OneArmPhaseTwoStudy (version 1.0.3)

getCP: Calculates the conditional power.

Description

Calculates the conditional power for a given Simon's two-stage design in the interim analysis if the number of patients which should be enrolled in the second stage is altert to "n2".

Usage

getCP(n2, p1, design, k, mode = 0, alpha = 0.05)

Arguments

n2

number of patients to be enrolled in the second stage of the study.

p1

response probability under the alternative hypothesis

design

a dataframe containing all critical values for a Simon's two-stage design defined by the colums "r1", "n1", "r", "n" and "p0".

• r1 = critical value for the first stage (more than r1 responses needed to proceed to the second stage).

• n1 = number of patients enrolled in the first stage.

• r = critical value for the whole trial (more than r responses needed at the end of the study to reject the null hypothesis).

• n = number of patients enrolled in the whole trial.

• p0 = response probability under the null hypothesis.

k

number of responses observed at the interim analysis.

mode

a value out of {0,1,2,3} dedicating the methode spending the "rest alpha" (difference between nominal alpha level and actual alpha level for the given design).

• 0 = "rest alpha" is not used.

• 1 = "rest alpha" is spent proportionally.

• 2 = "rest alpha" is spent equally.

• 3 = "rest alpha" is spent only to the worst case scenario (minimal number of responses at the interim analysis so that the study can proceed to the second stage).

alpha

overall significance level the trial was planned for.

References

Englert S., Kieser M. (2012): Adaptive designs for single-arm phase II trials in oncology. Pharmaceutical Statistics 11,241-249.

See Also

getN2

Examples

# NOT RUN {
#Calculate a Simon's two-stage design
design <- getSolutions()$Solutions[3,] #minimax-design for the default values.

#Assume 3 responses were observed in the interim analysis.
#Therefore the conditional power is only about 0.55.
#In order to raise the conditional power to 0.8 "n2" has to be increased.

#get the current "n2"
n2 <- design$n - design$n1

#set k to 3 (only 3 responses observed so far)
k = 3

#get the current conditional power
cp <- getCP(n2, design$p1, design, k, mode = 1, alpha = 0.05)
cp

# }
# NOT RUN {
#increase n2 until the conditional power is larger than 0.8
while(cp < 0.8){
n2 <- n2 + 1
# Assume we spent the "rest alpha" proportionally (in the planning phase)
# therefore we set "mode = 1".
cp <- getCP(n2, design$p1, design, k, mode = 1, alpha = 0.05)
}

n2
# }