A COMPARISON OF EHR AND CLAIMS-BASED COHORT DEFINITIONS ON REAL-WORLD RESEARCH RESULTS FOR ATHEROSCLEROTIC CARDIOVASCULAR DISEASE (ASCVD)
Author(s)
Michael Buck, PhD1, Michael Wong, BEng2, Jared Kamauu, BA1, Aaron Kamauu, MPH, MS, MD1.
1Navidence, Salt Lake City, UT, USA, 2OneMedNet, Eden Prairie, MN, USA.
1Navidence, Salt Lake City, UT, USA, 2OneMedNet, Eden Prairie, MN, USA.
OBJECTIVES: In prior evaluations [ISPOR 2026], it was found that different definitions for complex conditions like Atherosclerotic Cardiovascular Disease used often in real-world research, depend more on the right specific codes being included rather than maximizing the number of codes. To pragmatically validate this finding further we applied the same methodology used on claims data only to two overlapping claims and EHR datasets.
METHODS: Based on a targeted literature review, we leveraged five different ASCVD cohort definitions containing the following number of ICD-10-CM codes: Cohort1 at 267, Cohort2 at 627, Cohort3 at 637, Cohort4 at 544, and Cohort5 at 202. The cohort definitions cover common conditions like peripheral artery disease (PAD), ischemic stroke and myocardial infarction. Using two different sources, OneMedNet Billing claims and EHR data, we produced an UpSet plot analysis of each cohort definition.
RESULTS: There were 237,983 patients matching at least one of the five definitions across the Billing or EHR sources. Billing directly identified 217,901 and uniquely identified 95,670. The EHR source directly identified 142,313 and uniquely identified 20,082. Billing patient counts ranged from 76,784 to 207,188. EHR patient counts ranged from 61,855 to 133,813. For Billing, the most frequent cohort intersections were 65,724 patients for all 5 Cohorts (30%), 57,845 in Cohort3 only (27%), 33,124 in Cohort2_3_4 (15%). For EHR, the most frequent cohort intersections were 52,305 patients for all 5 Cohorts (37%), 33,241 in Cohort3 only (23%), and 18,855 in Cohort1_2_3_4 (13%).
CONCLUSIONS: The same top three Billing cohort intersection groups in this study identified the same percentage of patients as the prior Billing source study (~70%). The top three EHR cohort intersection groups account for 73% of patients and differ only by Cohort1’s identification of more patients with H43.1x and H43.2x retinal occlusions compared with Billing. Consistent results across multiple nationwide sources greatly supports future research.
METHODS: Based on a targeted literature review, we leveraged five different ASCVD cohort definitions containing the following number of ICD-10-CM codes: Cohort1 at 267, Cohort2 at 627, Cohort3 at 637, Cohort4 at 544, and Cohort5 at 202. The cohort definitions cover common conditions like peripheral artery disease (PAD), ischemic stroke and myocardial infarction. Using two different sources, OneMedNet Billing claims and EHR data, we produced an UpSet plot analysis of each cohort definition.
RESULTS: There were 237,983 patients matching at least one of the five definitions across the Billing or EHR sources. Billing directly identified 217,901 and uniquely identified 95,670. The EHR source directly identified 142,313 and uniquely identified 20,082. Billing patient counts ranged from 76,784 to 207,188. EHR patient counts ranged from 61,855 to 133,813. For Billing, the most frequent cohort intersections were 65,724 patients for all 5 Cohorts (30%), 57,845 in Cohort3 only (27%), 33,124 in Cohort2_3_4 (15%). For EHR, the most frequent cohort intersections were 52,305 patients for all 5 Cohorts (37%), 33,241 in Cohort3 only (23%), and 18,855 in Cohort1_2_3_4 (13%).
CONCLUSIONS: The same top three Billing cohort intersection groups in this study identified the same percentage of patients as the prior Billing source study (~70%). The top three EHR cohort intersection groups account for 73% of patients and differ only by Cohort1’s identification of more patients with H43.1x and H43.2x retinal occlusions compared with Billing. Consistent results across multiple nationwide sources greatly supports future research.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
SA40
Topic
Epidemiology & Public Health, Real World Data & Information Systems, Study Approaches
Disease
No Additional Disease & Conditions/Specialized Treatment Areas