Discussion
Our study quantified changes in depression-related health care utilization after Medicaid expansions in NY and AZ in 2001. Overall, the balance of evidence indicated that Medicaid expansion was associated with decreased reliance on the VA for depression-related services. There was an exception: income-eligible veterans in AZ did not shift their hospital care away from the VA in a statistically discernible way, although the point estimate was lower. More broadly, these findings concerning veterans’ reliance varied not only in inpatient vs outpatient services and income- vs service-connected eligibility, but also in the state-level contexts of veteran dual users and per capita utilization.
Given that the overall per capita utilization of depression care was unchanged from pre- to postexpansion periods, one might interpret the decreases in VA reliance and increases in Medicaid-VA dual users as a substitution effect from VA care to non-VA care. This could be plausible for hospitalizations where state-level analyses showed similarly stable levels of per capita utilization. However, state-level trends in our outpatient utilization analysis, especially with a substantial 2.32 pp increase in annual per capita visits among income-eligible veterans in AZ, leave open the possibility that in some cases veterans may be complementing VA care with Medicaid-reimbursed services.
The causes underlying these differences in reliance shifts between NY and AZ are likely also influenced by the policy contexts of their respective Medicaid expansions. For example, in 1999, NY passed Kendra’s Law, which established a procedure for obtaining court orders for assisted outpatient mental health treatment for individuals deemed unlikely to survive safely in the community.26 A reasonable inference is that there was less unfulfilled outpatient mental health need in NY under the existing accessibility provisioned by Kendra’s Law. In addition, while both states extended coverage to childless adults under 100% of the Federal Poverty level (FPL), the AZ Medicaid expansion was via a voters’ initiative and extended family coverage to 200% FPL vs 150% FPL for families in NY. Given that the AZ Medicaid expansion enjoyed both broader public participation and generosity in terms of eligibility, its uptake and therefore effect size may have been larger than in NY for nonacute outpatient care.
Our findings contribute to the growing body of literature surrounding the changes in health care utilization after Medicaid expansion, specifically for a newly dual-eligible population of veterans seeking mental health services for depression. While prior research concerning Medicare dual-enrolled veterans has shown high reliance on the VA for both mental health diagnoses and services, scholars have established the association of Medicaid enrollment with decreased VA reliance.27-29 Our analysis is the first to investigate state-level effects of Medicaid expansion on VA reliance for a single mental health condition using a natural experimental framework. We focus on a population that includes a large portion of veterans who are newly Medicaid-eligible due to a sweeping policy change and use demographically matched nonexpansion states to draw comparisons in VA reliance for depression care. Our findings of Medicaid expansion–associated decreases in VA reliance for depression care complement prior literature that describe Medicaid enrollment–associated decreases in VA reliance for overall mental health care.
Implications
From a systems-level perspective, the implications of shifting services away from the VA are complex and incompletely understood. The VA lacks interoperability with the electronic health records (EHRs) used by Medicaid clinicians. Consequently, significant issues of service duplication and incomplete clinical data exist for veterans seeking treatment outside of the VA system, posing health care quality and safety concerns.30 On one hand, Medicaid access is associated with increased health care utilization attributed to filling unmet needs for Medicare dual enrollees, as well as increased prescription filling for psychiatric medications.31,32 Furthermore, the only randomized control trial of Medicaid expansion to date was associated with a 9-pp decrease in positive screening rates for depression among those who received access at around 2 years postexpansion.33 On the other hand, the VA has developed a mental health system tailored to the particular needs of veterans, and health care practitioners at the VA have significantly greater rates of military cultural competency compared to those in nonmilitary settings (70% vs 24% in the TRICARE network and 8% among those with no military or TRICARE affiliation).34 Compared to individuals seeking mental health services with private insurance plans, veterans were about twice as likely to receive appropriate treatment for schizophrenia and depression at the VA.35 These documented strengths of VA mental health care may together help explain the small absolute number of visits that were associated with shifts away from VA overall after Medicaid expansion.
Finally, it is worth considering extrinsic factors that influence utilization among newly dual-eligible veterans. For example, hospitalizations are less likely to be planned than outpatient services, translating to a greater importance of proximity to a nearby medical facility than a veteran’s preference of where to seek care. In the same vein, major VA medical centers are fewer and more distant on average than VA outpatient clinics, therefore reducing the advantage of a Medicaid-reimbursed outpatient clinic in terms of distance.36 These realities may partially explain the proportionally larger shifts away from the VA for hospitalizations compared to outpatient care for depression.