A conversation came up recently with someone on the use of and adherence to healthcare industry standards inside of a particular robust solution in healthcare. I have seen many an organization model their internal standard data model after an industry-wide data model such as HL7 CDA or HL7 FHIR. What seems to be often missed is the importance of an organization to realize that they will always need their own “flavor” of that standard for internal use.
The idea to follow the industry standard is one that is very well-intentioned, but it is also extremely difficult to implement and maintain. It is also not always the best choice because industry-wide standards are intended to handle use cases across many different organizations (hence the name “industry-wide”) and while they may meet the specific needs of the organization desiring to implement the standard, they may also include additional “baggage” that is not helpful to the organization. Conversely, they may require extensions or adaptations to the model to fully support the organization’s specific use cases. The effort required to implement the content model with either of these considerations can become burdensome.
We must realize that industry standards are quite important to drive health IT applications toward common ways to exchange and communicate data, but they must be at the guidance level, and not the end-all-be-all way to represent data that needs to be shared. This is now being realized in the data warehousing market as ‘schema-on-read’ is becoming a more popular approach to dealing with analytics on large data sets as opposed to ‘schema-on-write.’ The optimism on ‘one data model to rule them all’ is shrinking. A good example of this would be a solution that leverages metadata for the anlaytical data points rather than the concrete source data structures. This allows an application to focus on writing good queries, and lets the metadata model deal with the underlying differences in the source data model. It provides an effective layer of abstraction on the source data model, and as long as that abstraction layer properly maps to the source data model, then we have an effective ‘schema on read’ solution. This sort of approach is becoming more and more necessary as the rate of change in the technology and in healthcare IT is still increasing.
Internal standards are more manageable. Organizations can design and implement a content model for a set of use cases in a reasonable time frame with a reasonable amount of resources. This model may even be based on an industry-standard model, but it must not BE the industry standard model! What I mean by that is that expectations must set clear from the outset that the model WILL change over time as the organization changes, as the business opportunities change, as laws change, etc. As the decision is made as to what the internal model is to be, it must be understood that it is for that organization only and mapping shall be provided to and from as needed, while looking for opportunities of reuse across specific data elements or data element groups.
What this all drives toward is having interoperability as a first class citizen in an orgnization’s solution set. The content model is important, but the content model is designed for internal usage, with mappings to external systems’ content models. In addition to the content model, an organization must also include their implementation approach in their overall strategy to ensure that external systems can be mapped to the internal content model effectively (on time, under budget, and meeting the business needs). A great strategy without an execution plan will die on the vine.
In summary, the intent of this post is an attempt to clarify the difference between this idea of an external data standard and an internal data standard, and the overlap between these ideas. Interoperability is not a clear cut landscape. Interoperability is hard. We must realize and accept that fact and look for ways to work effectively within it to drive toward higher quality communication between health IT systems, leading to improved patient health outcomes.