Silva
☆    

Portugal,
2016-07-28 11:13
(2800 d 07:58 ago)

Posting: # 16523
Views: 5,474
 

 Using only one sequence [General Sta­tis­tics]

Dear all

In a crossover study using only one sequence (RT for ex.) does it make sense to include a sequence effect in the ANOVA? And what about the subject nested within sequence?

Thanks in advance
ElMaestro
★★★

Denmark,
2016-07-28 11:51
(2800 d 07:20 ago)

@ Silva
Posting: # 16524
Views: 4,714
 

 Using only one sequence

Hi Silva,

❝ In a crossover study using only one sequence (RT for ex.) does it make sense to include a sequence effect in the ANOVA? And what about the subject nested within sequence?


There is no sequence effect here because Sequence only has one level.
Period and Treatment will have the same meaning; if you include one then addition of the other does not make sense. Looks like a normal linear model with subject and treatment as factors.

But are you sure this is a design you really want to use???

On the positive side regulators will probably not flag you for inspection on grounds of worry about distribution of the randomisation code :-D

Pass or fail!
ElMaestro
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2016-07-28 13:09
(2800 d 06:02 ago)

@ ElMaestro
Posting: # 16525
Views: 4,802
 

 Paired design

Hi ElMaestro & Silva,

❝ ❝ In a crossover study using only one sequence (RT for ex.) […]



This is not a crossover but a paired design.

❝ There is no sequence effect here because Sequence only has one level.

❝ Period and Treatment will have the same meaning;


Right.

❝ But are you sure this is a design you really want to use???


The only application of a paired design I’m aware of is to assess whether PK is linear by comparing AUC0–τ (steady state = T) with AUC0–∞ (single dose = R). Nobody would perform such a study as a crossover since the logistics would be terrible. ElMaestro correctly stated

❝ Period and Treatment will have the same meaning;


Hence, it should be unambiguously stated in the protocol that the model assumes (!) no period effect, i.e., the difference is related solely to the treatment. Contrary to a crossover where period effects mean out (unless sequences are extremely unbalanced) in a paired design great care must be taken to avoid a true period effect – which would confound the treatment effect.
Stupid example: No air conditioning in the clinics and a heat wave during SD. ;-)

@Silva: Is this was you are planning?

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
Silva
☆    

Portugal,
2016-07-28 14:17
(2800 d 04:54 ago)

@ Helmut
Posting: # 16526
Views: 4,652
 

 Paired design

❝ @Silva: Is this was you are planning?


Many thanks to both El Maestro and Helmut for the rapid input.

@Helmut, my plans were actually to breath some enriched air! Maybe I could become smarter!! But for now I need to have this question clarified in my mind!

According to PI, it is a "dose finding" study for a new route of adm. PI pretends to start with the Ref injection dose and then study the other route. All subjects will perform all study treatments (all planned doses), unless safety concerns. It is an exploratory study, not a true bioequivalence study.

I agree with you that it is a "paired study" and not a common "crossover study". Facing dosing administration procedure, what would be the ANOVA model: Treatment (dose normalized parameters)+Subject?


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post #5! [Helmut]
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2016-07-28 14:47
(2800 d 04:24 ago)

@ Silva
Posting: # 16528
Views: 4,722
 

 Dose escalation

Hi Silva,

❝ @Helmut, my plans were actually to breath some enriched air! Maybe I could become smarter!!


Nitrox in Porto Santo?
I love this stuff when diving in “special” conditions. :party:Headaches from Vinho Verde Tinto gone in minutes. Beware of the “virtual bottom” by pO2!

❝ According to PI, it is a "dose finding" study for a new route of adm. PI pretends to start with the Ref injection dose and then study the other route. All subjects will perform all study treatments (all planned doses), unless safety concerns. It is an exploratory study, not a true bioequivalence study.


OK, makes sense. Haven’t thought of it. Dose escalation studies can’t be done in a crossover.

❝ Facing dosing administration procedure, what would be the ANOVA model: Treatment (dose normalized parameters)+Subject?


Yep. Then the PEs will give you the Fabs for each dose level.


PS: Can you send me your mobile No (in case I’ll be arrested for misbehavior in my forthcoming vacation and need help). ;-)

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
DavidManteigas
★    

Portugal,
2016-08-01 13:17
(2796 d 05:54 ago)

@ Silva
Posting: # 16532
Views: 4,465
 

 Using only one sequence

If the purpose of the study is to assess the bioequivalence between different dose formulations, wouldn't in that case a true cross-over design more appropriate instead of a single-sequence (not cross-over) design? I'm thinking of carry-over effects in this case. Imagine that one specific dose of the test formulation has pk properties you don't antecipate and all the subjects have pre-dose concentrations for the next dose. How would you handle that? Remove the entire cohort for that dose? :-D

I believe the ANOVA model with subject as random effect as you proposed is ok. For this kind of design, I'm used to linear mixed models (random intercept & slope models). Nevertheless, I'm not sure if this is adequate for a bioequivalence study since you don't pretend to study a trend but to test for equivalence in each "period" or "dose level". Since it is a dose escalation study maybe you should present alpha-adjusted CI for bioequivalence between doses.
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,638 registered users;
77 visitors (0 registered, 77 guests [including 9 identified bots]).
Forum time: 18:12 CET (Europe/Vienna)

Nothing shows a lack of mathematical education more
than an overly precise calculation.    Carl Friedrich Gauß

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5