Standardizing the 'Meaning' of Data in HIEs | Jennifer Prestigiacomo | Healthcare Blogs Skip to content Skip to navigation

Standardizing the 'Meaning' of Data in HIEs

| Reprints
Moving beyond normalizing content, format, and getting to the meaning of the data

Recently, I was doing research and interviews on the August cover story package that will be delving into the ever complex and evolving topic of health information exchanges. As the topic is as deep as the ocean and wide as the sea, I decided to focus on the different data architecture models and how different business models were influencing different data architectures. I want to know why certain exchanges are choosing to adopt a central repository model as opposed to a federated or hybrid approach or vice versa. While conducting an interview with Patrick Rossignol, principal, Technology, Deloitte Consulting LLP, makes a salient point that I also need to focus on standardizing the meaning of the data that HIEs intend to share.

Rossignol notes that the industry has come a long ways in terms with standardizing the content that HIEs are sharing, using standard content types like the CCD and CCR. The industry has also done much in the way of standardizing the format of that content by adopting HL7. But he notes that we still have a ways to go on standardizing the meaning of data. For example, what does Patient A’s allergy to penicillin really mean? To what degree is that patient allergic to penicillin and does it mean the same thing when compared to Patient B’s allergy when doing population health analytics and reporting? Rossignol says that until we have national conversations about this topic and reach a national consensus to normalize the meaning of the data that HIEs share, we won’t ever truly be interoperable to enable the ‘meaning’ful exchange of health information.

Dev Culver, executive director, HealthInfoNet, Maine’s statewide HIE, says that topic was one of the big ones discussed in 2005 when HealthInfoNet was getting its start. By architecting a centralized data model for HealthInfoNet, Culvers says that the exchange maps all the local codes that the participating organizations use and align them with LOINC identifiers to ensure that each piece of data is standardized across the exchange.

I’m curious what kinds of discussions your organization has been having around this topic. Leave a comment below with your thoughts.




Great post. Without question, the heterogeneous world of HIEs play a critical role in the storage, transport, and business of semantic interoperability.  The context and intent around captured data is often not captured and implied at best.  This isn't good enough, as was acknowledged in HBR last month (May 2011) in  "The Big Idea: The Wise Leader" by Ikujiro Nonaka and Hirotaka Takeuchi:

"... However, all social phenomena—including business—are context dependent, and analyzing them is meaningless unless you consider peopleʼs goals, values, and interests along with the power relationships among them. Yet executives fail to do just that."

I picked up this idea with an example in my Thanksgiving blog post.  In short, the turkey being raised for consumption at Thanksgiving does not have access to knowledge of the intent of the humans she sees.  Collecting information about feeding, temperature, friendliness and cleanliness has essentially no explanatory power to predict the future.  Such observations, out of context of the intent of the turkey farmer, are not only meaningless, they are distracting from getting to the truth.  That's always been an issue with claims data in healthcare.  HIEs, like any powerful tool, can lead to either good or bad impacts, depending on how appropriately data is captured and interpreted.

In Mark Hagland's recent cover story "Balancing Act" on Physician Documentation, and my posts on the topic, we elaborate some of the needed context that we're not capturing today.  These include explicit connection between what it documented and what is ordered, why specific medications are continued or discontinued during medication reconciliation, and why various Medicare Quality Measures were apparently ignored.  And those are just a few obvious examples.  This undocumented context is an important frontier in healthcare informatics. 

That's for your provocative elaboration of the problem.