Log in
Register|
Silva ☆ Portugal, 2014-02-26 13:27 (4070 d 04:46 ago) Posting: # 12502 Views: 6,680 |
|
|
Dear all Does it make any sense covering in the randomization procedure in SAS, gender stratification for BE studies? Despite FDA refers that BE study data is not expected to have sufficient power to draw conclusions for each subgroup and therefore statistical analysis of subgroups is not recommended, do you think in a near future this may overcome? Best regards Silva |
|
Helmut ★★★   Vienna, Austria, 2014-02-26 14:34 (4070 d 03:38 ago) @ Silva Posting: # 12505 Views: 5,684 |
|
|
Hi Silva, ❝ Does it make any sense covering in the randomization procedure in SAS, gender stratification for BE studies? If the sample size is sufficiently large, why not? I doesn’t hurt. If you don’t stratify – by stupid chance – you might end up with extremely imbalanced sequences within subgroups. That’s not desirable. ❝ Despite FDA refers that BE study data is not expected to have sufficient power to draw conclusions for each subgroup and therefore statistical analysis of subgroups is not recommended, do you think in a near future this may overcome? I don’t think so. But you never know what may come into the mind of assessors.  — 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 |
|
Silva ☆ Portugal, 2014-02-26 14:41 (4070 d 03:31 ago) @ Helmut Posting: # 12507 Views: 5,672 |
|
|
Hi Helmut, Regarding your comment: ❝ If the sample size is sufficiently large, why not? I doesn’t hurt. If you don’t stratify – by stupid chance – you might end up with extremely imbalanced sequences within subgroups. That’s not desirable. What do you consider a sufficiently large sample size? ≥34? ≥46? ≥56? Regards |
|
Helmut ★★★ Vienna, Austria, 2014-02-26 14:53 (4070 d 03:19 ago) @ Silva Posting: # 12509 Views: 5,664 |
|
|
Hi Silva, ❝ What do you consider a sufficiently large sample size? ≥34? ≥46? ≥56? “Sufficiently large” is the term commonly used by statisticians if they want to stay vague and don’t have the balls to come up with a particular number.  Simply give it a try. I guess that for small sample sizes (e.g., 12) in the worst case you will have to start the randomization-algo more than once – with different seeds – in order to get everything balanced. — 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 |
|
Silva ☆ Portugal, 2014-02-26 15:05 (4070 d 03:07 ago) @ Helmut Posting: # 12510 Views: 5,699 |
|
|
Hi Helmut, Many thanks. It's a precisous help! Best regards |
|
ElMaestro ★★★ Denmark, 2014-02-26 15:45 (4070 d 02:27 ago) @ Helmut Posting: # 12512 Views: 5,699 |
|
|
Hi Hötzi, ❝ Simply give it a try. I guess that for small sample sizes (e.g., 12) in the worst case you will have to start the randomization-algo more than once – with different seeds – in order to get everything balanced. I think FDA would issue a 483 if anyone does that. — Pass or fail! ElMaestro |
|
Helmut ★★★ Vienna, Austria, 2014-02-26 15:52 (4070 d 02:20 ago) @ ElMaestro Posting: # 12513 Views: 5,674 |
|
|
Hi ElMaestro, ❝ I think FDA would issue a 483 if anyone does that. Write a macro in SAS which loops until balance is obtained. Write an SOP. Get many signatures (aka dissipation of responsibilities). Or would you prefer to push the button once and end up with all females in sequence RT and the males in TR?  — 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 |
|
ElMaestro ★★★ Denmark, 2014-02-26 18:10 (4070 d 00:02 ago) @ Helmut Posting: # 12516 Views: 5,656 |
|
|
Hi Hötzi, ❝ Write a macro in SAS which loops until balance is obtained. Write an SOP. Get many signatures (aka dissipation of responsibilities). ❝ Or would you prefer to push the button once and end up with all females in sequence RT and the males in TR? Yah, it is a borderline case. I think I would word the protocol and study title carefully. The study is not really randomised if some randomisations are better than others. This is when I think stratification must be part of the wording. For Xovers I wouldn't normally have a worry as long as we talk non-Brazilian and non-Japanese submissions. For parallel studies I might be inclined to stratify formally if two genders are necessary. — Pass or fail! ElMaestro |
|
mittyri ★★ Russia, 2014-02-26 20:23 (4069 d 21:50 ago) @ ElMaestro Posting: # 12520 Views: 5,686 |
|
|
Hi ElMaestro & all! ❝ I think I would word the protocol and study title carefully. The study is not really randomised if some randomisations are better than others. This is when I think stratification must be part of the wording. Could you give an example of wording? "Multicentre, single-dose, randomised, open label, two-sequence, two-period, crossover, bioavailability study in male and female patients under fed conditions with gender stratification" - something like that? Would stratification be applicable in BEQ studies from assessors point of view? Any experience? — Kind regards, Mittyri |
|
ElMaestro ★★★ Denmark, 2014-02-26 23:31 (4069 d 18:42 ago) @ mittyri Posting: # 12522 Views: 5,703 |
|
|
Hi Mittyri, ❝ Could you give an example of wording? ❝ "Multicentre, single-dose, randomised, open label, two-sequence, two-period, crossover, bioavailability study in male and female patients under fed conditions with gender stratification" - something like that? Yes, or simply "Multicentre, single-dose, gender-stratified, randomised, open label, two-sequence, two-period, crossover, bioavailability study under fed conditions." ❝ Would stratification be applicable in BEQ studies from assessors point of view? Any experience? I think so but I am hard pressed to think of a common situation where it would be really relevant. — Pass or fail! ElMaestro |
Ing. Helmut Schütz