Krithika Rajagopalan, PhD; Mariam Hassan, PhD; Kimberly Boswell, MD; Evelyn Sarnes, PharmD, MPH; Kellie Meyer, PharmD, MPH; and Fred Grossman, MD, PhD
Approximately 1 in 4 (26.2%) adults in the United States suffers from a diagnosable mental disorder in a given year,1
and owing to the high prevalence, mental disorders are associated with substantial financial burden on the healthcare system. The Substance Abuse and Mental Health Services Administration (SAMHSA) estimated that total direct costs for mental healthcare were $100 billion in 2003, representing 6.2% of all medical expenditures in the United States.2
In 2003, public payers provided more than half (58%) of the funding for mental health, with Medicaid representing the largest source of public funding (45%).2
Due to rising spend on pharmaceutical treatments in the last 2 decades, state Medicaid programs have increasingly been implementing pharmaceutical cost-containment strategies, such as prior authorizations (PAs), preferred drug lists, dispensing limits, mandatory use of generics, supplemental rebates, cost sharing, and step therapy.3
Although the goal is to encourage appropriate use of medications and to contain costs, these strategies may have unintended consequences; for example, PA requirements have been shown to increase the probability of drug access problems by 20%.4
Treatment decisions for patients with mental illness are driven by a complex interplay that includes efficacy, tolerability, prior drug history, patient preference, access, and drug interactions; successful pharmacological treatment is often only achieved after multiple therapeutic trials.5,6
Atypical antipsychotics (AAPs), which were introduced in the early 1990s, are considered standard pharmacotherapy for schizophrenia and bipolar disorder. Although AAPs are considered to belong to the same class, these drugs are quite heterogeneous as they exhibit different pharmacological profiles and may not be interchangeable. Patient responses to these agents can be variable—a lack of response to one agent does not necessarily predict a lack of response to another, even within the same drug class or vice versa.5
Compounding the variable nature of the clinical response, treatment resistance is a challenge to successful long-term management of the mentally ill. Given the absence of genetic biomarkers that can predict response with one agent compared with another, treatment optimization requires adequate trials of multiple antipsychotics.5
As such, the potential disruption of stable pharmacological management due to medication access challenges or restrictions may have serious clinical consequences, including treatment discontinuation, relapse, and deterioration of symptoms and/or function.5
Our objective was to conduct a thorough review and synthesis of the literature to evaluate the impact of restricted access to AAPs on healthcare costs and health outcomes in individuals with either schizophrenia or bipolar disorder.
A narrative review of all English-language studies published in peer-reviewed journals between January 1993 and December 2013 was conducted using MEDLINE (via PubMed). Inclusion criteria were as follows: studies conducted in the United States involving an evaluation of healthcare costs or health outcomes with regard to restrictions on access to antipsychotic medications in health plan populations, and encompassing schizophrenia or bipolar disorder. The search was limited to original research articles; therefore, commentaries, editorials, and reviews were excluded.
Search terms included the following: (schizophrenia
OR serious mental illness
OR psychiatric illness
) AND (atypical antipsychotic
) AND (step edit
OR step therapy
OR formulary restriction
[tier, control, cost
] OR drug coverage restriction
OR prior authorization
OR drug utilization
OR managed care
OR Veterans Affairs
OR tiered therapy
OR prescription limit
OR open access
). The following filters were applied: publication date (from 01/01/1993), humans, and English.
The search resulted in 230 hits, of which 208 were excluded based on review of title and/or abstract (Figure
). The remaining 22 were further assessed for eligibility by review of full-text article. Studies were excluded based on the following: non-schizophrenia or bipolar patient population, no abstract available, non-US study, nonprimary research article, or no evaluation of restricted access on antipsychotics with regard to healthcare costs or health outcomes.
A total of 15 studies met the strict inclusion and exclusion criteria and were thus included in the review (eAppendix
, available at www.ajmc.com
). Among the 15 included studies, 11 assessed PAs, 2 assessed mental health benefit carve-outs, 1 assessed a payment limit (cap), and 1 assessed Medicare Part D cost sharing; of these, 1 study evaluated costs only. The remaining studies considered both costs and other outcomes, such as healthcare and drug utilization, treatment discontinuation, and quality of care. Studies were assessed based on their impact on pharmacy costs, nonpharmacy costs, overall costs, and clinical event rates (Table
Impact on Expenditures
Findings from studies reporting the impact of cost-containment policies on healthcare expenditures are summarized below. Soumerai et al (1994) evaluated the effects of a New Hampshire Medicaid-imposed 3-prescription limit (cap) on the use of psychotropic drugs for schizophrenia during a study period from 1980 to 1983.7
The cap was associated with an immediate drop in the utilization of these drugs. However, it was also associated with a corresponding increase in visits to community mental health centers (CMHCs) of 1 to 2 visits per member per month (PMPM) and an increase in utilization of emergency mental health services and partial hospitalization from 1.2 to 1.4 episodes PMPM. It was estimated that during the 11-month period the cap was in effect, the increase in the statewide cost of mental health services was $390,000. The estimated increase in mental health services exceeded the savings in drug costs by a factor of 17.
Farley et al conducted a retrospective assessment of a step-therapy PA for AAPs implemented in the Georgia Medicaid program (1996-1997),8
where patients were required to fail 2 typical antipsychotics before receiving an AAP. In the subset of patients with schizophrenia, the PA policy demonstrated a $19.62 PMPM decrease in AAP expenditures; however, the savings were outweighed by a $31.59 PMPM increase in expenditures for outpatient services. The authors further explored the effect of the PA policy on the frequency of outpatient visits and found that the number of outpatient visits declined, albeit outpatient payments increased. This finding suggested the unintended consequence of greater payments per outpatient visit following the PA policy implementation.
McCombs et al evaluated the impact of a change in policy to allow open access (ie, remove formulary restrictions) to AAPs in the California Medicaid program during the period of 1994 to 2000.9
In the year following the initiation of open access, there was an increase in drug utilization and expenditures—this was especially noted in institutionalized patients; however, the increases in drug expenditures may have been offset by decreases in nursing home utilization. Additionally, total monthly costs increased in both the restricted and open-access periods, although the increase was greater in the restricted period compared with the open-access period (19.5% vs 15.2%, respectively).
Abouzaid et al developed a decision-analytic model to compare the economic impact of a PA for AAPs in the treatment of schizophrenia versus no PA over a 1-year time horizon.10
Researchers concluded that only modest potential cost savings would be recognized, without considering the potential increase in hospitalizations (ie, best-case scenario). Cost savings equated to $29 per patient per year when including increased costs of outpatient care and administrative costs associated with the PA program (2008 US dollars). The model was sensitive to the rate of hospitalizations; an increase in the subsequent hospitalization rate of only 0.5% would make the PA arm the more costly option.
Although most studies demonstrated that cost-containment policies reduce pharmacy expenditures, some demonstrated minimal savings in drug cost. Law et al conducted an evaluation of PA policies for AAPs in West Virginia and Texas Medicaid programs from 1991 to 2005. The West Virginia PA required a 14-day trial of a preferred agent, with the following drugs subject to PA: aripiprazole, brand name clozapine, olanzapine/fluoxetine, and olanzapine. The Texas policy required treatment failure, contraindication, or allergic reaction to a preferred agent, with the following drugs subject to PA: brand name clozapine, olanzapine/fluoxetine, and olanzapine. Results demonstrated a reduction in market share for nonpreferred AAPs, but no appreciable decrease in pharmacy expenditures.11
Notably, there were extensive grandfathering provisions for patients already receiving AAPs. However, researchers hypothesized that the lack of change in pharmacy expenditures could be a result of alternative medications also being costly. Furthermore, providers may respond to PA policies by increasing the dose of less-optimal medications or combining medications (polypharmacy); both of these scenarios would reduce any savings.
In another study, from 2005 to 2007 in the Vermont Medicaid program, Simeone et al evaluated the effects of a rescission of a PA exemption on cost and utilization of behavioral health medications.12
In this phase-out of the PA exemption for patients with severe and persistent mental illness, beneficiaries were exempt from PA requirements for medications not on the Medicaid preferred drug list. The phase-out demonstrated modest increases in formerly exempt behavioral health medications (2.1% PMPM) and was associated with a reduction in the rate of mental health–related inpatient hospitalizations (from 0.6% to 0.4%).
Impact on Medication Adherence and Clinical Outcomes
Disruption of treatment is an unintended and potentially serious consequence of restricted access to AAPs among patients with schizophrenia or bipolar disorder, especially considering that nonadherence and poor persistence to therapy is an ongoing challenge for patients with mental illness5
—with or without formulary restrictions. Zhang et al evaluated the association of PAs for psychiatric drugs with medication discontinuation among patients with bipolar disorder in the Maine Medicaid program from 2001 to 2004.13
The PA policy was associated with an 8% reduction in utilization of nonpreferred AAPs and anticonvulsants without a corresponding increase in utilization of preferred agents, indicating that patients were discontinuing treatment rather than switching to the program-preferred agents. Rates of medication discontinuation were over twice as high with implementation of the PA policy (hazard ratio, 2.28). Researchers concluded that a modest reduction in pharmacy spending ($27 per patient during the 8-month policy period) may have largely been driven by higher rates of medication discontinuation. Using the same data for patients with schizophrenia, Soumerai et al (2008) found that initiation of AAPs during the implementation of the Maine Medicaid PA policy resulted in a 29% greater risk of treatment discontinuation versus before the policy, while providing only minimal pharmacy savings.14
Similarly, Lu et al evaluated the association of PAs (affecting new prescriptions for nonpreferred AAPs [including olanzapine, olanzapine/fluoxetine, and aripiprazole] and anticonvulsants) with medication discontinuation among patients with bipolar disorder in the Maine Medicaid program (2001-2004).15
Compared with pre-policy, implementation of the PA policy was associated with increased medication discontinuation for patients who visited CMHCs and patients who did not: CMHC attenders (ie, ≥2 visits) (38% vs 31%, respectively) and non-attenders (ie, <2 visits) (41% vs 33%, respectively). The frequency of psychiatric visits was reduced among CMHC attenders (−64 per 100 patients per month); however, emergency visits were increased among non-attenders (+16 per 100 patients per month). Researchers suggested that the administrative barriers associated with PA policies could affect continuity of care, leading to relapse while off treatment. Also, for CMHC non-attenders, attempts were being made to manage medication access issues in outpatient emergency departments (EDs).
The impact of formulary restrictions on AAP drug utilization was evaluated among 30 state Medicaid programs.16
Among 11 states that instituted PAs between 1999 and 2008, AAP utilization per enrollee increased by 14%—a smaller increase than that observed in the 19 states without PA requirements (19%). However, overall antipsychotic utilization decreased by 3.1% in states with PA policies, which suggests that the reduction in targeted AAPs was not offset by the substitution of other antipsychotics. A similar finding was reported in a 2001 to 2004 study by Lu et al of patients with bipolar disorder in the Maine Medicaid program.17
The PA policy was associated with an immediate decrease in the rate of initiation of bipolar treatments (1.8%). By 4 months after policy implementation, the relative rate of bipolar treatment initiation decreased by 32.3% without a corresponding increase in the initiation of preferred agents, indicating a potential failure to treat.
Busch et al conducted a study to evaluate the effect of a managed behavioral health carve-out on the quality of outpatient care for Medicaid patients with schizophrenia from 1994 to 2000.18
Quality of care was measured by treatment recommendations established by the schizophrenia Patient Outcomes Research Team. Results showed that the carve-out was negatively associated with achieving quality care, with reduced odds of receiving any individual therapy (odds ratio [OR], 0.27; 95% CI, 0.22-0.33), group therapy (OR, 0.19; 95% CI, 0.14-0.25), and psychosocial rehabilitation (OR, 0.31; 95% CI, 0.26-0.38). However, the likelihood of receiving AAPs was the same whether or not a patient was in the carve-out program, and prescribing these medications had no direct economic consequence for the carve-out vendor. In another study from 2005 to 2007, a mental health carve-out in the Florida Medicaid program was assessed to determine if the implementation was associated with changes in drug utilization.19
The introduction of the carve-out plan was associated with both positive (increased penetration and reduced polypharmacy) and negative changes (reduced adherence). Since overall Medicaid expenditures for antipsychotics declined, researchers cautioned whether the cost savings may have contributed to reduced quality of care.
Fung et al evaluated Medicare Advantage Part D cost sharing on antipsychotic drug spending, adherence, and clinical outcomes from 2006 to 2007 for beneficiaries with schizophrenia, bipolar disorder, or no mental health diagnosis.20
Among beneficiaries with a coverage gap, greater patient cost sharing during the gap decreased Medicare antipsychotic drug expenditures while increasing patient out-of-pocket costs. However, hospitalizations and ED visits increased during the gap, and adherence rates (as measured by proportion of days covered) decreased during the gap (schizophrenia: −20.6%; bipolar disorder: −18.1%).
Walthour et al demonstrated an improvement in patient outcomes after implementation of the PA policy for AAPs among patients with schizophrenia in the Georgia Medicaid program from 2003 to 2006.21
Olanzapine, aripiprazole, and olanzapine/fluoxetine were nonpreferred AAPs under the PA policy. New prescriptions were required to obtain a PA; patients already taking any of these medications at the time of implementation were eligible for grandfathering. Criteria for PA approval included trial of a generic typical antipsychotic agent; however, new and continuing prescriptions for other AAPs (ziprasidone, risperidone, clozapine, and olanzapine injection) did not require a PA. During the PA policy period, decreases in the mean number of all-cause ED visits (absolute difference = −0.042 PMPM; relative difference = −20.92%) and all-cause hospital admissions (absolute difference = −0.010 PMPM; relative difference = −22.27%) were observed. The analysis of utilization was not restricted to schizophrenia-related ED visits and hospitalizations. Of note, this Georgia Medicaid PA policy was far less restrictive than many other PA policies, providing both grandfathering of existing medications and multiple AAPs not requiring a PA.
Our review of the literature revealed that, when evaluated, formulary restrictions more often than not were associated with pharmacy cost savings. Eight of the 15 studies included an assessment of the impact of AAP formulary restrictions on both changes in pharmacy costs and clinical outcomes. Among these, 5 studies suggested that pharmacy cost savings were accompanied by increases in healthcare utilization (eg, hospitalizations, ED visits, and nursing home visits) or treatment discontinuation.7,9,13,14,20
Two studies suggested that pharmacy cost savings were accompanied by decreases in healthcare utilization (eg, hospitalizations, ED visits, and outpatient visits).8,21
An additional study, which evaluated a rescission of a PA exemption in a mental health subgroup, suggested that pharmacy cost increases were associated with a reduction in hospitalizations.12
Assessment of overall cost burden of formulary restrictions was limited in the literature: 2 studies reported increases in overall costs after the implementation of formulary restrictions,7,8
another study reported reduced overall costs after a shift from closed formulary access to open access,9
and a decision analytical model estimated potential modest cost savings associated with the PA policy in another study.10
The intent of a PA is to encourage appropriate use of medications and to contain costs without negatively affecting clinical outcomes. Overall, although PA programs are common, there is a lack of adequately evaluated outcomes of these programs, including the long-term impact.22
Based on an underlying concern that formulary restrictions may result in a similar or higher overall cost burden after implementation, states, such as California, have reversed their restriction policies. In these situations, pharmacy spending may be reduced following implementation of formulary restrictions, but the healthcare system’s overall cost burden may have simply shifted to other parts of the system (eg, medical resource utilization).7,8,20
In addition, increases in medication discontinuation rates have been reported after implementation of PA policies, suggesting that for some healthcare systems, a component of the pharmacy cost savings may be a result of patient nonadherence and a decline in medication utilization rather than a migration to lower-cost therapies.13-15
The potential impact on adherence is critical when considering cost-containment approaches, as medication nonadherence is already a concern for patients with mental illness. It is estimated that half of patients with schizophrenia are nonadherent to therapy, with poor adherence linked to worse functional outcomes and increased risk of hospitalization.23-26
Restricted access may have an untoward effect on patients with mental illness by creating an administrative barrier to refilling medications.13
Further, restrictions can be burdensome to administer and expensive to implement. A survey of Medicaid providers in Texas assessed the effects of a PA for psychiatric medications.27
Administrative issues associated with PA policies included costs, time away from patients, uncompensated administrative time, and an extra step in providing care. In the Texas preferred drug list program evaluated in this review, the administrative costs were $4.4 million for all drug classes in 2005.11
Moreover, restricting access for a drug class with a high treatment response variability could negatively affect patient care.28,29
A patient who fails to respond to initial treatment with one drug may have a positive response to treatment with another drug; therefore, open access is essential for a drug class that requires such individualized patient care.
This literature review is bound by the limitations of each respective study. Many of the studies are based on administrative and pharmacy claims data, which may be limited by potential inaccuracies and the inability to establish causality for treatment discontinuation. Also, estimates of drug expenditures likely did not account for manufacturer rebates offered to state Medicaid programs. Further, although we did seek to be more inclusionary given the limited evaluations in the literature, there was much heterogeneity among the studies reviewed (eg, study period, evaluation period, geographic area, inclusion of control, outcomes measured, and type of cost containment policy evaluated). Seven of the 15 studies evaluated did not include a control comparison from another state. In most studies, the follow-up period after policy implementation was relatively short (≤1 year). Although the studies represented multiple states and geographic regions, the policies may not be representative of all states and may not be generalizable. In addition to the policy restrictions imposed, outcomes can be affected by a number of other factors, including the underlying patient population (eg, disease state, gender, age, socioeconomic status), provider prescribing behavior, and patient response.
Formulary restrictions are often used as a cost-containment strategy in an environment of rising prescription drug costs. Although most studies that evaluated a pharmacy cost change demonstrated a reduction in costs, the overall cost change in most cases was not measured. A reduction in pharmacy costs after the implementation of formulary restriction policies is often accompanied by increases in healthcare utilization, such as hospitalizations and treatment discontinuation. There exists a gap in the literature as to whether restricted access to AAPs results in overall cost savings or, rather, shifts the cost burden from pharmacy spending to other parts of the healthcare system. Further investigation is warranted to fully understand the clinical and economic impacts of formulary restriction of AAPs for individuals with schizophrenia or bipolar disorder.