Package 'CTP' reference manual

Title:	Closed Testing Procedure (CTP)
Description:	This is a package for constructing hypothesis trees for treatment comparisons based on the closure principle and analysing the corresponding Closed Testing Procedures (CTP) using adjusted p-values. For reference, see Marcus, R., Peritz, E, and Gabriel, K.R. (1976) <doi:10.2307/2335748> and Bauer, P (1991) <doi:10.1002/sim.4780100609>.
Authors:	Paul Jordan [aut, cre], Matthias Herpers [aut]
Maintainer:	Paul Jordan <[email protected]>
License:	GPL (>=2)
Version:	3.0.2
Built:	2024-12-02 04:13:36 UTC
Source:	https://github.com/openpharma/CTP

Adjusting p-values, auxiliary function

Description

Adjusting p-values, auxiliary function

Usage

Adjust_p(ctp.struc, ctp.pval)
Adjust_p(ctp.struc, ctp.pval)

Arguments

`ctp.struc`	Object of class `ctp.str`.
`ctp.pval`	Data frame with hypothesis names and unadjusted p-values. Output from `CTPcompare`.

Value

Adjusted p-values

Adjusting raw p-values of a CTP

Description

Function that adjusts the raw p-values of the elementary hypotheses of a closed testing procedure. The raw p-values are adjusted according to the closure principle. The adjusted p-value is calculated as the maximum of the raw p-value from the current hypothesis in question and the raw p-values from all subsequent hypotheses that contain the current hypothesis.

Usage

Adjust_raw(
  ctp.struc,
  p.value,
  dataset.name = NULL,
  factor.name = NULL,
  factor.levels = NULL,
  model = NULL,
  test.name = NULL
)
Adjust_raw(
  ctp.struc,
  p.value,
  dataset.name = NULL,
  factor.name = NULL,
  factor.levels = NULL,
  model = NULL,
  test.name = NULL
)

Arguments

`ctp.struc`	Object generated by `IntersectHypotheses`
`p.value`	Vector of raw p-values in the order of the hypotheses created by `summary.ctp.str`
`dataset.name`	Character string naming the analysis dataset (optional - only for documentation purposes).
`factor.name`	Character string naming the factor whose levels are compared (optional - only for documentation purposes).
`factor.levels`	Vector of type "character" containing the levels of the treatment factor (optional - only for documentation purposes).
`model`	Model used in the analysis (optional - only for documentation purposes).
`test.name`	Character string naming the statistical test applied.

Value

An object of oldClass = "ctp" to be used for summarizing and plotting the results.

Examples


Pairwise <- IntersectHypotheses(list(c(1,2), c(1,3), c(1,4), c(2,3), c(2,4), c(3,4)))
Display(Pairwise)
summary(Pairwise)

# the vector of p-values calculated by another software

  p.val <- c(
            0.4374,
            0.6485,
            0.4103,
            0.2203,
            0.1302,
            0.6725,
            0.4704,
            0.3173,
            0.6762,
            0.7112,
            0.2866,
            0.3362,
            0.2871,
            0.4633)

 result <- Adjust_raw(ctp.struc=Pairwise, p.value=p.val)

 ## details may be documented

 result <- Adjust_raw(Pairwise, p.value=p.val
          ,dataset.name="my Data", factor.name="Factor"
          ,factor.levels=c("A","B","C","D"), model=y~Factor
          ,test.name="my Test")

 summary(result)
 Display(result)


Pairwise <- IntersectHypotheses(list(c(1,2), c(1,3), c(1,4), c(2,3), c(2,4), c(3,4)))
Display(Pairwise)
summary(Pairwise)

# the vector of p-values calculated by another software

  p.val <- c(
            0.4374,
            0.6485,
            0.4103,
            0.2203,
            0.1302,
            0.6725,
            0.4704,
            0.3173,
            0.6762,
            0.7112,
            0.2866,
            0.3362,
            0.2871,
            0.4633)

 result <- Adjust_raw(ctp.struc=Pairwise, p.value=p.val)

 ## details may be documented

 result <- Adjust_raw(Pairwise, p.value=p.val
          ,dataset.name="my Data", factor.name="Factor"
          ,factor.levels=c("A","B","C","D"), model=y~Factor
          ,test.name="my Test")

 summary(result)
 Display(result)

Analysing a Closed Testing Procedure

Description

Calculation of p-values of a closed testing procedure (CTP). The function returns an object of oldClass "ctp"; summary() and Display() can be applied.

Usage

AnalyseCTP(ctp.struc, model, data, factor.name = NULL, test.name = "F", ...)
AnalyseCTP(ctp.struc, model, data, factor.name = NULL, test.name = "F", ...)

Arguments

`ctp.struc`	Object generated by the function `IntersectHypotheses` (structure of CTP)
`model`	model of the form response~treatment. If `testname="F"`, the model can be extended by covariates and other factors. In the case of a Logrank test the response must be imputed as `Surv(time,status)`.
`data`	Dataframe, missing values in the response or treatment variable are not allowed!
`factor.name`	Character string naming the factor whose levels are compared (treatment factor). By default the first variable of the right-hand side of the model formula is used.
`test.name`	One of the following strings `"F"` - F-Test (ANOVA from linear model, default) `"glm"` - generalised linear model `"kruskal"` -Kruskal-Wallis-Test `"chisq"` - Chi square test `"prob"` - Fisher's exact test for total number of observations <200 else Chi square test `"lgrank"` - Logrank-test `"jonckheere"` - Jonckheere-Terpstra test of ordered alternatives `"glm"` - generalised linear model, using function `glm` from `stats`.
`...`	Additional arguments for the chosen test.

Details

The hypothesis tree of the closed testing procedure must be created using IntersectHypotheses. For more details on the theory and the implementation as well for many examples, see the vignettes.

Value

An object of old class(ctp), consisting of a list with:

CTPparms: List with objects describing the CTP setup.
pvalues: Dataframe with all tested hypotheses, raw and adjusted p-values.

Note

This procedure is constructed for testing differences and two-sided hypotheses, but not for equivalence tests. It is further based on independent samples from the population involved (i.e. on parallel group designs, but not on cross-over designs).

Examples

	data(pasi)
	three.to.first <- IntersectHypotheses(list(1:2,c(1,3),c(1,4)))
	Display(three.to.first)
	pasi.ctp.F1 <- AnalyseCTP(three.to.first,pasi.ch~dose,pasi)
	summary(pasi.ctp.F1)
	Display(pasi.ctp.F1)

data(pasi)
	three.to.first <- IntersectHypotheses(list(1:2,c(1,3),c(1,4)))
	Display(three.to.first)
	pasi.ctp.F1 <- AnalyseCTP(three.to.first,pasi.ch~dose,pasi)
	summary(pasi.ctp.F1)
	Display(pasi.ctp.F1)

Data from a Study in Colorectal Cancer

Description

A dataframe containing the responders to the treatment with two doses of an experimental drug or a standard treatment.

Usage

data(colorectal)
data(colorectal)

Format

A data frame with 104 observations and 2 variables:

dose: a factor with levels control low high
responder: a factor with levels no yes

CTP, a package for closed testing procedures

Description

Library of functions to design and analyse closed testing procedures for the comparison of population parameters based on independent samples.

Details

Library of functions to design and analyse closed testing procedures for treatment comparisons. The comparison of the population parameters are based on independent samples. The function IntersectHypotheses creates the hypothesis tree (closure set) of a closed testing procedure. All possible intersecting hypotheses derived from the list of elementary hypotheses(i.e. the hypotheses to be tested) are produced. The resultant hypothesis tree will consist of the elementary hypotheses as well as all subsequent intersected hypotheses. The procedure ends when one hypothesis (the global hypothesis) remains. In this way, for each elementary hypothesis all hypotheses implying it can be found.

The analysis is performed using AnalyseCTP. First the raw p-values are computed for all hypotheses of the hypothesis tree, then the p-values of the elementary hypotheses are adjusted according to the closure principle i.e. the adjusted p-value is calculated as the maximum of the raw p-value of the hypothesis in question and of the raw p-values of all hypotheses implying it.

Instead of applying AnalyseCTP, the raw p-values can be computed using any other available software. For this purpose, the functions Adjust_raw is provided.

and summary.ctp.str generates a data frame comprising all the hypothesis of the hypothesis tree. The function Adjust_raw calculates the p-values for the given hypothesis tree from the raw p-values provided in the same order as the hypotheses occur in the data frame created by summary.ctp.str.

The results are presented using the generic functions summary and Display.

Author(s)

J. Bock & P.Jordan

Comparisons of treatment levels

Description

Performes all required comparisons of treatment levels

Usage

CTPcompare(CTPparms, ...)
CTPcompare(CTPparms, ...)

Arguments

`CTPparms`	Object of class `CTPparms` (i.e. generated by the function `getCTPparms`
`...`	Additional arguments for the chosen test

Value

A data.frame with the hypotheses and the unadjusted p-values

Auxiliary function

Description

Extracting CTP - parameters

Usage

getCTPparms(
  ctp.struc,
  model,
  dataset,
  factor.name = NULL,
  test.name = "F",
  ...
)
getCTPparms(
  ctp.struc,
  model,
  dataset,
  factor.name = NULL,
  test.name = "F",
  ...
)

Arguments

`ctp.struc`	Object of class `ctp.str`.
`model`	Formula of the form response~treatment.
`dataset`	Dataframe.
`factor.name`	Character string naming the factor whose levels are compared (treatment factor). By default the first variable of the right-hand side of the model formula is used.
`test.name`	One of the following strings `"F"` - F-Test (ANOVA, default) `"kruskal"` -Kruskal-Wallis-Test `"chisq"` - Chi square test `"prob"` - Fisher's exact test for total number of observations <200 else Chi square test `"lgrank"` - Logrank-test `"jonckheere"` - Jonckheere-Terpstra test of ordered alternatives `"glm"` - generlaized linear model
`...`	Additional arguments for the chosen test

Value

A list with CTP parameters

Auxiliary function

Description

Getting the response vector out of an object of class 'formula'

Usage

getResponse(formula, data)
getResponse(formula, data)

Arguments

`formula`	Model formula
`data`	Dataframe

Value

Vector of response

Examples


data(pasi)

data(pasi)
getResponse(pasi.ch ~ dose, data=pasi)

library(survival)
data(ovarian)
ovarian$subgroups <- as.factor(10*ovarian$ecog.ps+ovarian$rx)

surv <- getResponse(Surv(futime,fustat)~rx, data=ovarian)
print(surv)
class(surv)
data(pasi)

data(pasi)
getResponse(pasi.ch ~ dose, data=pasi)

library(survival)
data(ovarian)
ovarian$subgroups <- as.factor(10*ovarian$ecog.ps+ovarian$rx)

surv <- getResponse(Surv(futime,fustat)~rx, data=ovarian)
print(surv)
class(surv)

Data from a study in Diabetes II patients

Description

The dataframe glucose comprises the baseline values of fasting plasma glucose (mmol/L) and their changes from baseline to the end of the study.

Usage

data(glucose)
data(glucose)

Format

A data frame with 89 observations and 3 variables:

GLUCOSE.BLA: Fasting plasma glucose at baseline
GLUCOSE.CHANGE: Changes in fasting plasma glucose from baseline
DOSE: An ordered factor with levels PLACEBO < LOW < MEDIUM < HIGH

Intersection of hypotheses

Description

Generation of the hypotheses tree of a closed testing procedure (CTP). The function returns an object of oldClass "ctp.str"; summary and Display can be applied.

Usage

IntersectHypotheses(hyplst)
IntersectHypotheses(hyplst)

Arguments

hyplst

A list of integer vectors representing the elementary hypotheses.

Details

Sets of elementary hypotheses are described by lists of integer vectors eg. list(1:2, c(1,3), c(1,4)) if the populations 2, 3 and 4 have to be compared to population 1. For the generation of the hypothesis tree of a closed testing procedure first all intersections of the elementary hypothese; then all intersections of these intersections etc. have to be created. The set of hypotheses at each intersection level must be reduced by deleting double hypotheses and creating unions. The size of the hypothesis tree increases rapidly with growing number of elementary hypotheses, which can lead to memory and/or time problems! The intersection procedure ends if a single hypothesis (the global hypothesis) is left.

Value

hypothesis The hypothesis tree described as lists hypotheses at each intersection level.
relations The relation structure of the hypotheses tree used for plotting.
hypnames Character representations of the hypotheses used for summary tables and plotting.
connections The connections used for calculating the adjusted p-values.

Note

This procedure is constructed for null-hypotheses describing the equality of the same parameter for different populations.

Warning: The size of the hypothesis tree increases rapidly with growing number of elementary hypotheses. This can lead to memory and/or time problems!

Examples


hlist  <- list(1:2, c(1,3), c(1,4))
htree  <- IntersectHypotheses(hlist)
summary(htree)
Display(htree)


# compare in a set of six means all others to the fourth mean
#
five.to.fourth <- IntersectHypotheses(list(c(1,4),c(2,4),c(3,4),c(4,5),c(4,6)))
Display(five.to.fourth)

hlist  <- list(1:2, c(1,3), c(1,4))
htree  <- IntersectHypotheses(hlist)
summary(htree)
Display(htree)


# compare in a set of six means all others to the fourth mean
#
five.to.fourth <- IntersectHypotheses(list(c(1,4),c(2,4),c(3,4),c(4,5),c(4,6)))
Display(five.to.fourth)

Data from a Psoriasis study

Description

The dataframe pasi comprises the changes in PASI-score (Psoirasis Area and Severity Index) from Baseline within two months in 72 patients treated with different doses of Etretin or Placebo in a double blind study.

Usage

data(pasi)
data(pasi)

Format

A data frame with 72 observations and 3 variables:

treatment: a factor with levels Etretin10mg Etretin25mg Etretin50mg Placebo
pasi.ch: Changes in PASI score within two months
dose: Dose of Etretin as ordered factor with levels Placebo < ET.10mg < ET.25mg < ET.50mg

Summarising and plotting the results of hypothesis trees and closed testing procedures

Description

The functions summary.ctp.str and Display.ctp.str summarise and plot the hypothesis tree of a closed testing procedure, whereas the functions summary.ctp and Display.ctp summarise and plot the results of a closed testing procedure.

Usage

## S3 method for class 'ctp'
summary(object, digits = 4, ...)

## S3 method for class 'ctp.str'
summary(object, ...)

Display(object, ...)

## S3 method for class 'ctp'
Display(object, Type = "s", arrow = FALSE, ...)

## S3 method for class 'ctp.str'
Display(object, Type = "s", arrow = FALSE, ...)
## S3 method for class 'ctp'
summary(object, digits = 4, ...)

## S3 method for class 'ctp.str'
summary(object, ...)

Display(object, ...)

## S3 method for class 'ctp'
Display(object, Type = "s", arrow = FALSE, ...)

## S3 method for class 'ctp.str'
Display(object, Type = "s", arrow = FALSE, ...)

Arguments

`object`	Object generated by `AnalyseCTP` or `IntersectHypotheses`.
`digits`	The minimum number of significant digits to be printed for the p-values in the summary.
`...`	Additional arguments passed to `Display`.
`Type`	Type of the plotted tree: `"s"` straight lines, `""` tree type.
`arrow`	If `TRUE`, an arrow is plotted in the connections.

Details

@name Generic_Functions

@aliases summary Display

Value

summary.ctp.str provides lists with dataframes comprising the hypothesis names as strings and the relationship between the hypotheses.
summary.ctp provides a table of hypotheses, raw p-values and adjusted p-values.

Note

Disply.ctp.str plots the hypothesis tree.
Display.ctp plots the hypothesis tree together with the unadjusted and adjusted p-values.

Testing set for an elementary hypothesis

Description

Testing set for an elementary hypothesis

Usage

TestingSet(ctp.struc, Hyp)
TestingSet(ctp.struc, Hyp)

Arguments

`ctp.struc`	Object of class `ctp.str`.
`Hyp`	Elementary hypothesis (character variable).

Value

The testing set for the elementary hypothesis (character vector).

Examples


	Pairwise <- IntersectHypotheses(list(c(1,2), c(1,3),
	                                c(1,4), c(2,3), c(2,4), c(3,4)))
   Set24    <- TestingSet(Pairwise,"[24]")
   Set24
   
Pairwise <- IntersectHypotheses(list(c(1,2), c(1,3),
	                                c(1,4), c(2,3), c(2,4), c(3,4)))
   Set24    <- TestingSet(Pairwise,"[24]")
   Set24

Package 'CTP'

Help Index

Adjusting p-values, auxiliary function

Description

Usage

Arguments

Value

Adjusting raw p-values of a CTP

Description

Usage

Arguments

Value

See Also

Examples

Analysing a Closed Testing Procedure

Description

Usage

Arguments

Details

Value

Note

See Also

Examples

Data from a Study in Colorectal Cancer

Description

Usage

Format

CTP, a package for closed testing procedures

Description

Details

Author(s)

Comparisons of treatment levels

Description

Usage

Arguments

Value

Auxiliary function

Description

Usage

Arguments

Value

Auxiliary function

Description

Usage

Arguments

Value

Examples

Data from a study in Diabetes II patients

Description

Usage

Format

Intersection of hypotheses

Description

Usage

Arguments

Details

Value

Note

See Also

Examples

Data from a Psoriasis study

Description

Usage

Format

Summarising and plotting the results of hypothesis trees and closed testing procedures

Description

Usage

Arguments

Details

Value

Note

See Also

Testing set for an elementary hypothesis

Description

Usage

Arguments

Value

Examples