To err is Julious [Regulatives / Guidelines]
Dear All!
In my previous post I have noticed that I could reproduce the power and sample size for non-inferiority trials only if I used
Seems even the pope can err .
The formulas given in the book
are in error here, not only for the non-inferiority case (chapter 6) but also for the superiority tests (chapter 3/4).
Compare them to those in the paper
where the formulas are yet given with power=1-probt(...).
Compare further the results of the normal approximation formulas
(f.i. formula 6.21
to see that 1-probt(...) is correct.
With tau=2.5, df=7, alpha=0.05 (in R syntax):
BTW: Would it desirable to have non-inferiority power / sample size in PowerTOST (although it is not based on TOST but on OOST - one one-sided t-test)? I could ask the author if he had some spare time .
In my previous post I have noticed that I could reproduce the power and sample size for non-inferiority trials only if I used
1-beta = 1 - pt(t1-alpha,n-2,df=n-2,tau)
Seems even the pope can err .
The formulas given in the book
S.A. Julious
"Sample Sizes for Clinical Trials"
Chapman & Hall/CRC, 2010
are in error here, not only for the non-inferiority case (chapter 6) but also for the superiority tests (chapter 3/4).
Compare them to those in the paper
S.A. Julious
"TUTORIAL IN BIOSTATISTICS
Sample sizes for clinical trials with Normal data"
Statist. Med. 2004; 23: 1921-1986
where the formulas are yet given with power=1-probt(...).
Compare further the results of the normal approximation formulas
(f.i. formula 6.21
1-beta = pnorm(tau-t1-alpha,n-2
)to see that 1-probt(...) is correct.
With tau=2.5, df=7, alpha=0.05 (in R syntax):
(6.21) pnorm(2.5 - qt(1-0.05, df=7)) =0.72755
(6.22) pt(qt(1-0.05,df=7), df=7, 2.5) =0.27283
1 - pt(qt(1-0.05,df=7), df=7, 2.5) =0.72717 q.e.d.
BTW: Would it desirable to have non-inferiority power / sample size in PowerTOST (although it is not based on TOST but on OOST - one one-sided t-test)? I could ask the author if he had some spare time .
—
Regards,
Detlew
Regards,
Detlew
Complete thread:
- Primary and secondary parameters for SS study for EU ratnakar1811 2012-03-19 12:49 [Regulatives / Guidelines]
- MR MD studies (EMA) Helmut 2012-03-19 14:18
- Non-superiority Jaime_R 2012-03-19 16:26
- Non-superiority Helmut 2012-03-19 16:59
- 'Non-superiority' - sample size d_labes 2012-03-20 15:48
- To err is Juliousd_labes 2012-03-22 10:20
- PowerTOST 0.9-6 on CRAN Helmut 2012-03-30 14:13
- Non-superiority Helmut 2012-03-19 16:59
- Non-superiority Jaime_R 2012-03-19 16:26
- MR MD studies (EMA) Helmut 2012-03-19 14:18