Health Information System Architecture: Harmonized and Interoperable

By Manish Kumar, MPH, MS, Senior Technical Specialist for Health Systems Strengthening, MEASURE Evaluation

Photo by Manish Kumar, MEASURE Evaluation

In many low- and middle-income countries, collected health data more often serve the needs of national policymakers and are less often used at the clinical level to improve or monitor specific client outcomes. In a new article, Federated Health Information Architecture: Enabling healthcare providers and policymakers to use data for decision-making,[1] I and my co-authors argue that the architecture of existing health information systems is partially to blame and that a “federated” system architecture could enable the capture and use of health data to serve both national and clinical needs. Besides, it can overcome barriers associated with centralized and decentralized health information architecture.

A centralized health information architecture offers economies of scale and reduces costs through a central data repository. But it is less responsive to local needs and has a longer implementation timeline. A decentralized architecture is quick to deploy and adaptable to local needs, but it leads to redundancies and higher costs while limiting the ability to create a larger picture of any one health information need.

We conducted research in Bihar, India, where, even though aggregated individual healthcare data are communicated with increasing efficiency to national public health databases, clinical information on clients still is usually stored in multiple, nonstandard paper formats. The inaccessibility of data becomes more problematic when clients visit multiple clinics or multiple providers. The resulting data fragmentation impacts cost and quality of care. Not surprisingly, in the absence of national or state-level health information architecture policies, regulations, and standards, each health department or health program uses the technology of its choice, separately managed by the department or program and no policy or regulation exists to protect and secure patient data.

The solution we propose is a federated architecture based on standards and communication capacity among software platforms that collect data. Minimum data sets (MDSs) for clinical clients would include demographic information such as unique patient identification, date of birth, and address, plus past health history, clinic encounters, provider details, diagnosis, and treatment plans. The MDS also would allow query to discover specific past clinical decisions and could be aggregated in an anonymous fashion to inform policy decisions.

We conclude that this type of architecture would—over time—ensure the standardization of data collection and reporting formats and establish standards for data interchange to allow the consistent, accurate, and reproducible capture of clinical, administrative, diagnostic, and drug data across diverse health information systems. Federated architecture offers an opportunity to reduce duplication in data collection and sharing, solve inconsistency among data, and improve efficiency and quality. Extracting, transforming, and aggregating data will allow them to be pulled from different health information systems and the census database. Those data would be organized in a data warehouse where managers can use data analytics and data visualization to enable evidence-informed decision making.

This move would enable progress toward the gold standard: where policymakers and managers are making decisions and adjusting programs based on good-quality data from multiple sources, purposefully collected and organized to enable strategic decisions for better health.