Helmut
★★★
avatar
Homepage
Vienna, Austria,
2020-10-18 11:09
(284 d 23:29 ago)

Posting: # 22022
Views: 1,332
 

 FARTSSIE v2.5 [Software]

Dear all,

up to v2.4 of 14 March 2019 FARTSSIE’s tab Bioequivalence, Replicate contained two boxes for reference-scaling. The sample size was wrong because no analytical solution for power exists and simulations are required instead.

In v2.5 of 13 October 2020 Dave deleted the boxes and suggests to install PowerTOST. He gives in two boxes the arguments for PowerTOST’s functions sampleN.scABEL() for Average Bioequivalence with Expanding Limts (EMA and many others, Health Canada) and sampleN.NTIDFDA() for the FDA’s reference-scaling for NTIDs.
However, in the former don’t use the argument regulator="FDA" as he suggests since RSABE  ABEL. Not only that the regulatory constants are different, these are different approaches (upper limit of the linearized criterion ≤0 vs expansion of the BE limits).

library(PowerTOST) # show the regulatory conditions
reg_const(regulator = "EMA")
EMA regulatory settings
- CVswitch            = 0.3
- cap on scABEL if CVw(R) > 0.5
- regulatory constant = 0.76
- pe constraint applied

reg_const(regulator = "HC")
HC regulatory settings
- CVswitch            = 0.3
- cap on scABEL if CVw(R) > 0.57382
- regulatory constant = 0.76
- pe constraint applied

reg_const(regulator = "FDA")
FDA regulatory settings
- CVswitch            = 0.3
- no cap on scABEL
- regulatory constant = 0.8925742
- pe constraint applied


Use the function sampleN.RSABE() instead. Examples with comments at the end.

Since in the survey 20% of participants reported to never update their software: Not a good idea.


library(PowerTOST)
sampleN.scABEL(CV = 0.6, design = "2x2x4", regulator = "EMA") # correct

+++++++++++ scaled (widened) ABEL +++++++++++
            Sample size estimation

   (simulation based on ANOVA evaluation)
---------------------------------------------
Study design: 2x2x4 (4 period full replicate)
log-transformed data (multiplicative model)
1e+05 studies for each step simulated.

alpha  = 0.05, target power = 0.8
CVw(T) = 0.6; CVw(R) = 0.6
True ratio = 0.9

ABE limits / PE constraint = 0.8 ... 1.25
EMA regulatory settings
- CVswitch            = 0.3
- cap on scABEL if CVw(R) > 0.5
- regulatory constant = 0.76
- pe constraint applied


Sample size search
 n     power
30   0.7851
32   0.8101


sampleN.scABEL(CV = 0.6, design = "2x2x4", regulator = "HC") # correct

+++++++++++ scaled (widened) ABEL +++++++++++
            Sample size estimation

(simulations based on intra-subject contrasts)
----------------------------------------------
Study design:  2x2x4 (full replicate)
log-transformed data (multiplicative model)
1e+05 studies for each step simulated.

alpha  = 0.05, target power = 0.8
CVw(T) = 0.6; CVw(R) = 0.6
True ratio = 0.9

ABE limits / PE constraint = 0.8 ... 1.25
HC regulatory settings
- CVswitch            = 0.3
- cap on scABEL if CVw(R) > 0.57382
- regulatory constant = 0.76
- pe constraint applied


Sample size search
 n     power
24   0.7505
26   0.7851
28   0.8118


sampleN.scABEL(CV = 0.6, design = "2x2x4", regulator = "FDA") # wrong

+++++++++++ scaled (widened) ABEL +++++++++++
            Sample size estimation
(simulations based on intra-subject contrasts)
----------------------------------------------
Study design:  2x2x4 (full replicate)
log-transformed data (multiplicative model)
1e+05 studies for each step simulated.

alpha  = 0.05, target power = 0.8
CVw(T) = 0.6; CVw(R) = 0.6
True ratio = 0.9

ABE limits / PE constraint = 0.8 ... 1.25
FDA regulatory settings
- CVswitch            = 0.3
- no cap on scABEL
- regulatory constant = 0.8925742
- pe constraint applied


Sample size search
 n     power
16   0.7017
18   0.7476
20   0.7813
22   0.8071


sampleN.RSABE(CV = 0.6, design = "2x2x4") # correct

++++++++ Reference scaled ABE crit. +++++++++
           Sample size estimation
---------------------------------------------
Study design: 2x2x4 (4 period full replicate)
log-transformed data (multiplicative model)
1e+05 studies for each step simulated.

alpha  = 0.05, target power = 0.8
CVw(T) = 0.6; CVw(R) = 0.6
True ratio = 0.9

ABE limits / PE constraints = 0.8 ... 1.25
FDA regulatory settings
- CVswitch            = 0.3
- regulatory constant = 0.8925742
- pe constraint applied


Sample size search
 n     power
16   0.67580
18   0.72735
20   0.76531
22   0.79589
24   0.81947


Dif-tor heh smusma 🖖
Helmut Schütz
[image]

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

Toronto,
2021-03-10 16:43
(141 d 16:55 ago)

(edited by ddubins on 2021-03-10 18:34)
@ Helmut
Posting: # 22256
Views: 826
 

 FARTSSIE v2.7

Dear everyone,

I have updated FARTSSIE to version 2.7, and migrated it (finally) to github:

https://github.com/dndubins/FARTSSIE

It's been a while since I've posted to this forum (oh it has to be years now!) so I thought I would give you a bit of background on where FARTSSIE originated. When I was at Allied Research International Inc., Dr. Richard Moisan trained me as a PK scientist. He was a patient teacher, always with a cigarette in his mouth and a keen eye for science. He was a data purist really, and was passionate in the way he ran and interpreted Phase I and II trials. "If you don't love your job, then get the f**k out" was one of the many phrases he would repeat to us. And by the way, we did love our jobs in the department, especially working for Richard. He would tell us stories about his years at Roche, and about his PhD that took so long because he investigated the effects of aging on pharmacokinetics in rats. "I had to wait until the rats got old", he'd say. When HR left the building, he would smoke inside. He was a character!

One day Richard told me there was a talk he was supposed to give at SoCRA, the Society of Clinical Research Associates. He asked me if I could present in his place. "What would I talk about?" I asked him. I had just come up with a crude Excel spreadsheet to estimate sample size for a 2-way crossover BE trial, so he responded "how about sample size?" I took on the challenge, and adapted the spreadsheet for the talk, presenting it (at the time) as a free alternative to what was available. The talk was fun, and the spreadsheet was born. Richard was amenable to posting it as freeware, although when Cetero Research bought Allied, Murray Ducharme suggested selling it. No, I insisted, this needs to be a free tool, thinking about my students who needed a button to push for their problem sets.

As the regulations changed, along with my role leaving consulting work and teaching full time at UofT, I lost contact with my industry ties. So many good people - Yan Liu, our crazy smart biostatistician, for one. I struggled marginally successfully keeping up with some methods and changes in regulations, but recently finally realized my conception of reference-scaling was at best oversimplified, and not giving proper estimates. Meanwhile, PowerTOST was developed in the language I love, so rather than re-inventing the wheel, I outsourced this method to PowerTOST.

I have received a lot of kind emails about FARTSSIE over the years. If you have a version of it somewhere on your hard drive, please update it to the latest, as I really have made a number of important improvements and corrections. Helmut helped me correct the suggested PowerTOST code.

Richard passed away a while ago now (we all warned him about the chain smoking!). So many stories, so much house wine at the Open Cork restaurant, and so many hysterical conflicts with QA, Richard was without a doubt my favourite mentor. I miss him dearly.

Kind regards,
Dave

David Dubins, Ph.D., B.A.Sc.
Associate Professor, Teaching Stream
Director, Pharmaceutical Chemistry Specialist Program
Leslie Dan Faculty of Pharmacy
University of Toronto

144 College Street (room PB802), Toronto, ON M5S 3M2
Tel. +1 416-946-5303; FAX: +1 416 978-8511
dshah
★    

India,
2021-03-11 09:33
(141 d 00:04 ago)

(edited by dshah on 2021-03-11 11:53)
@ ddubins
Posting: # 22257
Views: 773
 

 FARTSSIE v2.7

Greetings Dr. Dubin!
Thank you for the wonderful work and contribution in the field.
I hope to see more from you in excel.
Still struggling with installation of R in organization and because of that I had asked for Sample size for NTI- FDA method in excel.
Thank you very much for your contribution.
Regards,
DShah
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2021-03-11 21:24
(140 d 12:14 ago)

@ dshah
Posting: # 22260
Views: 775
 

 FARTSSIE v2.7

Hi DShah,

» Still struggling with installation of R in organization …

It’t high time for your organization to learn that you need the right tools to perform your job.

I suppose it is tempting, if the only tool you have is a hammer,
to treat everything as if it were a nail.
  Abraham H. Maslow (Toward a Psychology of Being, 1962)


» … and because of that I had asked for Sample size for NTI- FDA method in excel.

You would need to simulate 100,000 studies in each iteration (cause you will not hit the target with a lucky punch). I don’t say that’s impossible in Excel. But why the hell would one do that, if PowerTOST can give you the result in less than a second?

library(PowerTOST)
st <- proc.time()[[3]]
sampleN.NTIDFDA(CV = 0.1, theta0 = 0.975, targetpower = 0.8, design = "2x2x4")
cat("Runtime:",  proc.time()[[3]] - st, "seconds\n")

+++++++++++ FDA method for NTIDs ++++++++++++
           Sample size estimation
---------------------------------------------
Study design:  2x2x4 (TRTR|RTRT)
log-transformed data (multiplicative model)
1e+05 studies for each step simulated.

alpha  = 0.05, target power = 0.8
CVw(T) = 0.1, CVw(R) = 0.1
True ratio     = 0.975
ABE limits     = 0.8 ... 1.25
Implied scABEL = 0.8000 ... 1.2500
Regulatory settings: FDA
- Regulatory const. = 1.053605
- 'CVcap'           = 0.2142

Sample size search
 n     power
14   0.717480
16   0.788690
18   0.841790

Runtime: 0.24 seconds


Dif-tor heh smusma 🖖
Helmut Schütz
[image]

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

India,
2021-03-16 06:22
(136 d 03:16 ago)

@ Helmut
Posting: # 22267
Views: 665
 

 FARTSSIE v2.7

Thank you Helmut, Detlew Labes and Benjamin Lang!

I am really thankful for your PowerTOST package in R. It is fast, accurate and required information limited to study design, ISCV, Power and expected ratio. Because of this I am able to do the sample size estimation after working hours in personal system.
But as of now, my organization's IT system is not allowing me to install R.
And I want to explore bear and ivivc and many other things of R, for which I am banging my head.
Thank you for wonderful PowerTOST and Dave's contribution as well.
Regards,
Dshah
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2021-03-11 21:08
(140 d 12:30 ago)

@ ddubins
Posting: # 22259
Views: 742
 

 FARTSSIE v2.7

Hi Dave,

» I have updated FARTSSIE to version 2.7, and migrated it (finally) to github:
» https://github.com/dndubins/FARTSSIE

Great!

» It's been a while since I've posted to this forum (oh it has to be years now!)…

September 2010…

THX for the moving background-stories! :smoke:

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

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

Berlin, Germany,
2021-03-12 18:29
(139 d 15:08 ago)

@ ddubins
Posting: # 22264
Views: 718
 

 Richard Moisan

Dear Dave,

thank you so much for your background story.
After reading it it was such a pity to me not to know Dr. Richard Moisan.
You glad boy, met him and know him.
May he rest in peace.

All the best

Regards,

Detlew
Activity
 Admin contact
21,592 posts in 4,514 threads, 1,532 registered users;
online 12 (0 registered, 12 guests [including 6 identified bots]).
Forum time: Friday 10:38 CEST (Europe/Vienna)

It is a capital mistake to theorise before one has data.    Arthur Conan Doyle

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