Page 64 line 20 sampcc.surv should be sampcl.surv
Page 80 The function ssizecl.rate uses an iterative procedure to decide the number of clusters when the number of clusters is expected to be less than 20. The published function uses the procedure for binary comparisons, which gives an error message since the proportions have not been defined. In fact the formula by Hayes and Moulton(2009), which simply adds one cluster per arm to correct for small numbers is probably OK for small numbers of clusters and so the iterative procedure is unnecessary and should be removed.
Thus the function is
ssizecl.rate<-function(alpha, beta, lamdac, propchange, pyear ,covt, covc)
{
phi<-(qnorm(1-alpha/2)+qnorm(1-beta))^2
lamdat<-lamdac*(1-propchange)
k<-1+phi*((lamdat+lamdac)/pyear+(covt^2*lamdat^2+covc^2*lamdac^2))/(lamdat-lamdac)^2
# sample size for 2 x alpha sig and betax100 power
k<-ceiling(k)
cat("number of clusters per arm", k, "\n")
}
Page 81 In the function ssizecl.cross, line 3 '+m*eta' should be '-m*eta'
The example in Section 3.18.2 uses this correction
Page 80 The function ssizecl.rate uses an iterative procedure to decide the number of clusters when the number of clusters is expected to be less than 20. The published function uses the procedure for binary comparisons, which gives an error message since the proportions have not been defined. In fact the formula by Hayes and Moulton(2009), which simply adds one cluster per arm to correct for small numbers is probably OK for small numbers of clusters and so the iterative procedure is unnecessary and should be removed.
Thus the function is
ssizecl.rate<-function(alpha, beta, lamdac, propchange, pyear ,covt, covc)
{
phi<-(qnorm(1-alpha/2)+qnorm(1-beta))^2
lamdat<-lamdac*(1-propchange)
k<-1+phi*((lamdat+lamdac)/pyear+(covt^2*lamdat^2+covc^2*lamdac^2))/(lamdat-lamdac)^2
# sample size for 2 x alpha sig and betax100 power
k<-ceiling(k)
cat("number of clusters per arm", k, "\n")
}
Page 81 In the function ssizecl.cross, line 3 '+m*eta' should be '-m*eta'
The example in Section 3.18.2 uses this correction