Publication|Articles|November 21, 2026

The American Journal of Managed Care

  • Online Early
  • Volume 32
  • Issue Early

Value-Based Pilot for Medicaid Members With Substance Use Disorder

Listen
0:00 / 0:00

Key Takeaways

  • A payer-attributed, capitated SUD model transferred full financial risk to the provider while tracking HEDIS quality, enabling longitudinal care coordination across settings beyond fee-for-service constraints.
  • Total PMPM costs fell from $2196 to $1728 in participants versus stability in matched controls, producing a –$487 difference-in-differences effect (95% CI, –$925 to –$48).
SHOW MORE

A value-based care pilot within Kentucky’s Medicaid program reduced total costs for adults with substance use disorder while improving utilization of evidence-based treatments.

ABSTRACT

Objectives: To evaluate a value-based care (VBC) pilot for substance use disorder (SUD) in Kentucky Medicaid, implemented via a payer-provider partnership between WellCare of Kentucky and Wayspring, a specialty SUD provider.

Study Design: Retrospective cohort analysis of Medicaid claims data (February 2023-July 2024).

Methods: The intervention group included 572 WellCare members with SUD who engaged in the pilot. A 1:1 matched control group (n = 572) was selected using Mahalanobis distance matching within propensity score calipers. Outcomes were assessed 6 months before and after enrollment. The primary outcome was total cost of clinical care, measured per member per month. Secondary outcomes included cost stratified by inpatient and outpatient settings and utilization of evidence-based services such as medication for addiction treatment (MAT). Impacts were estimated using a difference-in-differences framework.

Results: Per-member per-month costs declined from $2196 to $1728 in the intervention group and increased from $2062 to $2081 in the control group, yielding a difference-in-differences impact of –$487 (95% CI, –$925 to –$48). MAT use increased from 48.2% to 52.2% among intervention participants and decreased from 50.8% to 48.3% among control participants, a net difference-in-differences of 6.5 percentage points (95% CI, 0.05-12.95). Findings were robust across statistical specifications.

Conclusions: This pilot offers proof of concept for SUD-focused VBC in Medicaid, where adoption has been limited despite strong conceptual promise, and suggests that payer-provider partnerships can serve as a pathway toward broader implementation.

Am J Manag Care. 2027;33(2):In Press

doi:10.37765/ajmc.2027.90055

_____

Takeaway Points

A value-based care pilot between WellCare of Kentucky and Wayspring, a specialty substance use disorder (SUD) provider, reduced total costs for adults with SUD and increased uptake of evidence-based treatments.

  • These results demonstrate the promise of value-based care for high-need Medicaid patients in a low-income, rural state disproportionately affected by the opioid crisis.
  • Greater payment flexibility supported services beyond traditional reimbursement, including peer recovery and social needs navigation.
  • The findings underscore the potential importance of payer-provider partnerships to scale payment innovation in SUD care.

_____

Substance use disorder (SUD) remains a persistent public health threat, affecting more than 48 million individuals in the US.1 Although more than 80,000 people died from drug overdoses in 2024,2 these deaths represent only the most visible consequence. The broader burden of SUD extends across health, employment, and social domains.3 Although SUD is treatable, existing payment systems often fail to support the kind of care that works best for affected individuals. For example, services such as care coordination and recovery coaching, which are associated with improved outcomes, are often not reimbursed under traditional fee-for-service models.4,5 Recognizing this misalignment, policy makers view payment reform as not just a financing fix but also an important form of SUD treatment innovation, essential to improving access to evidence-based services across the continuum of care.6 Earlier payment reform initiatives focused on reimbursing noncovered services and tying small financial incentives to quality metrics.7,8 Although some programs such as Medicare’s Value in Opioid Use Disorder Treatment have had encouraging results,9 others, including a pay-for-performance demonstration in Maine, failed to demonstrate meaningful improvements.10

Recent reforms have embraced value-based care (VBC) in alignment with broader trends across the health sector.5,11,12 VBC builds on earlier models by introducing greater operational flexibility for providers alongside more substantial financial accountability. In practice, VBC arrangements exist along a spectrum, with the most advanced relying on capitated or global payment supported by quality requirements.13 This stands in contrast to bundled payment models that have historically structured opioid treatment program reimbursement around discrete treatment episodes, rather than incentivizing longitudinal, integrated care across settings with quality performance incentives.7 The principles of VBC are well-suited to the structure and challenges of SUD treatment.5,14 Individuals with SUD often need varied care across disparate providers and settings, making care coordination both essential and challenging.15 Moreover, recovery is often impeded by multiple unmet health-related social needs, such as housing and food insecurity, that could be addressed through wraparound services.16 Finally, VBC quality targets can help drive improvements in effective but underused treatments, notably medication for addiction treatment (MAT), the term endorsed by the American Society of Addiction Medicine to refer to pharmacotherapy for SUDs.17,18 Although the promise of VBC is clear, considerable implementation challenges stand in the way, with research indicating variable success to date.19 Limited information technology capacity makes it difficult for many practices to track and report outcomes,17 and high up-front costs often put necessary investments beyond reach.14 Similarly, without a strong referral network, it can be difficult to create and maintain clinic-to-community linkages.17

These and other resource limitations pose especially significant barriers within Medicaid, the largest payer of behavioral health services and a critical safety net for populations most affected by fragmented care and unmet social needs.21 Although some state Medicaid agencies have launched VBC initiatives for SUD services, a landscape analysis by the Substance Abuse and Mental Health Services Administration found that none have achieved statewide adoption.12 However, recent successes of purpose-built VBC providers serving Medicaid members provide hopeful clues about potential paths forward.22,23 These providers blend technology with community outreach and take on full risk in their payer contracts. For example, an evaluation of community-based care provider Waymark demonstrated reductions in inpatient admissions and emergency department visits for Medicaid members in Washington state and Virginia.23 Of particular relevance to our study is a recent evaluation of another community-based care provider, Cityblock Health. Its SUD-focused community health worker model for Washington, DC, Medicaid members was associated with reductions in total cost of care and inpatient admissions.24 Building on these successes, there is a need to test whether similar VBC approaches can be replicated in new settings with distinct demographic and delivery system contexts. Kentucky offers a compelling context given its high SUD prevalence, large rural population, and disproportionately high opioid-related mortality.25,26 Wayspring, a VBC provider operating in Kentucky, offers integrated clinical, behavioral health, and social support services to treat SUD. This study reports a pilot evaluation of Wayspring’s partnership with WellCare of Kentucky, a Medicaid payer, to implement an SUD-focused VBC model for WellCare’s members.

METHODS

Intervention

The Wayspring care delivery model seeks to lower inpatient costs and improve health outcomes by improving access to evidence-based SUD treatment and social services (Figure 1). A central emphasis is increasing uptake of MAT, the use of which remains suboptimal despite its having positive impacts on SUD-related morbidity and mortality.17 The Wayspring model integrates the provision of MAT with other behavioral health services, physical health care, community-based supports (eg, peer recovery), and social service navigation (eAppendix [eAppendix available at ajmc.com]). It also provides transition support across inpatient and outpatient settings. In addition to referring members to high-quality in-network providers, Wayspring operates its own clinics designed specifically for SUD treatment and wraparound services. Three such clinics were available to intervention group members during the pilot program, offering both in-person and telehealth options. Although members remained eligible for services beyond the pilot, the program was structured around a 6-month study period intended to support stabilization.27 All intervention group members were eligible for care coordination, social service navigation, and community-based supports, regardless of whether they received care at a Wayspring clinic.

The pilot program was financed through a capitated VBC arrangement in which WellCare of Kentucky attributed a member population to Wayspring, established a baseline spending target, and transferred full financial risk to Wayspring. Quality performance was monitored using Healthcare Effectiveness Data and Information Set measures, including initiation and engagement of SUD treatment and follow-up after emergency department visits or inpatient hospitalizations for behavioral health conditions. In contrast, traditional payment for Kentucky Medicaid SUD treatment follows a fee-for-service model, in which providers receive a fixed payment for each service delivered and bear no downside financial risk. Eligible members for the pilot were those with an SUD diagnosis within the past 22 months (eAppendix). The value-based payment structure provided flexibility in the type of services provided and created incentives to appropriately step down costly services as patients improved.

Data and Study Population

We conducted a retrospective cohort analysis using WellCare of Kentucky (Centene Corporation) Medicaid enrollment and claims data and Wayspring administrative data from February 2023 to July 2024. Aligning with similar evaluations,22,24 the intervention sample included 572 adults eligible for the pilot who enrolled between August 2023 and January 2024, had at least 1 program engagement, and had continuous WellCare coverage during both the pre- and post periods. Pilot eligibility was based on a member attribution algorithm jointly agreed upon by WellCare and Wayspring, including having a recent SUD diagnosis and being in a geography served by Wayspring (eAppendix).The pre- and post periods were defined as the 6 months before and after program enrollment, consistent with the intervention theory of change (Figure 1). A sample inclusion criteria flowchart is displayed in Figure 2.

To identify the control group, we used Mahalanobis matching within propensity score calipers.28 We chose this approach because of its greater ability to balance the joint distribution of covariates relative to other methods.29 The matching process began by estimating a binary propensity score predicting treatment likelihood. Assignment to the intervention group was measured using Wayspring’s internal administrative data system, which captured detailed information on all patient outreach, enrollment, and postenrollment engagement encounters. We then restricted potential control group members to those within 0.25 SD of the propensity score30 and within 30% of baseline total costs of care (the “calipers”). Among this restricted pool, we computed the Mahalanobis distance for all remaining potential sample pairs using a small set of clinically relevant variables (eAppendix Table 2). Nearest-neighbor matching was performed at the individual level using a 1:1 match without replacement, yielding a total study sample of 1144 individuals across intervention and control groups (572 in each). For each control group member, we assigned a pseudoenrollment date identical to the enrollment date of their matched intervention group member, which defined their pre- and post periods. Additional methodological details about the matching procedure are available in the eAppendix.

This study was conducted as an internal quality improvement analysis by Wayspring, so it was not subject to institutional review board oversight.

Statistical Analysis

We used a difference-in-differences framework to identify program impacts. This regression-based approach includes dummy variables for the post period and for intervention group membership.The key estimate is the coefficient on the associated interaction term, capturing the difference in pre-post change between the intervention and control groups.

By differencing out pre-post changes in the control group, the resulting impact estimate accounts for secular trends, including mean reversion common in evaluations of high-cost patients. The key assumption is that, absent the intervention, both groups would have followed similar trends. Although this is ultimately untestable, similarity in preperiod trends can offer some reassurance.

Our primary outcome was the total cost of clinical care, operationalized as a per-member per-month (PMPM) measure, capturing the sum of all health care spending across all services and settings. We also examined costs separately for inpatient and outpatient care. Secondary outcomes reflected evidence-based utilization,31 including engagement with outpatient care and prescription drug use. Specific measures included the percentage of individuals who used outpatient SUD services, received high-intensity outpatient care, filled at least 1 MAT prescription, filled at least 1 behavioral health prescription (non-MAT), and filled at least 1 non-behavioral health prescription. Collectively, the secondary outcomes help ensure that any potential reductions in cost were not associated with diminished quality of care.

Difference-in-differences regression models were estimated at the person level using generalized linear models, with cost outcomes modeled using a Gaussian distribution with an identity function and dichotomous outcomes modeled using a linear probability link, both for ease of interpretation. Robust SEs were estimated to account for person-level correlation in the error term. Further details are provided in the eAppendix.

We conducted several sensitivity analyses to explore robustness across alternative statistical specifications. First, we explored matching with replacement. Next, we estimated a simpler propensity score matching approach without applying caliper restrictions, as is common in similar evaluations.22,32 We also examined impacts among the highest-cost subgroup (defined as PMPM > $500; intervention group and control group: both n = 385), which the pilot was originally designed to serve, although Wayspring extended services to all eligible individuals. We also conducted stratified analyses by enrollment channel (institutional referrals vs community-based outreach), which differ meaningfully in preperiod cost levels and postperiod trends. Finally, we assessed a 12-month post period to evaluate the durability of program impacts, using a longer data window of August 2022 to January 2025.

RESULTS

Over the 6-month study period, approximately one-third of intervention participants received care from a Wayspring clinic. Among these participants, the average number of completed clinic appointments was 3.5, including a mean of 2.3 MAT visits. On average, participants received 3 social service navigation interactions and 6 peer support interactions.

Table 1 presents baseline characteristics for the 572 intervention and 572 control group members (N = 1144). Overall, the 2 groups were well balanced, with the exception of the intervention group being approximately 1 year older than the control group (41.5 years vs 40.3 years, P = .04). The sample included individuals with complex health needs, as reflected in baseline PMPM costs exceeding $2000, more than 2.5 times the Kentucky state Medicaid average of $776.34Participants had a mean of approximately 3 SUD diagnoses, and comorbid mental health conditions were highly prevalent. For instance, more than two-thirds had a diagnosis of anxiety (70.6% in the intervention group and 69.0% in the control group). Approximately half of the sample had received MAT during the preperiod (48.2% in the intervention group and 50.8% in the control group). Cost trends over the pre- and post periods are shown in Figure 3. Both groups experienced similar trends, with costs peaking prior to baseline and declining thereafter.

Table 2 displays the difference-in-differences impact estimates. Taken as a whole, they indicate that the program reduced costs and improved utilization of evidence-based care. For the primary outcome of total cost of care, PMPM spending in the intervention group dropped from an average of $2196 in the preperiod to $1728 in the post period, a reduction of $468. In contrast, the control group had a modest increase of $19 (from $2062 in the preperiod to $2081 in the post period), yielding a difference-in-differences impact of –$487 (95% CI, –$925 to –$48), a 22.2% decrease over the intervention group baseline (P = .03). When expenditure trends were decomposed by care setting, intervention group members showed steeper declines in inpatient costs and smaller increases in outpatient costs compared with the control group. These differences correspond to estimated program impacts of a 21.4% reduction in inpatient expenditures and a 23.2% reduction in outpatient expenditures, although only the outpatient result was statistically significant.

The program had positive impacts across the utilization measures, with impact estimates ranging from 7.4% (non–behavioral health, non-MAT prescription refill) to 27.7% (for high-intensity outpatient care). For example, the percentage of intervention group members receiving MAT increased from 48.2% in the preperiod to 52.2% in the post period, a gain of 4.0 percentage points. MAT use declined in the control group, from 50.8% to 48.3%, a decrease of 2.5 percentage points. The resulting difference-in-differences impact was 6.5 percentage points, representing a 13.5% increase over the intervention group baseline.

The sensitivity analyses are summarized in eAppendix Table 3. Impact estimates were consistent across statistical specifications, offering reassurance regarding robustness. When measured over a 12-month period, the estimated impact on total cost of care was smaller (–$255 PMPM) but still meaningful, though not statistically significant (P = .065). Impact estimates for evidence-based utilization remained consistent in magnitude. An additional sensitivity test addressed potential concerns about mean reversion.33 Results, presented in eAppendix Table 4, show consistent difference-in-differences estimates of approximately 20% net total cost-of-care reductions across both subgroups, including the community channel where the mean reversion concern is less applicable.

DISCUSSION

This study examined the impacts of a VBC pilot designed to reduce total costs and improve the quality of care for Medicaid enrollees in Kentucky with SUD. Over the study period, intervention group members experienced steeper declines in inpatient expenditures and smaller increases in outpatient expenditures relative to a well-matched control group, resulting in overall cost savings. Importantly, these cost savings were not associated with declines in quality. Intervention group members demonstrated greater use of evidence-based care compared with the control group. Notably, prescription fills improved in the intervention group, including both MAT and other types of medications. In contrast, the control group experienced declines in fills across all prescription categories.

Although an exhaustive exploration of longer-term treatment trajectories is beyond the scope of this study, it is notable that total costs begin to converge for the 2 groups after the initial 6-month stabilization period. This is consistent with the pilot’s design, which prioritized short-term stabilization, so the concentration of effects in this early window is perhaps unsurprising. That costs remain approximately $250 PMPM lower when measured over a 12-month period is encouraging (eAppendix), although the estimate does not reach statistical significance. The convergence nonetheless suggests that engagement strategies supporting initial crisis stabilization differ from those needed to sustain longer-term recovery. Notably, the closest comparable study, a quasi-experimental evaluation of a pilot of Cityblock’s behavioral health program, similarly reduced total costs of care24 but also saw member engagement fall over its 10-month study period, underscoring that sustaining member engagement is a key design challenge for future programs.

Limitations

Although the intervention and control groups were balanced on observable characteristics, group assignment was not randomized, a shared feature of related evaluations.22-24 Our key limitation, therefore, is the potential for residual confounding due to selection on unobservables. Because enrollment in the program was voluntary, members of the intervention group may differ from those in the control group in ways not captured in claims and administrative data, such as motivation to engage in treatment or capacity to follow through. Although no observational design can fully resolve this concern, several features of our approach offer some reassurance: Preperiod trends were similar across groups, findings were robust across statistical specifications, and our criteria for defining the intervention group align with those used in the most comparable studies in the literature.23,24

An additional limitation is our exclusive reliance on claims-based outcomes. Ideally, we would incorporate more patient-centric measures, such as the Response to Addiction Recovery and the Brief Addiction Monitor.35,36 Prior research shows that although utilization patterns often regress to the mean in high-need populations, patient-reported outcomes do not necessarily improve in parallel.37 Future work should assess whether the observed reductions in costs and increases in evidence-based outpatient care were accompanied by improvements in patients’ health and well-being. Also of note is that this evaluation was conducted internally by Wayspring. To help limit any potential bias, an independent academic evaluator was included on the study team (Leininger); the evaluator received no payment. Finally, this study draws on data from a single state, limiting its generalizability.

CONCLUSIONS

This study offers early evidence that VBC models tailored to Medicaid members with SUD can improve both cost and quality. Lessons learned from the Wayspring-WellCare experience, consistent with related literature, underscore several factors important to the success of future efforts.11,12,14,17 Foremost is a payer-provider partnership model with aligned financial and quality incentives and a shared commitment to ensuring integration across the entire continuum of addiction treatment care. A long-term mindset is also essential, as identifying the right SUD diagnostic criteria and clinical inclusions/exclusions, establishing operational workflows, and navigating compliance requirements for new contracting arrangements can take more than a year. Bidirectional data flows, including payer claims data and provider administrative data (eg, social needs navigation and resolution), support transparent performance monitoring and ongoing adjustments. Clear and open communication around actuarial assumptions is similarly important because calculations related to baseline spending trends and patient attribution are both sensitive and consequential. Regular touchpoints are also necessary to support coordinated operations and help resolve potential areas of overlap, such as case management for addiction recovery. In sum, trust and close collaboration are imperative for building successful payer-provider partnerships.

State and federal Medicaid agencies have taken meaningful steps to create an environment more conducive to novel payer partnerships in addiction treatment, including funding data infrastructure, actuarial modeling, SUD outcomes measurement, learning collaboratives, and impact evaluation.12,14,38,39 As the largest payer of behavioral health services, Medicaid is uniquely positioned to help redesign the fragmented SUD treatment landscape into a more coordinated, whole person–focused system while holding down total costs of care. This study and other recent examples of innovative payer-provider collaborations offer evidence that progress remains possible.24


Author Affiliations: Wayspring (DM, RC, AD, MF), Wilmington, DE; WellCare of Kentucky (Centene Corporation) (CP), Louisville, KY; Tuck School of Business at Dartmouth (LL), Dartmouth College, Hanover, NH.

Source of Funding: None.

Author Disclosures: Mr Morris, Ms Close, Mr Dunning, and Dr Frost report employment at Wayspring. The remaining author has no disclosures to report.

Authorship Information: Concept and design (DM, AD, MF, CP, LL); acquisition of data (DM, AD, CP); analysis and interpretation of data (DM, AD, MF, CP, LL); drafting of the manuscript (DM, RC, MF, LL); critical revision of the manuscript for important intellectual content (DM, MF, CP, LL); statistical analysis (DM, AD, LL); provision of patients or study materials (DM, CP);administrative, technical, or logistic support (DM, RC, AD, LL); and supervision (DM, RC, MF, LL).

Address Correspondence to: Lindsey Leininger, PhD, 100 Tuck Hall, Hanover, NH 03755. Email: lindsey.j.leininger@tuck.dartmouth.edu.

REFERENCES

  1. Highlights for the 2024 National Survey on Drug Use and Health. Substance Abuse and Mental Health Services Administration. Accessed September 18, 2025. https://www.samhsa.gov/data/sites/default/files/NSDUH%202024%20Annual%20Release/2024-nsduh-nnr-highlights.pdf
  2. Ahmad FB, Cisewski JA, Rosen LM, Sutton P. Provisional drug overdose death counts. National Center for Health Statistics. 2023. Updated May 15, 2024. Accessed September 18, 2025. https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data-copy2024.htm
  3. Ashworth M, Johnson D, Thompson R. Adaptable tool for modeling the benefits and costs of substance use disorder recovery programs. J Benefit Cost Anal. 2024;15(2):335-350. doi:10.1017/bca.2024.26
  4. Shoulders A, Andrews CM, Westlake MA, Abraham AJ, Grogan CM. Changes in Medicaid fee-for-service benefit design for substance use disorder treatment during the opioid crisis, 2014 to 2021. JAMA Health Forum. 2023;4(8):e232502. doi:10.1001/jamahealthforum.2023.2502
  5. Kaplan RS, Wakeman SE. Value-based health care can transform the treatment of patients with substance use disorder. Health Affairs Forefront. January 19, 2024. doi:10.1377/forefront.20240116.689820
  6. Saloner B. The long arc of substance use policy innovation in Medicaid: looking back, looking forward. Milbank Q. 2025;103(suppl 1):280-296. doi:10.1111/1468-0009.70007
  7. Medicaid and CHIP Payment and Access Commission. Recovery support services for Medicaid beneficiaries with a substance use disorder. July 2019. Accessed September 18, 2025. https://www.macpac.gov/wp-content/uploads/2019/07/Recovery-Support-Services-for-Medicaid-Beneficiaries-with-a-Substance-Use-Disorder.pdf
  8. Hodgkin D, Garnick DW, Horgan CM, Busch AB, Stewart MT, Reif S. Is it feasible to pay specialty substance use disorder treatment programs based on patient outcomes? Drug Alcohol Depend. 2020;206:107735. doi:10.1016/j.drugalcdep.2019.107735
  9. Value in Opioid Use Disorder Treatment (VIT-OUD) Demonstration Evaluation: Intermediate Report to Congress. CMS; April 2024. Accessed September 18, 2025. https://www.cms.gov/priorities/innovation/data-and-reports/2024/vit-intermediate-rtc
  10. Reif S, Stewart MT, Torres ME, Davis MT, Dana BM, Ritter GA. Effectiveness of value-based
    purchasing for substance use treatment engagement and retention. J Subst Abuse Treat. 2021;122:108217. doi:10.1016/j.jsat.2020.108217
  11. Kaye N. How states are leveraging payment to improve the delivery of SUD services. National Academy for State Health Policy. March 15, 2022. Accessed September 18, 2025. https://nashp.org/how-states-are-leveraging-payment-to-improve-the-delivery-of-sud-services/
  12. Substance Abuse and Mental Health Services Administration. Exploring Value-Based Payment for Substance Use Disorder Services in the United States. Substance Abuse and Mental Health Services Administration; 2023. Accessed September 18, 2025. https://library.samhsa.gov/sites/default/files/pep23-06-07-001.pdf
  13. APM framework. Health Care Payment Learning and Action Network. Accessed September 18, 2025. https://hcp-lan.org/apm-framework
  14. Duke-Margolis Center for Health Policy. Addressing key barriers to value-based payment for substance use disorder (SUD) treatment and recovery: current gaps and opportunities for advancing quality measurement: meeting summary. October 4, 2021. Accessed September 18, 2025. https://healthpolicy.duke.edu/sites/default/files/2021-11/VBP%20for%20SUD%20Roundtable%20Meeting%20Summary%20FINAL1123.pdf
  15. The ASAM Criteria. Vol 1. 4th ed. American Society of Addiction Medicine; 2023.
  16. Lin C, Cousins SJ, Zhu Y, et al. A scoping review of social determinants of health’s impact on substance use disorders over the life course. J Subst Use Addict Treat. 2024;166:209484. doi:10.1016/j.josat.2024.209484
  17. Schulman M, O’Brien J, Pierre-Wright M, Thomas-Henkel C. Exploring value-based payment to encourage substance use disorder treatment in primary care. Technical Assistance Collaborative; Center for Health Care Strategies. June 2018. Accessed September 18, 2025. https://www.chcs.org/media/VBP-for-SUD_Final_June-2018.pdf
  18. American Society of Addiction Medicine. Definition of Addiction. September 15, 2019. Accessed October 7, 2026.https://www.asam.org/quality-care/definition-of-addictionCarlo AD, Benson NM, Chu F, Busch AB.
  19. Association of alternative payment and delivery models with outcomes for mental health and substance use disorders: a systematic review. JAMA Netw Open. 2020;3(7):e207401. doi:10.1001/jamanetworkopen.2020.7401
  20. Banta-Green CJ, Owens MD, Williams JR, Floyd AS, Williams-Gilbert W, Kingston S. Community-based medications first for opioid use disorder — care utilization and mortality outcomes. Subst Abuse Rehabil. 2024;15:173-183. doi:10.2147/SAR.S475807
  21. Saunders H, Euhus R, Burns A, Rudowitz R. SUD treatment in Medicaid: variation by service type, demographics, states and spending. KFF. March 28, 2024. Accessed September 18, 2025. kff.org/mental-health/sud-treatment-in-medicaid-variation-by-service-type-demographics-states-and-spending/
  22. Pinnock C, Rothen J, Carlough T, Shah NR. Improving value for underserved populations with a community-based intervention: a retrospective cohort study. Arch Public Health. 2023;81(1):96. doi:10.1186/s13690-023-01117-z
  23. Baum A, Batniji R, Ratcliffe H, DeGosztonyi M, Basu S. Supporting rising-risk Medicaid patients through early intervention. NEJM Catal Innov Care Deliv. 2024;5(11). doi:10.1056/CAT.24.0060
  24. Mehta R, Dayan-Rosenman D, Sellinger D, Tang MH. Improving treatment and lowering costs for behavioral health patients through a value-based care program. NEJM Catal Innov Care Deliv. 2024;5(12). doi:10.1056/CAT.24.0082
  25. Avalere Health. The cost of addiction: opioid use disorder in the United States. May 2025. Accessed September 19, 2025. https://advisory.avalerehealth.com/wp-content/uploads/2025/05/Avalere-Health-White-Paper_The-cost-of-opioid-addiction_OUD-in-the-United-States.pdf
  26. America’s Health Rankings 2024 Annual Report. United Health Foundation; January 2025. Accessed September 19, 2025. https://www.americashealthrankings.org/publications/reports/2024-annual-report
  27. Lee JD, Nunes EV Jr, Novo P, et al. Comparative effectiveness of extended-release naltrexone versus buprenorphine-naloxone for opioid relapse prevention (X:BOT): a multicentre, open-label, randomised controlled trial. Lancet. 2018;391(10118):309-318. doi:10.1016/S0140-6736(17)32812-X
  28. Khan MTF, Lewis D, Kaelber DC, Winhusen TJ. Health outcomes associated with patterns of substance use disorders among patients with type 2 diabetes and hypertension: electronic health record findings. Prim Care Diabetes. 2023;17(1):43-47. doi:10.1016/j.pcd.2022.11.006
  29. Greifer N, Stuart EA. Matching methods for confounder adjustment: an addition to the epidemiologist’s toolbox. Epidemiol Rev. 2022;43(1):118-129. doi:10.1093/epirev/mxab003
  30. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat. 1985;39(1):33-38.
  31. Mosel S, Regan J, Fuller K. Treatment methods & evidence-based practices for addiction treatment. American Addiction Centers. Updated January 21, 2025. Accessed September 22, 2025. https://americanaddictioncenters.org/therapy-treatment/evidence-based
  32. Thapa BB, Li X, Galárraga O. Impacts of community-based care program on health care utilization and cost. Am J Manag Care. 2022;28(4):187-191. doi:10.37765/ajmc.2022.88862
  33. Daw JR, Hatfield LA. Matching and regression to the mean in difference-in-differences analysis. Health Serv Res. 2018;53(6):4138-4156. doi:10.1111/1475-6773.12993
  34. Euhus R, Chidambaram P. A look at variation in Medicaid spending per enrollee by group and across states. KFF. October 6, 2025. Accessed October 7, 2026. kff.org/medicaid/a-look-at-variation-
    in-medicaid-spending-per-enrollee-by-group-and-across-states/
  35. Okrant E, Reif S, Horgan CM. Development of an addiction recovery patient-reported outcome measure: response to addiction recovery (R2AR). Subst Abuse Treat Prev Policy. 2023;18(1):52. doi:10.1186/s13011-023-00560-z
  36. Cacciola JS, Alterman AI, DePhilippis D, et al. Development and initial evaluation of the Brief Addiction Monitor (BAM). J Subst Abuse Treat. 2013;44(3):256-263. doi:10.1016/j.jsat.2012.07.013
  37. Leininger LJ, Tomaino M, Meara E. Health-related quality of life in high-cost, high-need populations. Am J Manag Care. 2023;29(7):362-368. doi:10.37765/ajmc.2023.89396
  38. Aurrera Health Group; Coral Health Advisors. Creating a Sustainable Future for Value-Based Care: A Playbook of Voluntary Best Practices for VBC Payment Arrangements. America’s Health Insurance Plans; American Medical Association; National Association of Accountable Care Organizations; 2024. Accessed December 8, 2025. https://www.ama-assn.org/system/files/vbc-best-practices-playbook.pdf
  39. Chua PS, Lee A, Anise A. Multi-payer alignment on value-based care. National Academy of Medicine. June 8, 2022. Accessed September 24, 2026. https://nam.edu/wp-content/uploads/2022/06/Multi-Payer-Alignment-on-Value-Based-Care-discussion-proceedings.pdf

Related to this article