sequential then they ask for the other..................delays the review. That is why submitting both is pursued. On that note: Additionally Phoenix WinNonlin provides values for the model-predicted PK parameters (by the regression matrix approach) and then provides observed values and the residuals in a Table. I am not planning on submitting this information. Do you agree? I do not think it is relevant to bio equivalence Angus |
|
ElMaestro ★★★ Denmark, 2014-03-02 22:25 (4068 d 15:43 ago) @ AngusMcLean Posting: # 12550 Views: 9,089 |
|
|
Hi Angus, ❝ Additionally Phoenix WinNonlin provides values for the model-predicted PK parameters (by the regression matrix approach) and then provides observed values and the residuals in a Table. I am not planning on submitting this information. Do you agree? I do not think it is relevant to bio equivalence From the top of my head: Yes, I agree, the residual listings are not essential for FDA submission. But please also check the SAP, it may dictate what a relevant output is, then check your Study Report SOP which may dictate which listings go into your study report. FDA's reviewers will likely re-calculate the dataset to check if they can obtain the reported CIs.
— Pass or fail! ElMaestro |
|
AngusMcLean ★★ USA, 2014-03-04 21:55 (4066 d 16:12 ago) @ ElMaestro Posting: # 12556 Views: 8,832 |
|
|
Thank you I do follow the point about Type 1 and Type II. This is a good point to raise for fundamental understanding. That leads me to question about Least Square Geometric Means. How is that determined then? Is that from "the mode" or the ANOVA. I am thinking it is from the ANOVA? ANGUS |
|
ElMaestro ★★★ Denmark, 2014-03-04 22:42 (4066 d 15:25 ago) @ AngusMcLean Posting: # 12557 Views: 8,949 |
|
|
Hello Angus, ❝ That leads me to question about Least Square Geometric Means. How is that determined then? Is that from "the mode" or the ANOVA. I am thinking it is from the ANOVA? LS Means (it is questionable to call them LS Geometric Means) are determined from the b-vector of the model but only when you construct the model matrix as described above with 1's and 0's and when you opt out of the intercept and fit the factor of interest first; yes, I know, this is a cosmic mindf%cker but the sexy key word is 'contrast coding'.; try and look it up. I never found an explanation of it that made much sense to me, to be honest. The LS Means can also be calculated directly - simple formula given by Chow & Liu. Just think about it this way: For our purposes ls means for factor XYZ are the b-vector estimates that correspond to minimisation of the residual sums of squares. Remember: The b-vector conveys an estimate of all the effects (constants) we work with: treatments, periods, sequences, subjects. Minus those lost to df-considerations. Note also that the anova never determines effects (ls means or otherwise); those just come from the b-vector. A common -but not bad- misconception is also that the residual comes from the anova. It too comes from the model. The anova takes all the variance (all the variability) and tries to tell us where that variability comes from: treatment, sequence, subject, sequence or unaccounted for. Type I and Type III anova's are two different ways of getting the observed variability assigned to the factors. When things are balanced they are identical, by the way. — Pass or fail! ElMaestro |
|
AngusMcLean ★★ USA, 2014-03-05 03:59 (4066 d 10:08 ago) @ ElMaestro Posting: # 12558 Views: 8,908 |
|
|
Thank you! Seems you could write an article on this topic.......mixed effects muddling would indeed be a good title. You had me wondering why we need the ANOVA! I have the book by Chow and Liu. I tried to find least square means and how to calculate it. I have found it to be not useful, since it is too advanced for me. I see a letter Y and it has horizontal stroke above it and then there is a Y with a horizontal stroke above it with Y.11 (only like a subscript 11) and then Y.22 (each of the Y letters has a stroke above it. I do not understand. It is Pharsight that use the term Geo LS Mean: I wonder if this is accurate. Angus |
|
Ohlbe ★★★ France, 2014-03-05 09:18 (4066 d 04:49 ago) @ AngusMcLean Posting: # 12559 Views: 8,744 |
|
|
Dear Angus, ❝ I have the book by Chow and Liu. I tried to find least square means and how to calculate it. I have found it to be not useful, since it is too advanced for me. You may find this post by ElMaestro easier to follow  — Regards Ohlbe |
|
AngusMcLean ★★ USA, 2014-03-05 22:29 (4065 d 15:38 ago) @ Ohlbe Posting: # 12565 Views: 8,950 |
|
|
Thank you....I do find it most useful. I am trying to achieve a better understanding of these calculations and this post shows me the steps and associated problems. Angus |
|
ElMaestro ★★★ Denmark, 2014-03-05 17:02 (4065 d 21:06 ago) @ AngusMcLean Posting: # 12562 Views: 8,790 |
|
|
Hi Angus, ❝ I see a letter Y and it has horizontal stroke above it and then there is a Y with a horizontal stroke above it with Y.11 (only like a subscript 11) and then Y.22 (each of the Y letters has a stroke above it. I do not understand. Yijk is the response (e.g. log AUC) of the i'th subject in the k'th sequence at the j'th period. When you have a 'Y with a horizontal stroke above it with Y.11' then it means the average of observations from the first period in (all) subjects assigned to the first sequence (e.g. TR). — Pass or fail! ElMaestro |
|
AngusMcLean ★★ USA, 2014-03-05 22:31 (4065 d 15:36 ago) @ ElMaestro Posting: # 12566 Views: 8,840 |
|
|
Again many thanks......................I do not know such notations. I wish you would write 101 article on this topic. Angus |
Ing. Helmut Schütz