Learn R Programming

Reference Based Multiple Imputation (`rbmi`)

Overview

The rbmi package is used for the imputation of missing data in clinical trials with continuous multivariate normal longitudinal outcomes. It supports imputation under a missing at random (MAR) assumption, reference-based imputation methods, and delta adjustments (as required for sensitivity analysis such as tipping point analyses). The package implements both Bayesian and approximate Bayesian multiple imputation combined with Rubin's rules for inference, and frequentist conditional mean imputation combined with (jackknife or bootstrap) resampling.

Installation

The package can be installed directly from CRAN via:

install.packages("rbmi")

Note that the usage of Bayesian multiple imputation requires the installation of the suggested package rstan.

install.packages("rstan")

Usage

The package is designed around its 4 core functions:

draws() - Fits multiple imputation models
impute() - Imputes multiple datasets
analyse() - Analyses multiple datasets
pool() - Pools multiple results into a single statistic

The basic usage of these core functions is described in the quickstart vignette:

vignette(topic = "quickstart", package = "rbmi")

Validation

For clarification on the current validation status of rbmi please see the FAQ vignette.

Support

For any help with regards to using the package or if you find a bug please create a GitHub issue

Copy Link

Version

Install

install.packages('rbmi')

Monthly Downloads

860

Version

1.7.1

License

Apache License (>= 2)

Issues

Pull Requests

Stars

Forks

Repository

https://github.com/openpharma/rbmi

Homepage

https://openpharma.github.io/rbmi/

Maintainer

Lukas A. Widmer

Last Published

November 5th, 2026

Functions in rbmi (1.7.1)

Applies delta adjustment

antidepressant_data

Antidepressant trial data

Creates a draws object

ancova_group_labels

ANCOVA group level labels

ancova_contrast_name

Name a default ANCOVA treatment-effect contrast

Convert object to dataframe

Construct an analysis object

Convert indicator to index

Create an imputation object

as_cropped_char

as_cropped_char

Compute covariance matrix for some reference-based methods (JR, CIR)

Creates a "MMRM" ready dataset

complete_control_bayes

Completion of the Control Options List

Convert character variables to factor

as_mmrm_formula

Create MMRM formula

Diagnostics of the MCMC based on ESS

Implements an Analysis of Covariance (ANCOVA)

Calculate delta from a lagged scale coefficient

extract_covariates

Extract Variables from string vector

Create a delta data.frame template

Evaluate a call to mmrm

Expand and fill in missing data.frame rows

extract_data_mnar_as_na

Set to NA outcome values that would be MNAR if they were missing (i.e. which occur after an ICE handled using a reference-based imputation strategy)

check_hmc_diagn

Diagnostics of the MCMC based on HMC-related measures.

as_stan_fragments

Conversion of Character Vector into Stan Code Block List

Fit the base imputation model and get parameter estimates

Expand data.frame into a design matrix

Create vector of strata

as_simple_formula

Creates a simple formula object from a string

assert_variables_exist

Assert that all variables exist within a dataset

Ensure rstan exists

convert_to_imputation_list_df

Convert list of imputation_list_single() objects to an imputation_list_df() object (i.e. a list of imputation_df() objects)

Control the computational details of the imputation methods

Extract draws from a stanfit object

extract_imputed_df

Extract imputed dataset

get_conditional_parameters

Derive conditional multivariate normal parameters

formula_has_interaction

Check whether a variable participates in a formula interaction

generate_data_single

Generate data for a single group

Get imputation strategies

Extract parameters from a MMRM model

Diagnostics of the MCMC

extract_imputed_dfs

Extract imputed datasets

get_delta_template

Get delta utility variables

get_example_data

Simulate a realistic example dataset

Fit the base imputation model using a Bayesian approach

Does object have a class ?

get_visit_distribution_parameters

Derive visit distribution parameters

get_jackknife_stack

Creates a stack object populated with jackknife samples

Fit a MMRM model

Extract the Effective Sample Size (ESS) from a stanfit object

get_bootstrap_stack

Creates a stack object populated with bootstrapped samples

format_method_descriptions

Format method descriptions

Create a valid imputation_df object

get_unique_hash

Get unique hash

Invert and derive indexes

imputation_single

Create a valid imputation_single object

Create imputed datasets

Get Compiled Stan Object

frm_find_and_replace

Recursively find and replace symbols in a language object

Sample outcome value

longDataConstructor

R6 Class for Storing / Accessing & Sampling Longitudinal Data

impute_internal

Create imputed datasets

get_pattern_groups_unique

Get Pattern Summary

get_pool_components

Expected Pool Components

impute_data_individual

Impute data for a single subject

Is value absent

Calculate design vector for the lsmeans

is_valid_covariance_prior

Check for Valid Covariance and Prior Combination

Last Observation Carried Forward

Fit the base imputation model on bootstrap samples

get_mmrm_sample

Fit MMRM and returns parameter estimates

Von Hippel and Bartlett pooling of BMLMI method

imputation_list_df

List of imputation_dfs

Is character or factor

get_pattern_groups

Determine patients missingness group

is_in_rbmi_development

Is package in development mode?

is_num_char_fact

Is character, factor or numeric

imputation_list_single

A collection of imputation_singles() grouped by a single subjid ID

pool_bootstrap_percentile

Bootstrap Pooling via Percentiles

Pool analysis results obtained from the imputed datasets

make_rbmi_cluster

Create a rbmi ready cluster

Least Square Means

Internal Pool Methods

Internal MCSE Computations

Parallelise lapply

Calculate parametric confidence intervals

Is single character

Print analysis object

pool_bootstrap_normal

Bootstrap Pooling via normal approximation

pval_percentile

P-value of percentile bootstrap

R6 Class for printing current sampling progress

rbmi: Reference Based Multiple Imputation

random_effects_expr

Construct random effects formula

Set the multiple imputation methodology

print.imputation

Print imputation object

Print draws object

Capture all Output

Sample random values from the multivariate normal distribution

prepare_init_vals

Preparation of Initial Values for MCMC Sampler

Barnard and Rubin degrees of freedom adjustment

prepare_prior_params

Prepare Prior Parameters for Covariance Model Prior Distribution

resolve_contrast_weights

Resolve a numeric contrast weight vector to full group-level order

Sample Patient Ids

Original Rubin degrees of freedom approximation

Create object of sample_single class

simulate_test_data

Create simulated datasets

simulate_dropout

Simulate drop-out

Combine estimates using Rubin's rules

Sort data.frame

Simulate intercurrent event

Convert Transformed Correlation to Correlation

split_imputations

Split a flat list of imputation_single() into multiple imputation_df()'s by ID

prepare_stan_data

Prepare input data to run the Stan model

remove_if_all_missing

Remove subjects from dataset if they have no observed values

recursive_reduce

recursive_reduce

Transform array into list of arrays

theta_to_cs_cor

Convert Transformed Compound Symmetry Correlation to Correlation

transpose_imputations

Transpose imputations

validate_analyse_pars

Validate analysis results

validate.sample_single

Validate sample_single object

Does a string contain a substring

validate_strategies

Validate user specified strategies

validate.sample_list

Validate sample_list object

Set key variables

Create and validate a sample_list object

Set simulation parameters of a study group.

validate.analysis

Validate analysis objects

Generic validation method

validate_datalong

Validate a longdata object

validate.references

Validate user supplied references

Validate inputs for vars

validate.stan_data

Validate a stan_data object

scalerConstructor

R6 Class for scaling (and un-scaling) design matrices

validate.simul_pars

Validate a simul_pars object

transpose_results

Transpose results object

transpose_samples

Transpose samples

validate.is_mar

Validate is_mar for a given subject

Validate draws object

adjust_trajectories

Adjust trajectories due to the intercurrent event (ICE)

List of Stan Blocks

Analyse Multiple Imputed Datasets

Analysis of Covariance

adjust_trajectories_single

Adjust trajectory of a subject's outcome due to the intercurrent event (ICE)

QR decomposition

R6 Class for a FIFO stack

adjust_dimensions

Adjust Dimensions of Prior Parameters for Stan Usage

ancova_contrast_design

Build the ANCOVA contrast design matrix

ancova_resolve_contrasts

Resolve the set of ANCOVA contrasts to compute

ancova_linear_contrast

Compute a single ANCOVA linear contrast

Core ANCOVA Implementation