“Clinically nuanced” Medicaid cost-sharing

Cost-sharing plays a role in many state Medicaid programs. Requiring enrollees to contribute to the cost of their medical care is implemented to lower state medical expenditures, impart a sense of accountability in enrollees, and align public benefit designs with those in the private insurance market¹. While enrollee cost-sharing was present in Medicaid programs prior to the Affordable Care Act, its role in plan design has become more important as coverage has expanded and states look for strategies to deal with the fiscal realities of rising healthcare expenditures.

In a recent issue of Journal of Medical Economics, the paper by Doshi et al.² sheds light on the potential impact of increases in patient cost-sharing in a particularly vulnerable population, Medicaid enrollees with a diagnosis of schizophrenia. Their findings add to four decades of research—starting with the seminal RAND Health Insurance Experiment—examining how patients respond to changes in out-of-pocket costs. The results have important ramifications as states consider how to expand, restructure, and finance Medicaid programs.

Each state designs its own Medicaid program, with constraints imposed by federal laws and regulations. Although the regulations are complex³, in general, prescription drug copayments can be as high as $4 for each preferred medication and $8 for each non-preferred medication; people with incomes above 100% of the federal poverty level can be charged higher levels of cost-sharing. The federal government has granted waivers to some states to impose higher amounts of cost-sharing, including charging monthly enrollment fees, similar to premiums^4,5. States can impose other barriers to medication use as well, including limiting coverage of medications, requiring prior approval for certain medications or branded drugs, and capping the number of medication fills in a month.

Previous research in the Medicaid population has reported that enrollees reduce their utilization of prescription medications, even when cost-sharing is nominal. For example, an increase from $0 to $2 for prescription drugs in Utah’s Medicaid program decreased medication use by ∼8%⁶. Another analysis, using data from multiple states, found those eligible for both Medicaid and Medicare reduced prescription drug use in response to higher copayments⁷. Other analyses have suggested medication non-adherence related to cost sharing adds spending in other parts of the system. For example, one study examining Medicaid enrollees with cancer found that nominal increases in copayments both reduced prescription drug use and increased visits to emergency rooms⁸. Doshi et al.² add to this literature; they found that, when copayments were imposed on Medicaid enrollees diagnosed with schizophrenia, medication use decreased for both antipsychotics and, even more strongly, for medications prescribed for physical, co-morbid conditions.

One of many lessons we can take from this work and similar research in the commercial insurance market is that patients are not always rational; they often use too little of the medical care known to benefit health and overuse services with few benefits^9,10. This phenomenon has been described as a behavioural hazard¹¹, an augmentation of moral hazard that takes into account behavioural biases that result in the sub-optimal use of healthcare. For example, patients may overemphasize present needs or desires and skip a medication that causes their condition to worsen and later leads to higher healthcare costs.

An implication of behavioural hazard is that cost-sharing strategies must be better structured to provide patients clear signals regarding those clinical services and providers that are most likely to achieve better patient-centered outcomes (and those that don’t). That is, cost-sharing should not be applied uniformly across visits, medications and other services, but instead should be tailored to encourage the use of high-value services and discourage the use of low-value ones. This approach, known as value-based insurance design (V-BID), has been implemented for decades in the commercial insurance market¹², and V-BID demonstrations are underway in Medicare Advantage and the US military’s TRICARE program¹³.

V-BID plans are designed with the tenets of “clinical nuance” in mind. These tenets recognize that (1) medical services differ in the amount of health produced, and (2) the clinical benefit derived from a specific service depends on the person using it, as well as when, where, and by whom the service is provided. In the commercial market, V-BID plans have been shown to increase medication adherence¹⁴, reduce socioeconomic disparities, and lower adverse events associated with disease progression¹⁵. Still, many insurance plans, including most Medicaid plans, impose cost-sharing across the spectrum of care in a blunt, “one-size-fits-all” fashion. This approach lacks incentives that encourage enrollees to use healthcare in a more efficient way. Incorporating the idea of behavioural hazard into insurance benefit structures could improve enrollee health and the efficiency of how this population uses the healthcare system.

An interesting result from the work by Doshi et al.² is the differential price elasticities of medications to treat schizophrenia compared to those prescribed to manage other co-morbid conditions. The authors find that medications for physical ailments are much more responsive to the out-of-pocket price than those used for schizophrenia, which, as the authors note, contradicts other research showing higher price elasticities for mental health services. The finding is important and possibly ominous in light of evidence that, for patients with mental illness, observed reductions in lifespan are typically the result of the progression of physical, co-morbid conditions¹⁶.

One interpretation of this finding is that patients are more price responsive to medications for which they do not experience regular symptoms and for which stopping the medication does not result in immediately negative clinical consequences. Research from both commercially-insured enrollees and those in Medicaid has found differential demand responses to increased out-of-pocket costs between medication classes. In the commercial market, smoking deterrents had a higher price elasticity than other prescription classes, suggesting that patients will more readily quit a medication for which the primary benefit occurs over the long-run¹⁷. Within Medicaid, when caps on monthly medication fills were imposed, enrollees reduced their use of preventive medicines, but not medications that provide relief from disease symptoms, implying a higher value for symptomatic relief over prevention¹⁸.

As Americans are being asked to pay more for their medical care, future research should be undertaken to better understand the mechanisms driving patients’ response to the cost of healthcare services. When out-of-pocket costs rise, patients will likely make substitutions (rational or not) across medication classes based on financial constraints, including by self-triaging in a way that could impact later health and increase long-run costs.

Additionally, the population studied here may be particularly sensitive to policy changes impacting access or creating barriers to care. They are at risk for severely adverse outcomes, including frequent hospitalization and homelessness. While this study did not look at the non-medical choices made by this population, other work has shown that increases in medical costs force people to forego not just medications, but also basic necessities including food, housing, and transportation¹⁹.

State Medicaid programs cover some of the nation’s most vulnerable citizens, and account for a large and growing portion of state budgets. We need to think beyond the blunt instrument of wholesale increases in out-of-pocket costs to a more nuanced view of cost-sharing in this population. In 2013, the Centers for Medicare & Medicaid Services gave state Medicaid programs the flexibility to vary enrollee cost-sharing for drugs as well as certain outpatient, emergency department, and inpatient visits²⁰. If implemented successfully, a clinically nuanced cost-sharing model can enhance the use of high-value clinical services and reduce utilization of unnecessary and costly services, while fostering a sense of personal engagement and accountability. If V-BID principles are used to set enrollee cost-sharing levels, Medicaid programs can improve quality of care, remove waste, and mitigate the concern that cost-sharing may lead individuals to forgo clinically important care.

Transparency

Declaration of funding

There is no funding to report for this editorial.

Declaration of financial/other interests

EQC has no funding or financial relationships to declare. AMF has received funding from, or provided consulting services to, AbbVie, AHRQ, Amgen, Bayer, CalPERS, FMS Inc., Freedman Health, Gary and Mary West Health Policy Center, Janssen, Laura & John Arnold Foundation, Lilly, Mallinckrodt, MedZed, Merck, National Pharmaceutical Council, PCORI, PhRMA, Risalto, RWJ Foundation, Sempre Health, State of Michigan/CMS, State of Minnesota, State of New York, Takeda, TriZetto, Wellth, and Zansors. He is also a partner in V-BID Health, LLC. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.