eHealth Architecture-based Health Data Exchange
Ethiopia DHIS2 and SmartCare
Background and Purpose: A blueprint of the national-level arrangement of health system components in Ethiopia is depicted in the eHealth Architecture (eHA). At the time of the study, the lack of practical implementation experience limited Ethiopia’s ability to move toward maturing the architecture. In this study, the team set out to explore practical implementation and scaling solutions that leverage open standards and tools. Two major components of the eHA were used to demonstrate the health data exchange: the legacy Electronic Medical Record (EMR) system and the national electronic Health Management Information System (HMIS) instances. At the time of the research, the HMIS contained 53 data sets used by more than 35,000 health facilities serving at different layers of the health sector, including health posts, health centers, hospitals, and other facilities, to deliver reports of various types. It is the major source of information by the Federal Minister of Health. The main purpose of this study was to explore potential opportunities and overcome challenges related to health data exchange patterns of national eHA.
Methods: A thorough assessment of related works were conducted to examine global, national, and local perspectives. An eHA-based health data exchange model was developed, which leveraged open tools and standards. HAPI FHIR, an open-source and Java-based HL7 FHIR solution was adopted to extract local relational EMR data to FHIR messaging formats. With the main focus on design scalability, the team developed a strategy for mapping local data elements to FHIR resources and created an OpenHIM mediator. To demonstrate the developed model, existing architecture components with high impact on national-level health data quality were selected. Finally, an evaluation method was conducted using the top ten diseases which compared the computer-based solution and manual tally sheet-based data entry.
Results: The developed model was aligned with national eHA data exchange patterns and resulted in enhanced data quality. Data quality was examined using timelines, accuracy, completeness, and cost parameters over a 12-month period of production EMR data for the top ten disease classifications. Scalable local EMR data-to-FHIR resource data element mapping was developed using the HAPI FHIR library. Its effectiveness was proven as the model effectively mapped to the elements for the identified FHIR resources. Study results also showed the use of eHA interoperability patterns produced enhanced HMIS data quality and established reusable data exchange modules with national impact. Though Ethiopia has a standardized national classification of disease adopted from World Health Organization ICD-10, this extensive experiment revealed fragmented efforts toward health data reporting resulting in inconsistent data elements. To overcome the challenge, the study suggested using national terminology service instances with the capability to cascade concepts at a facility level.
Conclusions: A scalable health data exchange model was developed and tested with the major data source components of eHA, EMR, and the national HMIS instances. Interoperability of those systems significantly fostered national eHA maturity. Therefore, this study asserts the use of open global standards and tools facilitates the maturity of the national eHA.