Background

Multiple imputation is a powerful tool in presence of missing data.

However, especially in combination with propensity score analyses, multiple imputation can lead to challenges since analytic workflows can be become much more complex.

Objective

This repository showcases and evaluation different multiple imputation > propensity score > outcome analyses and associated implementation challenges.

Dependencies

This is a quarto book project and R package dependencies are managed through the _renv_requirements.sh file. Here, all packages and their versions can be viewed and re-installed by running the following line of code in the Terminal:

sh _requirements.sh

Reproducibility

Follow these steps in RStudio to reproduce this study:

1. Clone this repository via git clone <url>
2. Install all necessary dependencies (see above)
3. Run all scripts via quarto render or (optionally) in RStudio Build > Render Book (make sure quarto is installed)

The data used in this project is strictly simulated and no real patient-level data is used.