Implementing Multistate Behavioral Health Pay-for-Performance Initiatives in Medicaid Managed Care

ABSTRACT

Value-based payments are increasingly common in behavioral health care, but their impact to date has been inconsistent. Here, we describe the rollout of a multistate pay-for-performance initiative introduced by a large Medicaid managed care organization. We describe the introduction of the program, including attribution criteria and the selection of performance measures. Using administrative claims data from 7 states, we examine the characteristics of behavioral health providers and Medicaid beneficiaries who enrolled and did not enroll in this voluntary value-based arrangement. We demonstrate that participating providers were larger on average, with more clinicians and more Medicaid beneficiaries per clinician. We also demonstrate the large variation in patient characteristics, which is partly attributable to state-level differences in Medicaid eligibility. For example, participating providers treated an older population and were more likely to provide care to certain Medicaid eligibility types, including aged, blind, and disabled, which resulted in a higher-acuity population and a larger share of Medicaid beneficiaries with serious mental illness. In addition to showing that enrollment bias is a potentially important consideration when implementing pay-for-performance initiatives targeting behavioral health, we discuss other insights and areas of future inquiry, including the timing and structure of financial incentives and the use of performance measures that require collaboration between behavioral health and physical health providers.

Am J Accountable Care. 2022;10(3):16-22. https://doi.org/10.37765/ajac.2022.89232

The shift from volume to value in behavioral health care is ongoing, with varying degrees of success.^1,2 Studies have found that value-based payments (VBPs) such as bundled payments have little to no impact on key outcomes in behavioral health, including patient engagement and readmissions.³ Other studies of VBPs that directly incentivize providers to meet predetermined quality benchmarks are associated with modest improvements in some clinical and process measures, including treatment response and rates of follow-up care.^4-7

The inconsistent impact of VBPs in behavioral health care has not deterred their implementation by Medicaid managed care organizations (MCOs), which serve more than two-thirds of Medicaid beneficiaries.⁸ In 2020, nearly half of national Medicaid spending was on payments to MCOs, which are increasingly introducing value-based arrangements.⁸ In a 2021 Kaiser Family Foundation survey, more than three-fourths of responding Medicaid directors reported using at least 1 financial incentive, and half of the states using Medicaid MCOs had a specific target for the share of provider payments or the number of plan members to be covered by VBPs.⁹

Despite the proliferation of VBPs targeting behavioral health, studies rarely describe their implementation process in detail. An important element of VBP implementation is the extent of provider and patient selection, which has been demonstrated for VBPs targeting physical health.¹⁰ VBPs with voluntary participation, such as Medicare’s Bundled Payment for Care Improvement (BPCI) Initiative, may disproportionately select providers based on their readiness to provide high-quality care.¹¹ There was some evidence of this type of enrollment bias into the BPCI Initiative: Participating hospitals were larger on average, were more likely to be teaching hospitals, and had more access to relevant specialty care.¹¹ It could also be that providers opt into VBPs based on unobserved factors, such as how favorable they view their patient mix as being.¹²

In this Trends From the Field article, we describe a voluntary pay-for-performance initiative introduced by a large Medicaid MCO in 2016. Using administrative claims data, we document the variation in characteristics of behavioral health providers and their patient populations, stratifying the sample by whether they participated in the initiative. We also describe insights and areas of inquiry that may be relevant to other payers launching VBPs for behavioral health, such as the selection of performance measures and the structure and timing of financial incentives.

IMPLEMENTATION OF THE PAY-FOR-PERFORMANCE INITIATIVE

Background

The Behavioral Health Quality Improvement Program (BHQIP) was introduced by a large multistate Medicaid managed care organization in 2016 for Medicaid behavioral health providers located in Georgia, Indiana, Texas, and Virginia. It expanded to Kentucky, Louisiana, and Washington in 2017.

Inclusion Criteria

To be eligible, providers must specialize in behavioral health, which was determined using taxonomy codes (eAppendix Table [available at ajmc.com]), and have at least 50 Medicaid beneficiaries enrolled in the state’s Medicaid health plan (henceforth referred to as “members”) during the performance period, which was a calendar year. From 2016 through 2018, member attribution was determined during an 18-month period, ending on the last day of the performance period. In 2019, member attribution was determined using the 12-month calendar year. Members had to have a minimum of 2 behavioral health visits with the provider within 60 days (in 2016) or 90 days (in 2017 and 2018) during the performance period. Beginning in 2018, visits were required to be at least 15 days apart.

Performance Measurement

Three performance measures were required across all states: (1) behavioral health inpatient 30-day readmissions, (2) emergency department utilization, and (3) primary care visits. The Medicaid health plan in each state then selected from a list of variable measures for all providers in a given state: (4) 7-day follow-up visit after behavioral health inpatient discharge; (5) 30-day follow-up visit after behavioral health inpatient discharge; (6) follow-up care for children prescribed attention-deficit/hyperactivity disorder medication (initiation); (7) antidepressant initiation; (8) antidepressant continuation; (9) diabetic screening for members with schizophrenia or bipolar disorder using antipsychotics; (10) hemoglobin A1c screening for patients with diabetes; (11) initiation of alcohol and other drug dependence treatment; and (12) engagement of substance use treatment.

Points were assigned to the 3 required measures, which were worth half of the potential points. The variable measures were then assigned points as determined by state health plans. Points for both required and variable measures added up to 100. Performance improvement was assessed relative to a baseline period, which was defined as the prior calendar year. Providers received all points allocated for a given measure if they demonstrated a 10-percentage-point increase compared with their peer average rate from the baseline period. Half credit was given if they achieved a 5-percentage-point improvement over the provider’s own rate from the baseline period.

Incentive Structures

At the end of the performance period, points were tallied up across measures for a total score to determine the financial incentive for the performance period, which was multiplied by a provider’s total annual spending (all cause) for their services and a market-specific payout factor ranging from 3% to 20%. Payments were transferred to the provider 6 months after the end of the performance period. BHQIP used upside risk only, so no providers incurred losses.

Characteristics of Participating Providers and Members

Administrative claims data showed that BHQIP participants were different from nonparticipants in terms of both provider and member characteristics (Table 1). Overall, provider organizations that participated in BHQIP were larger on average, with more clinicians (defined by unique National Provider Identifiers) and more members per clinician. All differences, which were assessed using pairwise t tests, were statistically significant at the 99% confidence level.

BHQIP participants also treated an older population—BHQIP’s proportion of pediatric members was 46.6% compared with 54.5% for BHQIP nonparticipants. BHQIP participants were more likely to treat certain Medicaid eligibility types, including aged, blind, and disabled (ABD) (17.6% vs 14.2%) and family care (22.7% vs 13.0%). This resulted in a higher-acuity population with more members with serious mental illness in BHQIP, particularly schizophrenia (6.2% vs 3.5%) and substance use disorders (36.8% vs 29.0%).

Across the 7 states participating in BHQIP between 2016 and 2018, characteristics varied widely and group analyses of variance demonstrated that all differences were statistically significant at the 99% confidence level (Table 2). For example, BHQIP providers treated older populations in Texas and Louisiana. This was partly driven by differences in Medicaid eligibility, as these states also had the highest percentage of long-term services and supports and ABD eligibility.

DISCUSSION AND FUTURE DIRECTIONS

BHQIP was a voluntary pay-for-performance initiative targeting behavioral health providers who treated Medicaid beneficiaries. Initially launched in 7 states between 2016 and 2017, BHQIP extended to 7 more states after 2018 and is one of the largest VBP programs in Medicaid managed care that targets behavioral health.

The implementation of BHQIP yielded important insights and areas of inquiry that may be relevant to other payers introducing VBPs for behavioral health. We found differences in characteristics between BHQIP participants and nonparticipants, which aligns with prior research that documents enrollment bias in VBPs.^11-13 Prior research has cited policy environments (eg, the prohibition of federal Medicaid reimbursement for adults receiving behavioral health care in treatment facilities with more than 16 beds), workforce shortages, and the slow adoption of electronic health records in behavioral health care as potential contributors to enrollment bias.¹⁴ In the case of BHQIP, enrollment bias may also relate to recruitment variation, which occurs at the state level. Although BHQIP used similar data to identify eligible providers across states, future recruitment efforts could be adapted to account for this enrollment bias.

We also found differences in member characteristics across states, which likely stems from Medicaid eligibility requirements and the dynamic nature of Medicaid enrollment. This variation makes the selection of quality benchmarks in a multistate pay-for-performance initiative more difficult, as a state with a predominately pediatric population may require different measures than a state that recently expanded Medicaid to nonelderly adults. These differences may become more relevant as new states enter BHQIP. For example, Kentucky, Louisiana, and Washington, which joined in 2017, had smaller pediatric populations compared with Georgia, Indiana, Virginia, and Texas.

Current program changes in BHQIP include streamlining performance measures to those with the greatest provider participation and member volume, as well as adding an episodic total-cost-of-care measure, which includes both physical and behavioral health care costs. More research that examines the selection of performance measures could help guide decision-making and help establish an evidence base for VBP implementation, which extends to attribution criteria and the definition of quality improvement. Of note, this may require additional standardized quality benchmarks that address relevant behavioral health conditions (eg, there are no commonly used performance measures related to anxiety disorders).¹⁵

Another element of BHQIP, and pay-for-performance initiatives more broadly, that warrants further study is the optimal structure of financial incentives. Although the size of payments is dependent on participants’ total spending on members, it may not be sufficient for providers to institute meaningful changes to care delivery or make organizational changes that could better position providers to meet quality benchmarks, such as the adoption of electronic health records. There are also lags in the receipt of incentive payments—in BHQIP, providers received payments 6 months following the end of the performance year. However, shifting to more timely payouts is difficult due to delays in data availability.

A final insight and topic for future research relates to quality benchmarks that involve collaboration between behavioral and physical health providers, which is required for some performance measures. Because behavioral health care has been historically siloed in terms of payment and delivery, behavioral health providers may require additional support to successfully coordinate care.^16,17 For example, studies have shown that electronic health record adoption has lagged in behavioral health compared with physical health.¹⁸ And when electronic health records are adopted, they are often missing clinical information for psychiatric and substance use treatment, which may contribute to care quality.¹⁹

Coordinating care across behavioral and physical health providers remains a priority given the high incidence of behavioral and physical health comorbidities, which is a key driver of high levels of health care spending.²⁰ Moreover, studies have shown that more care coordination between primary and specialty providers can improve the management of chronic conditions, including depression.²¹ To better support behavioral health providers’ care coordination, BHQIP now provides a weekly report of member admissions, discharges, and transfers, which enables providers to connect with members more efficiently.

A key limitation of this Trends From the Field article is its focus on a single Medicaid managed care organization; findings may not be generalizable to other populations or settings. There were also concurrent changes in Medicaid policy, such as the Medicaid expansion to low-income nonelderly adults. No states expanded Medicaid under the Affordable Care Act while participating in the BHQIP program, so it was less likely to change the composition of providers or Medicaid beneficiaries who participated. However, it will be important to account for changes in Medicaid eligibility over time.

Another key limitation is that we did not assess the effectiveness of the BHQIP program in the present study, instead focusing on implementation and the selection of providers and their patient populations. An evaluation is ongoing, and the methodological issues described here (eg, enrollment bias) as well as unobserved factors (eg, behavior change associated with joining the BHQIP) make empirical approaches such as difference-in-differences difficult. There may be unmeasurable value of value-based arrangements that are not captured in the evaluation as well. For instance, given the low rates of provider participation in Medicaid (eg, less than half of psychiatrists accept Medicaid²²), it could be that VBPs encourage providers to treat more Medicaid beneficiaries.

More research is needed to understand the promises and pitfalls of value-based behavioral health care. There are also potential synergies with other trends in value-based care to consider, such as the increasing use of value-based contracts between Medicaid programs and MCOs.^23,24 As the shift from volume to value continues, a better understanding of VBP implementation will help payers increase the quality of behavioral health care.

Author Affiliations: Penn Center for Mental Health, Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania (MC), Philadelphia, PA; Department of Health Care Management, The Wharton School, University of Pennsylvania (MC), Philadelphia, PA; Leonard Davis Institute of Health Economics, University of Pennsylvania (MC), Philadelphia, PA; Anthem Inc (CG, JW, LE, BH), Indianapolis, IN.

Source of Funding: Anthem Inc.

Author Disclosures: Dr Candon had a services contract with Anthem Inc in 2019 and 2020, which supported consultation on the Behavioral Health Quality Improvement Program and the conceptualization of this study. Ms Girdish, Mr Walter, Ms Engel, and Dr Howell are employed by Anthem Inc, and Mr Walter and Dr Howell also report stock holdings in Anthem Inc as part of their compensation package.

Authorship Information: Concept and design (MC, CG, BH); acquisition of data (CG, BH); analysis and interpretation of data (MC, CG, JW, LE, BH); drafting of the manuscript (MC); critical revision of the manuscript for important intellectual content (CG, JW, LE, BH); statistical analysis (MC, CG); provision of study materials or patients (CG, JW, LE, BH); obtaining funding (MC, CG, JW, LE, BH); administrative, technical, or logistic support (MC, CG, JW, LE, BH); and supervision (BH).

Send Correspondence to: Molly Candon, PhD, University of Pennsylvania, 3535 Market St, Rm 3014, Philadelphia, PA 19104. Email: candon@upenn.edu.

REFERENCES

1. Stewart RE, Lareef I, Hadley TR, Mandell DS. Can we pay for performance in behavioral health care? Psychiatr Serv. 2016;68(2):109-111. doi:10.1176/appi.ps.201600475

2. Hobbs Knutson K, Wennberg D, Rajkumar R. Driving access and quality: a shift to value-based behavioral health care. Psychiatr Serv. 2021;72(8):943-950. doi:10.1176/appi.ps.202000386

3. Carlo AD, Benson NM, Chu F, Busch AB. 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

4. Garner BR, Godley SH, Bair CML. The impact of pay-for-performance on therapists’ intentions to deliver high-quality treatment. J Subst Abuse Treat. 2011;41(1):97-103. doi:10.1016/j.jsat.2011.01.012

5. Unützer J, Chan YF, Hafer E, et al. Quality improvement with pay-for-performance incentives in integrated behavioral health care. Am J Public Health. 2012;102(6):e41-e45. doi:10.2105/AJPH.2011.300555

6. Schmutte T, Van der Heide L, Szczygiel L, Phelan A, Davidson L, Plant R. A pay-for-performance initiative to reduce pediatric psychiatric inpatient length of stay. Psychiatr Serv. 2019;70(2):156-158. doi:10.1176/appi.ps.201800190

7. Garner BR, Godley SH, Dennis ML, Hunter BD, Bair CML, Godley MD. Using pay for performance to improve treatment implementation for adolescent substance use disorders: results from a cluster randomized trial. Arch Pediatr Adolesc Med. 2012;166(10):938-944. doi:10.1001/archpediatrics.2012.802

8. Hinton E, Stolyar L. 10 things to know about Medicaid managed care. Kaiser Family Foundation. February 23, 2022. Accessed August 17, 2022. https://www.kff.org/medicaid/issue-brief/10-things-to-know-about-medicaid-managed-care/

9. Hinton E, Stolyar L, Guth M, Nardone M. State delivery system and payment strategies aimed at improving outcomes and lowering costs in Medicaid. Kaiser Family Foundation. January 12, 2022. Accessed August 17, 2022. https://www.kff.org/medicaid/issue-brief/state-delivery-system-and-payment-strategies-aimed-at-improving-
outcomes-and-lowering-costs-in-medicaid/

10. Sheth SA, Saver JL, Starkman S, et al. Enrollment bias: frequency and impact on patient selection in endovascular stroke trials. J Neurointerv Surg. 2016;8(4):353-359. doi:10.1136/neurintsurg-2014-011628

11. Oseran AS, Howard SE, Blumenthal DM. Factors associated with participation in cardiac episode payments included in Medicare’s Bundled Payments for Care Improvement Initiative. JAMA Cardiol. 2018;3(8):761-766. doi:10.1001/jamacardio.2018.1736

12. Liao JM, Pauly MV, Navathe AS. When should Medicare mandate participation in alternative payment models? Health Aff (Millwood). 2020;39(2):305-309. doi:10.1377/hlthaff.2019.00570

13. McDowell MJ, Busch AB, Sen AP, et al. Participation in accountable care organizations among hospitals offering substance use disorder and mental health services. Psychiatr Serv. 2018;69(11):1131-1134. doi:10.1176/appi.ps.201800248

14. Herman Soper M. Integrating behavioral health into Medicaid managed care: design and implementation lessons from state innovators. Center for Health Care Strategies. April 2016. Accessed August 17, 2022. https://www.chcs.org/media/BH-Integration-Brief_041316.pdf

15. Kilbourne AM, Beck K, Spaeth-Rublee B, et al. Measuring and improving the quality of mental health care: a global perspective. World Psychiatry. 2018;17(1):30-38. doi:10.1002/wps.20482

16. Knutson KH. A value framework for transforming behavioral health. NEJM Catal Innov Care Deliv. 2021;2(8). doi:10.1056/CAT.21.0037

17. Lewis VA, Colla CH, Tierney K, Van Citters AD, Fisher ES, Meara E. Few ACOs pursue innovative models that integrate care for mental illness and substance abuse with primary care. Health Aff (Millwood). 2014;33(10):1808-1816. doi:10.1377/hlthaff.2014.0353

18. McGregor B, Mack D, Wrenn G, Shim RS, Holden K, Satcher D. Improving service coordination and reducing mental health disparities through adoption of electronic health records. Psychiatr Serv. 2015;66(9):985-987. doi:10.1176/appi.ps.201400095

19. Madden JM, Lakoma MD, Rusinak D, Lu CY, Soumerai SB. Missing clinical and behavioral health data in a large electronic health record (EHR) system. J Am Med Inform Assoc. 2016;23(6):1143-1149. doi:10.1093/jamia/ocw021

20. Thorpe K, Jain S, Joski P. Prevalence and spending associated with patients who have a behavioral health disorder and other conditions. Health Aff (Millwood). 2017;36(1):124-132. doi:10.1377/hlthaff.2016.0875

21. Smith SM, Cousins G, Clyne B, Allwright S, O’Dowd T. Shared care across the interface between primary and specialty care in management of long term conditions. Cochrane Database Syst Rev. 2017;2(2):CD004910. doi:10.1002/14651858.CD004910.pub3

22. Bishop TF, Press MJ, Keyhani S, Pincus HA. Acceptance of insurance by psychiatrists and the implications for access to mental health care. JAMA Psychiatry. 2014;71(2):176-181. doi:10.1001/jamapsychiatry.2013.2862

23. Kee A, Maio V. Value-based contracting: challenges and opportunities. Am J Med Qual. 2019;34(6):615-617. doi:10.1177/1062860619862756

24. Leddy T, McGinnis T, Howe G. Value-based payments in Medicaid managed care: an overview of state approaches. Center for Health Care Strategies. February 2016. Accessed August 17, 2022. https://www.chcs.org/media/VBP-Brief_022216_FINAL.pdf