R Markdown and Quarto costing-study workflow

This page documents the initial R binding posture for costing-study reports. The recommended path is a thin R wrapper over the shared CLI plus file interchange. The executable prototype uses CSV because the current CLI supports CSV today. Arrow/Parquet remains the target for larger cross-language batches once the shared file contract is implemented.

Synthetic fixtures only

The examples on this page use synthetic fixture data. Do not use them for policy decisions, submissions, or patient-level workflows.

Installation posture

The initial R path is an internal-use integration, not a CRAN-ready package claim.

• Install and keep the calculator core in the shared CLI toolchain.
• Use R packages for data shaping, report rendering, and file exchange.
• Treat any R package as a thin wrapper only.
• Revisit extendr or reticulate later only if the wrapper-only path is too limiting.

Recommended report flow

1. Load synthetic costing-study fixtures in R.
2. Validate the required columns in the report script.
3. Write a batch input file. Use CSV for the current prototype; use Parquet after the shared file contract is implemented.
4. Call the shared CLI from the R Markdown or Quarto document.
5. Read the output file and render summary tables or charts.

library(arrow)
library(dplyr)

inputs <- read_parquet("fixtures/synthetic_costing_inputs.parquet")

required_cols <- c("calculator", "pricing_year", "DRG", "LOS", "ICU_HOURS")
missing_cols <- setdiff(required_cols, names(inputs))
if (length(missing_cols) > 0) {
  stop("Missing required columns: ", paste(missing_cols, collapse = ", "))
}

batch_input <- inputs |>
  mutate(
    calculator = "acute",
    pricing_year = 2025L
  )

write.csv(batch_input, "work/batch_input.csv", row.names = FALSE)

system2(
  command = Sys.getenv("MCHS_CLI", "path/to/shared-cli"),
  args = c(
    "acute",
    "work/batch_input.csv",
    "--year", "2025",
    "--output", "work/batch_output.csv"
  )
)

results <- read.csv("work/batch_output.csv", check.names = FALSE)
results |>
  select(DRG, pricing_year, NWAU25, validation_status) |>
  head()

The same pattern works inside an .Rmd or .qmd document because the report simply calls the shared engine and then formats the returned outputs.

Limitations

• The R layer must not reimplement formulas, adjustment logic, or validation rules.
• The CLI dependency means portability depends on a local runtime install.
• This path is batch-oriented, not a substitute for a native interactive calculator API.
• Output parity should be checked against shared fixtures before any result is presented as authoritative.
• CRAN readiness is not claimed here.

What to try next

• Use the shared synthetic fixture pack to build a reusable report template.
• Add automated checks once the package skeleton and stable fixture set exist.