Bioequivalence and Bioavailability Forum

Main page Policy/Terms of Use Abbreviations Latest Posts

 Log-in |  Register |  Search

Back to the forum  Query: 2017-08-17 01:35 CEST (UTC+2h)
 
VSL
Junior

India,
2017-06-05 20:34

Posting: # 17444
Views: 1,492
 

 General BE topics [Design Issues]

Dear all,
Could you please give your views on the following topics?
  1. Half-life: Is half life independent of formulation effect? I mean, IR vs. MR products of the same substance. I am asking as protocol design depends a lot on it.
  2. High variable drug substance vs. High variable drug products, Is there is any demarcation for protocol design aspects?
  3. If there are two strengths, say, tablet, 10 and 20 mg and 20 mg is RLD. The company wishes to market only 10 mg strength but 10 mg is not RLD. What should be RLD in BE study? Is comparison of 10mg x2 tablets (test) vs. 20mg RLD right?
  4. What is the minimum sample size for parallel group BE study?
  5. For endogenous substances, how many and for how much time duration (-48 hours etc.) baseline samples should be collected?
  6. Why passing ratio is high for BE studies (especially pivotal studies) conducted in India compared to foreign countries? Is there any geographical/ethnic/staff expertise/deliberate effect?
  7. For some therapeutic class drugs BE study, I have not seen any PD effect. Example, For Antihypertensive drugs, no change in BP in vital recordings even at/around Cmax. Please explain.

Thanks and Regards,
VSL
ElMaestro
Hero

Denmark,
2017-06-06 00:20

@ VSL
Posting: # 17445
Views: 1,313
 

 General BE topics

Hello VSL,

» 1 Half-life: Is half life independent of formulation effect? I mean, IR vs. MR products of the same substance. I am asking as protocol design depends a lot on it.

Half-life can refer to the intrinsic half-life, or to the apparent half-life (what you observe in practice, given the formulation etc). The latter is what guides it.

» 2. High variable drug substance vs. High variable drug products, Is there is any demarcation for protocol design aspects?

No. Observed CV is what matters.

» 3. If there are two strengths, say, tablet, 10 and 20 mg and 20 mg is RLD. The company wishes to market only 10 mg strength but 10 mg is not RLD. What should be RLD in BE study? Is comparison of 10mg x2 tablets (test) vs. 20mg RLD right?

2x10mg, probably, but you should contact OGD.

» 4. What is the minimum sample size for parallel group BE study?

Wrong question. You will most likely not show BE in any parallel study at the minimum sample size. If you try, it might be unethical.

» 5. For endogenous substances, how many and for how much time duration (-48 hours etc.) baseline samples should be collected?

There is no rule if it isn't mentioned in the product-specific guidance.

» 6 Why passing ratio is high for BE studies (especially pivotal studies) conducted in India compared to foreign countries? Is there any geographical/ethnic/staff expertise/deliberate effect?

Wow, this one is toxic. :-D:-D:-D
Perhaps companies are powering their studies higher e.g. 90% rather than 80% when conducting them in India??
VSL, sometimes failure is not an option and sometimes shortcuts are taken to cut down effort/cost or to stay within agreed timelines. Google Semler, GVK Bio, MTR... I am sure you know the issues already.
Anyways, I never saw official passing statistics for India compared to other places. Can you show a reference?

» 7. For some therapeutic class drugs BE study, I have not seen any PD effect. Example, For Antihypertensive drugs, no change in BP in vital recordings even at/around Cmax. Please explain.

Some modern antihypertensives often have little effect in healthy volunteers. They are very, very safe. E.g. Telmisartan, Perindopril etc. But be aware you have nothing to compare with in BE trials since both arms are active and no placebo recording is available. If there is an effect you won't see it. Comparing BP at baseline to BP at Tmax is not particularly informative.

Have a good day.

I could be wrong, but…


Best regards,
ElMaestro

- since June 2017 having an affair with the bootstrap.
Helmut
Hero
Homepage
Vienna, Austria,
2017-06-06 14:24

@ VSL
Posting: # 17452
Views: 1,244
 

 General BE topics

Hi VSL,

I agree with all of what ElMaestro wrote. Hence, only some more remarks.

» 1. Half-life: Is half life independent of formulation effect? I mean, IR vs. MR products of the same substance.

Without IV data and after a PO dose we can only assume that the apparent half-life = the intrinsic half-life (elimination). This assumption is reasonable for IR products (ka > ke). When you slow down the absorption process you will reach ka = ke or “flip-flop PK”. If you go further (ka < ke) the apparent half-life (now the slowest phase) is no more the intrinsic half-life but absorption. That’s why the EMA (in contrary to IR) for MR products additionally to AUC0–t requires AUC0–∞ and does not allow using a truncated AUC.

» 2. High variable drug substance vs. High variable drug products, Is there is any demarcation for protocol design aspects?

No, but you have to know the drug and your formulation. If the high variability is a property of the drug (HVD) you can design the study as usual (but taking the high CV into account and possibly opt for reference-scaling). If you have to deal with a HVDP it depends. F.i., gastric resistant formulations exhibit a wide range of tmax-values (due to variability in tlag caused by the interplay of the formulation with gastric transit). In such a case you should increase the number of samples in order to “catch” Cmax.

» 5. For endogenous substances, how many and for how much time duration (-48 hours etc.) baseline samples should be collected?

As ElMaestro wrote: No rule rulez. There is a wide range of options.
  • Best case:
    If you know that the baseline is stable it might be sufficient to collect just three samples within one hour before the administration and subtract the mean (I prefer the median) from the profile.
  • Worst case:
    Circadian rhythm and levels of the endogenous compound depend on food. Have a run-in period where you standardize food. Sample an entire “blank” profile before administration and subtract concentrations at corresponding time points.

» 6. Why passing ratio is high for BE studies (especially pivotal studies) …

I have heard that pilot studies never fail. The CROs want to perform the pivotal ones as well.

» … conducted in India compared to foreign countries? Is there any geographical/ethnic/staff expertise/deliberate effect?

Fabricated data, plain fraud, etc. Sorry to say.
One of my clients (her boss’ decision) started to perform studies in India (before mainly in Canada, Central, and Eastern Europe). The company always powered studies to 90%. Since the sample size estimation is based on assumptions and with 90% power one can expect that 10% will fail by pure chance – and that was essentially what she observed in the past. In two years none (!) of the studies performed in India failed. Every study was monitored. No obvious explanation. She was concerned and called me up. Given the number of studies the chance that none failed (Fisher’s exact test) was not significant but scratching the limit.

Absence of evidence is not evidence of absence.    Carl Sagan


[image]All the best,
Helmut Schütz 
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
VSL
Junior

India,
2017-06-06 19:34

@ Helmut
Posting: # 17455
Views: 1,217
 

 General BE topics

Dear ElMaestro and Helmut,

Thank you very much for the explanation.

One more question, why regulatory agencies are allowing studies with deliberate high power? I mean, it is forced BE, unethical and violation of Declaration of Helsinki (Human exposure without scientific rationale). There has to be the scientific tradeoff in setting alfa and beta values while calculating sample size rather mere assumption. I have seen post analysis power more than 90% in many studies. If you ask biostatisticians, they say, we don't know, it comes, no justification is required as the study has passed.

Thank you again.
DavidManteigas
Regular

Portugal,
2017-06-12 12:35

@ VSL
Posting: # 17487
Views: 1,055
 

 General BE topics

Hi VSL,

In my opinion, from my knowledge of the portuguese regulators, there is a lack of members with statistical expertise to properly assess the adequacy of the sample size, or any other question regarding statistical methods, either in Ethics Committees or the regulator.

In my plans for 2018, I will attempt to meet with the portuguese regulator so they can gave me access to their approved studies in bioequivalence in the last years so I can quantify the degree of overpowering of bioequivalence studies when compared to the observed CV and the predicted/guessed/magically obtained predicted CV. My intent is to publish a paper with that information and to sensitize regulators and ethics committees to this issue, which mixes both the sample size calculation and the adequacy of the study design in light of the available information.
Helmut
Hero
Homepage
Vienna, Austria,
2017-08-04 15:51

@ DavidManteigas
Posting: # 17665
Views: 370
 

 forced BE?

Hi David & VSL,

sorry for excavating an old thread. :-D

» […] from my knowledge of the portuguese regulators, there is a lack of members with statistical expertise to properly assess the adequacy of the sample size, or any other question regarding statistical methods, either in Ethics Committees or the regulator.

According to Directive 2005/28/EC (6)

Provisions for the functioning of the Ethics Committees should be established in each Member State on the basis of common detailed guidelines, in order to ensure the protection of the trial subject while at the same time allowing a harmonised application in the different Member States of the procedures to be used by Ethics Committees.

Hence, ECs are set up according to National laws. E.g., in Austria acc. to §41 of the Medicines Law (my translation):
  • (2) The ethics committee must be composed in a balanced proportion of women and men and at least consist of:
    1. A physician who is entitled to pursue a self-employed profession and is not the investigator,
    2. A specialist physician in whose special subject the respective clinical examination falls […] and is the investigator,
    3. A representative of the senior service for health care and nursing,
    4. A lawyer,
    5. A pharmacist,
    6. A patient representative,
    7. A member of a representative disabled organization as well as a representative of the elderly, who is part of a senior citizens ‘organization in accordance with the Federal Senior Citizens’ Law, BGBl. I No. 84/1998,
    8. A person who has biometric expertise and
    9. Another person not covered by items 1 to 8, who is responsible for the perception of pastoral affairs or otherwise has the appropriate ethical competence.
Similar in Germany’s Medicines Law §41a(2)

[…] interdisciplinary composition of the ethics committee involving at least one lawyer, one person with scientific or professional experience in the field of ethics in medicine, one person with experience in the field of experimental planning and statistics, three physicians who have experience in clinical medicine, including a specialist in clinical pharmacology or pharmacology and toxicology, as well as a layperson.

Which are the rules for ECs in Portugal? BTW, I know some members of the Portuguese Agency whose knowledge of statistics is excellent.
It can even be that the EC is more knowledgeable than an agency. Happened to me. Joachim Röhmel (one of the pioneers of adaptive designs) is a member of the EC of Berlin. He was very happy with my first study in a Two-Stage Design. Later I learned that the Dutch MEB didn’t like (‼) it.

» In my plans for 2018, I will attempt to meet with the portuguese regulator so they can gave me access to their approved studies in bioequivalence in the last years …

Wow! Do you think that’s possible?

» » I have seen post analysis power more than 90% in many studies. If you ask biostatisticians, they say, we don't know, it comes, no justification is required as the study has passed.

» … so I can quantify the degree of overpowering of bioequivalence studies when compared to the observed CV and the predicted/guessed/magically obtained predicted CV.

Is that reasonable? You may fall into the trap of post hoc power.1 We know that θ0 and CV used in designing the study are assumptions or, at best, estimates. The applicant might have used conservative values.2 If the GMR turns out to be “better”, the CV and/or the dropout-rate lower than expected you will face high post hoc power. But this does not necessarily imply that the study was some kind of “forced BE”.

» My intent is to publish a paper with that information …

Great. But given what I wrote above, IMHO, it will not be sufficient to solely plugin the numbers. You have to dive into the protocol and report in order to get insights about what was going on.

» … and to sensitize regulators and ethics committees to this issue, which mixes both the sample size calculation and the adequacy of the study design in light of the available information.

Good! Members of an EC are in a difficult situation. There is no database giving the sample sizes of studies performed with a particular reference. I guess in many cases the assessment would require a member with a good memory. If he/she/it has seen studies of drug X with sample sizes of, say ~32 subjects and the protocol suggests 120, that should ring the alarm bells. I think it is difficult to draw the line.


  1. Hoenig JM, Heisey DM. The Abuse of Power. The Pervasive Fallacy of Power Calculations for Data Analysis. Am Stat. 2001;55(1):19–24. doi:10.1198/000313001300339897
  2. Not conservative but a waste of money & unethical: I know one company, which for years – when they had no clue about the CVw – simply used s∕x reported for the reference. No crossover, total CV instead of CVw. They called it “in order to be on the safe side”. Well, … In the meantime they opt for pilots or TSDs.

[image]All the best,
Helmut Schütz 
[image]

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

Portugal,
2017-08-07 12:42

@ Helmut
Posting: # 17666
Views: 230
 

 forced BE?

Hi Helmut,

From my knowledge, in the portuguese law and other guidances there is not any requirement on the backgroud of the participants other than the ones mentioned in GCP guideline and the portuguese law for health ethics committees, which states:

"1 - As CES têm uma composição multidisciplinar e são constituídas por sete
membros, designados de entre médicos, enfermeiros, farmacêuticos, juristas, teólogos,
psicólogos, sociólogos ou profissionais de outras áreas das ciências sociais e humanas."

Translated as
"1 - The EC have a multidisciplinary composition and are composed by 7 members, designated within physicians, nurses, pharmacists, jurist, theologians, psychologist, sociologists or other health and human sciences professionals"


Portuguese Ethics Committee has a lot of members with an amazing knolwedge (even a priest!), most of them experts in their fields of work. And they usually request the review from another "outside" expert on the therapeutic area when there is none in the permanent members. The only field where there is a lack of expertise (imo, of course) is statistics. IMO, that happens because portugal does not have a strong tradition in clinical trials, and even the SPE (Portuguese Statistics Society) does not have a strong interest in clinical trials, as one may see from their meetings and relevant conferences. Also, I had never found any comment from SPE to relevant guidelines nor have received any e-mail as member to provide insights into any sort of coordinated review. As regards to the portuguese authority, in my experience in the assessment of clinical trial applications (not MA), we never receive any questions from the statistical part of the study (either they don't have experts for this or they rely on the assessment of the EC). As statistician, I usually receive multiple questions regarding sample size and statistical methodology (for instance, missing data imputation method) when the investigators submit an application for financing from pharmaceutical companies, but usually none from either EC or regulators.

I will try, since I intend to use that data to my master thesis in biostatistics. I will hope that the relevance of the theme and the academic part of the job might sensitize them on this :-D

There are still a lot of questions on how to do that, of course, that is why I only have that planned for 2018 :-D Don't know if assuming the same GMR as in the planned sample size calculation and the observed CV might be a solution - still a long way to go from idea to practice. The primary purpose will be to sensitize the regulators/EC to studies that they approve with a justification like "Assuming a GMR of 0.95 and a CV of XX%, a total of XX patients will be included in the study" which after conduct have a result with 99.99% power. And not a single reference or a decent justification is provided on how they've got that CV. Sometimes is the "sponsor in-house data" or "published data, which we don't reference here and for some reason nobody asked for it and the estimated ISCV of 20% was in fact only 6%".
Helmut
Hero
Homepage
Vienna, Austria,
2017-08-07 13:31

@ DavidManteigas
Posting: # 17667
Views: 223
 

 forced BE?

Hi David,

» Portuguese Ethics Committee has a lot of members with an amazing knolwedge (even a priest!), …

Like in Austria (#9 of above). Catholic Church still strong. Not so in Germany. ;-)

» The only field where there is a lack of expertise (imo, of course) is statistics.

That’s a pity.

» […] I had never found any comment from SPE to relevant guidelines nor have received any e-mail as member to provide insights into any sort of coordinated review.

No surprise. I’m a member of the IBS. When the BE-GL was drafted I suggested to compile comments from the society. Response: Zero, niente, nada. Seemingly they considered BE too trivial.

» As regards to the portuguese authority, in my experience in the assessment of clinical trial applications (not MA), we never receive any questions from the statistical part of the study (either they don't have experts for this or they rely on the assessment of the EC).

One regulator once told me that he is interested in α not β.

» I will hope that the relevance of the theme and the academic part of the job might sensitize them on this :-D

Yep. As a side note: The European Community is a founding member of ICH. E9, Section 3.5 states:

The number of subjects in a clinical trial should always be large enough to provide a reliable answer to the questions addressed. […]
The method by which the sample size is calculated should be given in the protocol, together with the estimates of any quantities used in the calculations (such as variances, mean values, response rates, event rates, difference to be detected). The basis of these estimates should also be given. It is important to investigate the sensitivity of the sample size estimate to a variety of deviations from these assumptions and this may be facilitated by providing a range of sample sizes appropriate for a reasonable range of deviations from assumptions. In confirmatory trials, assumptions should normally be based on published data or on the results of earlier trials. […] Conventionally the probability of type I error is set at 5% or less or as dictated by any adjustments made necessary for multiplicity considerations; the precise choice may be influenced by the prior plausibility of the hypothesis under test and the desired impact of the results. The probability of type II error is conventionally set at 10% to 20%; it is in the sponsor’s interest to keep this figure as low as feasible especially in the case of trials that are difficult or impossible to repeat.

Did you ever see that in a protocol? I didn’t.

» Don't know if assuming the same GMR as in the planned sample size calculation and the observed CV might be a solution - still a long way to go from idea to practice.

Power is much more sensitive to the GMR than to the CV. Maybe looking at it the other way ’round would be better. If you are not a hard-core frequentist, going Bayesian might be an option.

» The primary purpose will be to sensitize the regulators/EC to studies …

I think this is very important indeed!

» … that they approve with a justification like "Assuming a GMR of 0.95 and a CV of XX%, a total of XX patients will be included in the study" which after conduct have a result with 99.99% power. …

Again: Post hoc power. How relevant is it?
IMHO, ECs should simply follow ICH E9. Which are the assumptions, how sensitive is the sample size to deviations, …? It is the job of the EC to protect the health of subjects in the study. Hence, it is desirable to keep the sample size as small as possible but as large as necessary.

[image]All the best,
Helmut Schütz 
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
ElMaestro
Hero

Denmark,
2017-08-07 13:44

@ Helmut
Posting: # 17668
Views: 212
 

 forced BE?

Hi Hötzi,

» (blahblah) The basis of these estimates should also be given. It is important to investigate the sensitivity of the sample size estimate to a variety of deviations from these assumptions and this may be facilitated by providing a range of sample sizes appropriate for a reasonable range of deviations from assumptions. In confirmatory trials, assumptions should normally be based on published data or on the results of earlier trials.

Yessir!!!!!!!!
Thanks for bringing it up.

That is indeed why we need to apply more bootstrapping in BE.

It, too, tells what sample size we need, but in contrast to ordinary power calcs it allows one to take into account that residuals may not be normal even though the evaluation itself makes that assumption. And that is a story for another day.

I could be wrong, but…


Best regards,
ElMaestro

- since June 2017 having an affair with the bootstrap.
Helmut
Hero
Homepage
Vienna, Austria,
2017-08-07 14:00

@ ElMaestro
Posting: # 17669
Views: 204
 

 forced BE?

Hi ElMaestro,

» That is indeed why we need to apply more bootstrapping in BE.

Duno whether we need it. Of course, one can do it since the GL only speaks about “an appropriate sample size calculation.” Whatever that might be.

» It, too, tells what sample size we need, but in contrast to ordinary power calcs it allows one to take into account that residuals may not be normal even though the evaluation itself makes that assumption.

Right. But IMHO, the quality of bootstrapping depends on the input. Remember what you once wrote?

OT: In your current signature you claim to have “an affair with the bootstrap”. Really an affair? I mean, does it respond to your love? Or is it just calf love?

[image]All the best,
Helmut Schütz 
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
ElMaestro
Hero

Denmark,
2017-08-07 15:58

@ Helmut
Posting: # 17672
Views: 184
 

 forced BE?

Hi Hötzi,

» Right. But IMHO, the quality of bootstrapping depends on the input. Remember what you once wrote?

I had forgotten that :-D

Truth be told, my aging brain seems to have very little ability to remember anything farther back than 2 weeks. That thread was from 2009, I think the Commodore 64 was still hot back then and movies were in black and white?!

» OT: In your current signature you claim to have “an affair with the bootstrap”. Really an affair? I mean, does it respond to your love? Or is it just calf love?

Manuscript accepted last week. Details to follow. I used the bootstrop to solve a BE-related problem that no Sponsor apparently was able to find a solution to over the last five years. One reviewer was pretty upset (I even got spanked for using the term Sponsor). I hope it wasn't you :-D:-D:-D
I am trying to re-arrange a meeting in September; if I am successful I can give a 20 min talk about it at the event in Prague if anyone cares? It requires, though, that the paper (probably a one-pager in letter form) is published at that time.

I could be wrong, but…


Best regards,
ElMaestro

- since June 2017 having an affair with the bootstrap.
DavidManteigas
Regular

Portugal,
2017-08-07 14:02

@ Helmut
Posting: # 17670
Views: 205
 

 forced BE?

Hi helmut

» The method by which the sample size is calculated should be given in the protocol, together with the estimates of any quantities used in the calculations (such as variances, mean values, response rates, event rates, difference to be detected). The basis of these estimates should also be given. It is important to investigate the sensitivity of the sample size estimate to a variety of deviations from these assumptions and this may be facilitated by providing a range of sample sizes appropriate for a reasonable range of deviations from assumptions. In confirmatory trials, assumptions should normally be based on published data or on the results of earlier trials. […] Conventionally the probability of type I error is set at 5% or less or as dictated by any adjustments made necessary for multiplicity considerations; the precise choice may be influenced by the prior plausibility of the hypothesis under test and the desired impact of the results. The probability of type II error is conventionally set at 10% to 20%; it is in the sponsor’s interest to keep this figure as low as feasible especially in the case of trials that are difficult or impossible to repeat.
» Did you ever see that in a protocol? I didn’t.

Never saw it either, although at the planning stage a lot of sensitivity analysis are done (at least, in the studies I work with)

» Power is much more sensitive to the GMR than to the CV. Maybe looking at it the other way ’round would be better. If you are not a hard-core frequentist, going Bayesian might be an option.

Not hard-core frequentist, but going bayesian is still a crime for some statisticians and I would only go that way if other alternatives are not available :-D

» Again: Post hoc power. How relevant is it?
» IMHO, ECs should simply follow ICH E9. Which are the assumptions, how sensitive is the sample size to deviations, …? It is the job of the EC to protect the health of subjects in the study. Hence, it is desirable to keep the sample size as small as possible but as large as necessary.

It might be relevant, if you note a trend in approved studies: for instance, if most of them are significantly (still to be defined what might be significantly) overpowered and the assumptions defined in sample size calculation are far way from the observed parameters, I think this might be an issue that should handled and it is being neglected. By the regulators, which should be (again, imo) more open to share their information on CV and GMR of already approved drugs, and by the sponsors that should present strong justifications (as per ICH E9) on their protocols, such as sensitivity analysis and references that support their assumptions (other than "available literature" and "in-house data") and should look more carefuly to other design strategies available instead of just plugging a CV based on intuition. Otherwise, we are assuming that sample size is magic performed by the statistician to acchieve the maximum budget of the project :-D
Helmut
Hero
Homepage
Vienna, Austria,
2017-08-07 14:26

@ DavidManteigas
Posting: # 17671
Views: 196
 

 forced BE?

Hi David,

» » [quote from ICE E9]
» » Did you ever see that in a protocol? I didn’t.
»
» Never saw it either, although at the planning stage a lot of sensitivity analysis are done (at least, in the studies I work with)

AFAIK, in phase III that’s regularly done. I know some people doing it in BE. However, it doesn’t make it to the protocol. Hence, the EC is left out in the rain.

» Not hard-core frequentist, but going bayesian is still a crime for some statisticians and I would only go that way if other alternatives are not available :-D

:-P

» » Again: Post hoc power. How relevant is it?
»
» It might be relevant, if you note a trend in approved studies: for instance, if most of them are significantly (still to be defined what might be significantly) overpowered and the assumptions defined in sample size calculation are far way from the observed parameters, I think this might be an issue that should handled and it is being neglected.

Agree. Concerning overpowering see this post and play around with the functions exppower.TOST() and expsampleN.TOST() in PowerTOST. Surprises guaranteed. See also this presentation (slide 8 and followings).

» By the regulators, which should be (again, imo) more open to share their information on CV and GMR of already approved drugs, and by the sponsors that should present strong justifications (as per ICH E9) on their protocols, such as sensitivity analysis and references that support their assumptions (other than "available literature" and "in-house data") and should look more carefuly to other design strategies available instead of just plugging a CV based on intuition.

Agree, again.

» Otherwise, we are assuming that sample size is magic performed by the sta­tis­tician to acchieve the maximum budget of the project :-D

Do you know Clarke’s third law?

Any sufficiently advanced technology is indistinguishable from magic.


[image]All the best,
Helmut Schütz 
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
Back to the forum Activity
 Thread view
Bioequivalence and Bioavailability Forum | Admin contact
17,199 Posts in 3,683 Threads, 1,054 registered users;
24 users online (0 registered, 24 guests).

To call the statistician after the experiment is done
may be no more than asking him to perform a postmortem examination:
he may be able to say what the experiment died of.    R.A. Fisher

The BIOEQUIVALENCE / BIOAVAILABILITY FORUM is hosted by
BEBAC Ing. Helmut Schütz
XHTML/CSS RSS Feed