The Effect of Medicaid Reimbursement for Psychiatrists on the Health Care Burden of Serious Mental Illness

ABSTRACT

Objective: Health care resource utilization (HCRU) and cost among Medicaid patients with serious mental illness (SMI) were compared in states with high and low reimbursement for psychiatric services.

Study Design: Retrospective cohort study.

Methods: Using 2021-2023 Kythera Labs Medicaid data, the 20 most frequently billed procedures by psychiatrists were identified. We selected states with the top 20% and bottom 20% mean reimbursement. The SMI population had 1 or more medical claims for SMI during the identification period and continuous enrollment for 1 year pre- and post diagnosis. The effect of reimbursement rates on health outcomes was estimated, controlling for patient-level and fixed-state characteristics.

Results: The high-reimbursement-rate cohort (n = 99,024) was 5% less likely to have psychiatric-specific inpatient and outpatient visits and 23% less likely to have psychiatric-specific emergency department (ED) visits than the low-reimbursement cohort (n = 154,437). The high-reimbursement cohort was less likely to have inpatient and ED visits and demonstrated 24% lower overall costs and 19% lower psychiatric-specific costs annually.

Conclusions: Higher Medicaid reimbursement rates for psychiatric services were associated with reduced HCRU and costs. Although this observed association may not imply causality, it could reflect beneficial behavioral changes (eg, improved provider participation and care continuity) or unintended mechanisms such as increased utilization review that may restrict access.

Am J Manag Care. 2026;32(6):In Press

Takeaway Points

• The adjusted results provide evidence that states with high Medicaid reimbursement rates for psychiatrists had lower psychiatric-specific and all-cause utilization rates and lower health care costs, which were associated with significant savings on inpatient and outpatient utilization and costs.
• These findings point to the need for additional investment in psychiatric care for patients with Medicaid to prevent costly inpatient and emergency department service utilization and to improve outcomes for patients with serious mental illness.
• Policy makers and state Medicaid administrators should assess the risk-benefit trade-offs of investing in mental health services and improving reimbursement for psychiatrists when setting Medicaid policies at the state level.

Serious mental illness (SMI) comprises conditions such as major depressive disorder, schizophrenia, bipolar disorder, obsessive-compulsive disorder, panic disorder, and borderline personality disorder.¹ One in 25 US adults experiences SMI, and Medicaid is the single largest payer for mental health services, covering approximately 26% of adults with SMI.^2,3

Medicaid reimbursement rates are determined by states through historical cost analysis, actuarial assessments, stakeholder input, and federal oversight, allowing for tailored approaches that address local health care needs and priorities.⁴ Medicaid has historically reimbursed health care professionals at a lower rate than any other insurer in the US health care market.⁵ For example, Medicaid reimburses 30% less for physician services, 22% less for hospital payments, and 19% less for psychiatrists than Medicare for the same services.^6-8 Furthermore, reimbursement rates vary substantially across states.⁷ Payments for mental health services in the lowest- and highest-paying states differ by a factor of more than 5,⁷ with psychiatrists receiving 13% to 20% less in payments for in-network services than nonpsychiatric physicians.⁹

Despite Medicaid’s critical role in mental health coverage, the relative lack of Medicaid enrollment among psychiatric providers is a key barrier to care for individuals with SMI.¹⁰ Low reimbursement rates may discourage provider participation, further limiting access to services. Studies have shown that fewer psychiatrists accept Medicaid,^10,11 restricting provider networks and contributing to delayed or unmet treatment needs among enrollees with SMI. This provider participation gap affects health care access and outcomes for this vulnerable population.

States in which Medicaid pays less may drive psychiatrists to sustain their financial margins through alternative methods such as cost shifting to private payers or patients paying in cash.^12,13 Moreover, higher reimbursement rates have been shown to improve the provision of care and patient outcomes, including better health status, fewer emergency department (ED) visits, and wider appointment availability.¹⁴

Medicaid reimbursement differentials have been examined for primary and obstetric care,⁶ opioid use disorder treatment,¹⁵ and mental health services.⁷ However, to our knowledge, the effect of variation in Medicaid reimbursement rates for psychiatrists on health care utilization and costs for patients with SMI has not been studied. This analysis compared health care costs and utilization of Medicaid patients with SMI between states with low and high reimbursement rates for psychiatric services.

METHODS

Data Source

We employed a retrospective cohort design to analyze a Kythera Labs Medicaid data population from January 2021 to December 2023, with 44,470,509 patients and 901,136,733 claims. The Medicaid enrollment proportion from Kythera data is provided in eAppendix Figure 1 (eAppendix available at ajmc.com) and includes both fee-for-service (FFS) and managed Medicaid data. The data include the unique deidentified number of patients; age; sex; insurance type (FFS vs managed care); zip code; International Statistical Classification of Diseases, Tenth Revision diagnosis codes; Current Procedural Terminology codes; and National Drug Codes. Because patient encounter data are linked via a unique identifier, longitudinal analysis was possible. The data details have been published elsewhere, and the health care outcomes derived from these data were compared with other data sets for their validity and consistency.^16-18

Construction of Index

The 20 most frequently billed procedures in 2022 by Medicaid-participating psychiatrists in every state (eAppendix Table 1) were identified using evaluation and management codes because they may contribute to variation in psychiatrists’ overall compensation.⁷ The most recent data available (2022) from public Medicaid physician fee schedules for the most frequently billed mental health services were utilized to identify cost of services (eAppendix Table 2). Two state cohorts were assessed: states with high (top 20%) and low (bottom 20%) reimbursement rates.

The SMI population in the selected states was defined as those with a diagnosis of schizophrenia and related spectrum disorders (including schizoaffective and schizophreniform disorders), bipolar disorder, mania and its related disorders, major depressive disorder (recurrent), and specific personality disorders. Patients were required to be 18 years or older, have had at least 1 medical claim for SMI during the identification period (1/1/2022-12/31/2022), and continuous enrollment for 1 year pre– and post index date.

Analysis

A descriptive analysis was conducted to compare patients within each cohort by age, sex, and comorbidities. Elixhauser Comorbidity Index (ECI)¹⁹ scores and flags for any mental health comorbidities and systematic comorbidities prior to SMI diagnosis were created to proxy differences in patient severity among the cohorts. Numbers and percentages were provided for dichotomous and polychotomous variables. Means and SDs were provided for continuous variables. P values were calculated according to the c² test for dichotomous variables and according to t tests for continuous variables. Standardized differences were also calculated for each variable.

Estimates were made using the following model with risk adjustment:

log(Total Cost or PSY Cost)_i,s = β₀ + β₁HIGH_s + β₂P_i,s + β₃X_s + u_i,s

where log(Total Cost or PSY Cost)_i,s was the all-cause health care costs or psychiatric-specific costs for individual i in state s. High is 1 if the state had a high reimbursement rate and 0 otherwise. P_i,s is a vector of individual characteristics such as age, sex, and comorbidities. X_s is a vector of state-level characteristics including a binary indicator for medical marijuana law,²⁰ a binary indicator for a Medicaid Health Home for mental illness and/or substance use disorder (SUD),²¹ a binary indicator for an Affordable Care Act (ACA)–related Medicaid expansion,²² unemployment rate,²³ the number of behavioral health care providers per 1000 state residents from US Census Bureau County Business Patterns,^7,24,25 and the level of Substance Abuse and Mental Health Services Administration (SAMHSA) block grant funding for mental illness and SUD prevention and treatment per capita.⁷ The β₁ coefficient estimate shows the percentage change of all-cause health care costs or psychiatric-specific costs due to high reimbursement relative to low reimbursement. SEs were clustered by state. Logistic models were used to estimate the likelihood of all-cause health care or psychiatric-specific inpatient, outpatient, ED, and pharmacy visits. We used the standard α level of 0.05 to denote statistical significance.⁷ To strengthen causal inference, several sensitivity analyses such as propensity score matching,²⁶ falsification testing,²⁷ and E-values²⁸ were conducted. All statistical analysis was performed with R 4.4.1 (R Foundation for Statistical Computing).

RESULTS

Unadjusted Descriptive Analysis

In total, 154,437 patients with SMI were identified in the low-reimbursement-rate cohort, and 99,024 patients with SMI were identified in the high-reimbursement-rate cohort (eAppendix Table 3). In low-reimbursement states, patients with SMI were older and more likely to be female. The ECI score was also significantly higher in low-reimbursement states. Consistent with higher comorbidity index scores, our results showed that systemic and mental health comorbidities were higher in low-reimbursement states (Table 1).

In terms of psychiatric-specific utilization, the proportion of patients with at least 1 inpatient admission (29.4% vs 24.5%; P < .0001) or outpatient visit (74.4% vs 71.3%; P < .0001) was higher in high-reimbursement states, but the proportion of patients with at least 1 ED visit was higher (1.6% vs 1.2%; P < .0001) in low-reimbursement states (Figure). Unadjusted psychiatric-related cost per patient per year (PPPY) followed a similar trend, with higher inpatient and outpatient costs in the high-reimbursement cohort and higher ED costs in the low-reimbursement cohort.

All-cause health care utilization was higher in low-reimbursement states for inpatient, outpatient, and ED visits (all P < .0001), with the largest difference in all-cause ED visits (Figure). Although corresponding all-cause PPPY outpatient ($13,169 vs $10,970; P < .0001) and ED ($2231 vs $2028; P < .0001) costs were higher in the low-reimbursement cohort, inpatient costs were slightly lower ($3233 vs $3342; P = .04).

Adjusted Multivariable Analysis

Adjusted for patient and state-level factors, the high-reimbursement cohort was 5% less likely to have psychiatric-specific inpatient visits, 4% less likely to have psychiatric-specific outpatient visits, and 23% less likely to have psychiatric-specific ED visits than the low-reimbursement cohort. The odds of having a psychiatric-specific pharmacy visit did not differ between the cohorts. The high-reimbursement cohort was also 6% less likely to have an all-cause inpatient visit and 6% less likely to have an all-cause ED visit than the low-reimbursement cohort. The likelihood of having an all-cause outpatient or pharmacy visit did not differ between cohorts (Table 2).

After risk adjustment, total annual psychiatric-specific costs were 19% lower in high-reimbursement states. Similar to total costs, annual psychiatric-specific inpatient costs were 14% lower and annual psychiatric-specific outpatient costs were 11% lower in high-reimbursement states. There were no significant differences in annual total and psychiatric-specific ED and pharmacy costs between the low-reimbursement and high-reimbursement states. Similar trends were seen for all-cause costs. High-reimbursement states had 24% lower total costs, 12% lower inpatient costs, and 29% lower outpatient costs annually than low-reimbursement states (Table 3).

Effect of Patient Characteristics

Several patient characteristics were associated with differences in health care utilization. Older age (≥ 65 years) was associated with greater all-cause and psychiatric inpatient and ED visits. Being male was associated with increased odds of an all-cause inpatient visit, odds of a psychiatric-specific inpatient visit, and the number of psychiatric inpatient stays but with decreased odds of a psychiatric ED visit. ECI scores of 2 or greater were associated with increased odds of a psychiatric-specific or all-cause inpatient visit and an increased number of inpatient visits. Comorbid conditions were also associated with increased odds of psychiatric-specific or all-cause ED visits and more ED visits. Unemployment was associated with increased odds of a psychiatric-specific inpatient visit and the number of psychiatric-specific inpatient visits (eAppendix Tables 4-5).

Sex and comorbidity were also associated with differences in cost. Being male was associated with 20% higher psychiatric-specific total costs (P = .0007) but 11% lower all-cause total costs compared with female patients (P = .0007) (eAppendix Table 6). Greater comorbidity (ECI score ≥ 2) was also associated with 22% higher psychiatric-specific total costs (P < .0001) and 46% higher all-cause total costs than lower ECI scores (P < .0001) (eAppendix Tables 4-5).

Effect of State-Level Policies

Regarding state-level policies, the availability of a Medicaid Health Home for mental illness/SUD was associated with decreased odds of all-cause and psychiatric-specific inpatient visits. Also, the odds of having a psychiatric-specific ED visit were reduced if a state had a medical marijuana law and higher SAMHSA block grant funding for mental illness/SUD per capita compared with other states. Increased SAMHSA block grants per capita were also associated with a lower number of all-cause ED visits but not with the odds of having one. More behavioral health providers were associated with a minor (nearly 1%) decrease in the odds of having all-cause and psychiatric-specific inpatient visits and all-cause ED visits (eAppendix Table 6).

Availability of a Medicaid Health Home for mental illness and/or an SUD was associated with a 22% reduction in psychiatric-specific total cost (P = .0256) and a 38% reduction in all-cause total cost (P = .0345). Similarly, ACA-related Medicaid expansion was associated with a 38% reduction in psychiatric-specific total costs, whereas the higher per capita SAMHSA block grant funding (estimate: 0.0733; P = .0453) and the presence of medical marijuana laws (estimate: 0.3413; P < .0001) were associated with increased psychiatric-specific total costs. More behavioral health care providers were associated with reduced all-cause costs, but the effect was minimal (estimate: –0.0029; P = .0011) (eAppendix Table 6).

Sensitivity Analysis

To reduce selection bias by creating comparable groups, we used propensity score matching to balance observed covariates (eg, state characteristics, demographics) between high- and low-reimbursed states, mimicking randomization in observational data (eAppendix Table 7). Similar associations between reimbursement and SMI utilization and costs suggest that this effect was not an artifact of methodical choices.²⁹

Additionally, we conducted falsification tests to assess potential unobserved confounding by examining nonpsychiatric utilization (dermatology and ophthalmology services) as negative controls during the study period. eAppendix Table 8 demonstrates the lack of significant differences in the likelihood of these utilizations between high- and low-reimbursement states, which weakens the argument that unobserved state-level factors (eg, general practice patterns, infrastructure) drive the results.²⁷ We calculated E-values for sensitivity to unmeasured confounding. E-values quantify the minimum strength of association an unmeasured confounder would need with high vs low reimbursement and SMI utilization to explain the observed effect. A relatively high E-value suggests the observed association is relatively insensitive to potential unmeasured confounding and would require the strongest unmeasured confounding to nullify its observed association (eAppendix Table 9).²⁸

DISCUSSION

Our adjusted analysis showed that patients in states with higher Medicaid reimbursement rates experienced lower psychiatric-specific and all-cause health care utilization, along with reduced costs. Although these findings are consistent with other research associating increased Medicaid reimbursement with improved outcomes and increased access, the interpretation is complicated by substantial heterogeneity in state Medicaid delivery systems, prevailing managed care arrangements, and the diverse array of provider types treating Medicaid patients with serious mental illness.⁵ For example, studies indicate that Medicaid reimburses psychiatrists at approximately 81% of the rate of other physicians for equivalent services and that acceptance rates for Medicaid among psychiatrists are much lower than for primary care physicians and have been declining over time.^7,30,31 However, many Medicaid patients receive behavioral health care in community mental health centers and from a range of licensed providers other than psychiatrists, including therapists, social workers, and nurse practitioners. Therefore, associations between psychiatrist reimbursement rates and health care use are likely confounded by the diversity of providers and care settings and should be interpreted with caution within heterogeneous Medicaid systems.

In our study, the 20 most frequently billed psychiatric procedures accounted for 27.3% of total costs, indicating that raising reimbursement for these services, particularly in low-reimbursement states, may have a limited direct impact on overall Medicaid spending (eAppendix Table 1). For example, a 20% increase in reimbursement for these services would affect only approximately 5.5% of total costs, assuming linear scaling. However, our findings suggest that higher reimbursement rates may yield significant indirect systemwide benefits because they were associated with a 24% reduction in total health care costs for patients with SMI, primarily through reduced ED and inpatient use.

Our results indicated that the high-reimbursement cohort was less likely to have psychiatric outpatient visits, which seems counterintuitive given the expectation that higher reimbursement should increase provider access and utilization of lower-acuity care. However, this unexpected finding warrants careful interpretation. Given that individuals with SMI typically require ongoing outpatient care, it is plausible that unmeasured factors, such as variations in managed care arrangements (eg, behavioral health carve-ins/carve-outs), regional delivery models, or unobserved care coordination initiatives, may have contributed to these findings. For example, higher reimbursement rates may have incentivized providers to prioritize patients with more severe/complex needs, thereby reducing the number of routine outpatient visits but potentially improving the quality of care for those with higher acuity. Higher reimbursement may enable the adoption of more efficient care delivery models, such as evidence-based practices or measurement-based care, allowing providers to achieve similar or better outcomes with fewer visits. Additionally, states offering higher reimbursement may simultaneously impose stricter prior authorization (PA) requirements or other administrative barriers to control costs, which could inadvertently limit outpatient visit frequency despite improved provider payment. Although our models adjusted for available confounders, our analysis did not account for variations in PA policies, provider mix, or alternative service delivery models (telehealth, group therapy), all of which could influence utilization patterns. Therefore, these results should be interpreted as identifying an important association rather than establishing causality.

Demographic factors also impacted service utilization. Patients with SMI in states with low reimbursement rates for psychiatrists were older and had more comorbidities. These patients often require more complex and frequent medical attention, which is less likely to be delivered due to inadequate compensation in states with low reimbursement of psychiatrists. Being older ( ≥ 65 years), having more comorbidities, and being male were associated with higher inpatient utilization and costs. In 2020, hospitals received $0.88 for every dollar spent on Medicaid patients, leading to a $24.8 billion shortfall.³² Safety-net hospitals, which rely heavily on Medicaid payments, are particularly affected.³³ These hospitals often operate on narrow margins and face financial instability, leading to service cuts or hospital closures, further limiting access to care for vulnerable populations, including older adults with a high comorbidity burden.

Beyond patient demographic and clinical characteristics, several state policy levers were also associated with lower health care utilization and costs, including the presence of Medicaid expansion and Medicaid Health Homes for mental health and SUD, higher per capita SAMHSA block grant funding, and more behavioral health providers available per patient with SMI. These policies likely reduce costs and improve outcomes by expanding access to preventive care, enhancing care coordination, and offering alternative treatments. Medicaid Health Homes support integrated health care, SAMHSA block grants fund community-based services, and medical marijuana laws may reduce reliance on high-cost medications and acute care.^21,34,35 These state-level levers help address systemic gaps and promote more efficient, continuous care, reducing the need for costly inpatient and ED services.

Strengths and Limitations

Strengths of this study include large sample size and coverage of Medicaid enrollees without a time lag. The study also reflects a broad range of reimbursement levels among different states and provides a comprehensive ascertainment of variables. However, as an observational study, there is potential for bias and confounding.

Retrospective outcomes research studies, although valuable for generating insights from existing data, have several limitations that affect the validity and reliability of their findings. Because our patients were not randomly selected, selection bias may have occurred. We used regression analysis to control all available factors to minimize bias. To strengthen causal inferences, 3 complementary sensitivity analyses were employed alongside regression analysis.

Because the data were collected for administrative purposes rather than research, some existing records may be incomplete, inaccurate, or inconsistently recorded. The validation of Kythera Labs data is detailed in eAppendix Figure 1. A diagnosis code on a medical claim is not a positive presence of disease, as the diagnosis code may be incorrectly coded or included as rule-out criteria rather than an actual disease. Certain information, such as clinical and disease-specific parameters, is not readily available in claims data, which may influence study outcomes.

This study’s timeline overlapped with the COVID-19 pandemic, when telehealth services expanded rapidly and states implemented varying policies. These changes may have influenced mental health service use among Medicaid enrollees, and some state-level differences may reflect temporary pandemic-era effects rather than reimbursement policies alone.

Additionally, the study did not account for variation in Medicaid delivery systems (FFS, managed care, behavioral health carve-outs). Because most enrollees receive care through managed care organizations (MCOs), MCO-negotiated rates may differ from state fee schedules, and reimbursement estimates may not fully reflect varying provider payments. The effects of bundled payments or delivery system structures were not assessed, which may have influenced utilization and costs and warrants further research.

CONCLUSIONS

States with higher Medicaid reimbursement rates for psychiatrists were associated with improved health care utilization and lower overall health care costs for patients with SMI. In addition to demographic and clinical characteristics, state-related policy initiatives indicative of greater investment in mental health (eg, presence of health homes, block grants, Medicaid expansion) were contributing factors to lower psychiatric-specific utilization and costs. However, these observed relationships are associative and may not be causal, as unmeasured confounding factors and differences in care delivery structures across states could also influence these outcomes. We recommend that future research utilize more granular data on care delivery models and policy environments to better clarify these relationships because understanding these dynamics remains critical for optimizing mental health systems and outcomes for patients with SMI. Policy makers and state Medicaid administrators should carefully assess the risk-benefit trade-offs of investing in mental health services and adjusting reimbursement for psychiatrists when setting Medicaid policies at the state level.

Author Affiliations: City University of New York (OB), New York, NY; Bogazici University (OB), Istanbul, Turkiye; Otsuka Pharmaceutical Development & Commercialization, Inc (HCW, XH), Princeton, NJ; Columbia Data Analytics (NY, GS, DF, LI), New York, NY; University of Cambridge (RP), Cambridge, UK

Source of Funding: Otsuka Pharmaceutical Development & Commercialization, Inc.

Author Disclosures: Dr Waters and Dr Han are employees of Otsuka Pharmaceutical Development & Commercialization, Inc, which manufactures antipsychotic medication and funded the study. Mr Yapar, Dr Samayoa, Ms Freedman, and Ms Isenman are employees of Columbia Data Analytics, which is a paid consultant to Otsuka. Dr Patel is a consultant to Columbia Data Analytics and Otsuka. Dr Baser reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (OB, HCW, NY, XH, DF, LI, RP); acquisition of data (OB, XH); analysis and interpretation of data (OB, HCW, NY, GS, XH, DF, LI); drafting of the manuscript (OB, HCW, NY, GS, XH, DF, LI, RP); critical revision of the manuscript for important intellectual content (OB, HCW, NY, GS, XH, DF, LI, RP); statistical analysis (OB, NY); provision of patients or study materials (OB); obtaining funding (HCW, XH); administrative, technical, or logistic support (HCW, XH); and supervision (OB, HCW, XH, RP).

Address Correspondence to: Onur Baser, PhD, MS, Graduate School of Public Health, City University of New York, 55 W 125th St, New York, NY 10027. Email: onur.baser@sph.cuny.edu.

REFERENCES

1. Kessler RC, Barker PR, Colpe LJ, et al. Screening for serious mental illness in the general population. Arch Gen Psychiatry. 2003;60(2):184-189. doi:10.1001/archpsyc.60.2.184

2. Mark TL, Levit KR, Yee T, Chow CM. Spending on mental and substance use disorders projected to grow more slowly than all health spending through 2020. Health Aff (Millwood). 2014;33(8):1407-1415. doi:10.1377/hlthaff.2014.0163

3. Terlizzi EP, Schiller JS. Mental health treatment among adults aged 18-44: United States, 2019-2021. NCHS Data Brief. 2022;(444):1-8.

4. Mitchell A, Baumrucker EP, Colello KJ, Napili A, Binder C, Keyser JA. Medicaid: an overview (CRS report for Congress). Congressional Research Service. Updated April 30, 2025. Accessed July 20, 2024. https://www.legistorm.com/reports/view/crs/514169/Medicaid_An_Overview.html

5. Maclean JC, McClellan C, Pesko MF, Polsky D. Medicaid reimbursement rates for primary care services and behavioral health outcomes. Health Econ. 2023;32(4):873-909. doi:10.1002/hec.4646

6. Zuckerman S, Skopec L, Aarons J. Medicaid physician fees remained substantially below fees paid by Medicare in 2019. Health Aff (Millwood). 2021;40(2):343-348. doi:10.1377/hlthaff.2020.00611

7. Zhu JM, Renfro S, Watson K, Deshmukh A, McConnell KJ. Medicaid reimbursement for psychiatric services: comparisons across states and with Medicare. Health Aff (Millwood). 2023;42(4):556-565. doi:10.1377/hlthaff.2022.00805

8. Report to Congress on Medicaid and CHIP. Medicaid and CHIP Payment and Access Commission; June 2021. Accessed July 20, 2024. https://www.macpac.gov/wp-content/uploads/2021/06/June-2021-Report-to-Congress-on-Medicaid-and-CHIP.pdf

9. Mark TL, Olesiuk W, Ali MM, Sherman LJ, Mutter R, Teich JL. Differential reimbursement of psychiatric services by psychiatrists and other medical providers. Psychiatr Serv. 2018;69(3):281-285. doi:10.1176/appi.ps.201700271

10. Holgash K, Heberlein M. Physician acceptance of new Medicaid patients. Medicaid and CHIP Payment and Access Commission. January 24, 2019. Accessed April 22, 2025. https://www.macpac.gov/wp-content/uploads/2019/01/Physician-Acceptance-of-New-Medicaid-Patients.pdf

11. Grimm CA. A lack of behavioral health providers in Medicare and Medicaid impedes enrollees’ access to care. HHS Office of Inspector General. March 2024. OEI-02-22-00050. Accessed April 22, 2025. https://oig.hhs.gov/documents/evaluation/9844/OEI-02-22-00050.pdf

12. Domino ME, Norton EC, Morrissey JP, Thakur N. Cost shifting to jails after a change to managed mental health care. Health Serv Res. 2004;39(5):1379-1401. doi:10.1111/j.1475-6773.2004.00295.x

13. Zuvekas SH, Rupp A, Norquist G. Cost shifting under managed behavioral health care. Psychiatr Serv. 2007;58(1):100-108. doi:10.1176/ps.2007.58.1.100

14. Alexander D, Schnell M. The impacts of physician payments on patient access, use, and health. National Bureau of Economic Research working paper 26095. July 2019. Accessed April 20, 2025. https://www.nber.org/system/files/working_papers/w26095/w26095.pdf

15. Clemans-Cope L, Lynch V, Payton M, Aarons J. Medicaid professional fees for treatment of opioid use disorder varied widely across states and were substantially below fees paid by Medicare in 2021. Subst Abuse Treat Prev Policy. 2022;17(1):49. doi:10.1186/s13011-022-00478-y

16. Baser O, Samayoa G, Yapar N, Baser E, Mete F. Use of open claims vs closed claims in health outcomes research. J Health Econ Outcomes Res. 2023;10(2):44-52. doi:10.36469/001c.87538

17. Baser O, Isenman L, Baser S, Samayoa G. Impact of semaglutide on osteoarthritis risk in patients with obesity: a retrospective cohort study. Obes Sci Pract. 2024;10(3):e762. doi:10.1002/osp4.762

18. Baser O, Samayoa G, Rodchenko K, Isenman L, Baser E, Yapar N. The association between weight loss medications and cardiovascular complications. Obesity (Silver Spring). 2024;32(7):1401-1409. doi:10.1002/oby.24037

19. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8-27. doi:10.1097/00005650-199801000-00004

20. Sabia JJ, Nguyen TT. The effect of medical marijuana laws on labor market outcomes. J Law Econ. 2018;61(3):361-396. doi:10.1086/701193

21. Health Home information resource center. Medicaid.gov. Accessed November 9, 2023. https://www.medicaid.gov/resources-for-states/medicaid-state-technical-assistance/health-home-information-resource-center/index.html

22. Status of state action on the Medicaid expansion decision. KFF. Accessed November 9, 2023. https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/

23. UKCPR national welfare data, 1980-2019. University of Kentucky Center for Poverty Research. Accessed July 1, 2024. https://cpr.uky.edu/resources/national-welfare-data

24. Bondurant SR, Lindo JM, Swensen ID. Substance abuse treatment centers and local crime. J Urban Econ. 2018;104:124-133. doi:10.1016/j.jue.2018.01.007

25. Deza M, Maclean JC, Solomon K. Local access to mental healthcare and crime. J Urban Econ. 2022;129:103410. doi:10.1016/j.jue.2021.103410

26. Baser O. Too much ado about propensity score models? comparing methods of propensity score matching. Value Health. 2006;9(6):377-385. doi:10.1111/j.1524-4733.2006.00130.x

27. Keele L, Zhao Q, Kelz RR, Small D. Falsification tests for instrumental variable designs with an application to tendency to operate. Med Care. 2019;57(2):167-171. doi:10.1097/MLR.0000000000001040

28. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268-274. doi:10.7326/M16-2607

29. Baser O. Choosing propensity score matching over regression adjustment for causal inference: when, why and how it makes sense. J Med Econ. 2007;10(4):379-391. doi:10.3111/13696990701646577

30. Appelbaum PS. The ‘quiet’ crisis in mental health services. Health Aff (Millwood). 2003;22(5):110-116. doi:10.1377/hlthaff.22.5.110

31. Andrews CM, Hinds OM, Lozano-Rojas F, et al. State funding for substance use disorder treatment declined in the wake of Medicaid expansion. Health Aff (Millwood). 2023;42(7):981-990. doi:10.1377/hlthaff.2022.01568

32. Underpayment by Medicare and Medicaid fact sheet. American Hospital Association. February 2022. Accessed April 20, 2025. https://www.aha.org/system/files/media/file/2020/01/2020-Medicare-Medicaid-Underpayment-Fact-Sheet.pdf

33. Cunningham P, Rudowitz R, Young K, Garfield R, Foutz J. Understanding Medicaid hospital payments and the impact of recent policy changes. KFF. June 9, 2016. Accessed April 24, 2025. https://www.kff.org/medicaid/understanding-medicaid-hospital-payments-and-the-impact-of-recent-policy-changes/

34. Substance use and mental health block grants. Substance Abuse and Mental Health Services Administration. Updated August 16, 2024. Accessed April 22, 2025. https://www.samhsa.gov/grants/block-grants

35. Significant and sustained reductions in health insurance premiums follow medical cannabis legalization. Drexel University Medical Cannabis Research Center. October 18, 2023. Accessed April 22, 2025. https://drexel.edu/cannabis- research/research/research-highlights/2023/October/medical-cannabis-leads-to-reductions-insurance-premiums/