Impact of formulary-related pharmacy claim rejections of cariprazine on health care utilization and treatment patterns among patients with bipolar I disorder

Abstract

BACKGROUND:

Formulary restrictions, intended to limit inappropriate medication use and decrease pharmacy costs, may prevent or delay patients with bipolar I disorder from initiating cariprazine, a dopamine D₃-preferring D₃/D₂ and serotonin 5HT_1A receptor partial agonist that is approved to treat manic/mixed or depressive episodes associated with bipolar I disorder. Little is known about the downstream consequences of formulary-related cariprazine prescription rejections.

OBJECTIVE:

To evaluate the impact of formulary-related cariprazine claim rejections on health care resource utilization (HCRU) and treatment patterns among patients newly prescribed cariprazine for bipolar I disorder.

METHODS:

Symphony Health Integrated Dataverse was used to identify commercially insured adults (aged ≥18 years) with bipolar I disorder and at least 1 pharmacy claim for cariprazine (rejected because of formulary restrictions or approved; date of the first claim is the index date) from March 2015 through October 2020. Formulary-related rejection reasons included noncoverage, prior authorization requirement, and step therapy requirement. Baseline characteristics were evaluated during the 12 months pre-index and balanced between rejected and approved cohorts using 1:2 propensity score matching. HCRU outcomes included all-cause and mental health (MH)–related hospitalizations, emergency department (ED) visits, and outpatient visits. Treatment patterns were analyzed descriptively and included treatment delay and atypical antipsychotic use. HCRU was reported per patient-year and compared between cohorts using rate ratios; 95% CIs and P values were calculated using nonparametric bootstrap procedures.

RESULTS:

The matched rejected and approved cohorts comprised 1,554 and 3,108 patients, respectively. The rejected cohort had 22% more all-cause and 24% more MH-related hospitalizations per patient-year vs the approved cohort (rate ratio [95% CI], all-cause: 1.22 [1.01-1.48], P = 0.024; MH-related: 1.24 [1.01-1.55], P = 0.044). ED and outpatient visits were numerically, but not significantly, greater in the rejected cohort. Of patients in the rejected cohort, 34.7% never received an atypical antipsychotic and 76.8% never received cariprazine. For those who later received cariprazine or another atypical antipsychotic, the average treatment delay was approximately 6 months (188 days) and approximately 4 months (123 days) after the initial rejection, respectively.

CONCLUSIONS:

Patients with bipolar I disorder and formulary-related cariprazine claim rejections experienced significantly more hospitalizations than patients whose initial claim was approved; ED and outpatient visits were similar between cohorts. Less than a quarter of patients whose initial claim was rejected later received cariprazine, and more than one-third never received any atypical antipsychotic. To our knowledge, this is the first study to evaluate the impact of formulary-related rejections of cariprazine on HCRU and treatment patterns in patients with bipolar I disorder.

Plain language summary

This study looked at how insurance rejections of cariprazine affect health care service use in patients with bipolar I disorder. Results showed patients who had their first cariprazine prescription rejected had more hospital visits after the rejection than patients who had their first cariprazine claim approved. Only 1 in 4 patients later received cariprazine after their first rejection, and less than two-thirds later received a similar medication. These findings suggest cariprazine rejections have unintended effects.

Implications for managed care pharmacy

This claims-based analysis evaluated the impact of formulary-related rejections of cariprazine on health care resource utilization and treatment patterns in patients with bipolar I disorder. Patients with an initial cariprazine claim rejection were more likely to have a subsequent hospitalization than patients with an initial cariprazine claim approval. These findings should be considered during formulary development, as restrictions on cariprazine may result in unintended consequences, such as higher health care resource utilization and treatment delays.

The estimated total annual economic burden of bipolar I disorder in the United States is approximately $219 billion (2018 US dollars), comprising direct, indirect, research, and substance abuse–related costs.¹ Evidence suggests that compared with patients without bipolar I disorder, those with bipolar I disorder use 3-4 times more health care resources and incur about 4 times the health care costs.² Atypical antipsychotics are the mainstay of treatment for several psychiatric conditions, including bipolar I disorder.^3,4 The efficacy and tolerability of atypical antipsychotics vary between patients, and many patients often cycle through multiple regimens before achieving the desired treatment effect because of psychotropic medications being less therapeutically interchangeable than other medication classes.^5,6 However, high acquisition costs often affect patient access to some atypical antipsychotics.⁶ To ensure safe and appropriate medication use as well as reduce pharmacy costs, third-party payers commonly apply formulary restrictions, such as noncoverage, prior authorizations rules, and step therapy requirements, to atypical antipsychotics.^7,8

Formulary restrictions are intended to reduce costs and mitigate potentially inappropriate use of prescription medications without compromising the quality of patient care; however, there is mixed evidence for the effectiveness of formulary restrictions on atypical antipsychotics in achieving this goal. Although some studies have shown formulary restrictions in bipolar I disorder reduce drug utilization and pharmacy costs,^5,9 others suggest that formulary restrictions may lead to unintended consequences, particularly increased overall costs and adverse events and decreased medication adherence.^5,6,9-11 In a literature review that examined the impact of restricted access to atypical antipsychotics, 4 studies that assessed overall costs were included; 1 study reported modest overall cost savings, but 3 studies reported an increase in overall cost burden.¹² One retrospective analysis of Medicaid beneficiaries found that patients with bipolar disorder who were subject to formulary restrictions had 20% higher inpatient costs and 10% higher total direct health care costs than those who were not subject to formulary restrictions.⁶ Further, a 2013 study by Brown et al examined Medicaid data from 22 states and found that patients with bipolar disorder in states with more prior authorization requirements had worse medication continuity than those in states with fewer prior authorization requirements.¹⁰ Another study of Medicaid data from 30 states concluded that atypical antipsychotics may not be a suitable target for some cost containment strategies given their significant heterogeneity in clinical efficacy as a class.¹³ Overall, it appears that formulary restrictions on atypical antipsychotics typically reduce pharmacy costs but may shift costs to other medical services and negatively influence patient adherence. This is particularly concerning in patients with bipolar I disorder, who are known to have high rates of health care resource utilization (HCRU)² and medication nonadherence.^14,15

Cariprazine is an oral, dopamine D₃-preferring D₂/D₃ and serotonin 5-HT_1A receptor partial agonist approved to treat manic/mixed or depressive episodes associated with bipolar I disorder. Little is known about the effects of formulary-related cariprazine claim rejections in patients with bipolar I disorder. The goal of this study was to compare HCRU and describe treatment patterns among patients diagnosed with bipolar I disorder whose first pharmacy claim for cariprazine was either approved or rejected for a formulary-related reason.

Methods

STUDY DESIGN AND DATA SOURCE

This was a retrospective claims-based study conducted using data obtained from Symphony Health, an ICON plc Company, Integrated Dataverse (IDV) from March 1, 2015, to October 31, 2020 (Figure 1). Symphony Health IDV is a nationally representative claims database that covers about 280 million lives annually and comprises medical, hospital, and pharmacy claims data, including claims submitted to all payer types, such as commercial plans, Medicare Part D, cash assistance programs, and Medicaid.¹⁶ At the time of the analysis, the database captured approximately 70% of retail and specialty pharmacy claims and approximately 60% of mail orders and provided pharmacy data on approved, reversed, and rejected claims, as well the reason for rejections (Supplementary Figure 1 ^{(435.7KB, pdf)} , available in online article). Reasons for rejections were provided according to the National Council for Prescription Drug Programs (NCPDP) standard. Furthermore, socioeconomic data, such as information on race or ethnicity, annual household income, and education level, were available for some patients. Symphony Health IDV data were deidentified and complied with the Health Insurance Portability and Accountability Act (HIPAA) of 1996; therefore, no institutional review board approval was required.

FIGURE 1.

Study Design

Study Sample. Adults (aged ≥18 years) with a bipolar I disorder diagnosis (Supplementary Table 1 ^{(435.7KB, pdf)} ) and at least 1 pharmacy claim (rejected or approved) for cariprazine were included in the analysis. The first pharmacy claim for cariprazine was defined as the index date. Patients were required to be insured with a commercial health plan on the index date and have at least 12 months of continuous clinical activity prior to the index date, classified as the baseline period. As Symphony Health IDV is a provider-based database, as opposed to an insurance database, eligibility files were not available. Therefore, in this study continuous clinical activity was defined as at least 1 pharmacy or medical claim per quarter and was used as a proxy for health plan enrollment. The follow-up (observation) period began on the index date and continued until the earliest date of either the end of continuous clinical activity or the end of data availability.

Patients were categorized into either the approved cohort or the rejected cohort based on the disposition of their initial (index) cariprazine claim. The approved cohort consisted of patients whose index cariprazine claim was approved or those with an initial rejection who had a subsequent approved cariprazine claim within 30 days of the initial rejection. The rejected cohort consisted of patients who had their index cariprazine claim rejected for a formulary-related reason, including noncoverage, prior authorization requirement, and/or step therapy requirement (Table 1), and did not have an approved cariprazine claim within 30 days of the initial rejection. Rejection codes and description in Symphony IDV were presented using NCPDP standards and comprised formulary- and contract-related reasons, nonformulary/missing information or system errors. Patients with an index cariprazine claim rejected for a non–formulary-related reason (eg, patient not covered) or an inconclusive reason (eg, duplicates) were excluded from the analysis (Supplementary Table 2 ^{(435.7KB, pdf)} ). Reversed initial cariprazine claims (eg, prescriptions that were sent to the pharmacy and approved by the payer/clearinghouse but never picked up by the patient) were not assessed.

TABLE 1.

Formulary-Related Rejection Reasons

Type of formulary restriction	Description of reason for rejection (NCPDP rejection code)
Formulary noncoverage	•Product not on formulary (MR) •Product/service not covered (070)
Prior authorization	•Prior authorization required (75) •Prior authorization denied (3Y) •Claim submitted does not match prior authorization (64) •Prior authorization type code submitted not covered (9T) •Product not covered non-participating manufacturer (AC)
Step therapy	•Step therapy, alternative drug therapy required prior to use of submitted product service ID (608)

ID = identification; NCPDP = National Council for Prescription Drug Programs.

Baseline Patient Demographics and Clinical Characteristics. The following patient demographics were collected at the index date: age, sex (male, female), race or ethnicity (Black, Hispanic, White, other/unknown), annual household income (<$30,000; $30,000-$39,000; $40,000-$49,000; $50,000-$74,999; $75,000-$99,999; >$100,000; unknown) and US geographical region (South, Midwest, Northeast, West, unknown). Additionally, comorbidities (major depressive disorder [MDD], select bipolar I disorder–related comorbidities, mental health [MH]–related comorbidities, Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition comorbidities, Elixhauser comorbidities,¹⁷ and Quan Charlson Comorbidity Index [Quan-CCI] score¹⁸), previous medication use, and all-cause and MH-related HCRU were clinical characteristics collected during the baseline period. Baseline demographic and clinical characteristics were limited to the information available in the dataset.

OUTCOMES

All-cause and MH-related HCRU was evaluated during follow-up, which spanned from the index date to the end of clinical activity or the end of data availability. HCRU was defined using claims data and included hospitalizations (identified as visits with an “inpatient” place of service, as categorized by Symphony IDV), ED visits (identified via noninpatient visits using procedure codes), and outpatient visits (identified as noninpatient and non-ED visits). MH-related HCRU was defined as visits with a claim containing a primary or secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)/ICD-10-CM diagnosis code for an MH-related condition (see Supplementary Table 3 ^{(435.7KB, pdf)} ). Treatment patterns for cariprazine and other atypical antipsychotics (aripiprazole, asenapine, clozapine, brexpiprazole, iloperidone, lurasidone, olanzapine, paliperidone, quetiapine, risperidone, and ziprasidone) were also evaluated. Cariprazine treatment patterns included treatment duration, number of dispensings, days of supply per dispensing, and time to first dispensing. Treatment duration was defined as the number of days from the first approved cariprazine dispensing to the end of the days supply of the last approved cariprazine dispensing prior to the end of follow-up. Atypical antipsychotic treatment patterns included the number and type of atypical antipsychotics used, time to first dispensing, number of rejected and reversed atypical antipsychotic claims, and atypical antipsychotics received after step therapy rejection (ie, any atypical antipsychotic, generic atypical antipsychotic, or branded atypical antipsychotic). Finally, the number of patients in the rejected cohort who did not receive any bipolar I disorder–related medication/therapy (atypical antipsychotic, mood stabilizer/anticonvulsant, serotonin and norepinephrine reuptake inhibitors/monoamine oxidase inhibitors, adjunctive selective serotonin reuptake inhibitors/bupropion, typical antipsychotics, clonazepam, armodafinil/modafinil, ketamine, select non-pharmacotherapies [ie, electroconvulsive therapy and transcranial magnetic stimulation]), and the number of patients who did not receive any atypical antipsychotics or mood stabilizers/anticonvulsants was also evaluated.

STATISTICAL ANALYSIS

Propensity score matching (1:2) was used to balance characteristics between patients in the rejected and approved cohorts. Variables used in the propensity score calculation included age at the index date, sex, race or ethnicity, annual household income, region, year of index date, insurance type, physician specialty, Quan-CCI score,¹⁸ baseline medication use, number of atypical antipsychotics used during the baseline period (0, 1, 2, or 3+), psychotherapy during the baseline period, psychiatric diagnostic evaluation, all-cause and MH-related HCRU during the baseline period, plan-paid pharmacy costs during the baseline period, and baseline comorbidities (eg, MDD, Elixhauser comorbidities,¹⁷ Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition comorbidities, select bipolar I disorder comorbidities [related to abnormal glucose levels, disorders of lipid metabolism, essential/primary hypertension, malaise/fatigue, nausea/vomiting, obesity/overweight, other muscle disorders, other functional intestinal disorders, sleep disorders, symptoms/signs of emotional state, and types 1 and 2 diabetes], and MH-related comorbidities) with a prevalence of 5% or higher. Baseline characteristics were compared between cohorts using standardized differences; a difference of 10% or more was considered an important difference between cohorts.¹⁹ HCRU outcomes were reported per patient-year and compared between cohorts using rate ratios with 95% CIs and P values calculated using nonparametric bootstrap procedures with 499 replications.²⁰ Treatment patterns were reported using descriptive statistics and included mean and SD for the continuous variables and frequencies and proportions for the categorical variables. The number of patients who received an atypical antipsychotic, the number of atypical antipsychotics used, and rejected and reversed atypical antipsychotic claims were compared between cohorts using standardized differences. Similar to baseline characteristics, a standardized difference of 10% or more was considered an important difference between cohorts.

Results

Out of the 17,418 total patients included in the analysis, 1,554 (8.9%) had their initial cariprazine claim rejected for a formulary-related reason and 15,864 (91.1%) had their initial cariprazine claim approved. Prior to matching, the average follow-up period for both cohorts was 1.6 years, the average age in both cohorts was 40 years, and 74.6% of patients in the rejected cohort were female and 71.6% of patients in the approved cohort were female. After matching, a total of 3,108 patients were included in the approved cohort (Supplementary Figure 2 ^{(435.7KB, pdf)} ). The average follow-up period for both cohorts remained 1.6 years, the average age of the matched cohorts remained 40 years, and 74.6% and 73.6% of patients were female in the rejected and approved cohorts, respectively. Other baseline demographic and clinical characteristics were balanced between matched cohorts, with no standardized differences greater than 10% on any of the baseline measures assessed (Table 2). Additionally, comorbidities were similar at baseline between matched cohorts (Supplementary Table 4 ^{(435.7KB, pdf)} ). The most common formulary-related rejection reasons were product/service not covered (46.5%) and prior authorization required (38.5%).

TABLE 2.

Baseline Demographic and Clinical Characteristics

Characteristic	Matched cohorts (1:2)		Std Diffa (%)
	First cariprazine claim rejected (n = 1,554)	First cariprazine claim approved (n = 3,108)
Observation period, years, mean ± SD	1.6 ± 1.2	1.6 ± 1.2	3.7
Age, years, mean ± SD	39.6 ± 12.7	39.9 ± 13.0	2.3
Sexb, n (%)
Female	1,159 (74.6)	2,288 (73.6)	2.2
Male	395 (25.4)	820 (26.4)	2.2
Race or ethnicity, n (%)
Black	111 (7.1)	226 (7.3)	0.5
Hispanic	86 (5.5)	161 (5.2)	1.6
White	912 (58.7)	1,839 (59.2)	1.0
Otherc/unknown	445 (28.6)	882 (28.4)	0.6
Annual household income, n (%)
<$30,000	302 (19.4)	586 (18.9)	1.5
$30,000-$39,999	122 (7.9)	244 (7.9)	0.0
$40,000-$49,999	124 (8.0)	219 (7.0)	3.5
$50,000-$74,999	167 (10.7)	357 (11.5)	2.4
$75,000-$99,999	169 (10.9)	363 (11.7)	2.5
≥$100,000	268 (17.2)	537 (17.3)	0.1
Unknown	402 (25.9)	802 (25.8)	0.1
Geographical region, n (%)
South	769 (49.5)	1,517 (48.8)	1.4
Midwest	405 (26.1)	853 (27.4)	3.1
Northeast	202 (13.0)	406 (13.1)	0.2
West	175 (11.3)	326 (10.5)	2.5
Unknown	3 (0.2)	6 (0.2)	0.0
Type of formulary-related rejection at index, n (%)
Product/service not covered	722 (46.5)	–	–
Prior authorization required	598 (38.5)	–	–
Product not on formulary	164 (10.6)	–	–
Step therapy requirement	69 (4.4)	–	–
Prior authorization denied	1 (0.1)	–	–
Quan-CCI, mean ± SD	0.36 ± 0.87	0.36 ± 0.85	0.7
Medication use, n (%)
Antidepressants	1,198 (77.1)	2,416 (77.7)	1.5
Mood stabilizers/anticonvulsants	1,113 (71.6)	2,230 (71.8)	0.3
Anxiolytics	1,106 (71.2)	2,217 (71.3)	0.4
Atypical antipsychotics	966 (62.2)	1,925 (61.9)	0.5
Psychostimulants	427 (27.5)	843 (27.1)	0.8
Thyroid hormones	217 (14.0)	421 (13.5)	1.2
Typical antipsychotics	78 (5.0)	156 (5.0)	0.0
All-cause HCRU, mean ± SD
Hospitalizations	0.50 ± 1.10	0.50 ± 1.23	0.1
ED visits	0.88 ± 2.38	0.83 ± 2.36	2.1
OP visits	14.47 ± 17.14	14.06 ± 14.37	2.6
MH-related HCRU, mean ± SD
Hospitalizations	0.37 ± 0.90	0.37 ± 0.98	0.4
ED visits	0.47 ± 1.46	0.44 ± 1.54	1.8
OP visits	8.17 ± 10.23	7.84 ± 9.32	3.4

^a A standardized difference of more than 10% was considered an important difference.

^b Information on patient sex was derived from Symphony Health Integrated Dataverse records.

^c The term “other” stands for all races and ethnicities other than Black, Hispanic, and White.

CCI = Charlson Comorbidity Index; ED = emergency department; HCRU = health care resource utilization; MH = mental health; OP = outpatient; Std Diff = standardized difference.

HCRU

Patients in both the approved and rejected cohorts were observed for a mean (SD) period of 1.6 (1.2) years (Table 2). Compared with the approved cohort, the rejected cohort experienced 22% more all-cause and 24% more MH-related hospitalizations per patient-year (rate ratio [95% CI], all-cause: 1.22 [1.01-1.48], P = 0.024; MH-related: 1.24 [1.01-1.55], P = 0.044) during follow-up (Table 3). Rates of ED visits and outpatient visits were numerically higher among rejected patients, albeit not significantly different between cohorts.

TABLE 3.

Health Care Resource Utilization During Follow-Up

HCRUa measure	Matched cohorts (1:2)		RR (95% CI)	P value
	First cariprazine claim rejected (n = 1,554)	First cariprazine claim approved (n = 3,108)
Total person-years	2,480	5,095	–	–
All-cause, rate, PPY
Hospitalizations	0.34	0.28	1.22 (1.01-1.48)	0.024
ED visits	0.62	0.55	1.12 (0.94-1.38)	0.253
OP visits	12.63	12.14	1.04 (0.96-1.15)	0.341
MH-related, rate, PPY
Hospitalizations	0.22	0.18	1.24 (1.01-1.55)	0.044
ED visits	0.32	0.26	1.21 (0.95-1.51)	0.096
OP visits	6.76	6.43	1.05 (0.94-1.18)	0.337

^a Hospitalizations were identified as visits with an inpatient place of service, as categorized by Symphony Integrated Dataverse; ED visits were identified via noninpatient visits using procedure codes; and OP visits were identified as noninpatient and non-ED visits.

ED = emergency department; HCRU = health care resource utilization; MH = mental health; OP = outpatient; PPY = per patient-year; RR = rate ratio.

Treatment Patterns. Among patients with an initial cariprazine rejection, 76.8% never received cariprazine, 34.7% never received any atypical antipsychotic during follow-up, 16.3% never received any mood stabilizer/anticonvulsant, and 13.6% never received any bipolar I–related medication (Table 4). The average treatment delay for patients who did receive cariprazine after the initial rejection was 188 days (6 months) following the first rejection. Patients in the rejected cohort who subsequently received cariprazine had a similar mean treatment duration to those whose first cariprazine claim was approved (254 vs 249 days). More than half (55.2%) of the patients in the rejected cohort subsequently received an atypical antipsychotic other than cariprazine, with an average treatment delay of 123 days (4 months) following the first cariprazine rejection. The most common atypical antipsychotics received after an initial cariprazine rejection were quetiapine, aripiprazole, olanzapine, and lurasidone (Supplementary Table 5 ^{(435.7KB, pdf)} ). Of patients in the rejected cohort who received an atypical antipsychotic, including cariprazine during follow-up (n = 1,015), 157 (15.5%) only received cariprazine, 654 (64.4%) only received an atypical antipsychotic other than cariprazine, and 204 (20.1%) received both cariprazine and another atypical antipsychotic (Supplementary Figure 3 ^{(435.7KB, pdf)} ). During follow-up, patients in the rejected cohort used a greater mean number of different atypical antipsychotics, excluding cariprazine, than patients in the approved cohort (0.86 vs 0.72, standardized difference = 14.0%) (Table 4). Differences between cohorts in the mean number of different atypical antipsychotics were driven mainly by higher use of aripiprazole, ziprasidone, and risperidone among the rejected cohort (Supplementary Table 5 ^{(435.7KB, pdf)} ).

TABLE 4.

Treatment Patterns During Follow-Up

Treatment pattern	Matched cohorts (1:2)		Std Diff (%)
	First cariprazine claim rejected (n = 1,554)	First cariprazine claim approved (n = 3,108)
Summary of treatment patterns
Patients who received an atypical antipsychotic (including cariprazine)a, n (%)	1,015 (65.3)	3,108 (100.0)	–
Cariprazine use among approved cohort
Cariprazine treatment duration, days, mean ± SD	–	249 ± 311	–
Number of dispensings, mean ± SD	–	5.1 ± 5.8	–
Days of supply per dispensing, mean ± SD	–	32.6 ± 11.8	–
Cariprazine use among rejected cohort
Patients who did not receive any atypical antipsychotic, n (%)	539 (34.7)	–	–
Patients who did not receive any atypical antipsychotic or mood stabilizer/anticonvulsant, n (%)	254 (16.3)	–	–
Patients who did not receive any bipolar I–related medication/therapyb, n (%)	211 (13.6)	–	–
Patients who subsequently received cariprazinec, n (%)	361 (23.2)	–	–
Time to first dispensing, days, mean ± SD	188 ± 227	–	–
Cariprazine treatment duration, days, mean ± SD	254 ± 285	–	–
Number of dispensings, mean ± SD	5.3 ± 5.9	–	–
Days of supply per dispensing, mean ± SD	33.1 ± 10.4	–	–
Atypical antipsychotic use among rejected cohort
Patients who subsequently received an atypical antipsychotic (excluding cariprazine), n (%)	858 (55.2)	–	–
Time to first dispensing, days, mean ± SD	123 ± 187	–	–
Atypical antipsychotic patterns
Number of atypical antipsychotics used (excluding cariprazine), mean ± SD	0.86 ± 1.00	0.72 ± 0.98	14.0
Rejected atypical antipsychotic claims
Number of rejected atypical antipsychotic claims (including cariprazine), mean ± SD	1.57 ± 0.99	0.35 ± 0.82	133.7
Number of rejected atypical antipsychotic claims (excluding cariprazine), mean ± SD	0.25 ± 0.68	0.14 ± 0.51	19.4
Number of rejected cariprazine claims, mean ± SD	1.32 ± 0.73	0.22 ± 0.60	164.1
Reversed (abandoned) atypical antipsychotic claims
Patients with ≥1 reversed atypical antipsychotic claim (including cariprazine), n (%)	213 (13.7)	447 (14.4)	1.9
Number of reversed atypical antipsychotic claims (including cariprazine), mean ± SD	0.22 ± 0.72	0.23 ± 0.69	1.1
Patients with ≥1 reversed atypical antipsychotic claim (excluding cariprazine), n (%)	151 (9.7)	197 (6.3)	12.4
Number of reversed atypical antipsychotic claims (excluding cariprazine), mean ± SD	0.15 ± 0.58	0.09 ± 0.43	11.8
Patients with ≥1 reversed cariprazine claim, n (%)	76 (4.9)	308 (9.9)	19.2
Number of reversed cariprazine claims, mean ± SD	0.07 ± 0.37	0.14 ± 0.49	15.7

^a Includes 157 patients and 1,641 patients who only received cariprazine in the rejected and approved cohort, respectively.

^b Bipolar I disorder–related therapy included any type of atypical antipsychotic, mood stabilizer/anticonvulsant, SNRI/MAOI, adjunctive SSRI/bupropion, typical antipsychotic, clonazepam, armodafinil/modafinil, ketamine, and select nonpharmacotherapies (ie, ECT and TMS).

^c Includes 204 patients who received both cariprazine and another atypical antipsychotic.

ECT = electroconvulsive therapy; MAOI = monoamine oxidase inhibitor; SNRI = serotonin-norepinephrine reuptake inhibitor; SSRI = selective serotonin reuptake inhibitor; Std Diff = standardized difference; TMS = transcranial magnetic stimulation.

Discussion

To our knowledge, this is the first study to evaluate the impact of formulary-related rejections of cariprazine on HCRU and treatment patterns in patients with bipolar I disorder. Patients whose first cariprazine claim was rejected for a formulary-related reason subsequently experienced significantly higher rates of all-cause and MH-related hospitalizations than those whose first cariprazine claim was approved, although there was no difference in ED or outpatient use. After a formulary-related cariprazine rejection, more than one-third of patients did not receive any atypical antipsychotics during an average follow-up of 1.6 years, suggesting that formulary restrictions may discourage medication use beyond the targeted medication. These findings indicate that there may be downstream consequences of formulary-related rejections of cariprazine in patients with bipolar I disorder.

The goal of formulary restrictions is to encourage appropriate medication use and minimize pharmacy costs without compromising the quality of patient care.⁸ However, there is mixed evidence in the literature evaluating the effects of formulary restrictions on atypical antipsychotics. For example, although some studies have suggested formulary restrictions result in reduced pharmacy expenditures, others have found a negative correlation among formulary restrictions and medication adherence, HCRU, and total medical costs.^6,12 Particularly concerning can be the decreased adherence rates seen in insurance programs enacting formulary restrictions, as persistence and continuation of therapy are ongoing challenges for patients with serious mental illness.^15,21 Contributing to decreased adherence rates is the heterogeneity of atypical antipsychotics and the variation from patient to patient in the efficacy and tolerability of these agents.^6,22 Because of this, many studies suggest that patients with bipolar I disorder may benefit from access to more diverse treatment options to optimize efficacy and decrease adverse events.^5,6,22 Our study results showed that, compared with patients whose first cariprazine claim was approved, patients with formulary-related cariprazine rejections experienced significantly higher rates of hospitalizations, one of the costliest resources in bipolar I disorder care that also increases patient burden.²³ Our finding of significantly increased hospitalization rates associated with formulary restrictions aligns with prior research from Seabury et al, who found that, even after adjusting for potential confounders, patients with bipolar disorder who were subject to formulary restrictions had a significantly greater risk of hospitalization compared with those who were not subject to formulary restrictions (odds ratio = 1.07; P = 0.016).⁶ Further, the authors noted higher inpatient costs and overall costs among patients who were subject to formulary restrictions.⁶ Overall, these results suggest that formulary restrictions on bipolar I disorder medications may unintentionally lead to increased hospitalization rates.

In the rejected cohort in our study population, about 55% of patients received an atypical antipsychotic other than cariprazine after the initial cariprazine claim was rejected. The average time from the cariprazine rejection to the first dispensing of another atypical antipsychotic was approximately 4 months (123 days). For patients who subsequently received cariprazine after an initial rejection, the average time to the first cariprazine dispensing was approximately 6 months (188 days). During these delays, patients could have been at risk for both worsening of their bipolar I disorder symptoms and potentially avoidable HCRU. Furthermore, although most patients in the rejected cohort went on to receive other bipolar I disorder medications, patients whose first cariprazine claim was rejected still had higher rates of hospitalizations after the rejection than those whose first cariprazine claim was approved. This pattern suggests that treatment delays may be an important risk factor for hospitalizations among patients with bipolar I disorder, with the implication that formulary restrictions may have negative spillover effects and that third-party payers face a complicated task in assessing the overall risk-to-benefit ratio prior to implementing formulary restrictions on pharmacologic therapies for patients with bipolar I disorder, including cariprazine.

Patients with bipolar I disorder commonly present with complex clinical profiles, and the real-world efficacy of pharmacologic therapeutic options can vary; these are both factors that complicate the management of their condition. In a bipolar disorder treatment patterns analysis, more than 2,000 unique regimens were identified during the first line of therapy for any type of initial bipolar episode, highlighting extensive heterogeneity in symptoms and prescribing strategies.²⁴ Patient- and disease-specific factors may contribute to the complexity of effectively treating bipolar I disorder. For example, bipolar I disorder is associated with high levels of functional disability,²⁵ which may limit the ability of patients to successfully navigate the health care system and obtain treatment. Additional barriers, such as the inability to obtain the specific medication prescribed by a provider because of coverage, cost, or administrative issues, may be discouraging to patients seeking treatment and limit their ability to effectively improve their symptoms and adequately manage their condition. Evidence of this may be seen in our study, as nearly 35% of patients never received any atypical antipsychotic following a formulary-related rejection of cariprazine. Given the challenges of managing bipolar I disorder and the complexity of this patient population, clinicians prescribing cariprazine may consider patient-specific factors, such as risk for adverse events and patient preference, for which formulary restrictions cannot account.

LIMITATIONS

Symphony Health IDV was selected for this analysis because it provided information on rejections reasons, as well as used pharmacy claims data to provide valuable information on medication patterns and physician choice. Additionally, propensity score matching allowed for adjustment for several sociodemographic variables (eg, annual household income, geographic region) that could potentially have an impact on formulary restrictions but are not commonly available in claims databases. However, there are inherent limitations with the use of claims data, and the findings of this study should be interpreted within the context of its limitations. First, pharmacy dispensing data on insurance claim denials and reasons for rejection available in Symphony Health IDV data, via the NCPDP standard, were used to identify formulary-related rejections. These data were used as a surrogate for health plan formulary data, which are not publicly available. The lack of eligibility data in Symphony Health IDV also may have limited the complete medical and pharmacy claims from being captured, as patients could have received their medication from a location not captured in the dataset (eg, a free clinic/pharmacy) or changed insurances. Furthermore, as Symphony Health IDV is a provider-based claims database, patients also may have been counted as multiple patients if they switched providers or were treated at multiple locations, although a linking algorithm was used to minimize this possibility. Additionally, as with all claims data, there was the potential for data entry/coding errors, including missed or misclassification of medication use. Moreover, data were limited to whether patients obtained prescriptions from their pharmacy—it was unknown whether patients took their medications or took them as prescribed. As there is evidence suggesting formulary restrictions may negatively impact medication adherence or lead to an increased risk of patients experiencing adverse events,^6,8,11 the lack of data on adherence patterns and patient-reported adverse events are limitations of this study. Additionally, only patients with commercial insurance were included in this analysis, which limits generalizability to patients with other insurance types (eg, Medicare or Medicaid) and to those who are uninsured or may have limited access to health care. Lastly, costs were unable to be captured in this study because of limitations of the dataset.

Conclusions

Among patients with bipolar I disorder, those whose initial cariprazine claim was rejected for a formulary-related reason experienced significantly higher rates of all-cause and MH-related hospitalizations than those whose initial cariprazine claim was approved, although there were no significant differences in ED or outpatient visits. More than 75% of patients in the rejected cohort never received cariprazine during follow-up and more than one-third never received any atypical antipsychotic. The results of this study suggest that cariprazine rejections due to formulary restrictions may have unintended consequences in bipolar I disorder and lead to higher rates of HCRU.

Funding Statement

This study was sponsored by AbbVie. Mr Laliberté, Dr Zanardo, Mr MacKnight, and Ms Urosevic are employees of Analysis Group, Inc., which received funding from Allergan (prior to its acquisition by AbbVie) to conduct this analysis.