Medical Concept Mapping: RIZIV Codes

RIZIV, an acronym for Rijksinstituut voor Ziekte- en Invaliditeitsverzekering (translated to English: National Institute for Health and Disability Insurance) is a Belgian state-run organization tasked with managing health care insurance, monitoring and registering health care providers, and overseeing legislative compliance by health insurance funds. In order to perform these tasks, RIZIV has developed a codification system that assigns unique codes to (1) health care providers and (2) health care events like visits, tests, or surgeries requiring specific reimbursement.

At edenceHealth, we are busy with several projects to map the event-based RIZIV codes to internationally recognized, standard medical concepts defined by the Observational Health Data Science and Informatics (OHDSI) community. Once the codes have been standardized, they can be integrated into EU-wide, or even global, network studies that create patient-level and population-level prediction models to determine risk and/or outcome measures for conditions ranging from colorectal cancer to COVID-19. By standardizing RIZIV codes, we can harness the tremendous potential in the data being recorded each day by medical centers across Belgium, and we can apply insights from those data to generate real-world impact in the lives of medical professionals and their patients.

We are collaborating with partners at AZ Klina in Brasschaat and Medaman in Geel to create links between the Dutch- or French-language RIZIV code descriptions and standard English-language medical concepts. A subset of these projects are funded by the EU-backed funding initiative European Health Data and Evidence Network (EHDEN), which aims to standardize medical data across the European continent.

Creating links, or ‘mappings’, between billing-oriented codes and precise medical concepts is, unfortunately, non-trivial. In addition to the complexity of translating nuanced meanings between languages, codes related to medical finance and reimbursement are often very granular and package multiple concepts into compact descriptions. For example, the RIZIV code with description cardiovasculaire echo: bloedvat. can be mapped to the standard procedure Doppler ultrasonography of blood vessel with three additional concept modifiers: Thoracic, Abdominal, and Pelvic. In order to address this challenge, we have developed two software platforms that (1) suggest and (2) facilitate the approval of source-to-standard mappings, which we refer to as edenceMapper and edenceReviewer, respectively.

edenceMapper is driven by our team’s custom-curated algorithms based on cutting-edge natural language processing (NLP) and machine-learning (ML) methods; it generates high accuracy mappings between source and standard codes and has produced mappings for more than 3000 RIZIV codes so far, with an ‘auto-mapped percentage’ (number of suggestions approved by reviewers without modification out of total suggestions) of greater than 70% on the RIZIV codebase. edenceReviewer is a powerful web application that enables collaborative review of these mappings by medical experts. The reviewers can work together to approve, reject, or modify mapping suggestions, and their actions and recommendations are fed back into our mapping pipeline to improve accuracies of subsequent suggestions.

If you’d like to learn more about our software, our mapping approach to the RIZIV codes, or if you’d like to get involved in medical code mapping projects, please don’t hesitate to reach out: we’d be happy to hear from you!