curerate: Power Calculation for Weighted Log-Rank Tests in Cure Rate Models

Description

This R-function performs power calculation in a group sequential clinical trial with censored survival data and possibly unequal patient allocation between treatment and control groups. The function can also be used to determine the study duration in a clinical trial with censored survival data as the sum of the accrual duration, which determines the sample size in a traditional sense, and the follow-up duration, which more or less controls the number of events to be observed.

Usage

curerate(rho = 0, numreps, cureobs, curerx, medobs, medrx, acrate, probrx, 
        actime, futime, info, crits, alpha = 0.025)

Arguments

rho

numeric specifying the value of rho in the G-rho test (Harrington and Fleming, 1982), rho=0 (default) gives the log-rank test, rho=1 the Peto-Peto Wilcoxon test, and rho=-1 the test discuss by Gray and Tsiatis, 1989.

numreps

integer > 0 specifying the number of replications.

cureobs

numeric specifying the probability of cure on the observation arm of the study.

curerx

numeric specifying the probability of cure on the treatment arm of the study.

medobs

numeric > 0 specifying the median survival among the non-cured on observation.

medrx

numeric > 0 specifying the median survival among the non-cured on treatment.

acrate

numeric > 0 specifying the rate at which patients accrue, combined over both observation and treatment arms of the study.

probrx

numeric specifying the probability of assigment to the treatment arm, must be in [0,1].

actime

vector specifying the accrual durations.

futime

vector specifying the follow-up durations.

info

vector of information times for the interim looks; must be an increasing sequence in (0,1].

crits

vector of critical values corresponding to the interim analysis time points specified in info.

alpha

numeric specifying the one-sided level of significance; must be in (0,1); default value 0.025; currently, it has no effect on the calculations, it is ignored in the current version of the software.

Value

An object of S4 class CureRate which has the following 22 components
cureobsnumeric
medobsnumeric
curerxnumeric
medrxnumeric
actimenumeric vector
futimenumeric vector
infonumeric vector
critsnumeric vector
alphanumeric
rhonumeric
acratenumeric
probrxnumeric
numrepsinteger
numobsnumeric matrix
timeptnumeric 3D array
deathsnumeric 3D array
testnamecharacter
powernumeric 3D array
numobsnumeric matrix
indacvector of integers
indfuvector of integers
printflaginteger

Details

The function curerate is used to calculate the power of weighted log-rank tests in cure rate models.

References

Ewell, M. and Ibrahim, J.G. (1997). The Large Sample Distribution of the Weighted Log Rank Statistic Under General Local Alternatives. Lifetime Data Analysis,3,5-12 Kim, K. (1992). Study Duration for Group Sequential Clinical Trials with Censored Survival Data Adjusting for Stratification. Statistics in Medicine, Vol. 11, 1477-1788 Harrington, D. P. and Fleming, T. R. (1982). A class of rank test procedures for censored survival data. Biometrika 69, 553-566. Gray, R.J. and Tsiatis, A.A. (1989). A linear rank test for use when the main interest is in differences in cure rates. Biometrics 45, 899-904.

Examples

Run this code

##---- Should be DIRECTLY executable !! ----
## Input Data
################
# rho - specifies the value of rho in the G-rho test (Harrington and Fleming, 1982). 
# rho = 0 gives the logrank test, and rho = 1 the Peto-Peto Wilcoxon test
#(and rho = -1 the test discussed by Gray and Tsiatis, 1989).
rho = 0 

numreps = 500
##cureobs - probability of cure on the observation arm of the study
cureobs = .359
##curerx - probability of cure on the treatment arm of the study
curerx = .459
##medobs - median survival among the non-cured on observation
medobs = .747
##medrx - median survival among the non-cured on treatment
medrx = .859
##acrate - rate at which patients accrue, combined over both 
##observation and treatment arms of the study
acrate = 232
##probrx - probability of assignment to the treatment arm
probrx = .5
##actime  - accrual duration
actime = c(3.3, 3.5, 3.8); 
##futime  - followup duration
futime = c(2.0,  2.5) 
##info - vector of information times for the interim looks
##must be an increasing sequence in (0,1]
info = c( .35, .61, .86, 1.0)
##crits - vector of critical values corresponding to the 
##interim analysis time points specified in info
crits = c(3.6128, 2.6506, 2.1894, 2.0536)#############################################################


### Log-rank test: rho=0 (default)
mycr<-curerate(rho,numreps,cureobs,curerx,medobs,medrx,acrate,
          probrx,actime,futime,info,crits)
mycr       ### (*)

show(mycr)   ### same as above
showcr(mycr) ### same as above 
showcr(mycr,full.result=TRUE)
showcr(mycr,indac=2,indfu=1)
showcr(mycr,indac=2:3,indfu=2:1)
showcr(mycr,indac=c(1,3),indfu=2)

show(mycr)   ### same as (*) above
showcr(mycr) ### same as above
mycr         ### same as above

### Peto & Peto modification of the Gehan-Wilcoxon test: rho=1
rho=1
mycr<-curerate(rho,numreps,cureobs,curerx,medobs,medrx,acrate,
          probrx,actime,futime,info,crits)
mycr

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning