Skip to contents

appendMCP

appendMCP generates analysis documentation for graphical multiple comparison procedures (MCPs) in group sequential design (GSD) clinical trials. Given a study configuration, it produces summary tables, diagnostic plots, and a fully formatted R Markdown report suitable for appending to a statistical analysis plan (SAP).

Installation 

# install.packages("remotes")
remotes::install_github("johnsonandjohnson/appendMCP")

Quick Start

The core workflow is three steps: load a configuration, process it, and generate a report.

library(appendMCP)

config  <- load_config_from_repository("example_study")
results <- process_config(config)
generate_report(results)

This produces an HTML report in your working directory containing summary tables, the graphical testing procedure diagram, information fraction plots, alpha spending visualizations, and operating characteristics.

What the Package Produces

  • Table 1 — Hypothesis summary (endpoint, type, spending function)
  • Table 2 — Analysis schedule by hypothesis (timing, information fractions)
  • Table 3 — Analysis schedule by analysis (all hypotheses at each look)
  • Table 4 — Weight scenarios under the graphical MCP
  • Table 5 — Boundary specifications (nominal p-values, information fractions, local power)
  • Table 6a / 6b — Simulation-based operating characteristics by analysis and overall
  • Plots — Graph diagram, information fractions, alpha spending, enrollment, time-to-event and binary endpoint distributions

Tables are returned as huxtable objects ready for Word or HTML output.

Exploring Results 

# Individual tables
results$tables$table1   # Hypothesis summary
results$tables$table2   # Schedule by hypothesis
results$tables$table5   # Boundary specifications (nominal p-values, local power)

# Plots (top-level fields on the results object)
results$graph_figure          # Graphical MCP diagram
results$information_figure    # Information fraction over time
results$alpha_spend_figure    # Alpha spending functions
results$tte_figure            # Time-to-event distributions
results$er_figure             # Enrollment rate

Scaffolding a New Study

create_study() copies a configuration and report template into a project folder and generates a ready-to-run analysis script:

create_study(
  config       = "example_study",   # built-in config, or path to your own .R file
  rmd_template = "gsd_default",     # built-in template, or path to your own .Rmd
  output_dir   = "my_study"
)
# Creates:
#   my_study/study_config.R    — edit this to define your study
#   my_study/render_config.R   — run this to execute the analysis
#   my_study/report.Rmd        — the report template

Available Built-in Configurations and Templates 

list_config_repository()        # built-in study configurations
list_rmd_template_repository()  # built-in report templates

Using a Custom Configuration

A configuration is an R list assigned to a single variable in a .R file. The required fields are:

Field Description
study_name Study identifier string
alpha One-sided significance level (e.g. 0.025)
analyses Data frame of analysis specifications
hypotheses Data frame of hypothesis definitions
enroll_rate Data frame of enrollment rates by stratum
graph List with g (transition matrix) and w (initial weights)
distribution_tte Time-to-event distributions (required if any TTE endpoint)
distribution_bin Binary endpoint distributions (required if any binary endpoint)

See the built-in example for a complete, annotated configuration:

config <- load_config_from_repository("example_study")
str(config, max.level = 1)

Or read the configuration guide vignette:

vignette("configuration-guide", package = "appendMCP")

Getting Help