Virginia, Maryland, and West Virginia are successfully utilizing substance use disorder (SUD) 1115 waivers to allow federal Medicaid matching to cover treatment in short-term residential treatment facilities with 16 beds or greater, which is otherwise prohibited under the Medicaid Institutions for Mental Disease exclusion (IMD). As evaluators who compared the impact of these waivers across the states, we found several insights relevant to other states who may be looking for ways to both improve the impact and the evaluations of their waiver initiatives.
Expanded access to services is more successful with state support
Maryland and Virginia both effectively expanded access to residential treatment services to Medicaid members with SUD with the use of waivers. For Virginia, where fewer Medicaid members had access to such services prior to the waiver implementation, access to treatment was significantly expanded. Despite this increase in SUD residential treatment services in Virginia, utilization of residential treatment services are considerably higher in Maryland compared to Virginia: about 9 percent of Maryland Medicaid members with substance use disorders used residential treatment in 2018, compared to 2 percent of Virginia Medicaid members with SUD. This reflects greater availability of residential treatment services prior to the waiver, as well as greater state support in Maryland for residential treatment both before and after the waiver.
Shortages of residential treatment providers were reported in both states post waiver implementation, but especially in Virginia where there was little state support for such services prior to the waiver. While overall access to addiction treatment services is greater overall among Medicaid members in Maryland, access has increased more significantly in Virginia since waiver implementation, likely due to the waiver being combined with a major reform of addiction treatment services. Rates of Medications for Opioid Use Disorder (MOUD) treatment doubled among Virginia Medicaid members with opioid use disorder (OUD) between 2016 and 2018, and are now comparable with MOUD treatment rates among Maryland Medicaid members with OUD. For both states, transitions to community-based care following discharge from residential treatment remains a problem. Lack of communication between providers, fragmentation of service delivery, and lack of patient motivation to follow through with treatment were cited by providers as some of the main reasons for lack of follow-up.
Mixed methods approach to evaluations provide a fuller picture
Analysis of Medicaid claims is the primary method for most of these evaluations. While this is essential, it is important that these evaluations use mix-methods approaches, including provider and patient surveys, as well as open-ended interviews with providers and patients to gain an in-depth understanding of the waiver impact. As it is often difficult to identify adequate comparison groups to use for more standard evaluation approaches, being able to triangulate findings from different data sources and methods may be key for the credibility of evaluation findings and results.
The comparison project for Maryland and Virginia was considered complementary to their state evaluations, deemed valuable in providing supplementary, rather than redundant information by the evaluation teams at Virginia Commonwealth University and The Hilltop Institute. The study combines semi-structured interviews with state officials, Medicaid health plans, and addiction treatment providers (including residential treatment providers), with quantitative measures of access to addiction treatment services as leveraged from the Medicaid Outcomes Distributed Research Network(MODRN) opioid use disorder project, a collaborative effort to analyze substance use disorder prevalence and utilization across 12 states using Medicaid claims data. Utilization of residential treatment providers is based on analysis of Medicaid claims data from Virginia and Maryland.
Our interviews with addiction treatment providers – including residential treatment providers – provides an important perspective on IMD waivers that has not been well covered in research. We focused these interviews in three communities – Baltimore (MD), Richmond, and Roanoke (VA), so that we could triangulate interview responses within a community to get a “system” perspective.
To gain trust and cooperation with providers, state officials, and other respondents, we assured them that they would not be identified with specific responses. Instead, we combined responses from all interviewees to get a balanced perspective on the addiction treatment system for Medicaid and did not quote responses without permission.
Rigorous policy review can be leveraged for approval of use of a comparator state
State-led evaluations of Medicaid waivers experience many barriers to making findings useful to policymakers. Several of these challenges were outlined in a 2018 report from the U.S. Government Accountability Office. Among the challenges outlined, many waiver evaluations do not include adequate comparison groups that can be used to isolate the effects of the demonstration project. West Virginia (WV) faced this challenge because our 1115 evaluation waiver was rolled out to the entire Medicaid population across the state at the same time – eliminating any control group. To overcome this challenge, we decided to pursue Medicaid claims data from another state that we could use as a control group within a differences-in-differences analysis.
We leveraged supportive leadership from WV Medicaid to make contact with Medicaid administrators in other states in an effort to find a state willing to share their claims data with us for this purpose then developed a process for securely receiving these data by building on current practices established within WV Medicaid. Finally, we worked with the Centers for Medicare and Medicaid Services (CMS) to consider our proposed comparator state as a good match.
This was particularly challenging because our comparator state wished to remain anonymous – even with CMS. To overcome this challenge, we conducted a rigorous policy review to assess similarities and differences in SUD treatment policies across the two states. We also developed descriptive tables comparing pre-trend measures in WV to our comparator state, which showed that both states were experiencing similar trends in SUD related outcomes.
We are still awaiting final approval on our evaluation design from CMS, but are looking forward to receiving de-identified claims data from our comparator state and aggregating outcome measures at the zip code level.
The authors presented on a recent webinar for AcademyHealth’s Medicaid Demonstration Evaluation Learning Collaborative, a group of evaluation researchers examining the policy questions, and study designs, methodologies, data sources and metrics used in Medicaid waiver initiatives.