Assessing randomization sequences
Assesses randomization sequences based on specified issues in clinical trials.
assess(randSeq, ..., endp) ## S4 method for signature 'randSeq,missing' assess(randSeq, ..., endp) ## S4 method for signature 'randSeq,endpoint' assess(randSeq, ..., endp)
randSeq: object of class randSeq....: at least one object of class issue or just a list of objects of the class issue.endp: object of class endpoint, or missing.S4 object of class assessment summarizing the assessment of the randomization procedure.
Randomization sequences behave differently with respect to issues like selection bias, chronological bias, or loss in power estimation. The assess function evaluates the behavior of randomization sequences with respect to these issues. The first argument should be a result of one of the functions genSeq or getAllSeq. The second argument should be any number of issues arising in a clinical trial. The last argument endp may be provided if the assessment should take the distribution of the treatment groups into account, e.g. for power evaluation.
# assess the full set of Random Allocation Rule for N=4 patients sequences <- getAllSeq(rarPar(4)) issue1 <- corGuess("CS") issue2 <- corGuess("DS") issue3 <- imbal("imb") issue4 <- imbal("maxImb") assess(sequences, issue1, issue2, issue3, issue4) # assess one sequence of the Big Stick Design with respect to correct guesses sequence <- genSeq(bsdPar(10, 2), seed = 1909) assess(sequence, issue1) # assess the same sequence with respect to selection bias and power for a normal endpoint endp <- normEndp(c(2, 2), c(1, 1)) issue5 <- selBias("CS", 4, "exact") issue6 <- setPower(2, "exact") assess(sequence, issue1, issue5, issue6, endp = endp) # assess the same sequence with respect to selection bias for an exponential endpoint endp <- expEndp(lambda = c(0.5, 0.5), cenRate=0.1, accrualTime=1, cenTime=5) issue7 <- selBias("CS", 0.1, "exact") assess(sequence, issue1, issue7, endp = endp) # recommended plot for the assessment of rejection probabilities RP <- getAllSeq(crPar(6)) cB <- chronBias(type = "linT", theta = 1/6, method = "exact") sB <- selBias(type= "CS", eta = 1/4, method = "exact") normEndp <- normEndp(c(0, 0), c(1, 1)) A <- assess(RP, cB, sB, endp = normEndp) D <- A$D desiredSeq <- round(sum(D[,2][D[,3] <= 0.05 & D[,4] <= 0.05]), digits = 4) colnames(D) <- c("Seq", "Prob", "SB", "linT") g <- ggplot(D, aes(x = SB, y = linT)) g <- g + annotate("rect", xmin = 0, xmax = 0.05, ymin = 0, ymax = 0.05, alpha=0.2, fill="green") g <- g + geom_point(alpha = 1/10, size = 3, col = "orange") g <- g <- g + geom_vline(xintercept = 0.05, col = "red") g <- g + geom_hline(yintercept = 0.05, col = "red") g <- g + geom_text(data = NULL, x = 0, y = 0, label = paste("Proportion:", desiredSeq), hjust=0, vjust=0, size = 7) g
Representation of randomization procedures: randPar
Generation of randomization sequences: genSeq
issues for the assessment of randomization sequences
Useful links