Analysis and Management of Computable Operational Definition (CODef) Concept Variation Based on a Large Real-World Studies Library
Author(s)
Michael D. Buck, PhD1, Craig G Parker, MD, MS1, Leanne Larson, MHA2, Aaron Kamauu, MD, MS, MPH1;
1Navidence, Salt Lake City, UT, USA, 2University of South Carolina - Beaufort, Wilmette, IL, USA
1Navidence, Salt Lake City, UT, USA, 2University of South Carolina - Beaufort, Wilmette, IL, USA
OBJECTIVES: Regulatory agencies are requiring life sciences companies to submit more-rigorously developed study protocols including computable operational definitions (CODefs). To date, large-scale CODef concept libraries have not been profiled or reported on, nor have maintenance requirements been delineated.
METHODS: Using a newly created CODef library currently leveraged by a few life-science companies, we profiled the subset of value sets used to support their studies. We aggregated the results at three levels: 1) counting discrete concepts from related value-set groups together; 2) counting related value-set groups for each code system that share an 80% or more code overlap; and 3) counting each unique value-set variation. For example, the discrete concept of “Cerebrovascular Disease Diagnoses” has seven related value-set groups containing 19 unique value-set variations with similar provenance, but slightly different code lists used by multiple studies.
RESULTS: As of January 2025, this new CODef library contains 912 discrete concepts defined by 2,586 related value-set groups comprising 27 different code systems (e.g. ICD-10-CM, SNOMED CT, CPT, ATC, LOINC, CVX, etc.) with 2,813 unique value-set variations overall; and growing. Details by both US and international terminologies will be presented. One interesting finding shows there are 98 discrete concepts with 227 value sets that are slight variations of the same code list. For example, the related value-set group “Cerebrovascular Disease (CCI) Diagnoses (ICD-10-CM)” contains three unique value-set variations for different studies (one with 100% overlap with the original literature review list and two with 96% overlap after dropping 23 of 528 codes).
CONCLUSIONS: As organizations strive to adhere to regulatory requirements for CODefs, managing the complex representations of CODefs can quickly become challenging when moving from the discrete concept to related groups for each code system, to unique variations used in real-world research.
METHODS: Using a newly created CODef library currently leveraged by a few life-science companies, we profiled the subset of value sets used to support their studies. We aggregated the results at three levels: 1) counting discrete concepts from related value-set groups together; 2) counting related value-set groups for each code system that share an 80% or more code overlap; and 3) counting each unique value-set variation. For example, the discrete concept of “Cerebrovascular Disease Diagnoses” has seven related value-set groups containing 19 unique value-set variations with similar provenance, but slightly different code lists used by multiple studies.
RESULTS: As of January 2025, this new CODef library contains 912 discrete concepts defined by 2,586 related value-set groups comprising 27 different code systems (e.g. ICD-10-CM, SNOMED CT, CPT, ATC, LOINC, CVX, etc.) with 2,813 unique value-set variations overall; and growing. Details by both US and international terminologies will be presented. One interesting finding shows there are 98 discrete concepts with 227 value sets that are slight variations of the same code list. For example, the related value-set group “Cerebrovascular Disease (CCI) Diagnoses (ICD-10-CM)” contains three unique value-set variations for different studies (one with 100% overlap with the original literature review list and two with 96% overlap after dropping 23 of 528 codes).
CONCLUSIONS: As organizations strive to adhere to regulatory requirements for CODefs, managing the complex representations of CODefs can quickly become challenging when moving from the discrete concept to related groups for each code system, to unique variations used in real-world research.
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
RWD90
Topic
Real World Data & Information Systems
Topic Subcategory
Data Protection, Integrity, & Quality Assurance, Reproducibility & Replicability
Disease
No Additional Disease & Conditions/Specialized Treatment Areas