Automating HTA Model Validation with Unit Tests

Health economic evaluations require rigorous validation to ensure that structural logic, cost accounting, and transition probabilities behave as expected. The testhta package provides a standardized framework to validate cohort Markov models against established criteria, specifically aligning with the tests outlined in the HTA Verification Registry (T01 - T16).

To make implementation as frictionless as possible, testhta includes scaffolding functions that instantly generate a comprehensive test suite tailored to your specific model data.

Setting Up Your Test Environment

If you are starting a new cost-effectiveness analysis (CEA) project or retrofitting an existing one, you can automatically generate the necessary testing infrastructure using the use_hta_unittests() function.

This function creates the standard tests/testthat/ directory structure required by the testthat package and injects a pre-configured test script.

Generating the Skeleton

In your R console, simply run:

library(testhta)

# Generate the default test skeleton
use_hta_unittests()

By default, this creates a file named test-hta_model.R that utilizes a placeholder data object named test_data.

Customizing the Data Reference

In practice, your CEA model parameters will likely be stored in a custom list or data frame. You can map the generated tests directly to your project’s naming conventions using the data_name argument.

For example, if your baseline parameters are stored in an object called basecase_params, you can run:

use_hta_unittests(data_name = "basecase_params")

The resulting test script will automatically substitute all internal references, ensuring tests like the one below are immediately executable:

test_that("T01: QALYs with discount_rate = 1 should be 0", {
  check_model_qalys(
    data = basecase_params, 
    discount_rate = 1, 
    expected_qalys = 0,
    label = "T01: QALYs with discount_rate = 1 should be 0"
  )
})

The Verification Registry Suite

The generated skeleton includes programmatic checks for 16 standardized HTA validation tests, categorized into four core areas. For the complete test matrix, detailed descriptions, and code examples, see the dedicated HTA Verification Framework Vignette (vignettes/verification-framework.Rmd).

Executing the Test Suite

Once your helper functions are defined and your parameter object is loaded into the environment, you can execute the validation suite.

Command Line Execution: You can run the entire suite from the R console using the devtools package:

devtools::test()

IDE Integration: If your programming workflow relies on Positron or VS Code, you can leverage their integrated testing environments. Both IDEs automatically detect testthat files, allowing you to run, debug, and track the success of individual T01-T16 test blocks directly from the testing pane without manually executing console commands.