Exploring openEHR-based clinical guidelines in acute stroke care and research

Abstract

Largely speaking, health information systems today are not able to exchange data between each other and understand the data’s meaning automatically by means of their information technology components. This lack of ‘interoperability’ also leads to patients experiencing an undesired discontinuity in their care. This thesis is a part of a health informatics field which tackles interoperability barriers by offering standardised information models for electronic health records. More specifically, this work explores possibilities of combining standardised information models offered by the openEHR interoperability approach with knowledge from evidence-based clinical practice guidelines. The applied methodology includes openEHR archetypes, the openEHR reference information model, standard medical terminologies such as SNOMED CT, the international stroke treatment registry SITS, a newly developed model for representing guideline knowledge (the ‘Care Entry-Network Model’), and rules authored in the Guideline Definition Language, a formalism recently endorsed by openEHR as a part of its specifications. The study design used is based on evaluating the work done by means of retrospectively checking the compliance of completed patient cases with guidelines from the domain of acute stroke management in Europe, both experimentally and using thousands of real patient cases from SITS. Our overall findings are that i) the Care Entry-Network Model facilitates an intermediate step between narrative guideline text and computer-interpretable guidelines to be deployed in openEHR systems, ii) the Guideline Definition Language is practicable for creating and automatically running openEHR-based computer-interpretable guidelines, where we also provide detailed accounts of our employed GDL technologies, and iii) the Guideline Definition Language combined with real patient data from patient data registries can generate new clinical knowledge, which in our case has benefited stroke carers and researchers working with acute stroke thrombolysis. In conclusion, using our methodology, health care stakeholders would get evidence-based knowledge components in their electronic health records based on shareable, well maintainable information and knowledge models in the form of archetypes and GDL rules respectively. However, our approach still needs to be tested at the point of clinical decision making and compared to other approaches for providing exchangeable computer-interpretable guidelines.