A function for computing the bias adjusted point estimate for a statistic observed to cross the efficacy boundary.
A function for computing the bias adjusted point estimate for a statistic, on the Brownian scale, observed to cross the efficacy boundary.
EX1gXK(xk, b.eff, frac)
xk |
The observed value of the statistic, on the “Brownian” scale. |
b.eff |
Efficacy boundary points at current and prior analyses |
frac |
Information fraction at current and prior analyses |
Returns the expected value of X_1 given X_K, which is the bias adjusted point estimate
This works for the unweighted, proportional hazards case, but also works in the case of the weighted log-rank statistic when we assume the chosen weights are proportional to the true shape.
Grant Izmirlian <izmirlig@mail.nih.gov>
Emerson, S. S. (1993). Computation of the uniform minimum variance unibiased estimator of a normal mean following a group sequential trialdiscrete sequential boundaries for clinical trials. Computers and Biomedical Research 26 68–73.
Izmirlian, G. (2014). Estimation of the relative risk following group sequential procedure based upon the weighted log-rank statistic. Statistics and its Interface 00 00–00
# if Z.K = U_K/V_K^0.5 is the log-rank statistic on the standard normal # scale, then we obtain an estimate of the logged relative risk as follows # Suppose we've stopped at analysis number K=4, and Z.K = 2.5 # suppose the end of trial variance of the log-rank statistic # (specified in design and used to compute 'frac') is V.end = 100 K <- 4 Z.K <- 2.5 V.end <- 100 # Information fraction frac <- c(0.15, 0.37, 0.64, 0.76) # Efficacy Boundary gsb <- GrpSeqBnds(frac=frac, EfficacyBoundary=LanDemets(spending=ObrienFleming, alpha=0.05)) # Efficacy boundary points be <- gsb$table[,"b.e"] # Brownian scale X.K <- Z.K*frac[K] # expected value of X_1 given X_K ex1gxk <- EX1gXK(X.K, be, frac) # Crude estimate of logged relative risk X.K/(frac[K]*V.end^0.5) # Bias adjusted estimate of logged relative risk ex1gxk/(frac[1]*V.end^0.5)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.