Estimate and Compare Healthcare Resource Utilization (HCRU) for RWE • hcruR

hcruR is an R package to help health economists and RWE analysts estimate and compare healthcare resource utilization (HCRU) from observational healthcare data, such as claims or electronic health records.

🚀 Features

Estimate patient-level HCRU by:
- Domain (inpatient, outpatient, pharmacy, etc.)
- Time relative to index date (pre/post)
Compare HCRU across cohorts
Visualize domain-wise HCRU statistics
Designed for flexible real-world evidence (RWE) workflows

📥 Installation

You can install the development version of hcruR from GitHub with:

# Install from GitHub (after you upload the repo)
# install.packages("devtools")
devtools::install_github("mumbarkar/hcruR")

# install.packages("pak")
pak::pak("mumbarkar/hcruR")

🧪 Example Usage

This is a basic example which shows you how to solve a common problem:

This is a basic example which shows you how to solve a common problem:

# Load library
library(hcruR)

## Generate HCRU summary using dplyr (this can be used for create HCRU plots)

# Load sample data
data(hcru_sample_data)
head(hcru_sample_data)

# Estimate HCRU
hcru_summary <- estimate_hcru(data = hcru_sample_data,
                             cohort_col = "cohort",
                             patient_id_col = "patient_id",
                             admit_col = "admission_date",
                             discharge_col = "discharge_date",
                             index_col = "index_date",
                             visit_col = "visit_date",
                             encounter_id_col = "encounter_id",
                             setting_col = "care_setting",
                             cost_col = "cost_usd",
                             readmission_col = "readmission",
                             time_window_col = "period",
                             los_col = "length_of_stay",
                             custom_var_list = NULL,
                             pre_days = 180,
                             post_days = 365,
                             readmission_days_rule = 30,
                             group_var = "cohort",
                             test = NULL,
                             gt_output = FALSE)

hcru_summary

## Generate HCRU summary using gtsummary (a publication ready output) 

# Estimate HCRU
hcru_summary_gt <- estimate_hcru(data = hcru_sample_data,
                             cohort_col = "cohort",
                             patient_id_col = "patient_id",
                             admit_col = "admission_date",
                             discharge_col = "discharge_date",
                             index_col = "index_date",
                             visit_col = "visit_date",
                             encounter_id_col = "encounter_id",
                             setting_col = "care_setting",
                             cost_col = "cost_usd",
                             readmission_col = "readmission",
                             time_window_col = "period",
                             los_col = "length_of_stay",
                             custom_var_list = NULL,
                             pre_days = 180,
                             post_days = 365,
                             readmission_days_rule = 30,
                             group_var = "cohort",
                             test = NULL,
                             gt_output = TRUE)

hcru_summary_gt

## Generate the HCRU plot for average visits per patient by cohort and 
time-line

p_avg_visit <- plot_hcru(summary_df = hcru_summary$`Summary by settings using dplyr`,
               x_var = "period",
               y_var = "Avg_visits_per_patient",
               cohort_col = "cohort",
               facet_var = "care_setting",
               facet_var_n = 3,
               title = "Per patient average visits by domain and cohort",
               x_lable = "Healthcare Setting (Domain)",
               y_lable = "Average visits",
               fill_lable = "Cohort"
)

p_avg_visit

## Generate the HCRU plot for average costs per patient by cohort and time-line

p_avg_cost <- plot_hcru(summary_df = hcru_summary$`Summary by settings using dplyr`,
               x_var = "period",
               y_var = "Avg_cost_per_patient",
               cohort_col = "cohort",
               facet_var = "care_setting",
               facet_var_n = 3,
               title = "Per patient average cost by domain and cohort",
               x_lable = "Healthcare Setting (Domain)",
               y_lable = "Average costs",
               fill_lable = "Cohort"
)

p_avg_cost

🧾 Function Reference

estimate_hcru() estimates of healthcare resource utilization (HCRU) from electronic health record data across various care settings (e.g., IP, OP, ED/ER). It provides descriptive summaries of patient counts, encounters, costs, length of stay, and readmission rates for pre- and post-index periods

Arguments

Argument	Type	Description	Default
`data`	`data.frame`	Input healthcare dataset containing admission, discharge, and visit information.	—
`cohort_col`	`character`	Name of the column that defines cohort groupings.	`"cohort"`
`patient_id_col`	`character`	Name of the column containing unique patient identifiers.	`"patient_id"`
`admit_col`	`character`	Name of the column containing admission dates.	`"admission_date"`
`discharge_col`	`character`	Name of the column containing discharge dates.	`"discharge_date"`
`index_col`	`character`	Name of the column containing the index (diagnosis) date.	`"index_date"`
`visit_col`	`character`	Name of the column for visit or claim dates.	`"visit_date"`
`encounter_id_col`	`character`	Name of the column containing encounter or claim IDs.	`"encounter_id"`
`setting_col`	`character`	Name of the column representing care settings (e.g., IP, OP, ED).	`"care_setting"`
`pre_days`	`numeric`	Number of days before index date to include in pre-period.	`180`
`post_days`	`numeric`	Number of days after index date to include in post-period.	`365`
`readmission_days_rule`	`numeric`	Number of days to define readmission following a discharge in the IP setting.	`30`
`gt_output`	`logical`	Whether to generate an additional `gtsummary` output.	`FALSE`
`cost_col`	`character`	Name of the column containing cost information.	`"cost_usd"`
`los_col`	`character`	Name of the column for length of stay values.	`"length_of_stay"`
`readmission_col`	`character`	Name of the column indicating readmission status.	`"readmission"`
`time_window_col`	`character`	Name of the column that categorizes records as pre or post index.	`"period"`
`custom_var_list`	`character[]`	Optional list of additional columns to include in summary tables.	`NULL`
`group_var`	`character`	Name of the grouping column for stratified summaries.	`"cohort"`
`test`	`list`	Optional named list of statistical tests (e.g., `list(age = "wilcox.test")`).	`NULL`

plot_hcru() provides the visualization of the events of the settings/domains grouped by cohort and time window.

Arguments:

Argument	Type	Description
`summary_df`	dataframe	Output from `estimate_hcru()` function.
`x_var`	character	Column name to plot on the x-axis (default `"period"`).
`y_var`	character	Column name to plot on the y-axis (default `"Cost"`).
`cohort_col`	character	Name of the column identifying cohorts (default `"cohort"`).
`facet_var`	character	Column to generate subplots for (default `"care_setting"`).
`facet_var_n`	numeric	Number of columns in the facet grid (default `3`).
`title`	character	Title of the plot.
`x_lable`	character	Label for the x-axis.
`y_lable`	character	Label for the y-axis.
`fill_lable`	character	Label for the fill legend.

📊 Sample Data

This package includes a demo datasets for easy testing:

hcru_sample_data: 200 patients across 2 cohorts

head(hcru_sample_data)

📚 Vignette

Run the following to access the full walkthrough:

vignette("hcru-analysis", package = "hcruR")

🔬 Use Cases

Cost burden studies before/after treatment
Resource comparison across patient populations
Outcome stratification based on utilization patterns

🛠️ Development

To contribute:

git clone https://github.com/mumbarkar/hcruR.git
cd hcruR

📜 License

This package is licensed under the MIT License.