Bioequivalence and Bioavailability Forum

Dear All!

The (official) version on CRAN, promised by Helmut for April, is out now.

Beside the functions scABEL.ad() and sampleN.scABEL.ad() now allowing simulations via subject data there is another goody:
power.scABEL.sdsims() has an argument design_dta which you may use to evaluate designs with missing data. specify here a data.frame with columns subject, sequence, period and tmt without the rows were missings occured. Optional specify an additional column logval (log-transformed PK metric) with NA indicating the missings.
This feature is experimental! Only use it if you know what you do.

Example:
# full data partial replicate design (CVwT==CVwR), n=36 (12 in each sequence group)
# power at theta0=0.9
power.scABEL.sdsims(CV=0.3, n=36)
[1] 0.67794
# TIE
power.scABEL.sdsims(CV=0.3, n=36, theta0=1.25, nsims=1e6)
[1] 0.071743

# create the data.frame (misuse a hidden function in PowerTOST)
dta <- PowerTOST:::prep_data2(seqs=c("TRR", "RTR", "RRT"), nseq=c(12, 12, 12), muR=log(10),
                              ldiff=log(0.9), s2wT=CV2mse(0.3), s2wR=CV2mse(0.3))
# missings for subject 1 and 2 in period 3, sequence group TRR
dta$logval[dta$subject==1 & dta$period==3] <- NA
dta$logval[dta$subject==2 & dta$period==3] <- NA
# power at theta0=0.9
power.scABEL.sdsims(CV=0.3, design_dta=dta)
[1] 0.67038
# TIE
power.scABEL.sdsims(CV=0.3, design_dta=dta, theta0=1.25, nsims=1e6)
[1] 0.072705

# Evaluation of subjects with full data (excl. subject 1,2)
dta0 <- dta[!(dta$subject %in% c(1,2)), ]
# power at theta0=0.9
power.scABEL.sdsims(CV=0.3, design_dta=dta0)
[1] 0.65273
# or equivalently
power.scABEL.sdsims(CV=0.3, n=c(10, 12, 12))
[1] 0.65273
# TIE
power.scABEL.sdsims(CV=0.3, design_dta=dta0, theta0=1.25, nsims=1e6)
[1] 0.071462
# or equivalently
power.scABEL.sdsims(CV=0.3, n=c(10, 12, 12), theta0=1.25, nsims=1e6)
[1] 0.071462