Occasional articles about what's coming in the next 12-24 months in health care IT.
Thursday, January 14, 2016
Interoperability: It’s difficult!
This article is the first of a series focused on interoperability, a key to continuing the momentum started by the Meaningful Use program, highlighted in “2016 HIT Trends: Consensus Predictions.”
Now that a critical mass of healthcare data is stored electronically rather than on paper, it's natural that we would expect it to flow freely among electronic systems. The healthcare community is learning what integration specialists have always known: interoperability is hard!
Why? Because there are many details that have to be exactly aligned in order to get two systems talking to one another. Here are four aspects of interoperability with a rating of how we're doing overall in each.
1. Protocol-level interoperability. This means just being able to get a message from one system to another regardless of the message content. For example, connections can be made over a simple TCP connection, a heavyweight SOAP web service, or any number of other standard protocols. After agreeing on a protocol, sending and receiving systems must agree on security methods. While there are a lot of details to handle, these are usually reasonably easy to resolve. Grade: A
2. Syntactic interoperability. Syntax is about transmitting messages that can be read. Like protocol interoperability, there are a number of standards in common use: XML, HL7, X12, and many others. Once there is agreement on general syntax (e.g., HL7 version 2.4), there is still a lot of work to do to ensure that, between sending and receiving systems, all the fields are in the same place (e.g. is it "firstname lastname"? or "lastname, firstname"?), field lengths match, etc. This typically takes much longer to negotiate, but usually the receiving system can get what it needs from the sender. Grade: B+
3. Workflow interoperability. After the systems are successfully exchanging messages, the applications need to agree on what is appropriate to say. For example, imagine an EMR that receives lab results from a laboratory information management system (LIMS). The EMR might reasonably require that a lab result must be in the final status before a correction is accepted. But perhaps the LIMS allows the user to skip the final lab result if a correction is available at the time of transmission. If the LIMS transmits the correction without first sending the final result, the EMR will reject the message due to the unexpected status. These types of interoperability mismatches tend to be difficult to smooth over because each system has different business rules to define expected messages. These rules are usually hard-coded in the application rather than handled in the interface layer, where more flexibility is available. Often a resolution is only possible with compromise, which can sometimes impact the seamless flow of data. Grade: C
4. Semantic interoperability. Semantic interoperability means the sender and receiver have achieved a common understanding by exchanging a message. This is only possible if the sender and receiver have the same meanings for each data field and use the same code sets to encode those fields. For example, does "diagnosis" mean admission diagnosis or discharge diagnosis? Is it coded with ICD-9 or ICD-10? (Or SNOMED?). Translations among coding systems are possible, but usually imperfect. They can only be implemented if an interface developer realizes there is a mismatch. Grade: D
Other industries have standards that make interoperability relatively easy. Unfortunately, such standards are not available in healthcare for two reasons. First, various vendors have used different data models and business rules to represent healthcare concepts in their systems. Rather than implementing standards with strict adherence to the specifications, vendors often implement standards loosely so that their model of healthcare more easily fits into the standard. (Using Z-segments in HL7 version 2.x is the textbook example of this type of loose implementation.) Second, the richness and importance of healthcare data make it difficult to capture the meaning and nuance electronically, let alone trying to have one system communicate with another in an understandable way.
The path to complete complete interoperability in healthcare is going to be long and difficult. But it’s not all doom and gloom. Looking on the bright side there has been remarkable progress so far, making healthcare interoperability worth the pursuit in achieving better care for all.
My next posts will drill a little deeper into the prediction that interoperability is the key to continuing the momentum started by the Meaningful Use program.