Facilitating the representation of EORTC QoL Modules and items in common data models

In recent years, big data has become a key driver of innovation in healthcare, creating new opportunities to improve patient care, outcomes, and experiences. Big data refers to large amounts of information collected from diverse sources, but one of the challenges in using it effectively is the alignment of disparate datasets so they can be analysed together. Common data models (CDMs) have emerged as a solution to this challenge by transforming data from different sources into a single, standardised format.

One widely used example is the OMOP CDM, developed by the OHDSI (Observational Health and Data Science Informatics) community. OMOP has been adopted internationally in healthcare research and underpins many big data initiatives funded by the European Commission. While OMOP and other CDMs incorporate established vocabularies such as SNOMED and LOINC to represent clinical data, they have so far lacked systematic representation of PROMs. This omission has limited the use of important instruments like the EORTC Quality of Life Questionnaires (QLQ-C30 and tumour-specific modules) in large-scale analyses.

To address this gap, the EORTC is funding this project to develop a dedicated vocabulary for EORTC questionnaires, making them interoperable within the OMOP CDM. This work ensures that patient-reported quality-of-life data can be aligned with other clinical information, enabling cross-study analyses and facilitating big data research in oncology. The project began with the development of vocabularies for the QLQ-C30, QLQ-PR25, and QLQ-HN43 modules, and has since expanded to additional questionnaires.
Since its launch in 2023, the project has achieved several milestones, including the successful acceptance of the EORTC vocabulary by the OHDSI community. Work is ongoing to validate its use in real-world datasets, expand coverage to further EORTC modules, and develop technical guidelines and user-friendly resources to support researchers, clinicians, and patients.

In WP1, A systematic review of the literature was conducted to analyse existing CDMs used for PROM data in cancer. This review identified various knowledge gaps which emphasised the need to standardise EORTC-specific quality-of-life data within a CDM to enhance its usability in big data initiatives. Additionally, insights from the BD4QoLo project were utilised to evaluate existing vocabularies for the EORTC QLQ-C30, forming the foundation for designing an interoperable and extensible vocabulary. Regular update meetings were coordinated to ensure a multidisciplinary and cohesive approach with the project collaborators.
In WP2, A core ontology for the EORTC QLQ-C30 was successfully designed, forming the foundation for interrelated extension ontologies such as EORTC QLQ-PR25, EORTC QLQ-HN43, and EORTC QLQ-CR29. To achieve this, web scraping of the EORTC Item Library was conducted to extract detailed information on each question item, including domains, response scales, and recall periods. The extracted JSON-organized data was flattened and stored in Postgress Database. After that, key elements’ identifiers as well as their relationships were defined and formalized according to OMOP Vocabularies Pallas (i.e. OMOP vocabulary ETL process). The ETL process is documented and available at OHDSI Vocabulary GitHub and is in the process of being uploaded to the EORTC webpages. The resulting terminology is what is called a source vocabulary in the OHDSI community. It is normalised and standardised in line with OMOP conventions and can be used to capture questionnaire data into OMOP CDM. As part of the quality assurance and validation measures, the EORTC vocabulary and its underlying model have been discussed with subject matter experts from EORTC headquarters to ensure term accuracy and completeness.
The developed vocabulary has been successfully reviewed and accepted by the OHDSI community, reinforcing its credibility and potential for widespread adoption. Ongoing work includes the integration and evaluation of the vocabulary within the OMOP CDM using test datasets provided by each collaborator to ensure its applicability in real-world settings. Additionally, efforts are underway to test the Extract, Transform, Load (ETL) process by developing research questions that utilise standardised EORTC data from collaborating sites. These tests aim to validate the vocabulary’s functionality and improve its usability in clinical research. These validation studies are in the process of being evaluated and will be written up into manuscript format.
The groundwork has been completed to extend the vocabulary beyond the EORTC QLQ-C30 and initial tumour specific Modules (EORTC QLQ-PR25, EORTC QLQ-HN43, EORTC QLQ-CR29) to additional EORTC modules and items, allowing for a more comprehensive representation of patient-reported outcomes. This shows that the OMOP Vocabulary ETL that was designed is a scalable and adaptable framework that can accommodate future expansions, enhancing the interoperability and usability of the vocabulary across various cancer contexts. The final version of the terminology is available for download and browsing via athena.ohdsi.org and will soon be accessible directly through the EORTC website.

Work is ongoing to publish the comprehensive systematic review of WP1 summarising key findings and positioning this project within the broader context of quality-of-life research. In parallel, technical guidelines are being drafted to support researchers and data scientists in the development of vocabularies for EORTC questionnaires. Additionally, user-friendly materials that explain the process of utilising CDMs in quality-of-life research have been created and are currently being published within the EORTC website. These resources aim to make the project’s outcomes more accessible to a wider audience, including clinicians, patients, and policymakers.

We are in the process of applying for sustainability funding to support on-going vocabulary maintenance and updates, onboard additional and newly developed EORTC modules and Items, and develop advanced tools and resources to support use of the quality of life vocabulary.

A lot of research projects are now based on using data about cancer patients to help improve patient care and experience. One problem which exists when using data is that different sites and hospitals collect data in different ways – for example one hospital might collect ethnicity data in 6 categories and another might collect it using 12 categories. Common data models, or standard ways of collecting data, have recently been developed so that many different datasets can be combined together into “big datasets”. These big datasets can then be analyzed together as part of big worldwide research studies. This project is looking at ways we can create a common data model for the EORTC quality of life questionnaires so that datasets of different cancer patients who have filled in the questionnaires can be combined and analyzed together. We will be able to find out a lot more about cancer patients’ quality of life if we are able to combine these datasets into one big dataset.