Post-Coordination and Pre-Coordination of Codified Concepts
Post-Coordination and Pre-Coordination of Codified Concepts
This article provides a brief overview of two fairly basic concepts in healthcare informatics referred to as post-coordination and precordination. For reasons that will be explained below, post-coordination is markedly underutilized and pre-coordination is markedly over-utilized in current healthcare IT systems, such as electronic health records.
Storing clinical concepts as codes has multiple advantages over free text. However, storing information as a code that represents a single unique concept does not always convey all of the information necessary for clinical care, population management, and research. For example, a symptom such as a headache can be minor or severe, and the severity of the headache may have significant implications as to its cause and prognosis. Storing not just that the patient has a headache, but also how severe it was, can be managed by the two related processes: post-coordination and pre-coordination.
Post-Coordination:
Post-coordination refers to using two codes to represent a clinical expression such as “severe headache.” For example a code exists for the symptom “headache” (e.g., SNOMED CT code 25064002) and a separate code exists for the modifier “severe” (e.g., SNOMED CT code 24484000). These two are then placed together in combination as 24484000 (severe) + 25064002 (headache), allowing the information system to identify when a patient presents with a severe headache vs. just a headache alone or a mild headache. The two codes are “post-coordinated” to create a single clinical expression with significantly greater value than when the only the base concept code for headache is stored. The challenges with a post-coordination approach are tied to the need for individuals doing the data entry (e.g., physicians using an EHR) to identify and select two codes instead of one. This may have a negative impact on usability and efficiency of documentation. Another issue that has arisen are difficulties with storing, retrieving and sharing (e.g., via a health information exchange) combinations of post-coordinated codes in a manner that protects the semantic integrity of the clinical expression. An additional example of this and a discussion about the status of the use of post-coordination is covered in another article on this website titled: “The Importance of Structured Data and Context in Healthcare.” Basically systems have not evolved to the point where they can handle post-coordinated expressions, despite their critical value in healthcare.
Pre-Coordination:
When complex clinical expressions representing two concepts or a concept and modifiers occur commonly in clinical settings, it becomes more efficient to create one code that represents both concepts. This is referred to as the process of “pre-coordination.” For example, a new code may be created that represented “severe headache.” This would prevent users from having to identify two codes (one for headache and another for severe) during the clinical documentation process. This also simplifies the process of storage and retrieval, as a single code is less complicated to store and retrieve than a combination of two codes. One of the challenges of pre-coordination is what is referred to a “terminology explosion” as additional codes are created from core concept and modifier codes to create long lists of concepts. For example, headache clinic might have value in creating multiple concepts related to headache (e.g., mild headache, mild-moderate headache, moderate headache, moderate to severe headache, severe headache, recurrent headache, refractory headache, migrating headache, grade 1/10 headache, grade 2/10 headache, etc.).
Unfortunately many proprietary systems have created their own sets of unique pre-coordinated concept codes. SNOMED CT and other terminologies offer processes whereby users can create their own local pre-coordinated codes that map to the corresponding SNOMED CT core codes, but this approach has not been readily adopted.
In summary, implemented standards that detail how systems should create, store, use locally, transmit and receive post-coordinated expressions is needed to protect the safety of patients and the integrity of healthcare information. Locally created pre-coordinated concept codes will need to be decomposed into their component clinical terms that can then be mapped to codes in SNOMED CT and other terminologies, allowing this information to have value outside of a single application.
This article represents the opinions of its author, Michael Stearns, MD
©Michael Stearns, all rights reserved