Toward Greater Accountability in Mental Health Care

Because many mental health providers’ educational and training practices still operate in apprenticeship models, effective teaching is often limited to the extent of what these teachers know. Too often, education on clinical advances is not built into standard practice. For example, the Accreditation Council for Graduate Medical Education requires resident physicians to be educated in concepts related to cultural competence, even though the field has evolved to consider structural competence and cultural humility as equally if not more important skills for psychiatric trainees. Structural competence is defined as the trained ability to understand how clinical presentations of disease are influenced by upstream social determinants of mental health (8), and cultural humility is the lifelong practice of self-reflection and self-critique in the examination of cultural identities (9). Moreover, few psychiatrists, psychologists, and social work trainees have received specialized training in evidence-based psychotherapies, because most required trainings in these programs are not in evidence-based therapies (10).

Lack of workforce diversity also limits accountability. For example, despite long-standing data highlighting that patients from minoritized backgrounds receive poorer quality of care, compared with White populations (11), no specific accountability metrics have been put in place to monitor, let alone improve, this specific care inequity, because those most responsible for implementing accountability metrics are often not directly affected by this lack of diversity and the resulting disparate outcomes.

The use of metrics is highly uneven and inconsistent across the mental health delivery system, and as a result, performance improvement has lagged behind the rest of health care (12). Metrics that assess quality of care exist, including the National Committee for Quality Assurance (NCQA) Healthcare Effectiveness Data and Information Set (HEDIS) measures and the federal government’s Healthy People 2030 Goals, which provide basic benchmarks for monitoring mental health outcomes; however, these are limited to a small number of mental health outcomes (i.e., suicide, depression screening, and access to care). These metrics fail to capture the full breadth of mental illnesses and substance use disorders.

A recent analysis of quality measures for behavioral health care in federal government programs, such as Medicare, Medicaid, and the health insurance marketplaces, concluded that “standardized BH [behavioral health] quality measures used in federal programs focus on narrowly specified conditions or processes and are misaligned and used variably across programs” (13). The authors reported that four measures are commonly used across federal programs: screening for depression, outpatient follow-up after hospitalization, initiation and engagement in substance use disorders treatment, and screening and intervention for tobacco use. Notably, none of these metrics report on the patient experience. Measures of patient experience are used in various programs, such as Medicare Advantage, but elsewhere in Medicare, patient experiences, such as in inpatient psychiatric care, are explicitly excluded from quality measurement. These measures also rarely include measures that examine discrimination or inequities in outcomes. Furthermore, these measures typically do not reflect a strategic view of the sources of quality failures that occur in behavioral health care.

In some health care circumstances, choosing appropriate metrics and reporting them publicly can serve to create economic consequences for strong or weak performance based on market responses. The capitated health plan model that is prevalent in commercial insurance is based on combining high-powered financial incentives to spend less with quality metrics reported publicly, which in theory drives consumers to plans that best balance quality and cost of care. However, in many cases, relying on consumer choices based on cost and quality in health care does not serve to discipline the performance of health plans or providers. Moreover, the behavioral health quality measures tend to be crude and relatively few. The Medicare Advantage program recognizes such limitations by paying bonuses to health plans achieving high scores on quality metrics, in addition to publicly reporting quality measures. However, the weight given behavioral health indicators in the bonus scheme is minimal compared with weights given to other indicators, such as those for diabetes care, and thus the consequences for performance in treating mental illnesses and substance use disorders are minimal (14).

Because relatively few psychiatrists take insurance (operating cash-only private practices), accessing mental health treatment can be challenging for people from lower socioeconomic status backgrounds (15). This is in part due to reimbursement rates that are quite low and administrative processes that can be cumbersome (16).

Regulations affecting treatment facilities and providers are frequently relics of outdated clinical science. As a result, they permit program structures and practices that are ineffective and sometimes harmful. Perhaps most exemplary is the licensure and accreditation of programs that permit the continued use of seclusion and restraints that extend beyond the minimal “last resort” standard. Likewise, hospitals that provide simple inpatient detoxification-only services for people with opioid use disorder are licensed and accredited despite offering a treatment that elevates mortality risk (17). The risk is elevated because patients are discharged with lower resistance to opioids, and without active engagement in postdischarge treatment, they are at risk of using opioids and overdosing.

Performance metrics must include patient-reported experiences to be appropriately inclusive and accountable. Given that key federal programs that pay for mental health care, such as Medicare, Medicaid, and the health insurance marketplaces, also support the range of other medical services, practical management concerns require that the measures used for behavioral health are quite circumscribed. Therefore, selecting measures that would reveal unwanted conduct where the incentives to do so are greatest might serve as an appropriate basis for measure selection. For instance, in identifying ineffective treatment for bipolar disorder, measuring the number of people treated for bipolar disorder and the percentage that received an antidepressant without a mood stabilizer would serve as an indicator of how attentive a health plan or a provider organization is in maintaining the minimum standard of appropriate clinical care.

Nevertheless, this type of “poor performance” metric represents a process measure. Some measures of the outcomes produced are important to include. Outcome measures come with additional complexity—if not carefully designed they risk creating incentives for plans and providers to choose the patients likely to yield the best outcomes (19). Thus attention to patient selection must accompany the use of outcome metrics.

An example of a strategically selected process-based quality indicator would focus on the treatment of borderline personality disorder, a debilitating mental health condition often characterized by chronic suicidality. There is strong evidence of the effectiveness of dialectal behavior therapy (DBT) for the treatment of borderline personality disorder. However, few public mental health systems have DBT services available for their patient populations (20). In addition, there are no Food and Drug Administration–approved pharmacological indications for borderline personality disorder, and although evidence shows that few medications have any beneficial effect, 96% of people with a diagnosis of borderline personality disorder take at least one psychotropic medication, and 20% take four or more such medications (21). This polypharmacy is often independent of comorbid psychiatric diagnoses (22). Therefore, a performance metric could include measuring the percentage of people with a diagnosis of borderline personality disorder who receive high-fidelity DBT services (a marker of “good care”) coupled with a measure of polypharmacy (a poor-performance marker). Together, these two performance elements would reflect a provider system’s appropriate use of evidence-based psychotherapy alongside the ability to appropriately manage the use of pharmacotherapies. Such a performance measurement approach would likely correlate with the provision of high-quality care for other illnesses. Incorporating patient experience, such as satisfaction with DBT services, would be an important additional performance metric.

In promoting high-quality care for people with mental illnesses, concern that reliance on market forces would promote quality is especially salient. Because of experiences of historical and current oppression and discrimination, people with mental illnesses have had challenges advocating for their needs when dealing with insurers and providers (23). The implication is that policy cannot count on market mechanisms alone to create appropriate incentives for good-quality mental health care. This means finding a way to couple payments and performance metrics. One approach is so-called pay for performance (P4P). P4P typically works by paying bonuses to health plans or providers that meet certain outcome standards for their enrolled or patient populations. The available evidence typically demonstrates small gains that are quite costly (24, 25). This is because P4P arrangements frequently pay based on the provider’s exceeding a threshold value on a quality index. However, in many cases, most of the payments are made to providers that were already above the threshold, thus greatly increasing the cost per provider that improved its quality rating. Moreover, some experiences highlight the potential for gaming of P4P schemes that use outcome measures, such as abstinence for people with substance use disorders, allowing providers to “cherry-pick” patients who are more likely to improve (19).

Another approach is gain sharing. Gain sharing is a payment system that establishes cost targets for providers or networks of providers in which financial gains or losses are shared between payer and provider. However, the provider would only realize the available gains if quality standards for key metrics are met. Several evaluations of various types of gain sharing arrangements in health care have shown realization of savings with no declines in access to or quality of care (26). The application of gain sharing to behavioral health in the context of Medicare accountable care organizations has shown little change, in part because the weight given to behavioral measures in the overall set of quality measures was small and the potential savings modest (27). In addition, several state-shared saving programs have shown mixed results for both spending and quality measures (28).

Finally, in some cases, risk-based capitation payments are constructed from historical data on utilization and spending. If, however, historical levels of utilization were low (because of benefit design or care management processes), then the capitation rate will reflect a situation in which the money available for mental health care is inadequate to achieve the desired clinical outcomes. Thus, in such programs, the behavioral health functions are subject to strong resource constraints that affect the size of the provider network and the ability to pay competitive fees for some professionals.

To effectively tie payments to performance, the levels of payment must be consistent with the ability to produce strong performance. The behavioral health capacity must be based on an up-to-date assessment of the resources necessary to appropriately serve an enrolled population, and that capacity must take contextual factors, such as the social determinants of mental health of communities, into consideration. Payment structures that allow flexibility to combine resources to treat most patient and community circumstances are likely to produce environments most conducive to high-value behavioral health care.

Flexibility, however, typically is accompanied by significant incentives to spend less. Existing quality measures are limited in their ability to identify substandard care and are typically given little weight in bonus schemes or gain sharing arrangements. Thus the rewards and penalties linked to the quality of behavioral health care are minimal and not given programmatic attention. Payment should be linked to performance via mechanisms such as shared savings or penalties for poor performance and bonuses for strong performance, among other approaches, in ways that create substantial consequences—both financially and in public perception related to performance in serving people with mental illnesses and substance use disorders. Additionally, payment and risk adjustment should consider various social determinants of mental health affecting patients and their communities, including area-level poverty, residential segregation, and food insecurity. This ensures appropriate access to care and provides “fair” payment to providers that must address more complex and likely more costly treatment circumstances.

The supply and structural characteristics of providers, along with the workforce that delivers care and supports for people with mental illnesses and substance use disorders, are subject to state regulations. Providers serving people covered by Medicare and Medicaid are also subject to reviews and conditions by accrediting organizations such as JCAHO and CARF. State regulations are typically slow to be updated, so they frequently reflect standards of care that are decades out of date. Likewise, accrediting bodies seldom incorporate evidence about structural features that drive quality in behavioral health providers (12). Recent evidence on residential substance use disorder programs showed that 29% of programs that did not use evidence-based, medication-assisted therapy to treat opioid use disorders and frequently engaged in practices that were financially exploitive of consumers were accredited by JCAHO or CARF (29). An analysis of the overall performance of accredited versus nonaccredited hospitals showed no significant differences in quality (30). The Government Accountability Office reported that JCAHO missed the identification of a substantial portion of hospitals with serious safety deficiencies in 2004 (31). Thus regulatory oversight must be provided in ways in which outcomes and the recovery of patients are central goals.

The establishment of regulatory standards for licensing and accreditation that are undergirded by modern clinical science and are regularly updated would represent an important step in promoting higher-quality behavioral health care. A second step would involve enforcement of standards through quality improvement processes and sanctions when necessary. These are fundamental steps to ensure consumer protections and adequate quality.