This package is created for analysts starting a claims-based research project, requiring a cohort definition using ICD diagnosis codes.

Figure: the proposed hsrdefbuilder workflow. Analysts run the hsrdefbuilder code on claims, and create a summary of the codes associated with that service. Collaborators and analysts can view this and use the hsrdefbuilder Shiny application to make decisions about which codes to include in their final cohort definition.

Users can select and investigate the diagnosis codes most associated with the service claims, and share this output with their collaborators outside of a SAS environment.

The hsrdefbuilder package allows users to run our Shiny R application in their own R session.

We built this application with a SAS to R workflow in mind, meaning the initial claims selection and processing is done using a SAS program, while its output is processed and displayed using R.

To install the hsrdefbuilder R package from github, run:

devtools::install_github("kelsey209/hsrdefbuilder")

The hsrdefbuilder shiny application can be run using the R function:

hsrdefbuilder::runDefBuilder()

The folder minimum_dataset_paper contains the output data from the SAS program for the services discussed in our submitted paper (medRxiv link).

• Table 2 was compiled from results that were output from running hsrdef_run.SAS.
• Table 3 was based on the saved output data set from carrier claims with CPT code 29877: results_table3_carrier29877.csv.
• Table 4 results were compiled using the SAS code hsrdef_run_urinalysis_check.R. The datasets containing the important diagnosis codes listed in the table are: results_table4_carrier81001.csv, results_table4_carrier81003.csv, results_table4_outpatient81001.csv, results_table4_outpatient81003.csv).
• Table 5 was created using the data set output from the SAS application with inpatient claims for procedure codes 0SG0x: results_table5_inpatient0SG0.csv.