Skip to contents

Changelog

Source: NEWS.md

rwevalidate 0.1.2

CRAN resubmission addressing a second review comment; no functional changes.

  • DESCRIPTION: write the HARPER and RECORD-PE framework names in normal capitalization (“Harmonized…”, “Reporting…”, “Pharmacoepidemiology”) instead of the mixed-case styling that the CRAN spell check flagged. The acronyms remain in parentheses.

rwevalidate 0.1.1

CRAN resubmission addressing reviewer feedback; no functional changes.

  • DESCRIPTION: all acronyms (OMOP, CDM, RWE, HTML, JSON, FDA, HARPER, RECORD-PE) are now explained at first use, and references follow the authors (year) <doi:...> format.
  • Removed the \dontrun{} example from validate_cohort(); the live-database call is now shown in the function’s Details section.
  • Added an executable example to cdm_disconnect(). The only remaining \dontrun{} example is cdm_connect(), which requires a live PostgreSQL server and credentials.
  • Expanded the Title to spell out “OMOP Common Data Model”.
  • example_cdm() now builds a 20-person, two-arm demo CDM (target and comparator), so every exported module — including run_covariates() — has a runnable example against the bundled database.
  • Added @examples to run_attrition(), run_density(), run_concepts(), and run_covariates().

rwevalidate 0.1.0

First public release. rwevalidate validates an instantiated OMOP CDM cohort for real-world-evidence use and produces a regulator-aligned HTML + JSON report from a single validate_cohort() call.

Features

  • cdm_connect() / cdm_disconnect() - open a PostgreSQL CDM connection and validate that the required clinical tables exist (split-schema aware).
  • Module 1 - run_concepts() - concept prevalence, ancestor coverage, and per-domain vocabulary mapping rate (FDA Relevance).
  • Module 2 - run_attrition() - cohort size, index-date distribution, prior-observation coverage, and demographics at index (HARPER / RECORD-PE).
  • Module 3 - run_density() - records per patient per month by domain and follow-up completeness with a censoring breakdown (FDA Reliability).
  • Module 4 - run_covariates() - standardized mean differences (love plot), per-arm prevalence, and a power calculation against a comparator (HARPER).
  • Reporting - render_validation_report() writes a self-contained HTML report with a traffic-light summary plus a machine-readable validation_results.json sidecar; write_results_json() works without Pandoc.
  • validate_cohort() - one entry point that wires connection, all modules, and the report. Accepts an open con or PostgreSQL connection arguments; concept_ids enables Module 1, comparator_id enables Module 4.

Notes

  • Targets OMOP CDM v5.3.1. Validated against MIMIC-IV on PostgreSQL and a synthetic DuckDB mock used by the test suite.
  • No dependency on CohortDiagnostics, DataQualityDashboard, or Achilles.
  • All thresholds are documented, configurable function arguments.