¹Berlin Institute of Health at Charité - Universitätsmedizin Berlin, Center of Health Data Science, Berlin, Germany
²Institute for Hygiene and Environmental Medicine, Charité – Universitätsmedizin Berlin, Berlin, Germany
³Robert Koch Institute, Berlin, Germany

Text

Introduction: Hospital-onset bacteremia (HOB) has been proposed as a novel metric for healthcare-associated infection surveillance. However, data on HOB epidemiology and underlying risk factors are scarce. The Medical Informatics Initiative (MII) use-case RISK PRINCIPE addresses both surveillance as well as risk assessment and prediction of HOB infections, leveraging its cross-institutional data integration to enhance infection control strategies [1]. This includes the development of a demonstrator application for federated surveillance and retrospective monitoring of HOB infections, utilizing the MII infrastructure to enable standardized data access and benchmarking across hospitals.

The aim is to design and implement a nationwide federated surveillance application that allows hospitals to explore, visualize, and benchmark HOB infection rates. As a first step, a demonstrator application has been implemented that provides healthcare professionals and epidemiologists with a user-friendly tool for post-infection monitoring, infection control measurement, benchmarking and documentation, while leveraging the MII-infrastructure to ensure standardized access to the required data.

Methods: The web-based demonstrator is designed to facilitate retrospective analyses of HOB dynamics, providing statistical insights and benchmarking functionality stratified by clinical units, pathogens and time periods. It integrates a data connector that retrieves clinical data from FHIR stores providing resources compliant to the specification of the MII National Core Dataset. Data sourcing and HOB episode processing are handled by a long-running, periodic background task. HOB episodes follow definitions from the RISK PRINCIPE project specifications, based on the PRAISE HOB algorithm [2]. The interaction between application components and the result representation was developed through close interdisciplinary collaboration between hygiene experts and software engineers. The structured infection data is stored in a relational database implementing a data warehouse-oriented scheme, optimized for efficient analytics, enabling real-time generation of statistical metrics and visualizations. The backend is developed using the Spring Boot framework and incorporates a PostgreSQL database alongside a RESTful API supporting scalable data handling and interoperable system communication. The frontend, built with React and Next.js, provides an interactive and seamless user experience aligned with the risk prediction and prevention applications of the project. Security is ensured through OAuth2 with Keycloak as the identity provider.

Results: The initial version of the dashboard supports the analysis of HOB trends and incidence rates, stratified by clinical units and time at a macro level. Users can compare infection rates within their institution and drill down to individual HOB events for more detailed insights. The dashboard was developed and tested using real-world data from a consortium hospital and large synthetic datasets to ensure applicability in clinical settings.

Conclusion: The RISK PRINCIPE surveillance dashboard represents a digital tool for infection control implemented on the MII infrastructure, promoting standardized reporting and cross-institutional benchmarking. Future developments will focus on the highlighting of outliers and unusual events, integration with the project's prevention application [1] and the federated exchange of aggregated reference data. This will allow participating hospitals to report and access HOB statistics from a central authority, thereby supporting nation-wide benchmarking and epidemiological analyses. Additionally, a usability survey will be conducted to gather user feedback and guide further refinement.

The authors declare that they have no competing interests.

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

References

[1] Marschollek M, Marquet M, Reinoso Schiller N, et al. Automatisierte Surveillance und Risikovorhersage mit dem Ziel einer risikostratifizierten Infektionskontrolle und -prävention (RISK Prediction for Risk-stratified Infection Control and Prevention). Bundesgesundheitsbl. 2024;67:685–692. DOI: 10.1007/s00103-024-03882-w
[2] Aghdassi M, Werff S, Catho G, et al. Hospital-onset bacteraemia and fungaemia as a novel automated surveillance indicator: results from four European university hospitals [Preprint]. medRxiv. 2024. DOI: 10.1101/2024.09.16.24310433