gms | German Medical Science

Introduction: Until recently, the focus of data analysis in German cancer registries has been primarily on epidemiological data, which provides information on cancer rates and patterns in populations. With the establishment of clinical cancer registries and the introduction of the oBDS dataset schema [1], there is now a growing shift towards analyzing clinical cancer data, which contains course-of-disease data for individual patients. This type of data is novel and complex, presenting new challenges for researchers and medical experts. An important aspect here is the definition and analysis of medical cohorts.

For the definition of cohorts the Clinical Quality Language (CQL) exists. CQL is a HL7 standard [2] used as a high-level, expression-based language designed to support quality improvement in clinical practice. In the course of the SePaMiM-project we created tooling to query oBDS-based data using CQL.

State of the art: Efforts in German cancer registries to analyze the oBDS dataset mainly include descriptive statistical measures that examine individual patient characteristics. The course of disease is only evaluated in the context of rigid quality indicators. For this purpose, cohort definitions are developed using SQL, R, or other languages. CQL as a query language for medical datasets has been widely adopted, e.g. in [3]. However, to the best of our knowledge this language has not been used in German cancer registries previously.

Concept: In SePaMiM we develop a novel tool, aimed at facilitating the analysis of cancer data. For this purpose, we have created a custom CQL model, based on the oBDS schema, as well as a set of scripts to transform cleaned and prepared oBDS data into this model. Users can then define CQL queries to find cohorts of patients based on specific criteria, which are then visualized in a comprehensive and easy-to-understand format.

The tool provides an overview of the found patients using charts with descriptive statistics, and a list of all patients is also displayed. Furthermore, users can select individual patients to view their medical history, including treatment and examination events that are visualized in a timeline. Detailed information about each event can also be accessed.

Implementation: We implemented the tool as a web application utilizing React and Material UI to build the frontend. The CQL-editor is based on the Monaco code-editor. The server is a Spring-based REST server application. At startup, the prepared oBDS data is loaded from a relational database into memory. For parsing and executing CQL queries an open source implementation [4] is used.

Lessons learned: We set up two test instances in two different German cancer registries, the cancer registry of North-Rhine-Westphalia as well as the KLast, which is part of the clinical cancer registry of Lower Saxony. It became apparent that the utilization of CQL represents a suitable strategy to assist cancer registry users in the formation of cohorts for internal analyses and computation of quality indicators, while also facilitating the fulfillment of data requests from external researchers. We also learned that visualizations for presenting results and temporal disease data were highly appreciated.

The authors declare that they have no competing interests.

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