Approximately 50% of US men aged 65 or older have served in the military and are therefore potentially eligible to obtain health care from the Veterans Affairs (VA) health care system. However, comprehensive data to compare the characteristics and long-term outcomes of Veterans who obtain health care within and outside the VA system are limited. Additionally, research on the health and health care of older US adults will miss a significant segment of the population if VA care is not considered. To better understand which older Veterans are using VA care, and how that care affects their long-term health and social outcomes, we developed a unique data source linking longitudinal data from the Health and Retirement Study (HRS) with VA health care data.
The HRS is conducted by the Institute for Social Research at the University of Michigan in order to provide nationally representative health and social science data for public scientific use. The HRS is funded as a cooperative agreement (U01) by the National Institute on Aging, with additional funding support from the Social Security Administration. The HRS began collecting data in 1992 and has conducted biennial interviews through 2014 with a representative sample of nearly 20,000 individuals aged 51+.
The main goal of this project was to create a unique and comprehensive data source to support innovative research on crucial issues regarding Veterans' health, their utilization of VA and non-VA health care services, and the impact of illness on Veterans and their families. We did this by linking Veterans Health Administration (VHA) administrative data for 1999-2013 with the already-established linkage of comprehensive survey interview data from the HRS for 1992-2012 and Medicare claims from 1991-2012.
For this data linkage project, the HRS identified Veterans (both living and deceased) based on their self-report of military service in a prior wave of the HRS. Names and social security numbers of HRS participants who are deceased Veterans, and of living Veterans who gave informed consent, were forwarded to the VA using mutually approved methods that ensure data security and privacy. The linkage of VA data to HRS data was performed by the VA Information Resource Center (VIReC) within VHA, under the direction of Elizabeth Tarlov, PhD, RN.
We conducted probabilistic matching of VA and HRS data using Link Plus software developed by the Centers for Disease Control (CDC). This process provided a match score, or sum of weighted matched data elements, for each pair of records. This score represents how likely it is that both records refer to the same person. The following variables were used in the matching process: social security number; first, middle, and last names; gender; date of birth, and date of death. The matching process resulted in five match classes (denoted as categories 1 through 5). Class 1 represented true matches. Class 2-4 represented intermediate matches which were subsequently examined by hand, and a threshold match score was identified. Class 5 represented non-matches.
The HRS finder file included a total of 4,055 deceased self-reported Veterans. Of those, 1,445 subjects (36%) were matched to the VA master file (indicating that the subject had either health care or other benefit records within the VA). 950 subjects (23%) had records indicating utilization of VA health care services between 1999 and 2013.
The HRS mailed VA-HRS data linkage data release authorizations (i.e., consent forms) to a total of 3,811 living self-reported Veterans in July of 2013. A total of 1,875 subjects (49%) consented to the linkage. Of those, 916 subjects (49%) were matched to the VA master file, and 719 (38%) utilized VA health care between 1999 and 2013.
The final total sample of self-reported Veterans in the HRS who were linked to VA utilization data was 1,669.
We expect that the linked VA-HRS dataset will be used by researchers to provide useful and important information about Veterans' health, their utilization of VA and non-VA healthcare services, and the long-term outcomes of illness on Veterans and their families. We expect that the VA-HRS linked datasets will be especially valuable for better understanding which Veterans made use of VA services in the years both before and after the implementation of the Affordable Care Act, a key policy question that has implications regarding the demand for and cost of VA health care services.
- Langa KM, Wolinsky F, Wallace R, Zhang Q, Geers A, Kabeto M, Kuhn L, Tarlov E. The Veterans Affairs (VA)--Health and Retirement Study (HRS) Data Linkage Project: A New Data Source for Studying Older Veterans. Poster session presented at: Gerontological Society of America Annual Scientific Meeting; 2014 Nov 6; Washington, DC.