Non-inferiority Sample Size Estimation [Power / Sample Size]
❝ […] 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 18.104.22.168.*) in
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)
… which agrees with FARTSSIE2.4
++++++++++++ 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)
❝ Can we assume the standard deviation …
❝ … 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 🖖🏼 Довге життя Україна!
The quality of responses received is directly proportional to the quality of the question asked. 🚮