Bioequivalence and Bioavailability Forum

Mahesh ☆ India, 2013-08-06 08:12 (4274 d 12:30 ago) Posting: # 11218 Views: 27,452	Acceptance by FDA/CDER [Two-Stage / GS Designs] Post reply
	Hi, Is two stage or Add-on BE design accepted by USFDA/CDER? Thak you, Mahesh Edit: Category and subject line changed. [Helmut]

ElMaestro ★★★ Denmark, 2013-08-06 12:05 (4274 d 08:38 ago) @ Mahesh Posting: # 11220 Views: 25,937	Two stage BE design Post reply
	Hello Mahesh, yes, in practice I think you should have highest chances of the FDA nod if you apply Potvin's method C. — Pass or fail! ElMaestro

Helmut
★★★

Vienna, Austria,
2013-08-06 17:31
(4274 d 03:12 ago)

@ Mahesh
Posting: # 11226
Views: 25,969

Two stage BE design

Post reply

Hi Mahesh,

❝ Is two stage …

Yes.

❝ … or Add-on BE design accepted by USFDA/CDER?

No (and never was – contrary to Canada and Japan).

As ElMaestro already pointed out FDA prefers variants of ‘Method C’.

FDA’s Bioequivalence Recommendations for Specific Products
- Draft Guidance on Loteprednol Etabonate; Tobramycin (06/2012)
- Draft Guidance on Lotepredenol Etabonate (04/2013)
- Interesting that the Draft Guidance on Dexamethasone/Tobramycin (06/2012) mentioned a sequential design as well, but the current revision (06/2013) states “The number of subjects to be enrolled should be determined from the results of a pilot study.” Why?
- There might be other guidances published in the meantime; please check yourself.
Davit BM. Sequential Designs and Interim Analyses in Bioequivalence: FDA’s Experience. Mini-Symposium on Adaptive Study Designs and Assessment Approaches for Bioequivalence. AAPS Annual Meeting, Chicago, IL, October 13–18, 2012.
Potvin D et al. Sequential design approaches for bioequivalence studies with crossover designs Pharm Stat. 2008;7(4):245–62. doi 10.1002/pst.294
Montague TH et al. Additional results for ‘Sequential design approaches for bio-equivalence studies with crossover designs’. Pharma Stat. 2011;11(1):8–13. doi 10.1002/pst.483
Fuglsang A.Sequential Bioequivalence Trial Designs with Increased Power and Controlled Type I Error Rates. AAPS J. 2013;15(3):659–61. doi 10.1208/s12248-013-9475-5

—
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

Mahesh ☆ India, 2013-08-07 08:51 (4273 d 11:51 ago) @ Helmut Posting: # 11229 Views: 25,764	Two stage BE design Post reply
	Thank you very much Helmut and ElMaestro for your kind suggestions.

kumarnaidu ★ Mumbai, India, 2013-10-17 17:08 (4202 d 03:35 ago) @ Mahesh Posting: # 11675 Views: 25,532	Two stage BE design Post reply
	Hi all, Can anybody tell me for using group sequential design which method (Potvin's C or B) should be used for EMEA. — Kumar Naidu

ElMaestro
★★★

Denmark,
2013-10-17 19:28
(4202 d 01:15 ago)

@ kumarnaidu
Posting: # 11679
Views: 25,697

Two stage BE design

Post reply

Hi Kumarnaidu,

❝ Can anybody tell me for using group sequential design which method (Potvin's C or B) should be used for EMEA.

I am receiving a lot of noise on the transmission frequencies that EU regulators are using. Sc. advice that I have experienced include the following feedback:

"You must use Potvin B."
"You must use Potvin C."
"You must use PROC GLM and not Potvin B" (luckily I had some Schützomycin tablets in my pocket and immediately swallowed a handfull. Otherwise the meeting could have ended with a mental breakdown or with a manslaughter.)

I think method B is the current EU winner due to alpha=0.0294 being used at both steps.

—
Pass or fail!
ElMaestro

kumarnaidu ★ Mumbai, India, 2013-10-18 07:59 (4201 d 12:43 ago) @ ElMaestro Posting: # 11681 Views: 25,540	Two stage BE design Post reply
	Hi ❝ I think method B is the current EU winner due to alpha=0.0294 being used at both steps. That means if I am adopting method C EU regulators will not accept or may raise serious questions on the design. In Canadian guideline they have recommended method C. Is there any recommendations for two stage design in EU guidelines like Canada. Please reply — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-10-18 16:42
(4201 d 04:00 ago)

@ kumarnaidu
Posting: # 11700
Views: 25,563

EMA vs. rest of the world

Post reply

Hi Kumar,

❝ That means if I am adopting method C EU regulators will not accept or may raise serious questions on the design.

Yes, that will be possible. See the case studies in this presentation (slides 43–45). EMA* obviously believes to be very clever – which awaits proof.

❝ In Canadian guideline they have recommended method C.

Correct – following recommendations of the authors. Method C is also recommended by the FDA. Not surprising, since this method was developed by the the non-profit organization PQRI (members: FDA/CDER, Health Canada, USP, AAPS, CHPA, PhRMA,…). FDA’s Donald Schuirmann is one of the authors.

❝ Is there any recommendations for two stage design in EU guidelines like Canada.

Not that I recall. When it comes to two-stage designs the GL is crap. See also Shuanghe’s exeriences and my reply.

To be correct some members states’ authorities. I had never [sic] any problems with Germany’s BfArM, for example.

—
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

Shuanghe
★★

Spain,
2013-10-18 12:05
(4201 d 08:38 ago)

@ ElMaestro
Posting: # 11692
Views: 25,805

Two stage BE design — cold water thrown on our head

Post reply

Hi guys,

❝ I think method B is the current EU winner due to alpha=0.0294 being used at both steps.

Well, if the feedback I just got is official, you would be wrong — All methods with prefix "Potvin" are losers, in EMA region I mean.

Here's some input from certain agency with regard to 2-stage design I posted a year ago that the methods are "... not valid in Europe".

and a few weeks back we asked again for one of our projects, given the fact that more studies were presented with Potvin's method. Helmut alone got quite a few. :cool:

Below are loose translations:

Q: "..., we'd like to ask about the current status of 2-stage design BE study, ... if the BE protocol with Potvin's Method C is acceptable now ..."

A: "Potvin's methods are not acceptable in EMA."

:confused:

so, first question, how could it be? Have some of you guy's dossiers with Potvin's method(s) been approved/accepted (no deficient letter, rejection letter etc)?

Secondly, it got me double check the guideline. I always had the impression that method B is practically mentioned in the guide. Not quite. :no:

There's no mention of interim power analysis in the BE guide while method B... you know. So I guess the power analysis is probably the "evil" here causing all the fuss...

There's rumour that some working party is currently discussing this topic. Anyone has any insight?

—
All the best,
Shuanghe

Helmut
★★★

Vienna, Austria,
2013-10-18 15:29
(4201 d 05:14 ago)

@ Shuanghe
Posting: # 11695
Views: 25,485

Open your mouth and forcefully spit the water out

Post reply

Hi Shuanghe,

❝ Q: "..., we'd like to ask about the current status of 2-stage design BE study, ... if the BE protocol with Potvin's Method C is acceptable now ..."

❝

❝ A: "Potvin's methods are not acceptable in EMA."

Simply mad. And I don’t mean the MAD.
Have you asked them which methods would be acceptable according to their impeccable wisdom instead? They should enlighten us simpletons.

❝ so, first question, how could it be? Have some of you guy's dossiers with Potvin's method(s) been approved/accepted (no deficient letter, rejection letter etc)?

Yes. For some recent (this year’s) experiences see slide 45 of this presentation. Method C [sic] accepted by Germany (RMS) and CMSs Austria, Denmark, Sweden, and The Netherlands. Spain asked for an all-fixed effects analysis (subjects within sequence were random in the original analysis).
Note that in a previous case (slide 44) MEB didn’t like intermediate power.

❝ I always had the impression that method B is practically mentioned in the guide. Not quite. :no: There's no mention of interim power analysis in the BE guide while method B... you know. So I guess the power analysis is probably the "evil" here causing all the fuss...

Maybe. I suspect the guys writing the GL failed to comprehend the published methods, which – with no exception! – depend on an intermediate power estimation step. I hope someone has the balls and goes for a referral. If a GL (not a law anyway) states 2×2=5 I would wholeheartedly ignore it for sure.

BTW, the additional term introduced in the Q&A Ref.7 is futile as well. I have heard that this section was excreted by EMA’s Biostatistical Working Party and the PK Working Party is just “hosting” it. What the heck? A paper “On the Statistical Model of the Two-Stage Design in Bioequivalence Assessment” by Karalis/Macheras is currently under review (J Pharm Pharmacol). From their key findings:

The overall performance in terms of % BE acceptance is identical. The additional term, ‘Sequence × Stage’, suggested in the EMA method is in most cases non-significant. The same results were obtained regardless of the type (fixed or random) of the effect applied to the ‘Subjects’ term.

And the conclusions:

Any BE study either finished or in progress which relies on the existing literature methodology leads to the same % BE acceptance as if it was analyzed with the recently proposed EMA method.

So what?

❝ There's rumour that some working party is currently discussing this topic. Anyone has any insight?

Not me.

—
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

kumarnaidu
★

Mumbai, India,
2013-10-19 09:52
(4200 d 10:51 ago)

@ Helmut
Posting: # 11708
Views: 25,399

Change in statistical design

Post reply

Hi all, Thank you for your valuable comments.
We have already included Potvin's method C in our statistical analysis plan in the protocol. According to our client we should change our design to normal 2x2 crossover design. Now the problem is we have dosed volunteers and I think CRO has started bioanalyis part (but not statistical analysis). In this situation can we ammend the protocol or we need to repeat the study with ammended protocol.

—
Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-10-19 17:04
(4200 d 03:39 ago)

@ kumarnaidu
Posting: # 11710
Views: 25,344

Bad idea

Post reply

Hi Kumar,

❝ We have already included Potvin's method C in our statistical analysis plan in the protocol. According to our client we should change our design to normal 2x2 crossover design.

The study’s power will drop – in some cases substantially – which may render the study unethical. I hope the study was planned in such a way, that the sample size in stage 1 was large enough to show BE if the assumption about the CV turns out to be correct (that’s the idea behind Method C). If the sponsor played Bingo (“We guess that the CV will be 20%, but maybe we will pass with only 16 – or even 12 – subjects. Let’s try.”) you will be in trouble if you switch to a conventional 2×2 design.
Example (T/R 95%, CV 20%, target power 80%): The sample size in a conventional 2×2 cross-over will be 20 (power 83.5%). If you perform the study in 16 subjects, power will be 73.5% (unethical). With only 12 power will be 56.6% (crazy). Compare this to Method C (10⁶ simulations, exact* power estimation):

Stage 1 sample size 20:
Overall power 86.3%, power in stage 1 75.1%.
Average total sample size 22 (n_5% 20, median 20, n_95% 32, n_max 64).
18.7% of studies proceed to the second stage.
Stage 1 sample size 16:
Overall power 85.2%, power in stage 1 61.9%.
Average total sample size 20 (n_5% 16, median 16, n_95% 36, n_max 80).
34.1% of studies proceed to the second stage.
Stage 1 sample size 12:
Overall power 84.0%, power in stage 1 41.2%.
Average total sample size 21 (n_5% 12, median 18, n_95% 40, n_max 98).
56.5% of studies proceed to the second stage.

In scenario I (stage 1 sample size = fixed design sample size) you have a fairly high chance (75.1%) to show BE already in the first stage; only 18.7% chance to proceed to the second stage. Note that power in the first stage (75.1%) is lower than in a fixed sample design (83.5%) because in some studies you have to use the wider 94.12% CI ⇒ less likely to pass than with the conventional 90% CI.
In scenario II (= gambling) the chance to show BE in the first stage is only 61.9%; chance to proceed to the second stage is 34.1%. Note the distribution of total sample sizes – a consequence of the penalty.
Let’s be silent about scenario III.

❝ Now the problem is we have dosed volunteers and I think CRO has started bioanalyis part (but not statistical analysis). In this situation can we ammend the protocol or we need to repeat the study with ammended protocol.

I would do neither; as shown above power may drop – ethical problems. If the sponsor is wary about EMA’s acceptance of Method C, consider switching to Method B instead.
Can you post your assumed CV and stage 1 sample size?

Slightly different values for n₁ 12 compared to results reported by Potvin et al. since they used in their simulations for speed reasons an approximate power estimation (based on the shifted central t-distribution).

—
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

kumarnaidu ★ Mumbai, India, 2013-10-21 08:01 (4198 d 12:42 ago) (edited on 2013-10-21 14:41) @ Helmut Posting: # 11712 Views: 25,203	Bad idea Post reply
	Hi Helmut, ❝ Can you post your assumed CV and stage 1 sample size? We have back calculated the CV from the available literature it is ~21%. And the sample size is 28 for stage I. — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-10-21 16:51
(4198 d 03:52 ago)

@ kumarnaidu
Posting: # 11719
Views: 25,344

Change to Method B?

Post reply

Hi Kumar,

❝ We have back calculated the CV from the available literature it is ~21%. And the sample size is 28 for stage I.

OK, that’s not too bad (10⁶ simulations, exact power):

Method C
Overall power 92.0%, power in stage 1 90.6%.
Average total sample size 28 (n_5% 28, median 28, n_95% 28, n_max 66).
2.0% of studies proceed to the second stage.
Method B
Overall power 88.7%, power in stage 1 86.2%.
Average total sample size 28 (n_5% 28, median 28, n_95% 30, n_max 66).
5.5% of studies proceed to the second stage.
Conventional fixed sample design
Power 91.4%

Methods B and C have already a high chance to show BE in the first stage. If the sponsor is wary about EMA’s acceptance, consider changing to Method B. If you want to change to a conventional 2×2 cross-over – contrary to what I said before – there shouldn’t be ethical issues (power >90%). If the actual CV is 25%, power would still be 80.7%. Up to you. On the other hand, if the CV is higher than 25% or you have drop-outs the producer’s risk increases.

—
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

kumarnaidu ★ Mumbai, India, 2013-10-22 08:15 (4197 d 12:27 ago) @ Helmut Posting: # 11723 Views: 25,145	Change to Method B? Post reply
	Thanks a lot Helmut for your valuable suggestions. can you please guide me how you have calculated the power above so that from next time I can do this exercise for two stage design. 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! [Helmut] — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-10-22 15:20
(4197 d 05:23 ago)

@ kumarnaidu
Posting: # 11726
Views: 25,378

Simulations

Post reply

Hi Kumar,

❝ can you please guide me how you have calculated the power above so that from next time I can do this exercise for two stage design.

My simulations were done in R using package PowerTOST. Set up the framework (decision scheme, alphas in both stages, acceptance rule) and generate N BE-studies of n₁ sample sizes with the expected CV. For Potvin use a T/R-ratio of 1.25 (for alpha) or 0.95 (for power). Simulate N=10⁶ studies for alpha (slow convergence) and at least N=10⁵ studies for power. Number of studies passing / N at 1.25 = empiric alpha and at 0.95 = empiric power.
Depending on your CPU and programming skills expect run-times of 15 seconds up to one day for 10⁶ sim’s. :-D

—
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

Shuanghe
★★

Spain,
2013-10-22 11:03
(4197 d 09:39 ago)

@ Helmut
Posting: # 11724
Views: 25,236

Open your mouth and forcefully spit the water out

Post reply

Hi Helmut,

❝ Have you asked them which methods would be acceptable according to their impeccable wisdom instead? They should enlighten us simpletons.

No, I contact them indirectly (through a colleague of mine).

❝ Yes. For some recent (this year’s) experiences see slide 45 of this presentation. Method C [sic] accepted by Germany (RMS) and CMSs Austria, Denmark, Sweden, and The Netherlands. Spain asked for an all-fixed effects analysis (subjects within sequence were random in the original analysis).

Seems the same one as I saw in Budapest in your talk this May. This is actually the very reason I wanted to ask the same question again since you have presented some studies already... Well, now we know how it goes...

❝ ... I hope someone has the balls and goes for a referral. If a GL (not a law anyway) states 2×2=5 I would wholeheartedly ignore it for sure.

If that's my own company, I probably will. :lol3:

Unfortunately I have boss to answer to...

❝ ... A paper “On the Statistical Model of the Two-Stage Design in Bioequivalence Assessment” by Karalis/Macheras is currently under review (J Pharm Pharmacol). ...

Interesting. Thanks. I'll get it when it is published.

—
All the best,
Shuanghe

Helmut
★★★

Vienna, Austria,
2013-10-22 15:05
(4197 d 05:38 ago)

@ Shuanghe
Posting: # 11725
Views: 25,127

Reach for the stars

Post reply

Hi Shuanghe,

❝ Seems the same one as I saw in Budapest in your talk this May.

The difference is that in May the case was pending and in September it was closed.

❝ ❝ ... I hope someone has the balls and goes for a referral. If a GL (not a law anyway) states 2×2=5 I would wholeheartedly ignore it for sure.

❝

❝ If that's my own company, I probably will. :lol3: Unfortunately I have boss to answer to...

This whole “Potvin not valid in Europe” business is some kind of applied hypocrisy. It’s like saying “You may fly to the moon. But it is not acceptable to use a chemical rocket propellant.” What else? Levitation? Warp-drive?

[image]

—
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

kumarnaidu
★

Mumbai, India,
2013-12-04 14:43
(4154 d 05:00 ago)

@ Helmut
Posting: # 11999
Views: 24,744

Alpha inflation in two stage design

Post reply

Hi all,
In one of BEBAC presentation Page 45 http://bebac.at/lectures/Prague2012-2.pdf
a below statement is written.
Deficiency 2: The Applicant should demonstrate that the
type I error inflation which can be expected from the
chosen approach, did not impact on the decision of
bioequivalence.

Response 2: One million simulations based on study’s
sample size and CV.
α_emp 0.0494 (95% CI: 0.0490 – 0.049)

Here I want to ask if I want to demonstrate that type error inflation did not impact on my study then can I demonstrate this by simply showing 90% CI as well as 94.12% CI for GMR of PK parameters within the acceptance limits (80% to 125%).
also I am not getting how to interpret Response 2

—
Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-12-04 15:50
(4154 d 03:52 ago)

@ kumarnaidu
Posting: # 12000
Views: 24,645

Deficiency letters

Post reply

Hi Kumar,

❝ […] if I want to demonstrate that type error inflation did not impact on my study then can I demonstrate this by simply showing 90% CI as well as 94.12% CI for GMR of PK parameters within the acceptance limits (80% to 125%).

This was actually the response to deficiency #1. However, post hoc switch of statistical methods (from Method C to B) is really bad practice, but the regulator insisted on it. :-(

If you expect problems (only in Europe!) plan the study already for Method B.

❝ also I am not getting how to interpret Response 2

The study’s CV was 30.65% observed in 49 subjects. Potvin’s closest table entry is for CV 30% and n₁ 48 (α_emp 0.0494). I simulated α_emp with the study’s values and got 0.0494 as well (10⁶ simulated studies). It’s easy to calculate the confidence interval based on the binomial test. In R:

sims <- 1e6 # number of simulated studies alpha <- 0.0494 # empiric risk type I binom.test(alpha*sims, sims, alternative="two.sided", conf.level=0.95)

getting:

Exact binomial test data: alpha * sims and sims number of successes = 49400, number of trials = 1e+06, p-value < 2.2e-16 alternative hypothesis: true probability of success is not equal to 0.5 95 percent confidence interval: 0.04897608 0.04982653 sample estimates: probability of success 0.0494

If you run a study with Method B and still get such a weird deficiency letter – and don’t have the means to run simulations – it should be possible to look at the 95% CI. IMHO Bonferroni’s α 0.025 should keep the overall risk type I <0.05 for any two-stage method. Ask a statistician.

—
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

kumarnaidu
★

Mumbai, India,
2013-12-05 07:18
(4153 d 12:25 ago)

(edited on 2013-12-05 12:19)
@ Helmut
Posting: # 12005
Views: 24,833

Deficiency letters

Post reply

Hi Helmut,
In my study CV is around 19% and sample size was 28 so can I use CV=20% and sample size stage 1(n1)=24 here as 28 is not there in the table.

1. By using CV=20% and n=24

        Exact binomial test



data:  alpha * sims and sims

number of successes = 49000, number of trials = 1e+06, p-value <

2.2e-16

alternative hypothesis: true probability of success is not equal to 0.5

95 percent confidence interval:

 0.04857771 0.04942490

sample estimates:

probability of success

                 0.049

2. By using CV=20% and n=36

      Exact binomial test



data:  alpha * sims and sims

number of successes = 49900, number of trials = 1e+06, p-value <

2.2e-16

alternative hypothesis: true probability of success is not equal to 0.5

95 percent confidence interval:

 0.04947404 0.05032856

sample estimates:

probability of success

                0.0499

In second case upper limit of type 1 error rate is 0.05032 (exceeding 0.05),Is it means that here type 1 error inflation is present and regulator may reject the application.

—
Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-12-05 16:51
(4153 d 02:52 ago)

@ kumarnaidu
Posting: # 12006
Views: 24,654

Simulations

Post reply

Hi Kumar,

❝ In my study CV is around 19% and sample size was 28 so can I use CV=20% and sample size stage 1(n1)=24 here as 28 is not there in the table.

❝ In second case upper limit of type 1 error rate is 0.05032 (exceeding 0.05),Is it means that here type 1 error inflation is present and regulator may reject the application.

I don’t understand what you want to achieve here. In 10⁶ simulations everything <0.05036 is not significantly >0.05. Potvin et al. covered the entire range of n₁ 12–60 and CV 10–100%. For Method B the largest inflation was 0.0485 and for Method C 0.0510. I would not step into this muddle unless I actually receive a deficiency letter. Did you?

Sometimes you get significant results. Why not? I simulated your CV 19% and n₁ 28. 10⁶ simulations each, Method C.

run   α_emp

 1  0.049872

 2  0.049777

 3  0.050218

 4  0.049988

 5  0.049882

 6  0.050440 *

 7  0.049946

 8  0.050084

 9  0.050139

10  0.050289

Now what?

—
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

kumarnaidu ★ Mumbai, India, 2013-12-05 17:07 (4153 d 02:36 ago) @ Helmut Posting: # 12007 Views: 24,671	Simulations Post reply
	Hi Helmut, actually we have not yet submitted the report to the regulatory but our client wants clarification that how we are going to address the defiency 2 in the presentation thats all. — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-12-05 17:55
(4153 d 01:48 ago)

@ kumarnaidu
Posting: # 12008
Views: 24,580

Simulations

Post reply

Hi Kumar,

❝ actually we have not yet submitted the report to the regulatory but our client wants clarification that how we are going to address the defiency 2 in the presentation thats all.

Maybe you will not receive a deficiency letter. :-D

In the meantime I succeeded with other studies (Method C; Germany, Austria, Denmark, Sweden, The Netherlands, Spain)…

If you want to be prepared:

Try Method B (94.12% CI in the first stage). Seems to be preferred by some European regulators.
Try Bonferroni’s correction (95% CI). Most conservative, IMHO.
Believe in my simulations. If you prefer higher numbers: 20·10⁶ sim’s, α_emp 0.050032 (ns >0.05, limit 0.05008).
Run your own.
Find someone how can do it for you (e.g., Detlew, ElMaestro, Vangelis, myself).

—
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

kumarnaidu ★ Mumbai, India, 2013-12-06 07:07 (4152 d 12:35 ago) @ Helmut Posting: # 12011 Views: 24,668	Simulations Post reply
	Thanks Helmut, Yes you are right we should go with the method B and wait for the deficiency letter. — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-12-06 15:30
(4152 d 04:13 ago)

@ kumarnaidu
Posting: # 12014
Views: 24,446

Ejaculatio praecox

Post reply

Hi Kumar

❝ Yes you are right we should go with the method B and wait for the deficiency letter.

Don’t be tempted to change the method. Submit the study evaluated according to the protocol (Method C). Internally you can explore the other options. For educational reasons I would include these evaluations in the submission not even in an exploratory manner.

—
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

kumarnaidu ★ Mumbai, India, 2013-12-06 16:13 (4152 d 03:30 ago) @ Helmut Posting: # 12015 Views: 24,601	Ejaculatio praecox Post reply
	Thanks a lot Helmut for your guidance on this issue. we will keep the method C as it was. — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-12-06 16:31
(4152 d 03:11 ago)

@ kumarnaidu
Posting: # 12016
Views: 24,654

All effects fixed

Post reply

Hi Kumar,

sorry, I forgot: For EMA (only!) use an all-fixed effects model (according to the BE GL Section 4.1.8).
For future studies: If you proceed to the second stage, add the term sequence(stage) to the model. Required since EMA’s Q&A Rev.7 (Feb 2013). Since this is not a within-subject term, CV, power, and the CI – as expected – are not affected.* Why EMA’s Biostatistical Working Party introduced it is beyond my intellectual reach.

Karalis V, Macheras P. On the Statistical Model of the Two-Stage Designs in Bioequivalence Assessment. J Pharm Pharmacol. (Epub ahead of print 31 Oct 2013). doi 10.1111/jphp.12164

—
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

kumarnaidu ★ Mumbai, India, 2013-12-09 08:33 (4149 d 11:09 ago) @ Helmut Posting: # 12024 Views: 24,362	All effects fixed Post reply
	Thanks Helmut. I think like new FDA draft guidance EMA should also include the provision of submitting complete protocol for review and comment before study initiation in sequential design to avoid such situations. — Kumar Naidu

Helmut
★★★

Vienna, Austria,
2013-12-09 13:45
(4149 d 05:58 ago)

@ kumarnaidu
Posting: # 12025
Views: 24,373

EMA…

Post reply

Hi Kumar,

❝ […] EMA should also include the provision of submitting complete protocol for review and comment before study initiation in sequential design to avoid such situations.

Don’t be too optimistic. All the deficiency letters in my presentations came from studies which were performed in the EU. Hence, the protocols were approved by regulatory authorities where the CROs are located. In the “worst case MRP” you might have to face 31 different opinions (28 member states + Iceland, Liechtenstein, Norway).