Estimate, Compare, and Visualize Healthcare Resource Utilization for Real-World Evidence.

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)
data <- hcru_sample_data
head(hcru_sample_data)

# Estimate HCRU
hcru_summary <- estimate_hcru(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_main = "cohort",
                          group_var_by = "care_setting",
                          test = NULL,
                          timeline = "Pre",
                          gt_output = FALSE)

hcru_summary

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

# Estimate HCRU
hcru_summary_gt <- estimate_hcru(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_main = "cohort",
                          group_var_by = "care_setting",
                          test = NULL,
                          timeline = "Pre",
                          gt_output = TRUE)

hcru_summary_gt

## Generate the HCRU plot for average visits by cohort and time-line
# Calculate the average visits
sum_df1 <- hcru_summary$`Summary by settings using dplyr` |>
    dplyr::group_by(
      .data[["time_window"]], 
      .data[["cohort"]], 
      .data[["care_setting"]]) |>
    dplyr::summarise(
      AVG_VISIT = mean(.data[["Visits"]], na.rm = TRUE), .groups = "drop")

# Load the plot_hcru function
p1 <- plot_hcru(
  summary_df = sum_df1,
  x_var = "time_window",
  y_var = "AVG_VISIT",
  cohort_col = "cohort",
  facet_var = "care_setting",
  facet_var_n = 3,
  title = "Average visits by domain and cohort",
  x_label = "Healthcare Setting (Domain)",
  y_label = "Average visit",
  fill_label = "Cohort"
)

p1


## Generate HCRU plot for average cost by cohort and timeline
# Calculate the total cost
df2 <- hcru_summary$`Summary by settings using dplyr` |>
    dplyr::group_by(
      .data[["time_window"]], 
      .data[["cohort"]], 
      .data[["care_setting"]]) |>
    dplyr::summarise(
      AVG_COST = sum(.data[["Cost"]], na.rm = TRUE), .groups = "drop")

p2 <- plot_hcru(
  summary_df = df2,
  x_var = "time_window",
  y_var = "AVG_COST",
  cohort_col = "cohort",
  facet_var = "care_setting",
  facet_var_n = 3,
  title = "Average cost by domain and cohort",
  x_label = "Healthcare Setting (Domain)",
  y_label = "Average cost",
  fill_label = "Cohort"
)

p2

🧾 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

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

Arguments:

📊 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.