The effects of early initiation of aripiprazole once-monthly on healthcare resource utilization and healthcare costs in individuals with schizophrenia: real-world evidence from US claims data

Abstract

Aim

To evaluate the impact of timing of aripiprazole once-monthly (AOM) initiation on healthcare resource utilization (HCRU), risk of hospitalization, and healthcare costs in patients with schizophrenia.

Methods

A retrospective cohort study was conducted using data from the Merative MarketScan database (01/01/2013–12/31/2019). Adults aged ≥18 years with a new episode of care for schizophrenia and an AOM claim were included. Patients were classified into two cohorts based on the time between the first schizophrenia diagnosis and the first AOM claim (early cohort: ≤1 year; late cohort: >1 year). All-cause and psychiatric-specific HCRU, risk of hospitalization, and healthcare costs were evaluated over 1-year post-AOM initiation. The relationship between the timing of AOM initiation and HCRU was evaluated using negative binomial regression, and healthcare costs using generalized linear models (log-link with gamma distribution). Logistic regression was used to estimate the likelihood of hospitalization during the follow up period for both all-cause and psychiatric-specific hospitalization.

Results

A total of 945 patients were included (early cohort: n = 525; late cohort: n = 420). At baseline, the early cohort had lower mean age, a greater proportion of males, and a lower mean Charlson Comorbidity Index score than the late cohort (all p < .05). After adjusting for baseline demographic and clinical characteristics, all-cause and psychiatric-specific hospitalization during the 1-year follow-up period were statistically significantly higher for the late cohort versus the early cohort (all-cause: incident rate ratio [IRR] = 1.63, 95% confidence interval [CI]: 1.28–2.07, p < .01; psychiatric-specific: IRR = 1.93, 95% CI: 1.46–2.55, p < .01). The early cohort had statistically significantly lower adjusted all-cause ($21,686 versus $29,033; p = .0002) and psychiatric-specific ($24,414 versus $32,461; p = .0002) healthcare costs versus the late cohort.

Limitations

This study utilized claims data, which are intended for administrative purposes rather than for research.

Conclusions

This analysis extends previous evidence for the benefits of AOM in patients with new episodes of schizophrenia, by demonstrating lower HCRU, risk of hospitalization, and healthcare costs with early AOM initiation compared with later initiation.

PLAIN LANGUAGE SUMMARY

Schizophrenia is a costly disease that impacts patients, caregivers, and the healthcare system. Antipsychotic medications are an important component of schizophrenia treatment. These medications reduce symptom severity, improve functioning and reduce costs. Aripiprazole once-monthly (AOM) is a long-acting injectable antipsychotic used to treat schizophrenia. This study evaluates whether starting AOM early in the disease course improves outcomes for people with schizophrenia. Outcomes include healthcare resource utilization, risk of hospitalization, and healthcare costs. The study team found that hospitalization and costs were lower for people who started AOM early in the disease course as opposed to later. This study points to the importance of early treatment to improve outcomes for people with schizophrenia.

Keywords:

• Aripiprazole once-monthly
• long-acting injectable
• schizophrenia
• healthcare utilization
• healthcare cost
• retrospective

JEL CLASSIFICATION CODES:

• I11
• I12

Introduction

Schizophrenia is a severe, chronic psychiatric disorder, characterized by a range of cognitive, behavioral, and emotional symptoms.^1,2 Schizophrenia affects approximately 23.6 million individuals worldwide, including 1.6 million individuals in the US.³ In 2020, the estimated annual cost of schizophrenia in the US was $281.6 billion, comprising direct healthcare costs as well as indirect costs resulting from factors such as unemployment, reduced quality of life, and caregiving.⁴ Monthly healthcare costs per patient are more than four times higher for patients with schizophrenia compared with the general US population.⁵ The higher costs are largely driven by rates of inpatient admission, which are thirteen times higher among patients with schizophrenia than the general US population.⁵ As may be expected given the functional impairments and high rates of unemployment that are associated with schizophrenia,^6,7 it is estimated that 67% of patients with schizophrenia in the US receive Medicaid health benefits.⁸

Antipsychotic medications are an important component of schizophrenia treatment, with the short-term goal of reducing acute symptoms, and the long-term goals of preventing symptom recurrence and maximizing functioning and quality of life.⁹ Randomized controlled trials have demonstrated the efficacy of oral antipsychotics in reducing schizophrenia symptoms and improving both functioning and quality of life¹⁰; however, non-adherence to medication is common in patients with schizophrenia,^11,12 and is associated with an increased risk of relapse and hospitalization.^13,14 Furthermore, multiple relapses have been associated with reduced response to antipsychotic therapy, and extended periods of relapse are potentially associated with the progression of the disease, characterized by brain volume loss.^15–17 Long-acting injectable (LAI) formulations of antipsychotic medications have been developed, in part, to address treatment non-adherence.¹⁸

Historically, treatment guidelines have recommended oral antipsychotics for the treatment of first-episode schizophrenia, with LAI antipsychotics reserved for more chronic patients with a history of poor or uncertain adherence, as well as patients who prefer this mode of administration.^19,20 However, LAI antipsychotics are increasingly being recognized as an effective first-line treatment for the early stages of schizophrenia.^21–26 LAIs may be of particular benefit to patients early in the disease course, by improving adherence to treatment, and reducing the risk of relapse and rehospitalization versus oral antipsychotics or usual care.^{22,24,25,27–29} The effects of the timing of LAI initiation have been evaluated in a retrospective cohort study of US claims data covering the period from 1 January 2007 to 30 June 2016, which showed that early initiation of LAIs (within 1 year of a new episode of schizophrenia) was associated with lower hospitalization rates and healthcare costs compared with later LAI initiation.³⁰ This study combined all LAIs including typical and atypical formulations and, thus, did not address any possible differences between specific LAIs.

The present study focuses on aripiprazole once-monthly (AOM), an LAI antipsychotic indicated for the treatment of schizophrenia (as well as maintenance monotherapy treatment of bipolar I disorder) in adults, and is manufactured by Otsuka America Pharmaceutical, Inc..³¹ A previous 2-year randomized controlled trial conducted in patients with early-phase schizophrenia demonstrated that AOM was associated with reduced healthcare resource utilization (HCRU), as shown by increased time to hospitalization, compared with usual care.²⁹ The effect of early initiation of AOM, compared with later initiation, on HCRU in patients with schizophrenia remains unknown. The prior claims analysis showing the benefits of LAIs earlier in treatment was conducted shortly after the launch of AOM, and therefore AOM was likely under-represented in the sample.³⁰ The choice of which LAI to prescribe is largely based on an individual’s history of response to the active moiety (if known), and the side effect profile of the different medications, as well as provider and patient preference.³² It was, therefore, important to determine whether or not results for AOM would be similar to the combined LAIs in the prior study³⁰ and also add to what was shown in the randomized controlled trial²⁹ in a larger real-world sample of patients. Therefore, the aim of the current study was to evaluate the impact of the timing of AOM initiation following a new episode of care for schizophrenia on HCRU, risk of hospitalization, and healthcare costs in a real-world setting.

Methods

This was a retrospective cohort study that utilized claims data from the Merative MarketScan Medicaid database. This database contains data relevant to the healthcare experiences of more than 47 million individuals from multiple US states, including records of patient demographics, HCRU, healthcare costs, and patient outcomes.³³ Whereas the prior retrospective cohort study included commercial insurance and Medicare data as well as Medicaid data, 80% of the sample were covered by Medicaid,³⁰ and so the inclusion of commercial data in the present study was not considered necessary. Data were evaluated for the period from 1 January 2013 to 31 December 2019. The study timeline is summarized in Figure 1.

Figure 1. Study timeline. Abbreviation. AOM, aripiprazole once-monthly.

Study sample

Patients were eligible if they had a new episode of care for schizophrenia during the study identification period (1 January 2014 to 31 December 2018) (Figure 1). Patients were classed as having a diagnosis of schizophrenia if they had ≥1 inpatient claim or ≥2 outpatient claims (on different days) for schizophrenia during the study identification period, based on specified International Classification of Diseases 9th or 10th Revision, Clinical Modification codes (ICD-9-CM codes: 295.xx; ICD-10-CM codes: F20.x and F25.x). To identify the subset of patients with new episodes of care for schizophrenia, patients were required to have a 1-year disease-free period (i.e. no claims for schizophrenia) before the first diagnosis of schizophrenia in the study identification period.

Patients were also required to have a claim for AOM (either 400 mg or 300 mg) during the study identification period. The date of the first AOM claim was defined as the index date. Patients were required to be aged ≥18 years on the index date; have a first diagnosis of schizophrenia prior to, or on, the index date; and have no AOM use in the year prior to the index date. In addition, eligible patients must have been continuously enrolled for ≥1 year prior to the index date (pre-index period) and ≥1 year after the index date (follow-up period).

Patients were excluded if they had a claim for clozapine at any time during the study period, as clozapine is indicated for the treatment of severely ill patients with schizophrenia who fail to respond adequately to standard antipsychotic treatment.³⁴ To ensure complete medication claims histories, patients with Medicare and Medicaid dual eligibility, patients with a Medicaid capitated plan, and patients who lacked pharmacy coverage during the study period were also excluded.

Eligible patients were classified into two groups based on the time between the beginning of a new episode of care for schizophrenia and the first AOM claim. The groups were: (1) patients who had their first claim for AOM within 1 year of schizophrenia diagnosis (early cohort) and (2) patients who had their first claim for AOM more than 1 year after schizophrenia diagnosis (late cohort). The 1-year timeframe was selected to align with the previous retrospective study.³⁰

Outcomes

Baseline patient demographic and clinical characteristics were evaluated during the 1-year pre-index period. Baseline comorbidities were characterized using the Charlson Comorbidity Index (CCI); higher scores indicate a greater number of, or more severe, non-psychiatric comorbidities.³⁵ The baseline occurrence of diagnoses of interest (other than schizophrenia) was also evaluated, for specific psychiatric conditions (major depressive disorder [MDD], bipolar disorder, anxiety disorder, personality disorder, substance use disorders, and post-traumatic stress disorder [PTSD]) and non-psychiatric conditions associated with antipsychotic use (obesity, diabetes mellitus, hyperlipidemia, hypertension, asthma, chronic obstructive pulmonary disease [COPD], and obstructive sleep apnea). Baseline use of psychotropic medications (antipsychotics, antidepressants, anti-anxiety medications, hypnotics, and mood stabilizers) and non-psychotropic medications associated with the treatment of metabolic disorders, which are common among patients taking antipsychotics (antidiabetic, lipid-lowering, and antihypertensive medications), were also evaluated (codes provided upon request).

All-cause and psychiatric-specific HCRU and healthcare costs were evaluated during the 1-year follow-up period. Psychiatric-specific claims were defined as those associated with a primary diagnosis of any mental disorder (ICD-9-CM codes: 290.xx–311.xx; ICD-10-CM codes: F01.xx–F99.xx). HCRU was measured as the number of inpatient hospitalizations, the length of inpatient stays (i.e. number of hospital days), the number of physician office visits, and the number of emergency department visits. The likelihood of hospitalization in the 12-month follow-up period were also estimated. Healthcare costs were evaluated for inpatient, emergency department, outpatient, and pharmacy settings, and the total of these settings.

Statistical analyses

Descriptive statistics were reported for the baseline demographic and clinical characteristics (means and standard deviations [SD] for continuous data, and patient numbers and percentages for categorical data). To assess differences between groups, t-tests, chi-square tests, and Wilcoxon tests were performed, as appropriate.

The relationship between the timing of AOM initiation and HCRU was evaluated using negative binomial regression, and the relationship between the timing of AOM initiation and healthcare costs was evaluated using generalized linear models. We employed the widely used log link with gamma distribution to assess the costs as this approach can adjust for the skewness and heteroscedasticity often observed in cost data.³⁶ HCRU data was analyzed using negative binomial regression as all types of HCRU (outpatient, emergency department, inpatient etc.) data were over-dispersed (and thus Poisson regression was not used). Logistic regression was used to assess the differences in the likelihood of all-cause and psychiatric hospitalization during the 12-month follow-up period. Models were adjusted for baseline demographic and clinical characteristics. All healthcare costs were adjusted to 2021 US dollars using the medical component of the annual consumer price index provided by the Bureau of Labor Statistics.³⁷

A sensitivity analysis was conducted that excluded patients who had a diagnosis of bipolar I disorder as well as schizophrenia during the study period. Dual diagnoses of schizophrenia and bipolar I disorder are not uncommon in claims data, and these patients tend to have higher HCRU and healthcare costs than patients with the respective single diagnoses.³⁸ For the cohort of patients included in the sensitivity analysis, the relationship between the timing of AOM initiation and HCRU and healthcare costs was evaluated as described for the overall study analyses.

All analyses were conducted in SAS Enterprise Guide version 7.1 (SAS Institute, Cary, NC).

Results

Patients

A total of 359,991 patients had a claims-based diagnosis of schizophrenia from 2013 to 2019; of these, 945 patients met the study eligibility criteria (Figure 2). When grouped based on the timing of AOM initiation, 525 patients (55.6%) were included in the early cohort (i.e. first AOM claim within 1 year after beginning a new episode of care for schizophrenia) and 420 patients (44.4%) were included in the late cohort (i.e. first AOM claim more than 1 year after beginning a new episode of care for schizophrenia).

Figure 2. Patient attrition. ^aThe index date was defined as the first AOM 400 claim. ^bThe identification period was 1 January 2014 to 31 December 2018. Abbreviation. AOM 400, aripiprazole once-monthly 400 mg.

Baseline demographic and clinical characteristics are presented in Table 1. The early cohort had a lower mean age (34.7 versus 36.5 years, p = .021), had a greater proportion of males (53.1% versus 45.5%, p = .019), and had a lower mean CCI score (0.75 versus 0.95, p = .039) at baseline compared with the late cohort. The most common diagnoses of interest during the baseline period were anxiety disorder (52.5%), bipolar disorder (51.7%) and substance use disorders (43.9%); however, there were no differences in baseline co-occurring psychiatric disorders (e.g. bipolar disorder, depression, anxiety, personality disorders or substance use) between the early and late cohorts. The late cohort had a higher baseline rate of hyperlipidemia (24.1% vs. 16.6%; p = .004) but no other differences in baseline physical health comorbidities existed (Table 1). The proportion of patients with prior antipsychotic use was lower in the early cohort compared with the late cohort (89.3% versus 94.0%; p = .010); this difference was partly driven by differences in prior use of LAIs between groups (18.7% versus 26.4%; p = .0043). Of note, there were no statistically significant differences in the use of any psychiatric medications during the follow-up period between the early and late cohorts.

Table 1. Baseline patient demographic and clinical characteristics.

	All patients (n = 945)	Early cohort (n = 525)	Late cohort (n = 420)	p-value^a
Age at index date, years, mean (SD)^b	35.5 (12.5)	34.7 (12.3)	36.5 (12.5)	.021
Male, n (%)	470 (49.7)	279 (53.1)	191 (45.5)	.019
Race, n (%)				.47
White	383 (40.5)	220 (41.9)	163 (38.8)
Black	409 (43.3)	219 (41.7)	190 (45.2)
Hispanic	13 (1.4)	8 (1.5)	5 (1.2)
Other	57 (6.0)	36 (6.9)	21 (5.0)
Unknown	83 (8.8)	42 (8.0)	41 (9.8)
Psychiatric diagnoses, n (%)
Anxiety disorder	496 (52.5)	275 (52.4)	221 (52.6)	.94
Bipolar disorder	489 (51.7)	272 (51.8)	217 (51.7)	.97
Substance use disorder	415 (43.9)	241 (45.9)	174 (41.4)	.17
MDD	413 (43.7)	241 (45.9)	172 (41.0)	.13
PTSD	196 (20.7)	112 (21.3)	84 (20.0)	.62
Personality disorder	166 (17.6)	96 (18.3)	70 (16.7)	.52
Non-psychiatric diagnoses, n (%)
Hypertension	312 (33.0)	161 (30.7)	151 (36.0)	.086
Obesity	213 (22.5)	114 (21.7)	99 (23.6)	.50
Asthma	190 (20.1)	101 (19.2)	89 (21.2)	.46
Hyperlipidemia	188 (19.9)	87 (16.6)	101 (24.0)	.0042
Diabetes	140 (14.8)	74 (14.1)	66 (15.7)	.49
COPD	75 (7.9)	38 (7.2)	37 (8.8)	.37
Obstructive sleep apnea	44 (4.7)	23 (4.4)	21 (5.0)	.65
CCI score, mean (SD)	0.84 (1.45)	0.75 (1.33)	0.95 (1.58)	.039
Psychotropic medications, n (%)^c
Antipsychotics	864 (91.4)	469 (89.3)	395 (94.0)	.010
Oral antipsychotics	821 (86.9)	449 (85.5)	372 (88.6)	.17
LAI antipsychotics	209 (22.1)	98 (18.7)	111 (26.4)	.0043
Antidepressants	622 (65.8)	334 (63.6)	288 (68.6)	.11
Anti-anxiety	427 (45.2)	237 (45.1)	190 (45.2)	.98
Hypnotics	243 (25.7)	128 (24.4)	115 (27.4)	.29
Mood stabilizers^d	173 (18.3)	94 (17.9)	79 (18.8)	.72
Non-psychotropic medications^e, n (%)
Antihypertensives	327 (34.6)	176 (33.5)	151 (36.0)	.44
Lipid lowering	189 (20.0)	89 (17.0)	100 (23.8)	.0088
Antidiabetics	105 (11.1)	53 (10.1)	52 (12.4)	.27

Abbreviations. CCI, Charlson Comorbidity Index; COPD, chronic obstructive pulmonary disease; LAI, long-acting injectable; MDD, major depressive disorder; PTSD, post-traumatic stress disorder; SD, standard deviation.

^at-test, chi-square test, or Wilcoxon test, as appropriate.

^bAverage time between schizophrenia diagnosis and index date was 130 (IQR = 32–216) and 779 (IQR = 505–1008) days for early and late initiators respectively.

^cThere were no statistically significant differences in psychiatric medication use during the follow-up period between cohorts.

^dLithium, carbamazepine, lamotrigine, or valproate.

^eAssociated with the treatment of metabolic disorders, which are common among patients taking antipsychotics.

Healthcare resource utilization

During the 12-month follow-up period, the early cohort had a significantly lower mean (SD) all-cause hospitalizations (0.47 [1.02] vs. 0.76 [1.45]; p = .0006) and inpatient days (2.74 [7.76] vs. 4.83 [9.98]; p = .0004) than the late cohort, but no differences were noted for outpatient or ED visits. Similarly, mean psychiatric-related hospitalizations (0.33 [0.77] vs. 0.62[1.32]; p < .001), inpatient days (2.16 [6.95] vs. 4.18 [9.17]; p = .0002) and ED visits (0.41 [1.14] vs. 0.75 [2.75]; p = .0178) were lower for the early cohort than the late cohort in the 12-month follow-up period, but no differences were noted in psychiatric-related outpatient visits.

After adjusting for baseline demographic and clinical characteristics (age, sex, CCI score, oral antipsychotic use, and other psychiatric conditions), the incidence of all-cause and psychiatric-specific hospitalizations were higher for the late cohort versus the early cohort (all-cause: incident rate ratio [IRR] = 1.63, 95% confidence interval [CI]: 1.28–2.07, p < .01; psychiatric-specific: IRR = 1.93, 95% CI: 1.46–2.55, p < .01). Similarly, the adjusted incidence of psychiatric-specific emergency department visits was statistically significantly higher for the late cohort compared with the early cohort (Table 2). The adjusted incidence of all-cause emergency department visits and all-cause and psychiatric-specific physician visits were not significantly different between groups (p > .05).

Table 2. Adjusted^a all-cause and psychiatric-specific^b HCRU during the 1-year follow-up period.

	IRR (late cohort versus early cohort) (95% CI)	p-value
All-cause HCRU^c
Number of hospitalizations	1.63 (1.28–2.07)	<.01
Number of physician visits	1.06 (0.90–1.25)	.49
Number of emergency department visits	1.17 (0.98–1.41)	.09
Psychiatric-specific HCRU^d
Number of hospitalizations	1.93 (1.46–2.55)	<.01
Number of physician visits	1.06 (0.84–1.34)	.62
Number of emergency department visits	1.84 (1.33–2.54)	<.01

Abbreviations. CI, confidence interval; HCRU, healthcare resource utilization; IRR, incident rate ratio.

^aNegative binomial regression analysis adjusted for baseline demographic and clinical characteristics.

^bClaims with a primary diagnosis of any mental disorder (ICD-9-CM codes: 290.xx–311.xx; ICD-10-CM codes: F01.xx–F99.xx).

^cFor early and late initiators mean number of hospitalizations per patient was 0.47 and 0.76, number of physician visits per patient was 23.19 and 24.17, and number of emergency department visits per patient was 2.27 and 2.83 respectively.

^dFor early and late initiators psychiatry specific mean number of hospitalizations per patient was 0.33 and 0.62, number of physician visits per patient was 18.99 and 19.34, and number of emergency department visits per patient was 0.41 and 0.75 respectively.

Risk of hospitalization

The late cohort was 62% more likely to be hospitalized for all causes in the 12-month follow-up period as compared to early initiators (OR = 1.62, 95% CI, 1.19–2.19, p = .001). Similarly, the late cohort was 78% more likely to be hospitalized for psychiatric-specific causes in the 12-month follow-up period as compared to early initiators (OR = 1.78, 95% CI 1.29–2.46, p = .0004) (Table 3).

Table 3. Likelihood of hospitalization between early and late initiators of AOM during the follow up.

Initiation	All cause		Psychiatric-specific
	OR (95% CI)	p-value	OR (95% CI)	p-value
Early	Ref		Ref
Late	1.62 (1.19–2.19)	.001	1.78 (1.29–2.46)	.0004

Abbreviations. CI, confidence interval; OR, odds ratio.

Healthcare costs

During the 12-month follow-up period, the early cohort had significantly lower mean unadjusted all-cause total ($22,375 vs. $30.085; p = .0019), inpatient ($2,365 vs. $4,269; p = .0146), outpatient ($8,641 vs. $12,273; p = .0357) and pharmacy costs ($10,654 vs. $12,674; p = .017) than the late cohort, but no differences were noted in all-cause ED visit costs. Likewise, mean unadjusted psychiatric-related total ($18,089 vs. $24,067; p = .0054), inpatient ($1475 vs. $2,984; p = .0071) and outpatient costs ($6,943 vs. $10,238; p = .0486) were lower for the early cohort in the 12-month follow-up period, but there were no differences between cohorts in psychiatric-related ED visit or pharmacy costs in the 12-month follow-up period.

After adjustment for baseline demographic and clinical characteristics, mean all-cause total healthcare costs during the follow-up year were statistically significantly lower for the early cohort compared with the late cohort ($21,686 versus $29,033, p = .0002) (Figure 3, Table 4). In addition to earlier initiation of AOM, covariates that influenced all-cause costs included baseline CCI score, and claims for MDD, bipolar disorder, personality disorder, and substance use disorders during the baseline period (data not presented in tabular form).

Figure 3. Adjusted^a mean all-cause and psychiatric-specific^b healthcare costs during the 1-year follow-up period. ^aGeneralized linear model adjusted for baseline demographic and clinical characteristics. Costs were adjusted to 2021 US dollars. ^bClaims with a primary diagnosis of any mental disorder (ICD-9-CM codes: 290.xx–311.xx; ICD-10-CM codes: F01.xx–F99.xx).

Table 4. Adjusted^a mean all-cause and psychiatric-specific^b healthcare costs during the 1-year follow-up period.

	Estimate (95% CI)		p-value
	Early cohort (n = 525)	Late cohort (n = 420)
All-cause healthcare costs, US$	21,686 (12,606–37,306)	29,033 (19,675–42,843)	.0002
Psychiatric-specific healthcare costs, US$	24,414 (14,333–41,589)	32,461 (22,146–47,586)	.0002

Abbreviation. CI, confidence interval.

^aGeneralized linear model. Total follow-up costs were modelled as the dependent variable and early versus late as the independent variable, adjusted for several demographic and clinical characteristics. Costs were adjusted to 2021 US dollars.

^bClaims with a primary diagnosis of any mental disorder (ICD-9-CM codes: 290.xx–311.xx; ICD-10-CM codes: F01.xx–F99.xx).

Similarly, adjusted psychiatric-specific total healthcare costs during the follow-up year were statistically significantly lower for the early cohort compared with the late cohort ($24,414 versus $32,461, p = .0002) (Figure 3, Table 4). Additional covariates that affected psychiatric-specific costs were age, sex, and claims for MDD, bipolar disorder, personality disorder, and substance use disorders during the baseline period (data not presented in tabular form).

Sensitivity analysis

Following the exclusion of patients with diagnoses of both schizophrenia and bipolar I disorder, a total of 456 patients were included in the sensitivity analysis (early cohort, n = 253; late cohort, n = 203). There were no differences between the early cohort and the late cohort in terms of demographic characteristics (data not presented in tabular form). The HCRU and healthcare costs findings were generally similar to base case analysis (data not presented in tabular form). There was no statistically significant difference between the two groups in terms of likelihood of all-cause (OR = 1.04 [0.69 − 1.56]; p = .80) and psychiatric-specific (OR = 1.44, 95% CI, 0.84 − 2.46, p = .1) hospitalization (data not presented in tabular form).

Discussion

This retrospective cohort study of real-world claims data provides evidence that early initiation of AOM (within the first year) in patients with newly identified episodes of schizophrenia is associated with lower HCRU, lower risk of hospitalization, and reduced healthcare costs compared with later initiation of AOM. This study builds on prior evidence showing the benefit of earlier use of LAIs in patients with schizophrenia,³⁰ which utilized data from 2007 to 2016, with a sample identification timeframe ending in mid-2015. Since AOM was launched in 2013 for the treatment of schizophrenia, AOM was likely under-represented in that sample. Assessing the effect of AOM was important to understanding whether the benefits first seen in the early use of LAIs would extend to AOM also. Furthermore, whereas clinical trials have demonstrated the efficacy of AOM on hospitalization outcomes in patients with early/new-onset schizophrenia,^29,39 the current study extends this evidence to show that the effectiveness of AOM is increased when it is initiated within 1 year of known diagnosis.

Considering healthcare costs, the adjusted total all-cause and psychiatric-specific costs during the 1-year follow-up period were each 25% lower following early AOM 400 initiation compared with late initiation. This study showed better results for AOM than a previous claims-based study of the timing of LAI initiation (any LAI) in patients with schizophrenia. In that study, all-cause and psychiatric-specific healthcare costs were 4% and 11% lower, respectively, with early initiation compared with later initiation³⁰ and in this study, costs were 25% lower. Comparisons are difficult between studies due to differences in study design. First, the present study utilized Medicaid data only, whereas the previous study utilized commercial insurance, Medicare, and Medicaid data. The majority of individuals with schizophrenia on LAIs are covered by Medicaid,³⁰ so this change in the current study was appropriate. Second, the extent to which the mix of LAIs, and differences between individual LAIs, affected the results is unknown. However, based on these two studies, it could be hypothesized that the early initiation of AOM has a potentially larger effect on healthcare costs than the early initiation of other LAIs in patients with new episodes of schizophrenia.

In the present study, the likelihood of all-cause and psychiatric-specific hospitalization were 62% and 39% higher, respectively, for the late cohort compared with the early cohort, which is comparable to the findings of the previous claims-based study of the timing of LAI initiation.³⁰ Of note, 56% of patients in the present study initiated AOM within 1 year after beginning a new episode of care for schizophrenia, which is higher than the proportion of patients who initiated any LAI within 1 year in the previous claim-based study (32%).³⁰ This difference could reflect higher confidence in AOM compared with other LAIs, which may be particularly relevant given that negative perceptions of LAIs are common among healthcare providers and patients with schizophrenia.^28,40,41 The difference could also result from changes in LAI prescription rates over time (the current study utilized data from 2013 to 2019, and the previous study from 2007 to 2016³⁰) or differences in prescribing patterns between health insurance databases (i.e. Medicaid versus commercial insurance). AOM was likely under-represented in the prior study, since that study’s sample identification timeframe ran through 30 June 2015,³⁰ and AOM launched in 2013, whereas other LAIs had been on the market for much longer.

Limitations

This study was affected by the inherent limitations of claims-based analyses since claims data are intended for administrative purposes rather than as a tool for research. For instance, it was not possible to determine whether a patient’s first new episode of care for schizophrenia during the study period was their true first diagnosis of schizophrenia, and data relating to baseline clinical characteristics (such as schizophrenia severity and duration of schizophrenia diagnosis) were not available. The study population was limited to patients with Medicaid health benefits (excluding those with Medicare and Medicaid dual eligibility). Patients younger than 26 years were, potentially, excluded if they received commercial insurance coverage through their parents’ health plan. The study included a washout period of 1 year without a schizophrenia diagnosis prior to the first diagnosis in the dataset for schizophrenia. However, we did not exclude patients who had a claim for an antipsychotic during that period, as they could have been taking an antipsychotic for a different indication (e.g. bipolar disorder). However, excluding these patients may have had an effect on the results. The threshold between early and late AOM initiation was set at 1 year, an arbitrary choice that allowed comparison with a prior claims database study.³⁰ The study also included patients with any number of claims for AOM, rather than excluding those patients with only one claim (and for whom long-term use may not be properly reflected) to be consistent with the prior claims database study.³⁰ The study also did not separate the cohorts based on the dose of AOM (e.g. proportion of patients in each cohort taking 400 mg or 300 mg doses of AOM) so the effect of dose on outcomes cannot be ascertained; clinical data have shown that approximately 99% of patients begin treatment using the 400 mg dose and 82% stay on the 400 mg dose throughout treatment⁴² so the effect of dose would likely not be a factor in this analysis. These differences in AOM use may be an area for further research. Finally, as this study involved analysis of administrative claims data, it was not possible to collect individual adverse event information. However, previous studies have shown that AOM is well tolerated in patients with schizophrenia.^43–45

Conclusions

Early initiation of AOM, within 1 year after beginning a new episode of care for schizophrenia, was associated with statistically significant reductions in all-cause and psychiatric-specific HCRU and healthcare costs, and reduced likelihood of all-cause and psychiatric hospitalization, compared with later AOM initiation. Thus, early initiation of AOM appears to provide meaningful benefits for both patients and healthcare providers.

Transparency

Declaration of funding

This work was supported by Otsuka Pharmaceutical Development & Commercialization Inc. (Princeton, NJ, USA) and Lundbeck LLC (Deerfield, IL, USA). Authors affiliated with the sponsors were involved in the design of the study, the analysis and interpretation of data, the writing and reviewing of this article, and the decision to submit the article for publication.

Declaration of financial/other relationships

HCW and SB are full-time employees of Otsuka Pharmaceutical Development & Commercialization, Inc. RS was a full-time employee of Otsuka Pharmaceutical Development & Commercialization Inc. at the time of this work. MT was a full-time employee of Lundbeck LLC. at the time of this work. HF is a full-time employee of Lundbeck LLC. LC has acted as a consultant for AbbVie/Allergan, Acadia, Adamas, Alkermes, Angelini, Astellas, Avanir, Axsome, BioXcel, Boehringer Ingelheim, Cadent Therapeutics, Eisai, Enteris BioPharma, HLS Therapeutics, Impel, Intra-Cellular Therapies, Janssen, Karuna, Lundbeck, Lyndra, Medavante-ProPhase, Merck, Neurocrine, Novartis, Noven, Otsuka, Ovid, Relmada, Reviva, Sage, Sunovion, Teva, and University of Arizona, and has performed one-off ad hoc consulting for individuals/entities conducting marketing, commercial, or scientific scoping research; has acted as a speaker for AbbVie/Allergan, Acadia, Alkermes, Angelini, Eisai, Intra-Cellular Therapies, Janssen, Lundbeck, Neurocrine, Noven, Otsuka, Sage, Sunovion, Takeda, Teva, and CME activities organized by medical education companies such as Medscape, NACCME, NEI, Vindico, and universities and professional organizations/societies; owns stocks (small number of shares of common stock) in Bristol-Myers Squibb, Eli Lilly, J & J, Merck, Pfizer (purchased >10 years ago), and has stock options for Reviva; and has received royalties from Wiley (Editor-in-Chief, International Journal of Clinical Practice, through end 2019), UpToDate (reviewer), Springer Healthcare (book), and Elsevier (Topic Editor, Psychiatry, Clinical Therapeutics).

Author contributions

HCW, MT, HF and LC contributed to the design of the study and interpretation of data. RS contributed to the design of the study, and was involved in the acquisition, analysis, and interpretation of data. SB was involved in the acquisition, analysis, and interpretation of data. All authors participated in the drafting or the critical review of the article, gave final approval of the version to be published, and agree to be accountable for all aspects of the work.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Previous presentations

Parts of this study were presented at the American Society of Clinical Psychopharmacology (ASCP) Annual Meeting; May 31–June 03, 2022; Scottsdale, AZ.

Acknowledgements

Writing support was provided by Gemma Shaw, VetMB, assisted by her colleagues at Cambridge – a Prime Global Agency (Cambridge, UK), and funded by Otsuka Pharmaceutical Development & Commercialization Inc. and Lundbeck LLC. The authors would like to acknowledge Maxine Chen, PhD, a former employee of Otsuka Pharmaceutical Development & Commercialization Inc., for her contributions to the study methodology.

Data availability statement

The data that support the findings of this study are available from IBM. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the authors with the permission of IBM.