Outcomes of Medicaid Beneficiaries With Schizophrenia Receiving Clozapine Only or Antipsychotic Combinations

Antipsychotic polypharmacy, defined as prescribing two or more antipsychotic agents simultaneously, has little evidentiary support in the treatment of patients with schizophrenia and is generally discouraged by practice guidelines and consensus statements (1–3). Nonetheless, the practice of using antipsychotic polypharmacy for schizophrenic disorder is common and increasing. Although an antipsychotic polypharmacy rate of only 13.7% was reported in a Department of Veterans Affairs study of second-generation antipsychotic use in 2004 (4), rates of 40% or more have been documented in studies of a wide variety of settings, time periods, and drugs (5–7).

One naturalistic study of U.S. patients initiating treatment for schizophrenic disorder with a second-generation antipsychotic (olanzapine, quetiapine, or risperidone) from 1997–2003 found that 58% had at least one period of antipsychotic polypharmacy exceeding 60 days (7). In samples of Medicaid beneficiaries with schizophrenia, Clark and colleagues (8) found that the rate of antipsychotic polypharmacy increased from 5.7% in 1995 to 24.3% in 1999. Ganguly and colleagues (9) documented an antipsychotic polypharmacy rate increase from 32% in 1998 to 41% in 2000. Ganguly and colleagues also found that the mean duration of antipsychotic polypharmacy was 149 days overall, with 23% of patients receiving long-term antipsychotic polypharmacy (more than two months) for a mean duration of 236 days.

Treatment resistance (that is, suboptimal response to at least one adequate trial of antipsychotic monotherapy) is a commonly cited justification for antipsychotic polypharmacy (10–13). However, two pragmatic randomized controlled trials have shown clozapine to be a first-line agent of choice in treatment-resistant schizophrenia (14,15). American Psychiatric Association guidelines for the treatment of schizophrenia (updated in May 2011) recommend that monotherapy with clozapine be considered “for a patient who has had no response or partial and suboptimal response to two trials of antipsychotic medication” (16). Thus it has been suggested that reluctance to prescribe clozapine when clinically indicated for treatment-resistant disease may contribute to antipsychotic polypharmacy (17–19).

Research evidence indicating that antipsychotic polypharmacy is often tried before clozapine provides support for this view. For example, Taylor and colleagues (20) reviewed medical histories for inpatients with schizophrenia who were prescribed clozapine for the first time and found that 65% had prior antipsychotic polypharmacy. In a more recent study of a sample that included both inpatient and outpatient clozapine users, Howes and colleagues (21) found a 36% rate of previous antipsychotic polypharmacy.

Despite the widespread use of antipsychotic polypharmacy, often in lieu of clozapine monotherapy, we are aware of no research comparing these regimens in the treatment of patients with schizophrenia. This study addressed that information gap by examining all-cause and disease-specific health care utilization rates and costs in a large, recent sample of Medicaid beneficiaries treated for schizophrenic disorder with either clozapine monotherapy or second-generation antipsychotic polypharmacy.

Methods

The study was a retrospective observational analysis of data from the Medicaid MarketScan database, which contains integrated administrative claims and eligibility data for approximately 30 million Medicaid beneficiaries located throughout the United States. Data fields for pharmacy claims included national drug code numbers, fill dates, and days’ supply. Medical claims included ICD-9-CM codes for up to 15 diagnoses, current procedural terminology and revenue codes, and Healthcare Common Procedure Coding System codes. Data were collected as part of normal business operations and deidentified before delivery to the investigators in accordance with HIPAA requirements. For this reason, institutional review board approval was not sought.

Sample

The data set delivered to the investigators represented a sample consisting of all Medicaid beneficiaries who had at least one pharmacy claim for a second-generation antipsychotic (aripiprazole, clozapine, olanzapine, paliperidone, quetiapine, risperidone, or ziprasidone) or at least one diagnosis of schizophrenic disorder (ICD-9-CM code 295.XX) in the Medicaid MarketScan database from January 2005 through December 2009. From this data set (N=611,622), all patients who initiated second-generation antipsychotic treatment between July 1, 2006, and January 1, 2009 (identification period) were selected.

The clozapine monotherapy cohort consisted of patients with at least one claim for clozapine during the identification period and no claims for any other second-generation antipsychotic from 90 days before through 90 days after the first clozapine claim fill (index) date. The antipsychotic polypharmacy cohort consisted of patients with two or more overlapping second-generation antipsychotic claims, defined on the basis of fill date plus days’ supply. For example, a patient with 30-day supply claims for aripiprazole on July 1 and August 1 who filled prescriptions for olanzapine on July 15 and August 15 was defined as receiving an aripiprazole-olanzapine combination. All cohorts were defined on an intent-to-treat basis; that is, the earliest drug (clozapine) or drug combination (antipsychotic polypharmacy) in the identification period was used to classify the patient throughout the study regardless of subsequent treatment augmentation or switching. Patients whose index regimen was antipsychotic polypharmacy with clozapine (that is, clozapine plus any other second-generation antipsychotic) were excluded from the sample (N=457).

From this group, study patients were selected if they were age 18 to 64 as of the index date (first second-generation antipsychotic claim), had continuous Medicaid coverage with both medical and drug benefits from six months before (preperiod) through 12 months after (postperiod) the index date, and had a diagnosis of schizophrenia or schizoaffective disorder, defined as having at least one medical claim with a diagnosis of a schizophrenic disorder (ICD-9-CM code 295.XX) in any diagnosis field in either the preperiod or postperiod.

Outcome Measures

Postperiod outcome measures included costs, use of inpatient care, and use of the emergency department. Costs were defined as gross payments to a provider for a service. This payment is equal to the amount eligible for payment under the medical plan terms after applying rules such as discounts, but before applying coordination of benefits, copayments, and deductibles. Inpatient care and the emergency department were measured as binomials. Each of these measures was reported in three diagnosis categories based on ICD-9-CM codes in any field: all cause (any diagnosis), mental disorder related (290.XX–319.XX), and schizophrenia related (295.XX). Mental disorder–related costs and schizophrenia-related costs were defined as all inpatient, outpatient, drug, and laboratory costs with an accompanying diagnosis of a mental disorder or schizophrenia, respectively. Total number of all-cause hospital inpatient days was an additional outcome measure.

Statistical Analyses

The primary study analyses were multivariate. Logistic regression analyses were used to model the binomial measures of emergency department and inpatient use, generalized linear models (GLMs) with gamma distribution and log link to model costs, and GLMs with negative binomial distribution and log link to model hospital days. All analyses controlled for demographic characteristics (age, sex, and race), preperiod Charlson Comorbidity Index (CCI) (22) score, and indicators of preperiod comorbidities common in schizophrenia. These indicators were based on ICD-9-CM diagnosis codes in any field and included substance abuse (303.XX, alcohol dependence; 304.XX, drug dependence; and 305.XX, nondependent abuse of drugs) (23,24), depression (300.4, dysthymic disorder; 309.XX, adjustment reaction; 296.2 and 296.3, major depressive disorder; and 311, depressive disorder not elsewhere classified) (23), and anxiety (300.0X, anxiety states; 300.2X, phobic disorders; and 300.83, somatoform disorders). In addition, the resource utilization models controlled for preperiod resource utilization.

In all analyses, antipsychotic polypharmacy was the reference category, and the coefficients represented clozapine monotherapy. For each study outcome, seven separate equations were run, one for antipsychotic polypharmacy overall and one for each antipsychotic polypharmacy cohort (aripiprazole, olanzapine, paliperidone, quetiapine, risperidone, and ziprasidone), compared with clozapine monotherapy (for example, aripiprazole polypharmacy versus clozapine monotherapy, olanzapine polypharmacy versus clozapine monotherapy, and so forth). Antipsychotic polypharmacy cohorts were not mutually exclusive; for example, a patient who used aripiprazole and olanzapine was included in both the antipsychotic polypharmacy aripiprazole and antipsychotic polypharmacy olanzapine cohorts. All analyses were conducted using SAS version 9.3, and an a priori significance level of .05 was specified.

Results

A total of 3,376 patients were treated with at least one second-generation antipsychotic and met the sampling criteria. Of these, 457 were treated with antipsychotic polypharmacy that included clozapine, and these patients were removed from the sample. Of the 2,919 remaining patients, 479 were treated with clozapine monotherapy, and 2,440 were treated with antipsychotic polypharmacy containing an antipsychotic other than clozapine (Table 1). The most common combination was risperidone and quetiapine (N=499), followed by risperidone and aripiprazole (N=272) (Table 2). Combinations of the top four antipsychotic drugs used by sample patients (aripiprazole, quetiapine, risperidone, and ziprasidone) accounted for 65% (N=1,584) of antipsychotic polypharmacy cases.

Patients treated with clozapine monotherapy differed at baseline from those treated with antipsychotic polypharmacy, both demographically and clinically (all differences significant at p<.001) (Table 1). Clozapine-treated patients were, on average, 3.4 years younger (36.3 versus 39.7 years, respectively) and had lower CCI scores (means of .3 versus .8). Among patients in the clozapine group, the proportion of white patients was smaller than in the polypharmacy group (52% versus 61%), as was the proportion of female patients (41% versus 51%).

Baseline comorbidity rates were lower for the clozapine group than for the polypharmacy group: 12% versus 31%, respectively, for substance abuse; 12% versus 35% for depression; and 3% versus 14% for anxiety (Table 1). In both all-cause and disease-specific preperiod utilization categories, patients in the clozapine group were approximately one-half as likely as those in the polypharmacy group to use medical services. For example, the percentages with at least one all-cause hospitalization and one emergency department visit, were 17% and 29%, respectively, for the clozapine group, compared with 37% and 57% for the polypharmacy group. Schizophrenia-related hospitalization and schizophrenia-related emergency department use occurred for 15% and 17%, respectively, of patients in the clozapine group, compared with 29% and 30% of those in the polypharmacy group.

When the analysis controlled for these baseline differences (Table 3), patients treated with clozapine monotherapy were less likely than those treated with antipsychotic polypharmacy to have mental disorder–related use of the emergency department (odds ratio [OR]=.75, 95% confidence interval [CI]=.60–.95) or schizophrenia-related use of the emergency department (OR=.70, CI=.54–.90) in the postperiod; however, the clozapine group was not significantly less likely to be hospitalized or have an all-cause emergency department visit in the postperiod. This pattern of results was similar for all antipsychotic polypharmacy combinations, but it was statistically significant only for the groups with polypharmacy involving quetiapine, risperidone, and ziprasidone. The number of all-cause hospital days for the clozapine group and the polypharmacy group did not significantly differ, either overall or for any specific polypharmacy combination (Table 4).

As shown in Table 5, in all utilization and drug categories, total Medicaid payments were significantly lower for clozapine-treated patients than for patients treated with antipsychotic polypharmacy, with all-cause costs reduced on average by $21,233, mental disorder–related costs reduced by $17,457, and schizophrenia-related costs reduced by $10,582. Across all antipsychotic polypharmacy categories, costs for clozapine-treated patients were on average $20,365–$26,557 lower for all-cause medical services, $15,530–$22,745 lower for mental disorder–related services, and $8,172–$11,923 lower for schizophrenia-related services.

Discussion

In a large, observational, 12-month study of Medicaid beneficiaries with a diagnosis of schizophrenia, those treated with clozapine monotherapy were less likely than those treated with antipsychotic polypharmacy to have a disease-related emergency department visit but not significantly less likely to have an all-cause emergency department visit or an all-cause hospitalization. Compared with antipsychotic polypharmacy, clozapine monotherapy was associated with disease-specific and all-cause Medicaid cost reductions of approximately 40%−45%.

These findings are generally consistent with treatment guideline recommendations that discourage the use of antipsychotic polypharmacy (1–3). Furthermore, although previous observational research did not directly compare antipsychotic polypharmacy with clozapine monotherapy, it has generally found benefits associated with monotherapy, even though polypharmacy is increasingly prescribed (25). For example, one study found that 69% of patients who switched to monotherapy remained on that therapy and that these patients experienced a loss in body mass index with no worsening of symptom control and no increase in hospitalizations (26). Another study found that switching from polypharmacy to monotherapy was associated with improvements in attention, which was associated with improvements in work skills (27). Finally, a naturalistic study conducted in Hungary found that monotherapy was associated with a longer time to discontinuation, although polypharmacy was associated with a lower likelihood of mortality and hospitalization (28). The authors concluded that the superiority of monotherapy for long-term sustained treatment, combined with the advantage of polypharmacy for mortality and psychiatric hospitalizations, suggests that combination treatments may be more efficacious during exacerbation of psychotic symptoms (28).

The potential economic and clinical implications of our findings should be considered in light of two main factors. First, our sample consisted of Medicaid beneficiaries, a population in which cost containment pressures are intensifying, the rate of antipsychotic polypharmacy is high (8,9), and the rate of clozapine use is generally low, estimated at only 2% of New York Medicaid beneficiaries with schizophrenia who initiated antipsychotic pharmacotherapy in 2009 (29). Although our results should not be viewed as definitive because of the study’s observational design, they suggest that efforts to increase conformity with guidelines for schizophrenia treatment have the potential to decrease Medicaid costs.

The second factor is the degree of unmet therapeutic need among patients with schizophrenia regardless of payment source. For example, Pickar and colleagues (5) conducted a cross-sectional survey of a community-based convenience sample of patients with a diagnosis of schizophrenic disorder (N=200), of whom 48% reported at least one involuntary hospitalization, 48% reported at least one suicide attempt, 49% reported previous incarceration, and 57% were in a supervised living arrangement or shelter. Overall, 43% were taking more than one antipsychotic medication, 70% were using an off-label regimen, and only 26% were using U.S. Food and Drug Administration–approved monotherapy. Pickar and colleagues commented that the “real world pharmacotherapy of schizophrenia has developed its own established practice that may have outstripped current data support” and argued that off-label use is partially attributable to a lack of effective treatments. In the first phase of the Clinical Antipsychotic Trials of Intervention Effectiveness study (CATIE), 74% of patients discontinued the study medication to which they were originally randomly assigned (30).

In light of the severe functional impairments that schizophrenia imposes on patients, as well as the prevalence of antipsychotic polypharmacy, it appears reasonable to intensify efforts both to identify more effective treatments and to encourage the use of treatments with demonstrated efficacy, such as clozapine, when clinically appropriate. Such efforts might target factors associated in previous research with reluctance to prescribe clozapine. These include concerns about side effects and comorbidities, lack of experience in prescribing clozapine, and reluctance to commit patients to ongoing blood level monitoring for agranulocytosis (31,32). Treatment inertia—that is, the failure to make a change in therapy even when clinically indicated—may also contribute to antipsychotic polypharmacy. For example, a semistructured-interview study conducted by Correll and colleagues (33) with 44 prescribers of combination antipsychotic therapy indicated that many “had inherited most of” their antipsychotic polypharmacy users “and were reluctant to convert patients to antipsychotic monotherapy.”

Given these concerns, educational efforts highlighting the results of pragmatic controlled trials may be helpful. For example, Essock and colleagues (34) randomly assigned patients with schizophrenia who were receiving antipsychotic polypharmacy to switch from polypharmacy to monotherapy (N=65) or to continue with antipsychotic polypharmacy (N=62). At a six-month follow-up, 86% of patients in the antipsychotic polypharmacy group were still taking the same medications and 69% of patients in the monotherapy group were still taking the same medication, with no significant group differences in symptom scores. Body mass index declined by .5 points in the monotherapy group and increased by .3 points in the antipsychotic polypharmacy group (p<.05 for group × time interaction). Although noting the limitations of their open-label design, Essock and colleagues (34) concluded that “most patients with continuing psychopathology while taking two antipsychotic medications are well able to tolerate switching from polypharmacy to monotherapy, and they reap the metabolic benefits associated with weight loss while avoiding the side effects and other burdens of taking an additional medication.” The generally favorable results observed for patients using clozapine in CATIE, the Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study, and the International Suicide Prevention Trial may also support educational efforts (14,35,36).

The limitations of this study should be noted. First, the substantial differences in the baseline characteristics of the antipsychotic polypharmacy and clozapine monotherapy cohorts suggest the possibility of confounding by indication. For example, physicians may be reluctant to prescribe clozapine for patients who appear to be particularly unlikely to adhere to blood monitoring requirements. The lower rates of substance abuse and depression among clozapine users compared with those in the polypharmacy group are consistent with that possibility. Manuel and colleagues (29) found that the odds of clozapine use were reduced by more than half for Medicaid beneficiaries with a diagnosis of substance abuse. Because of the possibility of drug-drug interactions, physicians may also be concerned about prescribing clozapine to patients with comorbidities, consistent with our finding of lower CCI scores among clozapine users. These results, combined with lower resource utilization in the preperiod, suggest that the clozapine cohort may have been more stable or may have been earlier in their course of treatment, compared with the antipsychotic polypharmacy cohort. However, the analyses controlled for observable differences between the two cohorts. Second, cohorts were defined on an intent-to-treat basis, and it is possible that treatment choices made after the initial cohort classification could have affected study outcomes. Third, the study sample consisted of Medicaid beneficiaries age 18 to 64, and results may not be generalizable to Medicare beneficiaries or commercially insured patients. Finally, the analyses focused on clozapine monotherapy and thus are not generalizable to all users of clozapine.

Conclusions

In a sample of adult Medicaid beneficiaries with schizophrenia, treatment with clozapine rather than antipsychotic polypharmacy was associated with reduced use of the emergency department for mental disorder or schizophrenia diagnoses and with reduced disease-specific and all-cause health care costs.