OBJECTIVES: To describe and compare the level of clinical capture for common, rare, and ultra-rare diseases using ICD-10-CM diagnosis codes vs. SNOMED CT codes among adults in a US ambulatory EHR database.

METHODS: Using data from the Veradigm Network EHR between 2017-2022, we identified four conditions (common: type 2 diabetes (T2D), polycystic ovarian syndrome (PCOS), rare: Turner’s syndrome, ultra-rare: Gaucher’s disease) using ICD-10-CM diagnosis and SNOMED CT coding separately and combined. The total number of unique visits with each diagnosis and the total number of unique adults were reported overall, and separately by SNOMED CT vs. ICD-10-CM.

RESULTS: We identified 13,085,811 adults with T2D, 516,054 with PCOS, 7,948 with Turner’s, and 727 with Gaucher’s between 2017-2022 using both coding systems. The patients corresponded to 93,923,669 (T2D), 1,609,807 (PCOS), 31,777 (Turner’s), and 3,876 (Gaucher’s) unique visits during the same time period. For T2D and PCOS, around a third (32-35%) of adults had a diagnosis in both ICD and SNOMED, compared to 55% of adults with Turner’s, and 60% of those with Gaucher’s. T2D and PCOS had higher proportions of patients with a diagnosis using ICD only (64-66%), compared to Turner’s and Gaucher’s (36-41%). Turner’s and Gaucher’s had the highest proportion of patients with diagnoses using SNOMED only (3-4%), compared to the other conditions (all <2%).

CONCLUSIONS: In the two common conditions examined (T2D, PCOS), a majority of adults were identified via ICD-10-CM diagnosis only, while in the two rare/ultra-rare conditions (Gaucher's, Turner's), a majority were identified via both. A small proportion of patients were identified via SNOMED CT only, with Turner’s having the highest proportion. ICD-10-CM diagnosis and SNOMED CT terminologies can be a complementary way to identify patients with common and rare diseases in ambulatory EHR data.