Healthcare Terminology Systems Explained (part 2)

Written By Brian Carlsen

How Clinical Terminologies are Released

In part one we talked about why Clinical Terminologies matter, some of the key terminologies in use today, and some of their basic components. In this article we will talk more about how Clinical Terminologies are released and how you can interact with them.

(For a deeper dive, read part three Common Clinical Terminology Components and part four What Makes Healthcare Data So Difficult?)

Clinical Terminology Releases

Clinical Terminologies that are used within healthcare are maintained and released by various organizations. Each country regulates and/or mandates the use of various combinations of clinical terminologies.

Many countries have a single entity/organization that is responsible for the maintenance and distribution of the relevant terminologies used within that jurisdiction (often called National Release Centers), for example NHS Digital in the United Kingdom.

In the United States, there are several organizations that release the terminology standards that are in use such as The National Library of Medicine (NLM) and Centers for Medicare & Medicaid Services (CMS).

You can receive all the US terminology standards via the NLM’s Unified Medical Language System (UMLS); however, the updates to the UMLS are delayed by the releases from the maintaining organizations such as SNOMED International and Regenstrief.

US Terminology Landscape

As seen in the table above, each of these terminologies are released in different formats and on different schedules. Since terminologies have different release formats, you will need to build an assortment of scripts and tools to view and work with them within your system.

Otherwise, you will need to transform them into a common format and build your tooling based off that. In fact, this what most terminology vendors do for their customers in building proprietary formats into their tooling.

Semantic Overlap

To further complicate things, each of these terminologies may overlap others to some degree and requires proper understanding of when to use concepts from each terminology. For example, the LOINC code 41995–2 Hemoglobin A1c [Mass/volume] in Blood is made up of six parts:

1. Component = Hemoglobin A1c

2. Property = Mass Concentration

3. Time = Point in Time

4. System = Blood

5. Scale = Quantitative

6. Method = not used for this code

Each of these parts of the LOINC code have a corresponding concept in SNOMED CT and, in fact, the LOINC code itself could be given a parent concept within SNOMED CT.

Health IT software and applications require the combination of various required clinical terminologies that are represented in different formats but with overlapping content. There have been a few large healthcare organizations that have tackled the combination of clinical terminologies in the past and have created their own integrated terminologies (see Kaiser Permanente’s CMT).

These combinations require a lot of resources to develop and maintain, so most organizations rely on commercial vendors.

However, most commercial vendors only provide the ability to store clinical terminologies in a common proprietary data format and do not provide much in the way of integrating the terminologies together.

As you can see, there are a lot of challenges, work to be done, and expertise needed to integrate terminologies together. This is due to the overall size of many of the clinical terminologies, the complexities involved with integration, and the underlying differences in the ways that the clinical terminologies are represented logically.

Since these various clinical terminologies are released in different formats, the first step to eliminate complexity would be to select a common representation that supports all of the features found in common clinical terminologies.

I helped author one of these common representations under a HL7 project called Standardized Terminology Knowledgebase (TINKAR) which began in 2020 and was released as an Informative Document in August 2021.

Working with Terminology Data

There are a variety of ways to work with the clinical terminology data that has evolved as the Medical Informatics field has matured. It starts from using basic text editors to industry-focused solutions. Here is a summary of the types of solutions available today.

Text Editors/Excel — Depending on the size of the terminology you are attempting to use, opening and browsing them using a Text Editor or spreadsheet software like Microsoft Excel is a quick and easy option. This would work well with simpler terminologies like ICD-10-CM or even LOINC. However, if you are attempting to use a more complex and large terminology like SNOMED CT, this is not the most optimal way of working with terminology data. If you are looking at doing a lot of work with terminology (mapping, creating value sets/reference sets, extending content) then using text editors and spreadsheets will become difficult and more error prone the more in-depth you get. Our experience is that terminologists that migrate from the Text Editors/Excel approach find that mature solutions make their jobs far more productive.

Database load scripts — Another option would be to load the data into your favorite database. This is a great option for those who understand the underlying data structures of each terminology and are capable of performing queries across those data structures. As mentioned above, each terminology has a different data format and would therefore require different database load script.

While this seems like a simple solution, the challenges arise quickly. Each terminology requires a unique database schema which users must learn (and appreciate the differences). Now, adding the need to query across multiple versions of each component within each terminology, the level of complexity increases rapidly. WCI has elected to make freely available SNOMED CT load scripts that allows you to run SQL queries against the latest version of SNOMED only (given the challenges in supporting multiple versions in a clean way within a database). You can find these load scripts, updated with each new version of SNOMED CT — US at: https://github.com/WestCoastInformatics/SNOMED-DB-Load-Scripts

Terminology servers (APIs) / browsers — Software dedicated to making clinical terminology available through an API. While these servers support system-to-system communication (such as providing the concept information associated with a given SNOMED Concept ID), these servers are also usually paired with a front-end user interface that allows you to search, view, and browse terminology and its components in a web browser. There are many terminology servers available with various capabilities. Here are a number of examples of each made available by terminology standards organizations:

  1. Browsers — There are many terminology specific browsers that will allow you to interact with a single terminology.

WCI has a terminology server application that can be populated with any standard or local terminology an organization requires and contains a browser that can be used to view the terminologies.

2. APIs — a means of connecting a computer or software to the terminology server so that terminology can be utilized. Below are a few examples of available APIs that can be used for various purposes:

Terminology specific APIs that support only one Clinical Terminology

Terminology Agnostic APIs

  • WCI’s terminology server includes an API that will support any standard or local terminology an organization requires. You can view a demonstration of our browser and API that is based on the terminologies available in the UMLS at: https://umls.westcoastinformatics.com/

  • National Cancer Institute Enterprise Vocabulary Services (EVS) makes available a wide variety of terminology tools at: https://evs.nci.nih.gov/tools

NLP APIs

These APIs are used to take text and extract medical concepts and map them to specified clinical terminologies.

Working with a single clinical terminology contains its own complexities and challenges. Adding in the need to work with multiple clinical terminologies with different release schedules, release formats and overlapping content necessitates the need to work with specialized software, API’s and browsers.

Visit our Resources page to find out more about the clinical terminology utilities we offer. Thanks for following along with our series on healthcare terminologies.

We have a few more planned in the upcoming weeks and months and hope you continue to find this all interesting and informative!

Brian Carlsen
Previous

Healthcare Terminology Systems Explained (part 1)

Next

What Is the USCDI?