The Department of Veterans Affairs (VA) provides health care to eligible veterans. In this study, we describe the current and projected characteristics and health care needs of the U.S. veteran population as whole, as well as the population of veterans who receive health care from VA. This analysis was conducted in response to the Veterans Access, Choice, and Accountability Act of 2014 (Section 201), which mandated “an independent assessment of current and projected demographics and unique health care needs of the patient population served by the Department.”
Study Purpose and Approach
We examined the demographic characteristics of the current and projected population of U.S. Veterans and patients of the VA health care system. In addition, we examined the unique health care needs of the patient population currently served by VA and projected the health care needs of Veterans who might become patients in the future. We use the term Veteran to describe all Veterans, whether or not they use VA health care services, and the term VA patients to describe Veterans who received at least some health care from VA in the past year.
This study addresses four overarching research questions:
- What are the demographic characteristics of the U.S. Veteran population, and how are these projected to change between 2015 and 2024?
- To what extent do Veterans, including VA patients, rely on VA for their health care?
- What are the current health care needs of the Veteran population, including both VA patients and non-VA patients, and how do these compare with the needs of the non-Veteran population? How will the needs of Veterans in general and the VA patient population specifically evolve over time given current policies?
- How might the projected number of Veterans and VA patients change because of external forces or changes in VA policies?
To address the research questions, we conducted a series of analytic activities: Using a cohort-based approach, we estimated the size and demographic composition of the Veteran population; using the projected number of Veterans as a baseline, we estimated future enrollment in the VA health care system, the future size of the VA patient population, and the share of health care services that current Veterans receive from VA; we combined several data sources to assess the unique health care needs of Veterans and VA patients compared with non-Veterans; we used a modeling approach to assess how the number of VA patients and their health conditions might evolve over time; and we conducted scenario testing to understand how VA policies and external factors might affect the size of the Veteran population and the number of VA patients.
VA provides health care services to enrolled Veterans who seek care at VA facilities, or—in some cases—through contracted care purchased from the civilian sector. Eligibility for VA health care has evolved over time, and today's eligibility rules are rooted in the Veterans Health Care Eligibility Reform Act of 1996. The law mandated health care for service-connected health conditions and for Veterans with a service-connected disability rated at 50 percent or higher. The Secretary of Veterans Affairs has legal discretion over the provision of all other care, but VA must maintain specialized treatment and rehabilitation programs for spinal injuries, blindness, amputations, mental illness, and other serious service-connected health conditions.
In general, a Veteran must have served in the U.S. military for at least 24 months and received an honorable discharge to enroll for VA health care. Some exceptions are permitted; for example, Veterans serving less than 24 months may be eligible if they were medically retired from military service due to a service-connected condition. To implement the 1996 law, VA established a priority system for determining which groups of Veterans will be authorized for care within the authorized budget. This structure places Veterans in one of eight priority groups based on their service-connected disability rating, income, and other factors. A Veteran's priority group designation affects his or her eligibility to receive care through VA, as well as his or her cost-sharing requirements (that is, whether co-payments are required and, if so, how much). Currently, enrollment is limited to recent combat Veterans, Veterans with qualifying incomes, and Veterans with service-connected or other disabilities. Based on our analysis of VA administrative data, about 9 million Veterans (42 percent of all Veterans) were enrolled in 2014. Non-enrolled Veterans include a mix of Veterans who are ineligible to enroll and Veterans who are eligible to enroll but choose not to do so.
Use of VA health care depends on a number of factors, including the total number of Veterans in the population, Veterans' eligibility to enroll for services, Veterans' enrollment decisions when eligible, and Veterans' decisions to seek VA health care services when enrolled. Because many Veterans have access to health care through other sources, such as employer insurance or Medicare, not all will choose to enroll, and those who do enroll may choose not to use VA for all of their health care needs. In addition, both VA policy and factors external to VA can affect Veterans' use of services. For example, a policy change enabling higher-income Veterans to enroll could increase demand for VA services. Similarly, a future military conflict could increase the number of Veterans in the pipeline and affect their health care needs. In our analysis, we distinguish VA enrollees from VA patients; a VA patient is an enrollee who has used VA health care in the past year.
In this assessment, we used data from VA and from other federal sources, such as the U.S. Census Bureau, to estimate the total number of Veterans and VA patients, to project the size of these populations over time, and to estimate the health care needs of these populations. Our baseline estimates and projections assumed that VA policies and other factors that might affect Veterans' demand for services are constant, with adjustments for policy changes that have already been announced (such as the President's plan to reduce the size of the U.S. military). In scenario testing, we considered how uncertain future events, such as a future conflict or a change in VA eligibility policy, might affect the size and health care needs of the Veteran and VA patient populations.
Current and Projected Demographic Trends in the Veteran Population
Today's Veterans generally enjoy favorable socioeconomic outcomes relative to their non-Veteran counterparts. Using data from the American Community Survey (ACS), we find that Veterans are less likely to be unemployed, less likely to be living below the poverty line, and more likely to have graduated from high school, on average, than non-Veterans (Figure 1). Veterans are also more likely than non-Veterans to have medical insurance; only 7 percent of female Veterans and 6 percent of male Veterans were uninsured during the 2009 to 2013 time period, according to the ACS. In contrast, 15 percent of female non-Veterans and 22 percent of male non-Veterans were uninsured during this time period. Rates of uninsurance among the Veteran population may be low in part because many Veterans have access to free insurance through VA. Insurance rates in the United States have increased since 2013, due to the implementation of the Patient Protection and Affordable Care Act (ACA) (Office of the Assistant Secretary for Planning and Evaluation, 2015). We did not have data, however, that allowed us to compare post-ACA insurance rates between Veterans and non-Veterans.
Homelessness is declining among Veterans. Homelessness remains a significant problem among Veterans. Veterans are overrepresented in the U.S. adult homeless population: In 2010, Veterans accounted for approximately 10 percent of the adult population; however, they represented a disproportionate share of the homeless adult (16 percent) and sheltered homeless adult (13 percent) populations (National Center for Veterans Analysis and Statistics, 2012). Notwithstanding this, the rate of homelessness among Veterans has declined since 2010. According to the U.S. Department of Housing and Urban Development, there were 49,933 homeless Veterans in 2014, representing less than 0.25 percent of the total Veteran population. Between 2010 and 2014, the number of homeless Veterans declined by 33 percent (U.S. Department of Housing and Urban Development, 2014).
VA patients tend to be older and less socioeconomically well off than Veterans who do not rely on VA for care. Using data from the Medical Expenditure Panel Survey (MEPS), we are able to compare Veterans who use VA care with Veterans who do not use VA care (Table 1). VA patients are older and less well-off from a socioeconomic standpoint than Veterans who do not use VA for care. For example, 9 percent of VA patients have less than a high school education, compared with 6 percent of Veterans. VA patients' average household incomes are more than 20 percent lower than incomes for non-patient Veterans. VA patients are also far less likely to be employed than non-VA patients. Partly, these differences are by design, because higher-income Veterans may not be eligible for VA services.
Table 1. Socioeconomic Characteristics of Veterans, by VA Patient Status, 2006–2012
|Characteristic||Veterans, VA Patients||Veterans, Non-VA Patients|
|Over age 65||52.2%||38.7%|
|Less than high school education||9.1%||5.8%|
|Average household income||$35,981||$45,278|
SOURCE: RAND analysis of MEPS, 2006–2012.
NOTES: Veterans, VA patients and Veterans, non-VA patients are mutually exclusive categories. Sample size, VA patients = 4,871, and sample size, non-VA patients = 7,442.
*Non-employed individuals include both people who are unemployed and people who are out of the labor force, such as retirees.
We project that the population of U.S. Veterans will decrease by 19 percent over the next 10 years. The U.S. Veteran population has been decreasing for the past three decades, and this trend will continue. There were 27.5 million Veterans in the United States as of the 1990 Census; we estimate that there were 21.6 million Veterans in 2014. Over the next 10 years, our projections, drawing on VA, U.S. Census, and U.S. Department of Defense (DoD) data, show that the Veteran population will decline to 17.5 million, a decrease of 19 percent relative to 2014 levels (Figure 2). Given the strong preexisting trends and the President's ongoing drawdown in the size of the active duty military population (Hagel, 2014; Parrish, 2011; Office of the Under Secretary of Defense [Comptroller], 2015), the reduction in the size of the Veteran population is inevitable, absent a major policy change to increase the size of the military (for example, if an unanticipated large-scale conflict were to materialize).
Geographic distribution of Veterans will shift slightly. We estimate that, geographically, the Veteran population will become more concentrated in urban areas, and the relative proportion of the Veteran population in the Ohio River Valley region will diminish.
There will be modest changes in the demographic mix, by sex and race/ethnicity. Currently, Veterans are more likely than non-Veterans to be male, and are on average much older. We estimate that approximately 92 percent of the Veteran population was male in 2014. We also estimate that 75 percent of Veterans were age 55 or older, compared with only 34 percent of the non-Veteran population. By 2024, this will shift somewhat: The proportion of female Veterans will increase 3 percentage points, from 8 to 11 percent, by 2024, and the share of non-Hispanic white males will decrease from 80 to 74 percent over the same period. Mean age will increase slightly; the population will have a higher proportion of both older and younger Veterans.
These projections are based on historic separation rates, the anticipated decrease in military end-strength over the next several years, and an assumption that there are no significant new conflicts during the projection period.
Enrollment and Reliance on VA Health Care
The number of Veterans receiving VA health care is projected to level off over the next 10 years. While the Veteran population is projected to decline by 19 percent over the next 10 years, the number of VA patients is projected to increase until 2019. Use of VA health care has increased across all demographic groups since 2005, and the fraction of Veterans under age 35 who are VA patients has increased threefold. The growth of VA use by Veterans may be related to outreach efforts on the part of VA, policies that have expanded the list of conditions granting presumptive eligibility for VA services, and streamlined enrollment processes. Continued increases in the rates of VA use are expected to slow the decline in the number of VA patients. Nevertheless, in years beyond 2019, VA may begin to experience slight declines in the volume of patients. Because VA will be coming off a period of more than a decade of expanded use, careful monitoring and new policies may be necessary to address the leveling-off and possible reduction in demand for services that could occur after 2019.
Health care planning for VA is complicated by the fact that most Veterans have more than one possible source of health coverage. The extent to which Veterans use VA care as opposed to care from other sources is captured in the concept of reliance, by which we mean the fraction of Veterans' total care that is provided by or paid for by VA. Reliance on VA versus other sources of care varies by type of care, but it averages below 50 percent for many routine services. Across all types of care, Veterans under age 30 are the most reliant on VA, and those over age 65 are least reliant.
Both VA policy, such as policies to enhance Veterans' access to VA services, and external trends, such as the cost and availability of private health insurance, can affect Veterans' reliance on VA. However, VA has limited visibility into patients' reliance. While VA has access to data on care obtained at VA facilities, it is difficult to track how much care Veterans consume outside the VA system—for example, through private health insurance. Yet understanding reliance is critical for planning, because shifts in reliance can affect the total amount of care that Veterans obtain from VA facilities.
We analyzed reliance using data from MEPS and compared these estimates with reliance estimates used in VA's Enrollee Health Care Projection Model (EHCPM). MEPS is a survey of health care utilization and spending conducted by the Agency for Healthcare Research and Quality (AHRQ). The EHCPM is a forecasting model sponsored by VA, which relies on VA survey data, Medicare claims data, and proprietary data from the actuarial firm Milliman.
Using MEPS data, we found that younger Veterans, lower-income Veterans, Veterans in rural areas, Veterans without other access to health insurance coverage, and Veterans with poorer self-reported health status rely more than other Veterans on VA. However, the estimated share of care obtained through VA is generally lower in the MEPS estimates than in the EHCPM estimates. For example, MEPS indicates that VA patients obtain 30 percent of their prescription drugs through VA, compared with ECHPM's estimate that enrollees obtain 66 percent of their prescriptions from VA. Because the EHCPM estimates are in part based on proprietary methods, we were unable to ascertain fully the reasons for these differences. However, a general conclusion is that VA might benefit from validating current reliance estimates and investing in survey approaches to better understand Veterans' total health care needs.
Unique Health Care Needs of Veterans and VA Patients
To identify the unique health care needs of Veterans and the VA patient population, we first compared the prevalence of key health conditions among the current Veteran population with those among the non-Veteran population. We then compared the prevalence of key health conditions among VA patients with those among Veterans who do not use VA health care and analyzed which characteristics (including the presence of particular health conditions) were associated with receiving care at VA facilities. Our analysis relied on MEPS, which collects information on all care received, regardless of payer, and information on Veteran status and use of VA services. With MEPS data, we can analyze all of a Veteran's diagnosed health conditions, regardless of whether the Veteran used VA health services. We can also use MEPS data to compare Veterans with non-Veterans
We examined both unadjusted prevalence rates of these health conditions and adjusted prevalence rates, which accounted for key demographic characteristics, such as age and sex. Both rates provide unique information with relevance to policy issues.
Unadjusted prevalence rates provide a snapshot of the overall Veteran population and enable us to compare how Veterans and VA patients may differ from civilians in terms of their health care needs. Unadjusted rates, however, do not account for the fact that Veterans are typically older and more likely to be male than civilians. Nevertheless, these numbers are useful for planning purposes. For example, the fact that Veterans have a much higher rate of diabetes than non-Veterans is useful for determining the types of providers and services that Veterans need, even if most of the difference between Veterans and non-Veterans can be explained by factors such as age and sex.
Adjusted prevalence rates help us understand how Veterans' and VA patients' health care needs may differ from the needs of demographically similar non-Veterans. As a result, these comparisons inform our understanding of how the experience of being a Veteran affects health. However, because they already account for demographic differences, without careful interpretation, the adjusted prevalence rates may appear to understate key differences in health care needs between Veterans and non-Veterans at the population level.
We also projected the prevalence of the health conditions of Veterans and VA patients forward over the next 10 years, accounting for predicted changes in their demographic composition and their service experiences.
Veterans have a higher unadjusted prevalence of diagnosed health conditions than non-Veterans. The diagnosed prevalence of many common chronic health conditions, unadjusted for differences in demographic characteristics, is higher among Veterans than non-Veterans. For example, the prevalence of diabetes and gastroesophageal reflux disease (GERD) disorders among Veterans is substantially higher than for non-Veterans (Figure 3). Veterans are more likely than non-Veterans to be diagnosed with cancer, hearing loss, and PTSD. Mental health conditions, generally, are equally prevalent in the Veteran and non-Veteran populations. Because Veterans are more likely to have insurance than non-Veterans, some of these differences could reflect that Veterans are more likely to receive diagnoses than non-Veterans. Nevertheless, understanding differences in diagnosed conditions sheds light on differences in conditions that Veterans and non-Veterans are being treated for under existing policies. As such, these analyses inform our understanding of whether Veteran providers are likely to treat a different mix of conditions than civilian providers. Our analyses suggest that VA providers are likely to be treating a sicker population with more chronic conditions, such as cancer, diabetes, and chronic obstructive pulmonary disease (COPD), than the population expected by civilian providers.
Veterans also have a higher adjusted prevalence of key health conditions than non-Veterans. For some conditions, adjusting for demographic characteristics substantially reduces the difference in prevalence rates between Veterans and non-Veterans. For example, Veterans are twice as likely to have diabetes as non-Veterans in the unadjusted model, but after adjusting for demographic characteristics, the prevalence rate among Veterans is only 13 percent higher. In the adjusted model, Veterans are more likely to have mental health conditions than non-Veterans, while differences were not statistically significant in the unadjusted model. Even though fewer than 5 percent of Veterans are diagnosed with PTSD, it is even rarer in the non-Veteran population. After controlling for age and other factors (Figure 4), Veterans are 13.5 times more likely than non-Veterans to be diagnosed with PTSD.
VA patients are typically less healthy than Veterans who do not use VA health care. Compared with Veterans seen by private health care providers, Veterans who received treatment from VA had higher rates of cancer, diabetes, hypertension, PTSD, ischemic heart disease (IHD), and other conditions (Figure 5). These differences reflect VA patients' older age, and also reflect the eligibility criteria for enrolling in VA care, which depend in part on health status. Among VA patients, the unadjusted prevalence of common chronic conditions (e.g., diabetes, cancer) is 51 to 96 percent higher for VA patients relative to Veterans who do not use VA care; however, rates of PTSD are several orders of magnitude higher among VA patients relative to non-patients. Adjusting for demographic characteristics slightly reduces differences in prevalence rates between patients and non-patients.
According to the MEPS data in Figure 5, about 25 percent of all patients who accessed care at VA had a mental health condition, and 3.3 percent had PTSD. Rates of PTSD are substantially higher among Veterans under age 35. When combined with the otherwise rare conditions related to combat—amputation, traumatic brain injury (TBI), blindness, and severe burns—and the vulnerable circumstances of some patients, VA handles a patient mix that differs from what community providers typically see.
The prevalence of many common conditions is projected to increase among Veterans over the next 10 years. We estimate that aging in the Veteran population will lead to increases in the prevalence of several common health conditions among Veterans over the next 10 years. Figure 6 shows projected unadjusted prevalence for hypertension, diabetes, IHD, and mental health conditions. Among all Veterans, we estimate that the prevalence rates for diabetes and hypertension will increase by about 12 and 8 percent, respectively, between 2015 and 2024. However, while aging will tend to increase the prevalence of IHD, we estimate that prevalence rates for IHD will decline during 2015–2024. This finding is consistent with long-standing trends toward decreasing prevalence of acute coronary syndrome across all age groups in the U.S. population (Krumholz, Normand, & Wang, 2014; Talbott et al., 2013). However, this decline largely represents an extrapolation of recent declines in the prevalence of IHD noted in MEPS. The relatively large confidence bands suggest that the trend is uncertain and actual prevalence may not decline as sharply. Mental health conditions increase moderately over time, with prevalence rates rising by about 6.8 percent.
Figure 6 reports prevalence rates among all Veterans, which we estimated using a combination of data sources, including MEPS and MHS data, on service members who recently converted from active duty to Veteran status. In general, prevalence rates among VA patients increase somewhat more than prevalence rates for all Veterans. As a result, the gap in prevalence rates between VA patients and Veterans who do no use VA health care is projected to increase over time.
We examined five scenarios, based on hypothetical future changes to VA policy or to the environment surrounding VA health care.
Scenario 1: Broader VA eligibility. Higher-income Veterans without disabilities are currently ineligible to enroll for VA coverage or to receive care at VA. Expanding eligibility to currently excluded groups could lead to more than 4.8 million newly eligible Veterans and as many as 2.1 million new VA patients, amounting to a 35-percent increase in the size of VA's patient population.
Scenario 2: Including hypertension presumptively as a service-connected condition for Vietnam Veterans. According to the Institute of Medicine of the National Academies (IOM), there is increasingly solid evidence that hypertension among Vietnam-era Veterans is related to service in the Vietnam Theater of Operations. As yet, VA has not added hypertension to the list of presumptive conditions for Vietnam-era Veterans. If hypertension were included, we estimate that this would translate into 363,000 new VA patients, an increase of 6.4 percent in VA's total patient population.
Scenario 3: Hypothetical future conflict. How would demand for VA health care services be affected by future military conflict? In examining 36 possible scenarios, we found that the vast majority of them project between 500,000 and 925,000 new VA patients by 2024. However, most low-conflict scenarios anticipate 500,000 and 600,000 new patients, while most high-conflict scenarios predict between 750,000 and 925,000. This suggests that even moderate levels of deployment could substantially increase the size of the incoming cohort of VA patients. In fact, our projections suggest that, for every new patient that would have entered the VA system in more-peaceful times, approximately 1.5 new patients will enter the VA system following a major conflict. However, previous cohorts, especially the Vietnam cohort, were much larger than recent cohorts, so the difference will be small relative to the entire VA patient population.
Scenario 4: Expanding access to VA care by extending the Veterans Choice Act or by other means. Surveys have shown that 1.8 million Veterans reported not using VA care due at least in part to access barriers. We estimate that if these barriers were removed, at most an additional 235,000 Veterans per year might use VA.
Scenario 5: Effects of the ACA. Policy changes associated with the ACA could have conflicting effects on Veterans' use of VA health care. The individual mandate, which requires most individuals to obtain health insurance coverage, could increase Veterans' propensity to enroll in the VA system. However, ACA's coverage expansions, which include expanded Medicaid eligibility and subsidies to buy individual market insurance, could cause some current enrollees to use fewer VA services. Our analysis found that the net impact of ACA coverage expansions on the number of VA patients is relatively modest: We estimate 98,000 fewer VA patients under base assumptions, although other plausible assumptions result in increases in VA patient counts.
Conclusions and Recommendations
The number of Veterans has been declining for three decades, and our analysis of the Veteran population over the next 10 years suggests that this trend will continue. The total number of Veterans is expected to decrease by 19 percent between 2014 and 2024. The median age of the population will continue to increase, and Veterans are projected to become more geographically concentrated over this period.
Veterans are more likely than non-Veterans to be diagnosed with health conditions, including those that are chronic and in some cases linked to service in the military. Although some of these differences may be explained by the age and sex characteristics of Veterans compared with non-Veterans, differences remain after these characteristics are taken into account. The higher adjusted prevalence rates may be related to the challenges of serving in the military and in combat.
Among all Veterans, those who receive at least some of their health care from VA are generally more likely than Veterans who do not use VA health care to be diagnosed with many of the conditions we examined in this assessment. This result may be related to the fact that eligibility to receive VA services is based, in part, on Veterans' disability status and whether they are rated for service-connected conditions.
Assessing trends in Veterans' health care needs is complicated, because assessing the need or demand for health care requires an understanding of the extent to which Veterans rely on VA to meet those needs. During the past three decades, the number of Veterans has decreased while the number of VA patients has increased. This is due in part to an increase in the number of Veterans who have been rated for service-connected disabilities, as well as policy changes that made more Veterans eligible for VA health care benefits (due to presumptive eligibility), that made it easier for Veterans to apply for benefits, and that gave the benefit of the doubt to Veterans in cases where there was uncertainty. Looking to the future, our patient projection models suggest that the number of VA patients will continue to increase through 2019, but could level off or decline in subsequent years.
In addition to VA policies, external policies (such as the ACA) and other trends (such as the cost of civilian health care) may influence the way Veterans interact with VA's health care system. Our analysis of five potential future scenarios illustrates the extent to which policy may affect the projected number of Veterans and VA patients. Among them, expanding eligibility rules to include higher-income Veterans, entering future conflicts, and improving access to VA health care generated the largest increases in new Veterans and VA patients.
Finally, VA data systems and U.S. data collection efforts more broadly have significant limitations that hinder planners' ability to assess how demand for VA services might change over time. For example, there has not been a full accounting of the U.S. Veteran population since the 2000 Census. In addition, current VA data collection systems do not assess detailed information on Veterans' health care conditions and health care utilization patterns. Important data gaps include that data are often completely unavailable for Veterans who are not currently eligible or enrolled in VA health programs. Even among those who use VA care regularly, VA has detailed information only on care paid for or provided by VA, or paid for by the federal Medicare program. Additional data collection would be needed to fully understand Veterans' total health care needs, including use of care currently provided by the civilian sector. Understanding these gaps is important because shifts in patient reliance and changes in eligibility rules could cause more Veterans to seek care at VA and could change the mix of care sought from VA versus civilian providers.
Recommendations for Consideration
Prepare for a Changing Veteran Landscape
The number of VA patients has been increasing since 2005, despite the three-decades-long decline in the size of the Veteran population. We estimate that this increase will continue through 2019. However, in 2020 and beyond, it is likely that the size of the VA patient population will level off or even decrease. Total demand for VA services during this time period will be heavily influenced by utilization patterns among patients; if the health care needs of the population are significant or the cost of outside options is high, patients may use more care than they have in previous years. Nonetheless, there is a possibility that demand for services will level off or decline as the continued growth in the patient population slows or even reverses. The likely short-term growth in demand, followed by a leveling-off or decline in the next decade, may make it difficult to ensure that the size of the VA health system is tailored to fully meet the needs of the population in the near term without becoming inefficiently large in the long run. Increasing the use of care purchased from the civilian sector may enable VA to meet short-run increases in demand without requiring costly investment in facilities, infrastructure, and personnel that could become less needed in the future.
Improve Tracking of Some Veteran Populations
Because the 2010 Census did not capture information on Veteran status, there has not been a full-scale accounting of the U.S. Veteran population since 2000. As a result, VA must estimate the size of the Veteran population using data from more than 15 years ago, coupled with smaller surveys and information on personnel losses from DoD. While ACS provides information on a sample of Veterans (1,197,923 Veterans in the 2009–2013 sample), this is not a full accounting of the Veteran population. An updated census of the Veteran population would enable a definitive count of all Veterans, while also helping to refine sampling procedures for the yearly surveys of samples of the population. Given that the events of September 11, 2001, set off prolonged U.S. engagement in oversees conflict and changed DoD accession and personnel retention policies that affect the flow of service members from active duty to Veteran status, it seems that the nation is overdue for an updated census of the Veteran population. We recommend asking about Veteran status in the 2020 Census.
In addition, little is currently known about how the utilization patterns and health care needs of Veterans from the conflicts in Afghanistan and Iraq will evolve over time. Yet, Afghanistan and Iraq Veterans are more likely to have service-connected disabilities than other Veterans and are automatically eligible for VA health care for five years after leaving the military. Historically, Veterans have relied less on VA health care as they age, gain access to other health insurance (e.g., through an employer), and start families. However, it is not clear the extent to which these patterns will hold for newer Veterans, who have different exposures and enhanced eligibility relative to previous cohorts. Closely monitoring this population may help VA planners to prepare as these Veterans age and their health care needs and utilization patterns shift.
Anticipate Potential Shifts in the Geographic Distribution of Veterans, and Align VA Facilities and Services to Meet These Needs
While our estimates suggest that the geographic distribution of Veterans will remain relatively stable over time, there may be several opportunities to streamline or shift VA resources to ensure adequate capacity in all parts of the country. Given projected declines in the size of the Veteran population living in the Ohio River Valley and upper Midwest, it may be possible to consolidate relatively proximal VA facilities in those regions. At the same time, some areas of projected Veteran population growth—including Montana, Wyoming, and Colorado—are not currently well covered by VA facilities. While the absolute size of the Veteran population in these areas will remain small, there may be opportunities to use telehealth and community-based outpatient clinic (CBOC) services to meet Veterans' needs in these areas. There may be a more pressing need to expand VA coverage in the Southwest, where Veterans Affairs Medical Centers (VAMCs) are currently widely spaced, and where growth in the Veteran population is expected to be significant. Finally, we estimate that the Veteran population under age 35 will increase in the regions around Los Angeles; Dallas; Washington, D.C.; and northern New Jersey by 2024. VA facilities in these areas might monitor growth in utilization among younger Veterans to ensure that they are able to meet the needs of this group.
Improve Collection of Data on Veteran Health Care Utilization and Reliance
Fully understanding the needs of the patient population served by VA will require data that do not currently exist. These data would capture information on all sources of health care that are used by an individual—including when and where care is delivered, what diagnoses are recorded and procedures performed, and who pays for the services—as well as what needs for care are unmet, and why. The data would also require a large enough sample of Veterans to ensure that it is representative of the population, and to allow VA to track the prevalence of relatively rare service-connected conditions. Creating these data would enable an analysis of the extent to which Veterans currently rely on VA for health care, as well as how that reliance may change as a result of internal VA policies or external factors. It would also provide insight into where VA succeeds in meeting the health care needs of its patient population and what types of obstacles exist in delivering needed care. In addition, by collecting information on Veterans who are not currently patients, the data would enable VA to better plan for changes in the demand for services that might occur if VA eligibility rules changed, or if additional Veterans chose to enroll.
Current surveys of Veterans do not capture comprehensive information on health care use, particularly among Veterans who are not currently eligible for or enrolled in the VA system. While MEPS contains information on all the care that respondents receive regardless of payer, the survey contains only a small sample of Veterans, and this sample may not be adequately representative of the population. VA might consider fielding a comprehensive survey of all Veterans, aimed at assessing their total health care use patterns, including use of non-VA care. Such a survey could be modeled on the MEPS Household Component (MEPS-HC), which collects utilization data across all sources of care for the general population. Potentially, VA could work with AHRQ (the organization that fields MEPS-HC) to include a more robust sample of Veterans in its survey.
Incorporate Separation Patterns and Health Care Needs of Current Service Members into Projections
In this assessment, we incorporated data on current service members—who will become Veterans in the future—in several of our analyses, including (1) counts of service member separations in our demographic analysis to augment Census data of Veterans from 2000, (2) diagnosed health conditions of separating service members who received care through the Military Health System (MHS), and (3) estimated number of service members who would separate and become Veterans in the case of a hypothetical future conflict.
At present, VA does not have access to DoD MHS encounter data. Such data could enable VA planners to analyze health care needs among current active duty service personnel who may become Veterans in the future. For this study, we utilized MHS data from 2008 to 2014 to explore whether current service members (future Veterans) have different health care needs from current Veterans. We estimate that service members are much more likely than current Veterans to have a diagnosed musculoskeletal condition or asthma at the time of separation from service. On the other hand, the prevalence of mental health conditions is higher in the existing Veteran population than among separating service members. This result may reflect a disincentive to seek care for mental health conditions while serving in the military. To the extent that individuals who separate from the military and become Veterans during the 2015–2024 projection window have different health care needs from the patients currently being served by VA, adding MHS data is critical for projecting the needs that VA must meet in the future.
Develop an Analytic Framework to Perform Scenario Testing
Our analysis of five future scenarios highlights the importance of developing methods and models that can respond quickly and agilely to policy changes. While some of the policy changes we considered resulted in modest changes in number of new Veterans and new VA patients, others estimated as many as hundreds of thousands of new Veterans and patients. The VA Office of the Actuary (OACT) has a Veteran Healthcare Scenario Model that is able to estimate, for instance, how changes in demographic characteristics or economic conditions (such as employment or income) may affect demand for VA services and related costs. Expanding this model to include such events as changes in the civilian health sector, unanticipated changes in perceptions about health care quality, and groundbreaking new technologies, to name a few, will enable VA to address the types of uncertainties that current models may not address. Having methods in place to estimate the effects of these types of changes on Veteran demand for health care services will improve VA's efforts to meet the health care needs of its patient population.
Hagel, C. (2014, February 24). Secretary of Defense speech: FY15 budget preview. Washington, DC: U.S. Department of Defense. Retrieved from
Krumholz, H. M., Normand, S. T., & Wang, Y. (2014). Trends in hospitalizations and outcomes for acute cardiovascular disease and stroke, 1999–2011. Circulation, 130(12), 966–975.
National Center for Veterans Analysis and Statistics. (2012). Profile of sheltered homeless Veterans for fiscal years 2009 and 2010. Washington, DC: U.S. Department of Veterans Affairs.
Office of the Assistant Secretary for Planning and Evaluation. (2015). Health insurance coverage and the Affordable Care Act. Retrieved from
Office of the Under Secretary of Defense (Comptroller). (2015). United States Department of Defense fiscal year 2016 budget request: Overview. Washington, DC: U.S. Department of Defense.
Parrish, K. (2011). DOD directs Army, Marine drawdowns for 2015, 2016. DoD News. Retrieved from
Talbott, E. O., Rager, J. R., Brink, L. L., Benson, S. M., Bilonick, R. A., Wu, W. C., & Han, Y.-Y. (2013). Trends in acute myocardial infarction hospitalization rates for US states in the CDC tracking network. doi: 10.1371/journal.pone.0064457
U.S. Department of Housing and Urban Development. (2014). The 2014 annual homeless assessment report (AHAR) to Congress. Washington, DC.
Veterans Access, Choice, and Accountability Act of 2014, Public Law 113-146 (August 7, 2014).
 The remaining 37.2 percent of non-VA patients and 58.7 percent of VA patients who are not employed include both unemployed individuals and people who are out of the labor force because, for example, they are retired or disabled and unable to work.
 We define the unique health care needs of Veterans as those that disproportionately affect Veterans relative to non-Veterans. These include both service-connected conditions, such as posttraumatic stress disorder (PTSD), and other conditions that are more prevalent among Veterans than non-Veterans, including diabetes and cancer.
 PTSD prevalence rates for VA patients are higher in the VA administrative encounter data than in the MEPS data. The higher prevalence in the encounter data could reflect that these data are more recent than MEPS, especially because we pool six years of data (2006–2012) in our MEPS analysis. In addition, MEPS data could be biased downward if people are reluctant to report mental health conditions in surveys, while VA encounter data could be biased upward if some encounters reflect “rule-out” diagnoses.
The research described in this article was sponsored by the U.S. Department of Veterans Affairs and conducted by RAND Health, a division of the RAND Corporation.