Unmeasured confounding is often raised as a source of potential biaswhen evaluating non-randomized study protocols, but evaluating suchconcerns during their design remains challenging. We propose a flexiblemethodology based on individual level simulations that can allowresearchers to characterize the bias arising from unmeasured confoundingwith a specified but modifiable structure during the study design.
sim.BA allows user to conduct a simulation-basedquantitative bias analysis using covariate structures generated withindividual-level data to characterize the bias arising from unmeasuredconfounding. Users can specify their desired data generating mechanismsto simulate data and quantitatively summarize findings in an end-to-endapplication using this package. Seevignette("sim.BA") fordetails.
You can install the development version ofsim.BA fromGitLab with:
# install.packages("remotes")remotes::install_gitlab("rjd48/sim.BA",host ="gitlab-scm.partners.org")You can install the published version from CRAN with:
install.packages("sim.BA")