bifactorial (version 1.4.7)

avemax: AVE- and MAX-test

Description

Compute global tests for factorial dose-response designs following Hung (2000) or by a bootstrap algorithm.

Usage

avetest(C,test=NULL,method="bootstrap",nboot=NULL,simerror=NULL,...)
maxtest(C,test=NULL,method="bootstrap",nboot=NULL,simerror=NULL,...)

Arguments

C

An object of class carpet or cube.

test

Either "ttest" or "ztest" - the test statistic for the inferences to be based on. Use "ztest" for binary data applications.

method

The calculation method - use "bootstrap" for a resampling-based approach and "hung" for calculations using the multivariate normal distribution.

nboot

The number of bootstrap iterations to use.

simerror

Prespecified simulation standard error.

...

Any further arguments.

Value

An object of class avetest or maxtext, respectively, with the following slots. The slot name is available for the MAX-test only.

p

p-value for the AVE- or MAX-test.

stat

Observed AVE- or MAX-statistic.

test

Type of test statistic which the AVE- or MAX-test was based on.

method

Algorithm used for the calculation.

nboot

Total number of resampling iterations.

simerror

Simulation standard error.

name

Combination group where the maximum of the min-statistics was observed.

duration

Total computing duration in seconds.

call

The function call.

Details

When handling with data from factorial clinical trial designs, one is often interested in the question whether dose combinations in the trial have got a better effect than all of their component drugs, because regulatoric requirements demand a contribution to the efficacy by all components. The decision if any of the tested combination drugs has got this property can be based on the AVE- or MAX-statistics proposed by Hung, Chi and Lipicky (1993). The hypothesis that this is true for none of the combinations is rejected if the largest or the average of the min-statistics is sufficiently high. The functions avetest and maxtest calculate the corresponding p-values on carpet or cube objects with a new bootstrap algorithm, which is default, or by the multivariate method for unbalanced designs from Hung (2000). A resampling-based method is available also for binary data applications. The desired simulation accuracy always needs to be specified by the number nboot of simulations to perform or an upper bound simerror for the simulation standard error. If both are given, the two constraints will be held simultaneously. Depending on the type of data, the calculations can be based on Student's t-test for metric data or the Z-statistic for binary applications.

References

Frommolt P, Hellmich M (2009): Resampling in multiple-dose factorial designs. Biometrical J 51(6), pp. 915-31

Hellmich M, Lehmacher W (2005): Closure procedures for monotone bi-factorial dose-response designs. Biometrics 61, pp. 269-276

Hung HMJ, Chi GYH, Lipicky RJ (1993): Testing for the existence of a desirable dose combination. Biometrics 49, pp. 85-94

Hung HMJ, Wang SJ (1997): Large-sample tests for binary outcomes in fixed-dose combination drug studies. Biometrics 53, pp. 498-503 Hung HMJ (2000): Evaluation of a combination drug with multiple doses in unbalanced factorial design clinical trials. Statist Med 19, pp. 2079-2087

See Also

bifactorial, carpet, cube, mintest, margint

Examples

#Hypertension example from Hung (2000)
n<-c(75,75,74,48,74,75,74,49,48,50,48,48)
m<-c(0,1.4,2.7,4.6,1.8,2.8,5.7,8.2,2.8,4.5,7.2,10.9)
s<-rep(7.07,12)
x<-list(12)
for(i in 1:12){
  x[[i]]<-rnorm(n[i],mean=0,sd=1)
  x[[i]]<-((x[[i]]-mean(x[[i]]))*(s[i]/sd(x[[i]])))+m[i]
}
hung<-carpet(x,D=c(2,3))
avetest(hung,test="ttest",nboot=20000)
maxtest(hung,test="ttest",nboot=20000)