Developing an Architecture: Building Foundational Registries and Exchange Components

Developing an Architecture

Developing an Architecture

The Malawi team is maturing and growing their registries which support data exchange by standardizing and linking key elements of health data. One of these components is the Master Patient Index (MPI) also known as a client registry (CR). This is one of the key registries to support a patient’s shared health record, or health data combined from different sources. The MPI does this by matching or linking a patient’s identity records from different point-of-care systems which creates a foundation for linking the patient’s health data from different systems. 

With the MPI, a few considerations were needed throughout the implementation. In Malawi, they already had the ART-specific implementation of an EMR that also had a way of standardizing patient identification. The system used a quality demographic data exchange and was rolled out in over 700 facilities across the country, out of a little more than one thousand facilities. 

Additionally, this team started with an implementation of the master health facility registry, a web-based application that has an API. This is the source of truth for health facility information in Malawi and was used to validate information as it was being migrated into this singular repository. 

Following the work on the master health facility registry, the team began to review the terminology service needs in order to harmonize all the medical terminologies within the health information system space. As the information was migrated, there was then one standard way of recognizing the terminologies.

The next project was setting up the Open Health Information Mediator (OpenHIM) that provides features and functionalities supporting secure transactions between systems. This team then created their own mediator, the Aggregate Data Exchange for exchanging aggregate values between systems.

Lessons Learned

Lessons Learned

There have also been challenges encountered during this implementation setup, including deduplication. There will be many branches for partner demographic data and so they anticipate finding a way to address this.