How to Create Admirable Plots with rpact

Utilities
Sample size
Power simulation
This document provides many different examples for creating plots with rpact and ggplot2.
Author
Published

October 28, 2024

Preparation

First, load the rpact package

library(rpact)
packageVersion("rpact") # version should be version 3.0 or later
[1] '4.1.0'

Design plots

One-sided design with futility bounds

design <- getDesignGroupSequential(
    kMax = 3,
    typeOfDesign = "OF", 
    sided = 1,
    futilityBounds = c(0, 0.1)
)

design |> plot(type = 1)

design |> plot(type = 5, nMax = 10)

design |> plot(type = 6, nMax = 10)

Two-sided design

design <- getDesignGroupSequential(
    kMax = 4, 
    typeOfDesign = "OF", 
    sided = 2, 
    twoSidedPower = TRUE
)

design |> plot(type = 1)

design |> plot(type = 5, nMax = 10)

design |> plot(type = 6, nMax = 10)

Two-sided design with futility bound

design <- getDesignGroupSequential(
    beta = 0.05, 
    kMax = 4, 
    typeOfDesign = "asOF",
    typeBetaSpending = "bsOF", 
    sided = 2
)

design |> plot(type = 1)

design |> plot(type = 5, nMax = 10)

design |> plot(type = 6, nMax = 10)

Sample size plots

Sample size means (continuous endpoint)

Sample size means for a one-sided design with futility bounds

sampleSizeMeans1 <- getDesignGroupSequential(
    sided = 1,
    futilityBounds = c(0, 0.2)) |>
  getSampleSizeMeans()

sampleSizeMeans1 |> plot(type = 1)

sampleSizeMeans1 |> plot(type = 2)

Sample size means for a two-sided design

sampleSizeMeans2 <- getDesignGroupSequential(sided = 2) |>
  getSampleSizeMeans()

sampleSizeMeans2 |> plot(type = 1)

sampleSizeMeans2 |> plot(type = 2)

Sample size rates (binary endpoint)

Sample size rates for a one-sided design with futility bounds

sampleSizeRates1 <- getDesignGroupSequential(
    sided = 1,
    futilityBounds = c(0, 0.1)) |>
  getSampleSizeRates()

sampleSizeRates1 |> plot(type = 1)

sampleSizeRates1 |> plot(type = 2)

Sample size rates for a two-sided design

sampleSizeRates2 <- getDesignGroupSequential(sided = 2) |>
  getSampleSizeRates()

sampleSizeRates2 |> plot(type = 1)

sampleSizeRates2 |> plot(type = 2)

Sample size survival (survival endpoint)

Sample size survival for a one-sided design with futility bounds

design <- getDesignGroupSequential(
    kMax = 3, 
    typeOfDesign = "OF", 
    sided = 1,
    futilityBounds = c(-1, 0.5)
)

piecewiseSurvivalTime <- list(
    "0 - <6" = 0.025,
    "6 - <9" = 0.04,
    "9 - <15" = 0.015,
    "15 - <21" = 0.01,
    ">= 21" = 0.007
)

sampleSizeSurvival1 <- getSampleSizeSurvival(
    design = design,
    typeOfComputation = "Schoenfeld",
    thetaH0 = 1, 
    allocationRatioPlanned = 1, 
    kappa = 1,
    piecewiseSurvivalTime = piecewiseSurvivalTime,
    hazardRatio = c(0.5, 0.9)
)

sampleSizeSurvival1 |> plot(type = 1)

sampleSizeSurvival1 |> plot(type = 2)

sampleSizeSurvival1 |> plot(type = 13, legendPosition = 1)

sampleSizeSurvival1 |> plot(type = 14)

Sample size survival for a two-sided design

design <- getDesignGroupSequential(
    kMax = 3, 
    typeOfDesign = "OF",
    sided = 2, 
    twoSidedPower = TRUE
)

piecewiseSurvivalTime <- list(
    "0  - <14" = 0.015,
    "14 - <24" = 0.01,
    "24 - <44" = 0.005,
    ">= 44" = 0.0025
)
sampleSizeSurvival2 <- getSampleSizeSurvival(
    design = design,
    typeOfComputation = "Schoenfeld",
    thetaH0 = 1,
    allocationRatioPlanned = 1, 
    kappa = 1,
    piecewiseSurvivalTime = piecewiseSurvivalTime,
    maxNumberOfSubjects = 0, 
    hazardRatio = c(0.1, 0.2, 0.5, 0.6)
)

sampleSizeSurvival2 |> plot(type = 1)

sampleSizeSurvival2 |> plot(type = 2)

sampleSizeSurvival2 |> plot(type = 13, legendPosition = 1)

sampleSizeSurvival2 |> plot(type = 14)

Power plots

Power means (continuous endpoint)

Power means for a one-sided design with futility bounds

powerMeans1 <- getDesignGroupSequential(
    typeOfDesign = "OF", 
    sided = 1,
    futilityBounds = c(0, 0.5)) |>
  getPowerMeans(
    groups = 1, 
    meanRatio = FALSE,
    thetaH0 = 0, 
    alternative = c(-1, 4),
    stDev = 2, 
    normalApproximation = FALSE,
    maxNumberOfSubjects = 40
)

powerMeans1 |> plot(type = 1)

powerMeans1 |> plot(type = 2)

powerMeans1 |> plot(type = 5)

powerMeans1 |> plot(type = 6)

Power means for a two-sided design

powerMeans2 <- getDesignGroupSequential(
    typeOfDesign = "OF", 
    sided = 2, 
    twoSidedPower = TRUE) |>
  getPowerMeans(maxNumberOfSubjects = 30)

powerMeans2 |> plot(type = 1)

powerMeans2 |> plot(type = 2)

powerMeans2 |> plot(type = 5)

powerMeans2 |> plot(type = 6)

Power rates (binary endpoint)

Power rates for a one-sided design with futility bounds

powerRates1 <- getDesignGroupSequential(
    kMax = 3,
    typeOfDesign = "OF",
    sided = 1,
    futilityBounds = c(-1, 0.5)) |>
  getPowerRates(
    groups = 2, 
    riskRatio = TRUE,
    thetaH0 = 0.2, 
    allocationRatioPlanned = 1,
    pi1 = c(0.1, 0.4), 
    pi2 = 0.2,
    maxNumberOfSubjects = 80
)

powerRates1 |> plot(type = 1)

powerRates1 |> plot(type = 2)

powerRates1 |> plot(type = 5)

powerRates1 |> plot(type = 6)

Power rates for a two-sided design

powerRates2 <- getDesignGroupSequential(
    typeOfDesign = "OF", 
    sided = 2, 
    twoSidedPower = TRUE) |>
  getPowerRates(maxNumberOfSubjects = 120)

powerRates2 |> plot(type = 1)

powerRates2 |> plot(type = 2)

powerRates2 |> plot(type = 5)

powerRates2 |> plot(type = 6)

Power survival (survival endpoint)

Power survival for a one-sided design with futility bounds

powerSurvival <- getDesignGroupSequential(
    kMax = 3,
    typeOfDesign = "OF", 
    sided = 1,
    futilityBounds = c(0, 0)) |>
  getPowerSurvival(
    typeOfComputation = "Schoenfeld", 
    thetaH0 = 1,
    allocationRatioPlanned = 1, 
    kappa = 1,
    maxNumberOfSubjects = 2480,
    maxNumberOfEvents = 70
)

powerSurvival |> plot(type = 1)

powerSurvival |> plot(type = 2)

powerSurvival |> plot(type = 5)

powerSurvival |> plot(type = 6)

powerSurvival |> plot(type = 7)

powerSurvival |> plot(type = 12)

Power for a piecewise exponential survival distribution

piecewiseSurvivalTime <- list(
    "<5" = 0.04,
    "5 - <10" = 0.02,
    ">= 10" = 0.008
)

powerSurvival1 <- getDesignGroupSequential(
    kMax = 3, 
    typeOfDesign = "OF",
    sided = 2, 
    twoSidedPower = TRUE
) |>
  getPowerSurvival(
    typeOfComputation = "Schoenfeld", 
    thetaH0 = 1,
    allocationRatioPlanned = 1, 
    kappa = 1,
    piecewiseSurvivalTime = piecewiseSurvivalTime,
    maxNumberOfSubjects = 2480, 
    maxNumberOfEvents = 70,
    hazardRatio = c(0.5, 2)
)

powerSurvival1 |> plot(type = 1)

powerSurvival1 |> plot(type = 2)

powerSurvival1 |> plot(type = 12)

powerSurvival1 |> plot(type = 13, legendPosition = 1)

powerSurvival1 |> plot(type = 14, legendPosition = 5)

Power for a piecewise exponential survival distribution and a design with futility bounds

piecewiseSurvivalTime <- list(
    "0 - <6" = 0.025,
    "6 - <9" = 0.04,
    "9 - <15" = 0.015,
    "15 - <21" = 0.01,
    ">= 21" = 0.007
)

powerSurvival2 <- getDesignGroupSequential(
    typeOfDesign = "OF", 
    sided = 1,
    futilityBounds = c(0, 0.1)) |>
  getPowerSurvival(
    typeOfComputation = "Schoenfeld", 
    thetaH0 = 1,
    allocationRatioPlanned = 1, 
    kappa = 1,
    piecewiseSurvivalTime = piecewiseSurvivalTime,
    maxNumberOfSubjects = 2480, 
    maxNumberOfEvents = 70,
    hazardRatio = 0.8
)

powerSurvival2 |> plot(type = 1)

powerSurvival2 |> plot(type = 2)

powerSurvival2 |> plot(type = 13, legendPosition = 1)

powerSurvival2 |> plot(type = 14)

powerSurvival3 <- getDesignGroupSequential(
    typeOfDesign = "OF", 
    sided = 1,
    futilityBounds = c(0, 0.1)) |>
  getPowerSurvival(
    typeOfComputation = "Schoenfeld", 
    thetaH0 = 1,
    allocationRatioPlanned = 1, 
    kappa = 1,
    piecewiseSurvivalTime = c(0, 5, 10),
    lambda2 = c(0.025, 0.04, 0.015),
    lambda1 = c(0.02, 0.032, 0.012),
    maxNumberOfSubjects = 2480, 
    maxNumberOfEvents = 70
)

powerSurvival3 |> plot(type = 1)

powerSurvival3 |> plot(type = 2)

powerSurvival3 |> plot(type = 13, legendPosition = 1)

powerSurvival3 |> plot(type = 14)

Power survival for one lambda

powerSurvival4 <- getPowerSurvival(
    accrualTime = 12,
    lambda2 = 0.04,
    hazardRatio = 0.6,
    maxNumberOfSubjects = 1400,
    maxNumberOfEvents = 300
)

powerSurvival4 |> plot(type = 13, legendPosition = 1)

powerSurvival4 |> plot(type = 14, legendPosition = 5)

Power survival for default pi1 and pi2

powerSurvival5 <- getPowerSurvival(
    maxNumberOfSubjects = 1400,
    maxNumberOfEvents = 300
)

powerSurvival5 |> plot(type = 13, legendPosition = 1)

powerSurvival5 |> plot(type = 14, legendPosition = 5)

Simulation plots

Simulation means (continuous endpoint)

simulationResults1 <- getSimulationMeans(
    design = getDesignFisher(kMax = 2),
    plannedSubjects = c(20, 40), 
    maxNumberOfIterations = 1000,
    seed = 12345
)

simulationResults1 |> plot(type = "all", grid = 0)

simulationResults1 |> plot(type = 4)

simulationResults1 |> plot(type = 5)

simulationResults1 |> plot(type = 6)

Simulation rates (binary endpoint)

simulationResults1 <- getSimulationRates(
    design = getDesignFisher(kMax = 2),
    plannedSubjects = c(20, 40), maxNumberOfIterations = 1000,
    seed = 12345
)

plot(simulationResults1, type = 4)

plot(simulationResults1, type = 5)

plot(simulationResults1, type = 6)

Simulation plots survival (survival endpoint)

simulationResults1 <- getSimulationSurvival(
    accrualTime = 12,
    maxNumberOfSubjects = 1405,
    plannedEvents = 300, 
    maxNumberOfIterations = 1000,
    seed = 12345
)

simulationResults1 |> plot(type = 5)

simulationResults1 |> plot(type = 6)

simulationResults1 |> plot(type = 7)

simulationResults1 |> plot(type = 9)

simulationResults1 |> plot(type = 10)

simulationResults1 |> plot(type = 11)

simulationResults1 |> plot(type = 12)

simulationResults1 |> plot(type = 13)

simulationResults1 |> plot(type = 14)

simulationResults2 <- getSimulationSurvival(
    accrualTime = 12,
    lambda2 = 0.03,
    hazardRatio = 0.8,
    maxNumberOfSubjects = 1405,
    plannedEvents = 300, 
    maxNumberOfIterations = 1000,
    seed = 23456
)

simulationResults2 |> plot(type = 13)

simulationResults2 |> plot(type = 14)


System: rpact 4.1.0, R version 4.3.3 (2024-02-29 ucrt), platform: x86_64-w64-mingw32

To cite R in publications use:

R Core Team (2024). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.

To cite package ‘rpact’ in publications use:

Wassmer G, Pahlke F (2024). rpact: Confirmatory Adaptive Clinical Trial Design and Analysis. R package version 4.1.0, https://CRAN.R-project.org/package=rpact.