Illustration of Continuous Enrollment and Beneficiary Categorization in DoD and VA Infrastructure for Clinical Intelligence

Abstract

ABSTRACT Introduction The DoD and VA Infrastructure for Clinical Intelligence (DaVINCI) data-sharing initiative has bridged the gap between DoD and VA data. DaVINCI utilizes the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) to map DoD and VA-specific health care codes to a standardized terminology. Although OMOP CDM provides a standardized longitudinal view of health care concepts, it fails in capturing multiple and changing relationships beneficiaries have with DoD and VA as it has a static (vs. yearly) person characteristic table. Furthermore, DoD and VA utilize different policies and terminology to identify their respective beneficiaries, which makes it difficult to track patients longitudinally. The primary purpose of this report is to provide a methodology for categorizing beneficiaries and creating continuous longitudinal patient records to maximize the use of the joint DoD and VA data in DaVINCI. Materials and Methods For calendar year 2000-2020, we combined DoD and VA OMOP CDM and source databases to uniquely categorize beneficiaries into the following hierarchical groups: Active Duty, Guard, and Reserve Service Members (ADSMs); Separatees; Retirees; Veterans; and Deceased. Once the cohorts were identified, we examined calendar year 2020 health care utilization data using the OMOP VISIT_OCCURRENCE, DRUG_EXPOSURE, MEASUREMENT, and PROCEDURE tables. We also used the Defense Enrollment and Eligibility Reporting System source table to derive enrollment periods for DoD beneficiaries. As VA does not have enrollment plans, we utilized the VA’s priority groups (1-5) in the SPATIENT source table to crosswalk the DoD’s enrollment concept to the VA. We then assessed lengths of continuous enrollments in DoD and VA and the impact of appending the longitudinal records together. Results The majority of the ADSMs utilized the DoD system, but about 60,557 (3%) were seen in the VA for varied types of care. The market share of care provided to ADSMs by the VA varied by specialty and location. For Retirees, the split between DoD (1,625,874 [75%]) and VA (895,992 [41%]) health care utilization was more significant. The value added for utilizing DaVINCI in longitudinal studies was the highest for researchers normally limited to DoD data only. For beneficiaries who had 5 years of continuous enrollment, DaVINCI increased the potential study population by over 202% compared to using DoD data alone and by over 14% compared to VA data alone. Among beneficiaries with 20 years of continuous enrollment, DaVINCI increased the potential study population by over 133% compared to DoD data and by nearly 39% compared to VA data. Conclusions DaVINCI has successfully combined DoD and VA data and utilized OMOP CDM to standardize health care concepts. However, to fully maximize the potential of DaVINCI’s DoD and VA OMOP databases, researchers must uniquely categorize the DaVINCI cohort and build longitudinal patient records across DoD and VA. Because of the low other health insurance rates among DoD enrollees and their choice to enroll to a DoD Primary Care Manager, we believe this population to be the least censored in the DoD. Applying a similar concept through VA’s priority groups (1-5) would enable researchers to follow ADSMs as they transition from the military.