Written by: Nicole Darden, Manager Solutions Management, Veradigm, and Janna Manjelievskaia, Director Health Econmics & Outcomes Research, Veradigm
Within clinical care settings, different coding systems are used depending on the coding’s purpose. Two of the more commonly used coding systems are the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) and the Systematized Nomenclature of Medicine, Clinical Terms (SNOMED CT). ICD-10-CM is one of the first generation of terminology systems, a mono-hierarchical system originally created to track diseases in a population. This coding system organizes content into meaningful standardized categories and is used in practice for billing and reimbursement. SNOMED CT, in contrast, is a poly-hierarchical, concept-based coding method, one of a new generation of coding systems where attributes or characteristics connect the different SNOMED CT concepts, building relationships between them. This system enables clinicians to represent detailed clinical information in a consistent, reliable, and comprehensive way.
Veradigm Network Electronic Health Record EHR (VNEHR) Data provides options for using either or both coding systems. In this article, we share recent research that used VNEHR Data, from 2017 to 2022, to gain a better understanding of how these 2 coding systems compare for identifying common, rare, and ultra-rare diseases.
Veradigm researchers used Veradigm Network EHR patient data from 2017 to 2022 to assess the frequency of diagnosis of 4 independent conditions using SNOMED CT and ICD-10-CM coding, both separately and in combination. The 4 conditions assessed were:
Researchers identified the total number of unique adults with each diagnosis and the total number of unique visits by each adult. These values were reported both overall and by separate use of SNOMED CT and ICD-10-CM.
Figure 1: Proportions of patients diagnosed using both ICD-10-CM and SNOMED CT in combination, only ICD-10-CM, or only SNOMED CT, for the common conditions type 2 diabetes mellitus (T2DM) and polycystic ovarian syndrome (PCOS); Turner’s syndrome (a rare condition); and Gaucher’s disease (an ultra-rare condition).
For the 2 common conditions, researchers identified more than 13 million adults with T2D and over 561,000adults with PCOS using both coding systems. For both conditions, a majority of adults were identified only via ICD-10-CM diagnosis—which could possibly be related to the greater use of ICD-10-CM compared to SNOMED CT coding in clinical practice. However, for the rare and ultra-rare conditions, researchers identified about 8,000 adults with Turner’s and 727 adults with Gaucher’s using both coding systems. For these 2 conditions, researchers found that a majority of patients were diagnosed with both coding systems rather than with just one or the other.
A small proportion of patients were identified via SNOMED CT only. Diagnosis using SNOMED CT only was highest for Turner’s and Gaucher’s patients (3% to 4%).
Based on these findings, we concluded that SNOMED CT and ICD-10-CM terminologies yield different rates of correct identification of patients with both common and rarer diseases. Because of this, our research demonstrates that ICD-10-CM diagnosis and SNOMED CT terminologies can provide complementary means of identifying patients with common and rare diseases in ambulatory EHR data.
The use of both terminologies to identify patients with specific diseases can provide a more complete picture of disease frequency than could be obtained using only one of the two systems. Our research also shows that it is especially valuable to assess SNOMED CT diagnoses when attempting to capture patients with rarer diseases, such as Turner’s and Gaucher’s, as some patients with these diagnoses would be missed if researchers assessed only ICD-10-CM diagnoses.
The 2024 International Society for Pharmacoeconomics and Outcomes Research Conference—ISPOR Europe, 2024—is fast approaching! We will be in Barcelona to present more detail about our research (ISPOR Europe, November 17-20, 2024). Find us there to discuss our results—or reach out today so we can tell you more.