Properties of a federated epidemiology query system

PurposeThe purpose of the study was to establish knowledge about how online access to epidemiological data from general practitioners (GPs) electronic health record (EHR) system should be provided. Before such systems are developed and deployed a decision about the appropriate system architecture must be made. Such a decision should ideally be based on knowledge about the properties of different system architectures. This choice is important because the system architecture may affect the willingness of GPs to participate in providing epidemiological data from their EHR system.MethodVerifying the performance and properties of an architectural approach by implementing and deploying a system on a trans-institutional level and performing evaluations studies is a very resource demanding method to establish a foundation for the decision of appropriate system architecture. Instead, we have tried to create this foundation by constructing a prototype system, establish knowledge about the properties of the system using experiments, and finally compare the properties of the federated approach to the properties of the centralised approach. By using this methodological approach we provide the best available knowledge, on this stage, for the appropriate system architecture to use for providing access to epidemiological data from the local population.ResultsOur experimental results show that it is possible to improve the timeliness and the temporal and spatial resolution of epidemiological data, compared to traditional centralised disease surveillance systems. Up-to-date epidemiological data from the local population may be provided directly from the source EHR system within 4 s. The responsiveness of the system is minimally affected (0.1 s) as the number of participating data providers grows from 1 to 49 data providers. The comparison of the federated approach to the centralised approach indicates that federated approaches avoid the privacy issues involved, as intended; it offers better scalability when computing speed is compared, and it provides better specificity because more data about the patient may be used.ConclusionThe conclusion from our study is that the federated approach to providing epidemiological data about the local population has many benefits over the traditional centralised approach. A federated approach to an epidemiology system may raise the GPs awareness of local disease outbreak because it is possible to share information about incidence rates of communicable diseases and use of laboratory requests in a geographical area that predates laboratory-based disease surveillance. The effects of the federated approach could be improved data quality in the EHR systems and improved representativeness of the epidemiological data for the areas covered by such systems.