by adgrooms on November 7, 2019

EHR systems aren’t very good at communicating with each other yet. Getting systems to communicate with each other is a common technology problem, but it’s still a little strange that on the cusp of 2020, with all of the technological advancements available to medicine, and the US government push into EHRs that we still don’t have ubiquitous communication between systems. Why is that?

We recently wrote a post about the basic concepts of interoperability. To recap, there are three base levels of interoperability: Physical, Syntactic and Semantic. Physical is the medium in which the data is transferred. Syntactic is the organization of the data in the message. These two levels appear to be resolving into a common standard, for the most part. This week, we are taking a closer look at semantic interoperability which means the ability to use a commonly understood, unambiguous vocabulary. Without a standard vocabulary, the systems may be able to share information, but for the people trying to share information, the meanings will be unclear.

To achieve this clarity, concepts that mean the same thing need to be identified and distilled to an appropriate standard. Reconciling different terms that have a shared meaning is called data normalization. For instance "trouble breathing" could also be "difficulty breathing". "Suicidal Ideation" could also be "thoughts of self-harm". If they are not properly interpreted, these inconsistencies can lead to inaccurate or missing data when transmitted. This could lead to inefficiencies such as retesting a patient to fill in the gaps or worse making incorrect decisions based on incomplete information.

The complexity of medical data is a human/technical puzzle to be solved. There are thousands of terms and data points for symptoms, diseases, observations, and medications and several sets of standards used to describe them including ICD-9/10, SNOMED CT, LOINC (for labs), and RxNorm (for drugs). (Each standard is expansive, for example, the ICD-10 system of diagnostic codes has 68,000 entries). For EHR systems to communicate, the data must be consistent and use defined terminology. To achieve consistency, the data must be entered uniformly, so that the same information can be found in the same fields of every record.

For each type of information (labs, medications, problems) an encoding standard must be chosen. The encoding standard can be unanimously selected for all organizations (the most efficient way) or mappings between standards must be applied between the organizations. This includes not just inpatient clinical data, but also data from labs, home health, pharmacies, and payers. It is by no means a simple or trivial problem. Either a large contingent of organizations must do the difficult work of selecting and implementing a standard approach, or every organization needs to be able to take data from disparate sets of terminology and reconcile it correctly in their EHRs according to the standards used by their clinicians.

EHR companies also bear responsibility for achieving semantic interoperability. They will need to implement these standards or translation maps between them. They have resisted working together on the issue of interoperability, but the day is coming where they will have to support an agreed-upon standard. Market demand will determine who survives once the EHR landscape approaches full interoperability. Those who don't play well with everyone else can learn a lesson on standard adherence taken from internet browsers:

There are a few main players in the world of browsers: Safari - by Apple; Firefox - by Mozilla; Chrome - by Google; and the now-defunct Internet Explorer (IE) - by Microsoft. The first three could be used, for the most part, interchangeably - because they implemented internet standards consistently. They were motivated to support interoperability. Microsoft wanted IE to be the sole internet browser on their systems. IE did not support internet standards completely or consistently.

The result was an unreliable and inconsistent experience between any other browser and IE. Developers would have to develop web apps twice so they would work on regular browsers and also IE. This problem was resolved by the users. Ultimately, people chose every other browser over IE. This has led Microsoft to discontinued IE and instead offer a browser, called Edge, which is built off of Google's (standard, interoperable) Chromium engine.

The EHR vendors that contribute the time and effort to find a common standard will come out ahead. The outliers are only concerned with the bottom line and market share. With some work and agreement, semantic interoperability will bring great value and improvement for all stakeholders, primarily patients and physicians.