Peeling back the onion of health data interoperability

Data interoperability is vital to today’s healthcare computing environment, allowing clinical information to be effectively and consistently exchanged, compared, and analyzed among healthcare partners such as insurers, pharmacies, affiliated providers, and public health departments. Put simply, data interoperability enables better decision making. The solution is clear: data standardization. But statistics show that the development and adoption of data standards is much easier said than done. Why is that, when we are clearly no stranger to standards in other aspects of our daily lives: railroads run on the same gauge track, we have a standard system of weights and measurements, and ISO 9000 has become the norm for quality systems management. Clearly, a set of common clinical data standards which would help reduce to the cost of healthcare and enable better outcomes, is long overdue.

As an architect, I usually find myself thinking top down and jumping straight into "clouds and arrows" mode when I think about a problem like this. But on occasion I find it useful to start at the bottom and work my way up. This is one of those times, so I'd like to start by peeling back of the rather large and layered onion of "data interoperability". First, a simple yet meaningful definition of data interoperability: the ability of two parties to exchange information that can be understood by both parties. That may seem like too basic a definition, but it masks a fair degree of complexity if we break it down.

Firstly, the notion of exchange infers agreement on a set of mechanisms for the transfer of the information. Considerations in an IT exchange include network, transport, application and messaging protocols. Thanks to the OSI layered model and the phenomenal growth of the Internet, most information exchanges conform to the TCP/IP transport protocol (not surprisingly, there are a few well known exceptions to this in healthcare). On top of this there are a wider variety of protocols in use: applications protocols like HTTP or FTP, and messaging protocols like ebXML, JMS or SOAP. Some older, established standards like HL7 v2.x are still in wide use in the healthcare industry, which require conformance to several of the layers described above. Either way, agreement on these protocols between two parties represents a contract for exchange of information.

The second part of the definition above deals with facilitating shared understanding between the parties. There are two key considerations: agreement on the structure and meaning of the data. Structure is defined as the formatting and order of information, and that it conforms to a set of expected rules. Those rules may be fairly loose in the case of requiring only valid XML, or more strict like conformance to an information model or schema as in the case of X12, HL7v3 or NCPDP. Meaning is defined as being interpretable, by humans but also by machines in some cases. After all, let's not lose sight of the true goal of data interoperability: to enable better decision making. Ultimately these decision fall to humans, but every efficient flow of information requires numerous machine-decisions along the way e.g. content routing, authentication, alerts, etc. Achieving a shared understanding requires agreement on the meaning of the information, referred to as "semantics". Semantic considerations for health data might include detail on the nuances of clinical diagnoses, care processes or lab tests.

Facilitating this shared understanding is often a challenge for healthcare organizations wishing to exchange health data, due to significant variation in use of medical vocabularies, level of customization, and geographic, cultural and professional differences. Data standardization can address these challenges to enable organizations to exchange information that can be understood by all parties. In practice, this standardization takes the form of a set of processes responsible for "normalization" of the clinical terms in data, which is the recognized way to remove ambiguity and improve comparability. In the next post I will describe an architecture to facilitate data normalization, with some specific examples that highlight the myriad of benefits to decision making.

Intel® SOA Expressway for Healthcare is a specific implementation of a new product category called a SOA "soft appliance", which delivers a breakthrough in simplicity, cost and scalability for enabling high-quality health information exchange.

Para obtener más información sobre las optimizaciones del compilador, consulte el aviso sobre la optimización.