## Non-inferiority Sample Size Estimation [Power / Sample Size]

Hi Sury,

» […] sample size estimation for the non inferiority clinical trials […] I have tried to do the same in FARTSSIE23 (Non-inferiority, Parallel), but in addition to that Standard deviation is necessary for the same and in SAS (PROC POWER), we need to provide the CV to estimate the sample size

No idea about Proc Power. Let’s try the example of FARTSSIE2.4 (which is based on Julious’ Example 4.1.1.1.*) in PowerTOST:

library(PowerTOST) design   <- "parallel"            # Well... desired  <- 0.90                  # Target power alpha    <- 0.025                 # Probability of type I error sigma    <- 40                    # Common (pooled) standard deviation margin   <- 10                    # Maximum allowed difference mean.A   <- 160                   # Test mean.B   <- 158                   # Reference theta0   <- mean.A - mean.B       # Expected difference if (theta0 > 0) theta0 <- -theta0 # Force non-inferiority logscale <- FALSE sampleN.noninf(alpha=alpha, CV=sigma, logscale=logscale, margin=margin,                theta0=theta0, targetpower=desired, design=design)

Gives

++++++++++++ Non-inferiority test +++++++++++++             Sample size estimation ----------------------------------------------- Study design:  2 parallel groups untransformed data (additive model) alpha = 0.025, target power = 0.9 Non-inf. margin = 10 True diff. = -2,  CV = 40 Sample size (total)  n     power 470   0.900652

… which agrees with FARTSSIE2.4

» Can we assume the standard deviation …

Sure.

» … or we need to provide the exact Standard deviation (obtained from the literatures on the drug) ?

That’s also an estimate. The true value is unknown.

» Dose the same criteria applicable as that of the ISCV concept for the bio equivalence studies?

Not sure what you mean here. Can you try to explain?

• Julious SA. Sample sizes for clinical trials with Normal data. Stat Med. 2004;23(12):1949–50. doi:10.1002/sim.1783

Dif-tor heh smusma 🖖
Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes Ing. Helmut Schütz 