¹Universitätsmedizin Greifswald, Klinik für Anästhesie, Intensiv-, Notfall- und Schmerzmedizin, Greifswald, Germany

Text

Introduction: The Clinical Quality Language (CQL) is a domain-specific language for expressing clinical logic in a human-readable, computable form. It is widely used in clinical decision support and quality measurement applications, particularly in combination with Fast Healthcare Interoperability Resources (FHIR). The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM), in turn, is a widely used data model for secondary research use of clinical data but lacks support for reusable clinical logic expressions comparable to FHIR-based CQL. We address this gap by enabling the CQL execution directly on OMOP data and demonstrate our approach using real-world data and guideline-based CQL libraries.

State of the art: Current CQL implementations are tightly coupled to FHIR, without native support for other data models. In OMOP, clinical logic is typically authored in ATLAS, which supports basic temporal logic and standardized vocabularies but lacks the expressiveness compared to a full domain-specific language like CQL. Direct querying of OMOP databases via structured query language (SQL) offers flexibility but is error-prone and lacks abstraction for common clinical concepts like unit conversion and time semantics – features inherent to CQL. To address these challenges, we previously developed an OMOP execution engine for logic expressed in Python or FHIR, but it remains incompatible with the broader CQL ecosystem [1].

Concept: Our approach enables native execution of CQL on OMOP data by implementing a custom data provider for the CQL reference engine. Unlike prior work that maps OMOP to FHIR, our solution directly exposes OMOP tables and terminology. Our implementation also addresses the challenge of binding CQL parameters and the CQL context to suitable values.

Implementation: Our implementation extends the CQL reference implementation [2] with:

1. A CQL-compatible representation of the OMOP CDM and Standard Vocabulary, including formal type mapping.
2. A data provider that translates CQL retrieve operations into SQL queries on OMOP databases.
3. A helper library to align OMOP and CQL semantics including explicit and implicit type conversion.
4. Support for binding CQL context and parameters to OMOP values and for per-patient CQL evaluation.

To enhance maintainability, we employ code generation based on a machine-readable OMOP CDM description [3] for data model implementation, retrieve operations and type conversions. Our implementation supports parallel processing across patients for scalability.

Lessons learned: We validated our approach by expressing previously studied guideline recommendations [1] as CQL libraries for OMOP and verifying results against our existing FHIR-based execution engine. We found that authoring CQL logic for OMOP was significantly more structured and concise than writing custom SQL. The ability to easily compose and test expressions was a substantial benefit during CQL development and debugging.

While our CQL-based implementation was slower than our previous solution (~10 seconds for 5,000 patients on standard hardware), it remains within a practical range. We plan to optimize performance by pushing more logic into generated SQL and restructuring the CQL libraries to facility efficient evaluation.

Overall, our work demonstrates that executing CQL on OMOP is feasible and beneficial, particularly when reuse, transparency and modularity of clinical logic are important.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.

References

[1] Lichtner G, Schiefenhövel F, Gashi B, Martin I, Jurth C, Vasiljewa L, et al. Automated Monitoring of Clinical Practice Guideline Adherence Using FHIR and OMOP: A Multi-Center Study in Intensive Care Units [Preprint]. medRxiv. 2025. DOI: 10.1101/2025.01.02.24319597
[2] Clinical Quality Language. 2025 [cited 2025 Apr 17]. Available from: https://github.com/cqframework/clinical_quality_language/
[3] Data Standardization – OHDSI. 2025 [cited 2025 Apr 17]. Available from: https://www.ohdsi.org/data-standardization/