Data Resource Profile: The Veterans Affairs—Health and Retirement Study Aging Veteran Cohort

Abstract

Constituting 72% of all living United States (US) Veterans, adults 50 and older represent the majority of Veterans Health Administration (VA) users [1]. VA administrative records represent a rich source of information on health status, healthcare utilization and costs, and demographics, but do not include non-VA funded healthcare use, detailed socioeconomic information, and other social determinants of health that are increasingly the focus of aging research [2]. The Health and Retirement Study (HRS), a longitudinal survey of individuals aged 50 or older across the US, provides high-quality data on these topics through in-depth queries about lifestyle, health and health service use, economic status, and family structure [3, 4]. We describe the Veterans Affairs—Health and Retirement Study (VA-HRS) linked data, which links these two unique sources for 2360 Veterans aged 50 and older.

Investigators at the VA Information Resource Center and the University of Michigan partnered to complete the VA-funded data linkage. Living and deceased HRS respondents who self-identified as United States Veterans in any HRS wave before 2013 were eligible for inclusion. Living Veterans were mailed linkage information and asked to return a signed informed consent form; deceased Veterans were automatically enrolled. A probabilistic matching algorithm was used to identify matches in VA records (Text S1). Institutional Review Boards at the University of Michigan and the Veterans Affairs Ann Arbor Healthcare System granted approval.

For successful matches, VA healthcare record data files were linked to key HRS respondent identification variables (i.e., household identification ID and person number). Complete HRS data from all waves is available, including future waves for living Veterans still participating in the HRS survey, which occurs every 2 years (Text S2) [3, 4].

The VA linkage is a curated dataset of diagnoses and VA services recorded between 1999 and 2013 (Table S1) [2]. Enrollment history data comprises co-pay information and enrollment priority group. Diagnosis and procedure information from all inpatient and outpatient care episodes provided by or paid for by VA is captured, including determinations regarding military service-related toxic exposures (e.g., Agent Orange). Pharmacy records include drug names and VA-specific drug classes. Laboratory and radiology information are included in files within the VA Decision Support System (DSS). Additionally, cost data is calculated for inpatient and outpatient encounters using the HERC Average Cost datasets which produce encounter-level cost estimates based on hypothetical Medicare reimbursement [5]. Payments made to non-VA providers for services to VA users are housed separately in Fee Basis data files [6].

Of 7866 self-reported Veterans in the HRS, 1865 living and 1445 deceased Veterans were matched to the VA master file (Figure S1). Among the linked cohort, 96.74% are male, and 79.87% are White/Caucasian (Table 1). The majority, 70.72%, utilized VA healthcare between 1999 and 2013. Most belong to the Asset and Health Dynamics Among the Oldest Old (AHEAD) (26.91%), the Children of the Depression (CODA) (21.65%), and the HRS (31.48%) birth cohorts, born before 1924, 1924–1930, and 1931–1941, respectively. HRS data are available for a subsample of linked Veterans for every HRS wave since study inception in 1992, with N = 426 linked respondents in the most recent HRS wave (2022; Figure 1).

The VA-HRS aging Veteran cohort leverages the strengths of administrative healthcare records and comprehensive longitudinal survey data to support research on aging and health. Constructing utilization and expenditure measures from VA claims data enables more accurate measurement of system-level costs, avoiding the error inherent in self-reported utilization measures. Moreover, as the VA-HRS sample includes Veterans aged 50 and older, the data provides insights into Veterans' use of non-VA services before Medicare enrollment (typically at age 65). Evidence suggests that prior military service may accelerate aging, making these additional years of data critical [7]. There are many research areas where the cohort can advance our understanding of Veteran outcomes. For example, it is a unique resource for those interested in studying the long-term effects of toxic exposures or traumatic brain injury, or understanding how policy changes impacting access to healthcare or social service affect Veterans' long-term social, occupational, and economic outcomes. Despite being limited to HRS self-reported Veterans who were deceased or consented to the linkage, the combination of administrative records with over three decades of longitudinal survey data makes the linked cohort an extensive resource on Veteran health.

A VA affiliation is not required to access this free dataset, broadening access to the larger research community. To apply for access, visit the HRS website (https://hrs.isr.umich.edu/data-products/restricted-data/vdi).

Data linkage project concept and design: K.M.L. and A.S. Acquisition of subjects and/or data: K.M.L. and A.S. Analysis and interpretation of data: L.S., M.V., J.J., and M.F.W. Preparation of manuscript L.S., M.V., J.J., A.S., K.M.L., A.L.B., and M.F.W.

The HRS (Health and Retirement Study) is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. The data linkage project was funded by the U.S. Dept of Veterans Affairs Health Services Research & Development Service (VA HSR&D), Office of Research & Development (SDR 10-180). Involvement in the preparation of this manuscript by L.S., M.V., and M.F.W. was supported under Career Development Award IK2 HX003080 to Dr. M.F.W. from the VA HSR&D. J.J. was supported by VA HSR&D Career Development Award IK2 HX002860. A.L.B. is the recipient of a VA Research Career Scientist award (IK6 CX002386). This manuscript has not been submitted or presented elsewhere. Funding agencies were not involved in the study design or preparation of this manuscript. The views expressed are those of the authors and do not necessarily reflect the position or policy of Veterans Affairs or the U.S. government.

The authors declare no conflicts of interest.