The incidence and economic burden of extrapyramidal symptoms in patients with schizophrenia treated with second generation antipsychotics in a Medicaid population

Abstract

Background

Extrapyramidal symptoms (EPS) are common side-effects of second-generation antipsychotics (SGA), that can negatively impact patient quality-of-life, and are associated with increased morbidity and mortality. This study examined the incidence and burden of EPS in patients with schizophrenia initiating SGAs.

Methods

Patients with schizophrenia initiating SGAs were identified in the MarketScan Multi-state Medicaid database from 1 January 2012 to 31 December 2018. Incidence of EPS (identified via ICD-9/ICD-10 diagnoses and medications) was assessed during the 12-months following SGA initiation. Cohorts with and without EPS were defined. Multivariate models were used to examine all-cause and schizophrenia-related hospitalizations and total costs in the 12 months following the first EPS claim (EPS) or randomly assigned index date (Non-EPS) while controlling for multiple baseline covariates. Logistic regression was used for hospitalization and two-part models were used for skewed cost data.

Results

A total of 11,642 patients with schizophrenia filled a prescription for an SGA; of which, 2,468 (21.2%) experienced EPS in the first year. The age- and gender-matched EPS group and non-EPS group included 2,295 and 5,607 patients, respectively. Multivariate analyses confirmed that EPS patients had 25% higher odds of all-cause (OR = 1.25; 95% CI = 1.11–1.40) and 75% increased odds of schizophrenia-related (OR = 1.75; 95% CI = 1.53–2.00) inpatient admissions compared to non-EPS patients. Multivariate adjustment of post-period costs between groups also found significant differences in both all-cause (EPS: $27,408 vs. non-EPS: $22,489, p < 0.001) and schizophrenia-related (EPS:$12,833 vs. non-EPS:$8,077, p < 0.0001) costs between the EPS and non-EPS cohorts.

Conclusions

Over one-fifth of patients initiating treatment with atypical antipsychotics in this study developed EPS in the 12 months following SGA initiation. Extrapyramidal side-effects associated with atypical antipsychotics increase the risk of hospitalization and contribute to higher healthcare costs. For patients with schizophrenia, treatment options that minimize the risk of EPS may reduce the economic burden associated with the disease.

Keywords:

• Schizophrenia
• atypical antipsychotic
• extrapyramidal symptoms
• healthcare utilization
• healthcare costs

JEL CLASSIFICATION CODES:

• C55
• C5
• C
• C12
• C1

Introduction

Schizophrenia is a serious, chronic disorder characterized by disturbances in perception, thought processes, social relations, and emotional expression, with a mean lifetime prevalence estimate of 1.1%¹. While the prevalence of schizophrenia is relatively low, it is among the top 15 leading causes of disability worldwide² and is also associated with significant direct and indirect costs that have increased dramatically over the past two decades. In 2002, the total costs attributed to schizophrenia in the United States were estimated at $62.7 billion, with $22.7 billion in direct healthcare costs³. In 2013, these same costs were estimated at $155.7 billion and $37.7 billion, respectively⁴. In the United States, two-thirds (67%) of individuals with schizophrenia are covered by Medicaid⁵.

Antipsychotic medications are the mainstay for acute and chronic pharmacologic treatment of patients with schizophrenia^6,7. Antipsychotics are mainly dopamine D2 and serotonin 5-HT2A antagonists and are associated with varying levels of side-effects, often relating to their D2 antagonistic effect, such as extrapyramidal symptoms (EPS) including acute dystonia, akathisia, and parkinsonism^6,8. Medications such as benztropine, trihexyphenidyl, amantadine, and biperiden have been used to treat EPS⁶. While not common, following sustained exposure to antipsychotic medications, the abnormal involuntary movement disorders can become persistent, a condition called tardive dyskinesia⁶. Medications such as deutetrabenazine, valbenazine, and tetrabenazine are used to treat tardive dyskinesia⁶.

While second generation antipsychotics (SGAs) have a lower EPS burden in general compared to the first generation or “typical” antipsychotics, they have not met initial expectations of being completely EPS-free medications^9–11. In fact, varying degrees of EPS have been reported in up to 37% of all patients taking SGAs^12–14. According to the latest version of the American Psychiatric Association treatment guidelines for schizophrenia, the risk of EPS among the SGAs is lowest for clozapine, iloperidone, and quetiapine, followed by aripiprazole, brexpiprazole, cariprazine, and ziprasidone, then olanzapine and asenapine, and finally lurasidone, paliperidone, and risperidone⁶. Extrapyramidal symptoms can negatively impact quality-of-life for patients and are associated with increased morbidity and mortality^13,15. Significant unmet needs remain to develop medications that provide efficacy in treating the spectrum of symptoms of schizophrenia while reducing side-effect burden, including that of EPS.

Side-effects of antipsychotic medications are significant, burdensome, and are associated with nonadherence to medication among patients with schizophrenia^16,17. A survey of adults with schizophrenia found 86% experienced at least one side-effect and only 43% were completely adherent to their medication¹⁷. Moreover, side-effect clusters including EPS/agitation, metabolic, sedation/cognition, and prolactin/endocrine were all significantly associated with decreased adherence. Medication adherence is of particular relevance to the economic burden of schizophrenia. Dibonaventura et al. found that patients reporting complete medication adherence were less likely to report being hospitalized for a mental or non-mental health reason or have an emergency room visit for mental health reason¹⁷. Medication non-adherence is also associated with greater risk of relapse. Relapse has been shown to be associated with higher inpatient, outpatient, and medication costs¹⁸.

While a previous study found that EPS was associated with increased utilization of healthcare resources and higher medical costs¹⁹, updated analyses based on more recent data is needed. The objectives of this study were to assess the 1-year incidence of EPS among schizophrenia patients after initiating a second generation antipsychotic and to assess the annual healthcare burden for patients who developed EPS compared to patients who did not develop EPS.

Methods

Dataset

A retrospective cohort study was designed using data from Marketscan Medicaid Multi-State Database from 1 January 2012 through 31 December 2018. This database includes the medical, surgical, and prescription drug experience of more than 48 million enrollees since 1999 from between five to 13 states as well as records of inpatient services and admissions, outpatient services, and prescription drugs²⁰. The analysis used de-identified data per the Health Insurance Portability and Accountability Act of 1996²¹; therefore, institutional review board approval was not required.

Patient selection

Incidence analysis cohort

Adult patients were selected if they filled a prescription for a SGA (aripiprazole, asenapine, brexpiprazole, cariprazine, clozapine, iloperidone, lurasidone, olanzapine, paliperidone, quetiapine, risperidone, or ziprasidone) between 1 January 2012 and 31 December 2018 (Figures 1 and 2). The index date for the incidence analysis was the first prescription for an SGA. Patients were included if they had medical, pharmacy, and mental health/substance abuse (MHSA) coverage for ≥6 months prior to and 12 months after the first filled SGA prescription and at least ≥1 inpatient or non-diagnostic outpatient claim for schizophrenia (ICD-9-CM:295x; ICD-10-CM; F20.x) in the 6 months prior to SGA initiation. To ensure patients were incident users of SGAs, patients with evidence of SGA use in the pre-period were excluded. Patients were excluded if they had evidence of EPS (EPS diagnosis [ICD-9-CM codes: 332.1, 333.1, 333.5, 333.72, 333.79, 333.81–333.83, 333.85, 333.89, 333.90, 333.99, 781.0; ICD-10-CM codes: G21.1x, G21.2, G21.8, G24.01, G24.02, G24.2–G24.9, G25.1–G25.4, G25.61, G25.7x, G25.89, G25.9, R25.xx] or a prescription filled for EPS or tardive dyskinesia medication [benztropine, trihexyphenidyl, amantadine, biperiden, deutetrabenazine, valbenazine, and tetrabenazine]) in the 6 months prior to SGA initiation. Additionally, patients with a fill or administration for a typical antipsychotic (chlorpromazine, fluphenazine, haloperidol, loxapine, perphenazine, pimozide, prochlorperazine, thioridazine, thiothixene, or trifluoperazine) or a claim with a diagnosis for Parkinson’s disease at any time were excluded.

Figure 1. Study schematic. EPS, extrapyramidal symptoms.

Figure 2. Patient selection. AAP, atypical antipsychotic; EPS, extrapyramidal symptoms.

Healthcare resource utilization and cost analysis cohort

The patients included in the healthcare resource use and cost analysis were a subset of the cohort from the incidence analysis.

EPS group

Patients were included in the EPS group if they had at least one EPS event in the twelve months following SGA initiation and had continuous medical, pharmacy, and MHSA eligibility for 6 months prior to and 12 months after the first EPS event. The first EPS event defined the index date.

Non-EPS group

Patients were included in the non-EPS group if they did not have any EPS events in the 12 months following SGA initiation. The index date for the non-EPS group was randomly assigned. The non-EPS group had continuous medical, pharmacy, and MHSA eligibility for 6 months prior to and 12 months after the randomly assigned index date. Patients who experienced an EPS event in the 12 months following their random non-EPS index date were also excluded from the EPS cohort due to their delayed development (>12 months following SGA initiation) of EPS. Patients in the non-EPS group were only included if they matched to a patient in the EPS group based on age and gender using a 1:1 to 1:3 ratio^22,23.

Variable definitions

Incidence analysis

For the incidence analysis, patients were followed from the index date (first SGA claim) until the earliest of the first EPS diagnosis, first EPS medication use, or the end of the 1-year follow-up. The rate per 100 patient years was calculated by dividing the number of patients experiencing EPS by the total number of eligible patient years. The incidence of EPS was reported as the percentage of patients who evidenced EPS over the 1-year follow-up as well as the rate of EPS per 100 person years.

Healthcare resource utilization and costs

Demographics were recorded on the EPS date for the EPS group and on the randomized index date for the non-EPS group. Baseline clinical characteristics were captured during the 6-month pre-period.

Second generation antipsychotic use and SGA days’ supply were assessed based on prescription claims during the 6-month pre-period or on the index date with days’ supply being truncated on index date. Healthcare resource utilization and costs were examined for the 6-month pre-period and 12-month post-period. Costs for all-cause and schizophrenia-related resource utilization were calculated for inpatient admissions, outpatient visits (emergency room visits, office visits, and other services), and outpatient pharmacy fills. Schizophrenia-related healthcare resource utilization and costs were based on claims with a diagnosis of schizophrenia or outpatient pharmacy claims for SGAs or EPS medications. Costs were calculated based on adjudicated claims and included payments made by insurers and patients. Costs for capitated plans were estimated by means of payment proxy. To ensure cost measures across years were comparable, all costs were adjusted for inflation using the Consumer Price Index and standardized to 2018 US dollars²⁴.

Statistical methods

Demographic and baseline clinical characteristics between the EPS and non-EPS groups were descriptively compared and differences between the EPS and non-EPS groups were analyzed using Chi-squared or Fisher’s Exact tests for dichotomous or categorical variables and t-tests or ANOVA for continuous variables.

Multivariate analysis

Risk of all-cause and schizophrenia-related hospitalizations was assessed via logistic regression. The cost variables were highly skewed; therefore, cost outcomes were assessed using two-part models. In part one, a firth logistic regression model compared the odds of having or not-having costs in the post period while controlling for baseline variables. All of the patients with EPS had some all-cause costs. Firth logistic regression uses a penalized maximum likelihood used to generate parameter estimates and standard errors, which allows for more stable estimates and model convergence when perfect separation occurs. In part two, gamma generalized linear models with log link were used to compare costs between the EPS and non-EPS groups for patients with some costs, while controlling for baseline variables. Covariate-adjusted costs were calculated using the recycled prediction method. Covariates for all models were measured during the pre-period or on index and included age, sex, region of residence, plan type Charlson comorbidity index (CCI)²⁵, SGA days’ supply, diagnoses of depression, bipolar disorder, anxiety, or alcohol/substance abuse, inpatient admission or ER visit, and total healthcare costs. Inpatient admission or ER visit was coded as an indicator variable and included as a proxy for severity of symptoms. WPS version 4.1 (World Programming, United Kingdom) was used to complete all analyses with an alpha level set at p < 0.05.

Results

Incidence analysis

Of 775,977 patients who filled a prescription for an SGA between 1 January 2012 and 31 December 2018, a total of 11,642 patients with schizophrenia newly-initiating SGAs for the first time qualified for the EPS incidence analysis (Figure 2). Of these patients, 2,468 (21.2%) experienced EPS, corresponding to an incidence rate of 25.5 per 100 person-years. Mean time to EPS within the incidence cohort was 76.5 (SD ±105.8) days.

Descriptive analyses

Demographic and clinical characteristics

The final EPS group in the healthcare resource utilization and cost analysis included 2,295 patients who were age- and gender-matched to 5,607 patients comprising the non-EPS group. The baseline demographic and pre-index clinical characteristics were overall similar for the EPS and non-EPS groups (Table 1). The EPS group was younger with a slightly higher proportion of males. The majority of patients in both groups were black. The non-EPS group had a higher physical comorbidity burden in the pre-period as evidenced by significantly higher Charlson Comorbidity Index scores. The non-EPS group had a higher proportion of individuals with cardiovascular disease. The EPS group was more likely to evidence pre-period psychiatric comorbidity, with significantly higher rates of anxiety and personality disorders. The most commonly filled comedications during the pre-period were antidepressants, anticonvulsants, and anxiolytics; a significantly greater proportion of non-EPS patients had a fill for all three commonly filled comedications (p < 0.0001). Nearly all of the EPS patients (90.2%, n = 2,070) had an index date based on a prescription for an EPS medication rather than a diagnostic code. EPS patients were prescribed benztropine (87.8%), trihexyphenidyl (5.1%), or amantadine (0.6%); there were no claims for the other EPS or tardive dyskinesia medications.

Table 1. Demographic and pre-index clinical characteristics.

Characteristic	EPS group (n = 2,295)	Non-EPS group (n = 5,607)
Demographics
Age, years, mean (SD)*	38.2 (13.0)	39.5 (12.7)
Male, n (%)*	1,397 (60.9)	3,189 (56.9)
Race, n (%)
White	663 (28.9)	1,747 (31.2)
Black	1,186 (51.7)	2,838 (50.6)
Hispanic	39 (1.7)	83. (1.5)
Missing	407 (17.7)	939 (16.7)
Insurance plan type, n (%)*
Comprehensive	1,566 (68.2)	3,428 (61.1)
HMO	711 (31.0)	2,138 (38.1)
PPO	1 (0.0)	6 (0.1)
POS with capitation	17 (0.7)	35 (0.6)
Clinical characteristics
Charlson Comorbidity Index, mean (SD)*	0.6 (1.2)	0.7 (1.5)
Physical Comorbidities, n (%)
Diabetes	275 (12.0)	759 (13.5)
Cardiovascular disease*	404 (17.6)	1,116 (19.9)
Hypertension	688 (30.0)	1,806 (32.2)
Hyperlipidemia	295 (12.9)	806 (14.4)
Obesity	198 (8.6)	521 (9.3)
Psychiatric Comorbidities, n (%)
Bipolar disorder	781 (34.0)	1,927 (34.4)
Anxiety*	284 (12.4)	541 (9.6)
Depression	789 (34.4)	1,919 (34.2)
Personality disorder*	167 (7.3)	335 (6.0)
Alcohol/substance abuse	1,169 (50.9)	2,750 (49.0)
Dementia	56 (2.4)	138 (2.5)
AAP fills in the pre-period or on index, n (%)*	2,262 (98.6)	5027 (89.7)
AAP days’ supply in pre-period or on index, mean (SD)*	38.9(64.9)	43.8(62.4)
Comedication use, n (%)
Antidepressants*	738 (32.2)	2,416 (43.1)
Anticonvulsants*	558 (24.3)	1,663 (29.7)
Anxiolytics*	420 (18.3)	1,218 (21.7)
Mood stabilizers	117 (5.1)	281 (5.0)
Healthcare resource utilization and costs
Inpatient admissions, n (%)*	1,064 (46.4)	1,792 (32.0)
Emergency room visits, n (%)	1,210 (52.7)	2,924 (52.1)
Total healthcare costs, mean (SD)*	$16,860 (33,571)	$13,425 (32,589)
Inpatient*	$12,186 (32,405)	$8,466 (30,235)
Outpatient	$3,672 (5,692)	$3,712 (6,043)
Outpatient pharmacy	$1,002 (3,644)	$1,246 (6,442)

Abbreviations. AAP, atypical antipsychotic; SD, standard deviation; HMO, health maintenance organization; PPO, preferred provider organization; POS, point of service.

Demographics were measured on index. Clinical characteristics were measured in the 6-month pre-index period.

*p < 0.05.

A greater proportion of EPS patients used SGAs during the pre-period or on the index date (98.6 vs. 89.7%, p < 0.0001). However, among SGA users, non-EPS patients exhibited a higher mean days’ supply at 43.8 (±62.4) vs. 38.9 (±64.9) days, p < 0.001 (Table 1). The majority of patients in both the cohorts had a supply of 1–30 days in the pre-period or on the index date. The SGAs used on the index date have been added to Supplementary Table S1. The EPS rate by specific SGA was not included because these rates were not corrected for selection bias and could be misinterpreted as risk.

Pre-period healthcare resource utilization and costs

A significantly greater proportion of EPS patients had ≥1 all-cause hospitalization (46.4 vs. 32.0%, p < 0.0001) compared to non-EPS patients during the 6-month pre-period (Table 1). In contrast, the non-EPS cohort had a greater proportion of patients having ≥1 office visit (70.8 vs. 61.2%, p < 0.01) or outpatient pharmacy fill (87.2 vs. 72.3%, p < 0.0001). Patterns in schizophrenia-related healthcare service utilization were similar, with a greater proportion of EPS patients having ≥1 schizophrenia related hospitalization (39.9 vs. 24.5%, p < 0.0001) or ER visit (25.8 vs. 19.2%, p < 0.0001) and a greater proportion of non-EPS patients with outpatient pharmacy fill (60.7 vs. 51.6%, p < 0.0001). All-cause total pre-period healthcare costs were higher in the EPS cohort at $16,860 ± $33,571 compared to $13,425 ± $32,589 in the non-EPS cohort, p < 0.05. Consistent with the increased rate of hospitalization in the EPS cohort, inpatient costs accounted for 72% of their total pre-period costs compared to 63% in the non-EPS cohort (Table 1).

Post-period healthcare utilization and costs

All-cause

Prior to controlling for baseline characteristics, the EPS group had significantly higher rates of all-cause (30.2 vs. 24.6%, p < 0.0001) hospitalizations during the 12-month post-index period when compared to the non-EPS group (Table 2). The EPS group also had significantly higher rates of all-cause outpatient visits (84.7 vs. 82.8%, p = 0.0486), other outpatient services (98.7 vs. 95.0%, p < 0.0001), and outpatient pharmacy fills (99.8 vs. 94.2%, p < 0.0001) compared to the non-EPS group. In addition, the EPS group had significantly greater all-cause outpatient ($10,082 vs. $7,827, p < 0.0001), outpatient pharmacy ($5,716 vs. $4,086, p < 0.0001), and total costs ($25,911 vs. $21,550, p = 0.0004) on average during the post-index period compared to the non-EPS group.

Table 2. Healthcare resource utilization and costs during the 12-month post-index period.

Characteristic	EPS group (n = 2,295)	Non-EPS group (n = 5,607)	p-value
All-cause healthcare utilization
Inpatient
Patients with an admission, n (%)	693 (30.2)	1,380 (24.6)	<0.0001*
Number of inpatient admissions, mean (SD)^†	2.2 (2.4)	1.9 (2.0)	0.0034*
Outpatient
ER visits
Patients with an ER visit, n (%)	1,333 (58.1)	3,268 (58.3)	0.8690
Number of ER visits, mean (SD)^†	4.0 (5.4)	4.0 (7.1)	0.7652
Office visits
Patients with a visit, n (%)	1,943 (84.7)	4,645 (82.8)	0.0486*
Number of visits, mean (SD)^†	7.5 (10.2)	7.2 (6.7)	0.2967
Other outpatient services
Patients with service, n (%)	2,265 (98.7)	5,329 (95.0)	<0.0001*
Number of services, mean (SD)^†	53.5 (66.6)	42.8 (65.4)	<0.0001*
Outpatient pharmacy
Patients with a prescription, n (%)	2,290 (99.8)	5,282 (94.2)	<0.0001*
Number of prescriptions, mean (SD)^†	52.3 (48.4)	43.7 (47.1)	<0.0001*
All-cause healthcare costs
Inpatient costs, mean (SD)	$10,113 (39,604)	$9,636 (44,953)	0.6584
Outpatient costs, mean (SD)	$10,082 (16,058)	$7,827 (12,764)	<0.0001*
ER visit costs, mean (SD)	$915 (2,623)	$898 (2,920)	0.8093
Office visit costs, mean (SD)	$540 (1,576)	$472 (698)	0.0087*
Other outpatient services costs, mean (SD)	$8,628 (15,229)	$6,457 (11,919)	<0.0001*
Outpatient pharmacy costs, mean (SD)	$5,716 (14,122)	$4,086 (13,216)	<0.0001*
Total medical costs (inpatient + outpatient), mean (SD)	$20,195 (47,300)	$17,464 (48,350)	0.0191
Total costs (medical + pharmacy), mean (SD)	$25,911 (47,300)	$21,550 (50,794)	0.0004*
Schizophrenia-related healthcare utilization
Inpatient
Patients with an admission, n (%)	516 (22.5)	724 (12.9)	<0.0001*
Number of inpatient admissions, mean (SD)^†	2.0 (2.1)	1.6 (1.5)	<0.0001*
Outpatient
ER visits
Patients with an ER visit, n (%)	585 (25.5)	934 (16.7)	<0.0001*
Number of ER visits, mean (SD)^†	2.2 (2.5)	2.1 (2.8)	0.2967
Office visits
Patients with a visit, n (%)	1,055 (46.0)	1,961 (35.0)	<0.0001*
Number of visits, mean (SD)^†	4.3 (11.6)	3.2 (4.0)	<0.0001*
Other outpatient services
Patients with service, n (%)	1,702 (74.2)	3,055 (54.5)	<0.0001*
Number of services, mean (SD)^†	34.6 (53.2)	23.6 (45.1)	<0.0001*
Patients with antipsychotic or EPS medication administration, n (%)	86 (3.7)	119 (2.1)	<0.0001*
Antipsychotic or EPS medication outpatient pharmacy
Patients with a prescription, n (%)	2,257 (98.3%)	4,406 (78.6%)	<0.0001*
Number of prescriptions, mean (SD)^†	15.6 (12.0)	7.8 (6.7)	<0.0001*
Schizophrenia-related healthcare costs
Inpatient costs, mean (SD)	$4,675($23,133)	$2,248 (13,451)	<0.0001*
Outpatient costs, mean (SD)	$3,815 (7,832)	$1,887 (5,260)	<0.0001*
ER visit costs, mean (SD)	$197 (807)	$118 (579)	<0.0001*
Office visit costs, mean (SD)	$188 (1,460)	$93 (390)	<0.0001*
Other outpatient services costs, mean (SD)	$3,313 (7,417)	$1,590 (4,957)	<0.0001*
Antipsychotic or EPS medication administration costs, mean (SD)	$117 ($1,205)	$86 ($1,031)	0.2449
Antipsychotic or EPS medication outpatient pharmacy costs, mean (SD)	$3,644 ($7,087)	$2,095 ($5,292)	0.0221*
Total medical costs (inpatient + outpatient), mean (SD)	$7,703 (20,238)	$3,663 (11,520)	<0.0001*
Total costs (medical + pharmacy), mean (SD)	$8,291 (20,322)	$4,152 (11,659)	<0.0001*

Abbreviations. EPS, extrapyramidal symptoms; SD, standard deviation.

^†Among patients with at least one service or prescription fill.

*p < 0.05.

Schizophrenia-related

Prior to controlling for baseline characteristics, the EPS group had significantly higher rates of schizophrenia-related (22.5 vs. 12.9%, p < 0.0001) hospitalizations during the 12-month post-index period when compared to the non-EPS group (Table 2). The EPS group also had a significantly higher proportion of patients with office visits (46 vs. 35%), schizophrenia-related ER visit (25.5 vs. 16.7%, p < 0.0001), other outpatient service (74.2 vs 54.5%), and outpatient pharmacy fills (98.3 vs. 78.6%, p < 0.0001) when compared to the non-EPS group. Finally, the EPS group had significantly greater schizophrenia-related inpatient ($4,675 vs. $2,248, p < 0.0001), outpatient ($3,815 vs. $1,887, p < 0.0001), ER visit ($197 vs. $118, p < 0.0001), outpatient pharmacy ($3,644 vs. $2,095, p < 0.0001), and total costs ($12,134 vs. $6,230, p < 0.0001) on average during the post-index period when compared to the non-EPS group.

Multivariate analysis

Inpatient admissions

Results from multivariate analysis that controlled for baseline differences found that patients with EPS had 25% increased risk of all cause (OR = 1.25; 95% CI = 1.11–1.40) and 75% increased risk of schizophrenia-related (OR = 1.75; 95% CI = 1.53–2.00) inpatient admission compared to non-EPS patients (Figures 3 and 4). Additional significant risk factors for all-cause inpatient admission included increased age (OR = 1.01; 95% CI = 1.00–1.01; p < 0.001), increased pre-period CCI score (OR = 1.13; 95% CI = 1.08–1.17, p < 0.001), pre-period bipolar disorder (OR = 1.15; 95% CI = 1.02–1.29, p < 0.001), depression (OR = 1.18; 95% CI = 1.05–1.32, p < 0.001), anxiety (OR = 1.25; 95% CI = 1.07–1.47, p < 0.001), or alcohol/substance abuse diagnoses (OR = 1.36; 95% CI = 1.21–1.54, p < 0.001), pre-index inpatient admissions or ER visits (OR = 2.07; 95% CI = 1.77–2.40, p < 0.001), and pre-index healthcare costs above the seventh decile compared to first decile (range: OR = 1.32; 95% CI = 1.01–1.71, p < 0.001 to OR = 2.27; 95% CI = 1.74–2.95; p < 0.0001). The supply of antipsychotic agents for 91+ days vs. 1–30 days (OR = 0.80; 95% CI = 0.68–0.94, p < 0.001) and pre-index healthcare costs in the third decile compared to first decile (OR = 0.70; 95% CI = 0.53–0.34, p = 0.00125) were associated with a lower risk of post-period all-cause inpatient admission. Additional significant risk factors for schizophrenia-related inpatient admission included black race (vs. white) (OR = 1.21; 95% CI = 1.04–1.40, p < 0.0124), pre-period anxiety (OR = 1.38; 95% CI = 1.15–1.65, p < 0.001), alcohol/substance abuse diagnoses (OR = 1.25; 95% CI = 1.09–1.45, p < 0.001), pre-index inpatient admissions or ER visits (OR = 2.07; 95% CI = 1.72–2.50, p < 0.001), and pre-period healthcare costs in the tenth decile vs. the first decile (OR = 1.75; 95% CI = 1.29–2.36, p < 0.001). Female gender (OR = 0.75; 95% CI = 0.65–0.85, p < 0.001), Medicaid plan type (HMO vs. comprehensive) (OR = 0.77; 95% CI = 0.67–0.89, p < 0.001) and pre-period healthcare costs in the third through sixth decile (compared to the first decile) (range: OR = 0.49; 95% CI = 0.35–0.69, p < 0.001 to OR = 0.72; 95% CI = 0.52–0.99; p < 0.0001) were associated with a lower risk of post-period schizophrenia-related inpatient admission.

Figure 3. Odds of all-cause inpatient admission. AP, antipsychotic; EPS, extrapyramidal symptoms.

Figure 4. Odds of schizophrenia-related inpatient admission. AP, antipsychotic; EPS, extrapyramidal symptoms.

Cost outcomes

In part one of the model, patients with EPS were significantly more likely to have any all-cause (OR = 6.93; 95% CI = 3.36–14.29) and schizophrenia-related (OR = 3.09; 95% CI = 2.55–3.73) costs compared to non-EPS patients (Supplementary Tables S2 and S3). In part two of the model, patients with EPS were significantly more likely to have higher all-cause (Cost Ratio = 1.22; 95% CI = 1.15–1.29) and schizophrenia-related (Cost Ratio = 1.59; 95% CI = 1.47–1.71) costs compared to non-EPS patients (Supplementary Tables S2 and S3 and Figures 5 and 6). The predicted all-cause ($26,998 vs. 21,792) and schizophrenia-related costs ($11,408 vs. $6,471) were higher for EPS patients compared to non-EPS patients.

Figure 5. Adjusted all-cause total healthcare costs. AP, antipsychotic; EPS, extrapyramidal symptoms.

Figure 6. Adjusted schizophrenia-related total healthcare costs. AP, antipsychotic; EPS, extrapyramidal symptoms.

Other significant baseline predictors of increased all-cause total costs in the post-period were increased age (OR = 1.01; 95% CI = 1.01–1.02; p < 0.001), increase in CCI (OR = 1.11; 95% CI = 1.08–1.13; p < 0.0001), pre-index depression (OR = 1.09; 95% CI = 1.03–1.16; p < 0.001) or alcohol/substance abuse (OR = 1.08; 95% CI = 1.02–1.15; p < 0.0001) diagnoses, and pre-index total all-cause costs above the third decile compared to the first decile (range: OR = 1.25; 95% CI = 1.11–1.41, p < 0.001 to OR = 6.64; 95% CI = 5.81–7.58; p < 0.0001). Reduced all-cause healthcare costs were associated with Medicaid HMO plan compared to comprehensive plan type (OR = 0.75; 95% CI = 0.62–0.91; p < 0.01), pre-index through the index date days’ supply category of 31 days or greater vs. 1–30 days (range: OR = 0.78; 95% CI = 0.72–0.86; p < 0.05 to OR = 0.86; 95% CI = 0.80–0.93), pre-period bipolar disorder (OR = 0.94; 95% CI = 0.88–1.00; p < 0.01), pre-index inpatient admission or ER visit (OR = 0.84; 95% CI = 0.78–0.90; p < 0.01).

Other significant baseline predictors of increased schizophrenia-related costs in the post-period were race. Black vs. White (OR = 1.09; 95% CI = 1.01–1.19; p < 0.0001) and pre-index total all-cause costs above the third decile compared to the first decile (range: OR = 1.20; 95% CI = 1.101–1.42, p < 0.001 to OR = 6.35; 95% CI = 5.26–7.66; p < 0.0001). Reduced schizophrenia-related healthcare costs were associated with female sex (OR = 0.80; 95% CI = 0.74–0.86; p < 0.01), HMO compared to comprehensive Medicaid plan type (OR = 0.83; 95% CI = 0.76–0.89; p < 0.01), pre-index through the index date days’ supply category of 31–60 days vs. 1–30 days (range: OR = 0.82; 95% CI = 0.73–0.92; p < 0.05), pre-index bipolar disorder (OR = 0.81; 95% CI = 0.75–0.88; p < 0.01), pre-index depression (OR = 0.92; 95% CI = 0.85–1.00; p < 0.01), and pre-index inpatient admission or ER visit (OR = 0.71; 95% CI = 0.65–0.78; p < 0.01).

Discussion

In this retrospective database study of Medicaid claims, 21.2% of patients with schizophrenia who were newly-prescribed an SGA developed EPS over a 12-month period. These results are consistent with those reported elsewhere and serve to reaffirm that EPS events are common among patients prescribed SGAs^9–11,14. This study also confirmed results from a previous study by Abouzaid and colleagues who found EPS among patients with schizophrenia to be associated with increased healthcare utilization and greater medical costs¹⁹.

Patients with EPS in our study had nearly twice the rate of schizophrenia-related hospitalization and almost double the cost of schizophrenia-related healthcare when compared to those without EPS. Results from the multivariate analysis that adjusted for baseline differences were consistent with the unadjusted findings. Compared to patients without EPS, patients with EPS were at significantly increased risk of all-cause (OR = 1.25) and schizophrenia-related (OR = 1.75) hospitalizations. In addition, patients with EPS were significantly more likely to have higher all-cause ($27,408 vs. $22,489) and schizophrenia-related ($12,833 vs. $8,077) costs. These findings are very consistent with those of Abouzaid and colleagues, who reported similar increased risks of all-cause (OR = 1.33) and schizophrenia-related (OR = 1.56) hospitalizations and higher total costs ($18,682 vs. $15,130) and schizophrenia-related total costs (9,377 vs. $6,688) for patients with EPS¹⁹. As previously highlighted, hospitalization among patients with schizophrenia is strongly correlated with future hospitalization¹⁸. This study provides additional evidence that EPS contribute to increased risk of hospitalization among patients with schizophrenia, thereby increasing the probability of subsequent hospitalizations.

In addition to the impact of EPS on healthcare utilization and costs, patients with EPS have been shown to have decreased QoL and increased morbidity and mortality^13,15. There is substantial variation among SGAs in their liability to cause EPS²⁶. According to the American Psychiatric Association, the choice of antipsychotic medication should be based on both efficacy and risks⁶. Thus, the potential of an antipsychotic medication to cause EPS appears to be an important risk to consider. This study was not designed to assess individual SGAs in terms of EPS and corresponding cost and resource utilization. Future studies are needed to further elucidate the role of specific SGAs with regard to frequency of EPS and impact on cost and resource utilization.

Limitations

The limitations of this study include those inherent to retrospective analyses. First, this study was limited to only those individuals insured through Medicaid. While two-thirds of patients with schizophrenia are covered by Medicaid⁵, results of this analysis may not be generalizable to patients with Medicare, commercial insurance, or the uninsured. Second, this study relied on administrative claims collected for billing purposes as opposed to medical records. As such, there is potential for misclassification of covariates and the data is subject to coding limitations and data entry errors. In addition, claims data are limited in scope to information relevant to reimbursement; therefore, analyses could not control for important clinical variables such as disease severity or specific symptoms of EPS. Third, medication utilization measures were based on medication administration and filled prescriptions, with the assumption that patients took their medications as prescribed. However, we could not confirm if patients took the prescribed medication fills. Fourth, the risk of EPS for oral antipsychotics vs. long acting injectables was not examined separately. Because treatment guidelines recommend the trial of an oral form of the antipsychotic prior to initiation of the long acting injectable to first ensure efficacy and tolerability⁶, such a comparison in a retrospective database analysis would not likely be confounded. Finally, there may be systematic differences between the study groups beyond the baseline covariates that were included in the models that account for differences found in healthcare costs and utilization. For example, patients with more severe or recalcitrant symptoms may have required higher doses of antipsychotic medications leading to higher incidence of EPS and more healthcare utilization. While the analyses corrected for baseline characteristics that are related to clinical severity (e.g. inpatient admissions, ER visits, comorbidities, and healthcare costs) there are no direct measures of symptom severity in administrative claims data.

Conclusions

In this analysis, over one-fifth of patients had a diagnosis or medication claim for EPS within 12 months of initiating treatment with SGAs. Extrapyramidal side-effects associated with SGAs increase the risk of hospitalization and contribute to higher healthcare costs. For patients with schizophrenia, treatment options that minimize the risk of EPS may reduce the economic burden associated with these side-effects.

Transparency

Declaration of funding

This study was sponsored by Sunovion Pharmaceuticals.

Declaration of financial/other relationships

AK, CD, and GRW are employed by Sunovion. JK was previously employed by Sunovion. BLB is employed by IBM Watson Health which received funding from Sunovion to conduct this study. The peer reviewers of this manuscript have received an honorarium from JME for their review work. One of the reviewers on this manuscript has disclosed that they have consulted for Sunovion in the past, but not for the last 2 years. Another reviewer has received manuscript or speaker’s fees from Astellas, Dainippon Sumitomo Pharma, Eisai, Eli Lilly, Elsevier Japan, Janssen Pharmaceuticals, Kyowa Yakuhin, Lundbeck, Meiji Seika Pharma, Mitsubishi Tanabe Pharma, MSD, Nihon Medi-Physics, Novartis, Otsuka Pharmaceutical, Shionogi, Shire, Takeda Pharmaceutical, Tsumura, Wiley Japan, and Yoshitomi Yakuhin, and research grants from Dainippon Sumitomo Pharma, Eisai, Mochida Pharmaceutical, Meiji Seika Pharma and Shionogi. The other peer reviewers have no relevant financial relationships or otherwise to disclose.

Acknowledgements

We thank Michael Stensland of Agile Outcomes Research Inc and Thomas Lee of Pandion Communications Inc who provided technical writing services on behalf of Sunovion Pharmaceuticals Inc.