A comparison of treatment patterns, healthcare resource utilization, and costs among young adult Medicaid beneficiaries with schizophrenia treated with paliperidone palmitate or oral atypical antipsychotics in the US

Abstract

Background: Much of the burden associated with schizophrenia is attributed to its early onset and chronic nature. Treatment with once monthly paliperidone palmitate (PP1M) is associated with lower healthcare utilization and better adherence as compared to oral atypical antipsychotics (OAAs). This study aimed to evaluate real-world effectiveness of PP1M and OAA therapies among US-based adult Medicaid patients with schizophrenia, overall and among young adults aged 18–35 years.

Methods: Adult patients with a diagnosis of schizophrenia and at least two claims for PP1M or OAA between January 1, 2010 and December 31, 2014 were selected from the IBM Watson Health MarketScan Medicaid Database. Treatment patterns and healthcare resource utilization and costs were compared between PP1M and OAA treatment groups following inverse probability of treatment (IPT) weighting to adjust for potential differences. Utilization and cost outcomes were estimated using OLS and weighted Poisson regression models.

Results: After IPT weighting, the young adult PP1M and OAA cohorts were comprised of 3,095 and 3,155 patients, respectively. PP1M patients had a higher duration of continuous treatment exposure (168.2 vs 132.5 days, p = .004) and better adherence on the index medication (proportion of days covered ≥80%: 19.0% vs 17.1%, p < .049). Young adults treated with PP1M were 37% less likely to have an all-cause inpatient admission (odds ratio [OR] = 0.63, 95% confidence interval [CI] = 0.53–0.74) and 33% less likely to have an ER visit (OR = 0.67, 95% CI = 0.55–0.81) compared to OAA young adult patients, but 27% more likely to have an all-cause outpatient office visit (OR = 1.27, 95% CI = 1.02–1.56). PP1M patients incurred significantly lower medical costs as compared to OAA patients.

Conclusions: Medicaid patients with schizophrenia treated with PP1M have higher medication adherence and have fewer hospitalizations as compared to patients treated with OAAs. PP1M may lead to reduced healthcare utilization and improved clinical outcomes.

Keywords:

• Paliperidone palmitate
• schizophrenia
• treatment patterns
• oral atypical antipsychotics
• healthcare resource utilization

JEL classification:

• I10
• I19

Introduction

Schizophrenia is a burdensome and costly mental illness, with an estimated prevalence of 0.51–1.1% among adults in the US, and is the most common psychotic disease globally^1,2. Schizophrenia is characterized by recurrent episodes of acute psychosis alternating with periods of full or partial remission³. The total cost of schizophrenia was estimated at $155.7 billion in 2013, with a substantial portion incurred from direct healthcare costs⁴. A large proportion of the burden associated with schizophrenia is also attributed to early onset of the disease and its chronic nature, specifically early-to-mid 20s for men and late 20s for women⁵. The disease lends itself to increased hospital admissions, physician visits, prescription medications, as well as indirect costs, such as unemployment and premature mortality, which result in a substantial cost burden². A recent study highlighted the substantial economic burden among young adults (aged 18–35 years) with schizophrenia⁶. Based on their analysis, further research is warranted to fully understand the association between treatment patterns, such as adherence in the early phases of the disease and longer-term health outcomes among young adult patients with schizophrenia.

The current recommended treatment for patients with schizophrenia in the US is antipsychotic (AP) therapy, which can reduce the severity and frequency of disease symptoms with adequate adherence. Oral antipsychotics, including oral atypical antipsychotics (OAAs), are the mainstay of schizophrenia treatment⁷. However, oral medication non-adherence is common, with an estimated 40–60% of patients with schizophrenia being non-adherent^8,9. Patients with schizophrenia may have different treatment trajectories, with younger patients being more likely to discontinue and less likely to be adherent as compared to older patients^6,10. Adequate treatment in the early stages of the disease is important in terms of clinical outcomes and disease course and may determine a patient’s long-term prognosis^11,12. Recent literature suggests that long-acting injectable therapies may be associated with lower rates of relapse and reduced morbidity as compared to OAAs^1,9. The long-acting injectable antipsychotic once-monthly paliperidone palmitate (PP1M) was approved by the US Food and Drug Administration in 2009 for the treatment of schizophrenia and in 2014 for schizoaffective disorder¹³. PP1M has been shown to be effective among an overall cohort of patients that are relatively older (mean age ≈ 41), on average¹⁴. Additionally, recent findings suggest that treatment with long-acting injectable therapies, such as once-monthly PP1M, is associated with lower healthcare resource utilization and better adherence as compared to treatment with OAAs among adult patients with schizophrenia^14,15. However, the literature that assesses the benefits of PP1M in patients across the age spectrum, particularly patients aged 18–35 years, is limited.

This study aimed to further evaluate treatment patterns, healthcare utilization, and cost outcomes in adult patients with schizophrenia and among a sub-set of young adults aged 18–35 years. Specifically, the study compared treatment patterns, healthcare resource utilization, and costs during a 12-month period following initiation of PP1M or OAA therapy between 2010 and 2015.

Methods

Study design and data source

This was a retrospective, observational cohort study of US adult Medicaid beneficiaries diagnosed with schizophrenia and receiving PP1M or OAA treatment between January 1, 2010 and December 31, 2014. Treatment patterns and healthcare resource utilization and costs were compared among patients who were initiated on PP1M vs OAAs. All outcomes were evaluated for the overall cohort (aged 18–64 years), as well as a sub-set of young adult patients (aged 18–35 years).

US administrative claims data were extracted from the IBM Watson Health MarketScan Medicaid Multi-State Database. This database comprises enrollment information, demographic information, and inpatient medical, outpatient medical, and outpatient pharmacy claims data for ∼44.2 million Medicaid enrollees between 1999 and 2016, including 8.2 million lives in 2016. Data come from multiple states that are geographically dispersed.

Data in the Medicaid Database are compliant with US patient confidentially requirements (including the Health Insurance Portability and Accountability Act of 1996 [HIPAA] regulations). The study was exempt from Institutional Review Board review, as it did not involve the collection, use, or transmittal of individually identifiable data.

Patient selection

While the study period spanned from January 1, 2009 through December 31, 2015, the patient identification period was January 1, 2010 through December 31, 2014. Adults (≥18 years of age) receiving PP1M or OAA treatment between January 1, 2010 and December 31, 2014 were identified within the database as patients with at least two outpatient prescription or medical claims for PP1M or the same OAA within 90 days of each other. OAA agents included: aripiprazole, asenapine maleate, iloperidone, lurasidone, olanzapine, quetiapine fumarate, paliperidone, risperidone, and ziprasidone. The date of the earliest claim was considered the index date, and the initial treatment (PP1M or OAA) was used to define the treatment cohorts. Patients were required to have at least 12 months of continuous enrollment with medical and pharmacy benefits prior to the index date (baseline period) to assess newly-treated PP1M and OAA patients by excluding previous medication use during the baseline period and at least 12 months of continuous enrollment after the index date (follow-up period) in order to evaluate annual healthcare resource utilization and costs while providing sufficient follow-up data to examine treatment patterns. Patients were also required to have at least one inpatient or outpatient medical claim with a diagnosis of schizophrenia (International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) 295.xx; International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) F20.xx) anytime during the study period. Patients with PP1M use during the baseline period were excluded. Additionally, OAA patients with previous OAA use (of the same index agent) during the baseline period or with PP1M use during the follow-up period were excluded¹⁴.

Study variables

Study sample demographics (age, gender, insurance plan type, race, length of follow-up, index year) and clinical characteristics (Deyo-Charlson Comorbidity Index (DCI))¹⁶, mental health-related comorbidities, non-mental health related comorbidities) were identified using ICD-9-CM and ICD-10-CM diagnosis codes measured during the 12-month baseline period.

Outcome variables included all-cause healthcare resource utilization and costs and treatment patterns. Outcomes were compared during the 12-month follow-up period between the PP1M and OAA treatment groups, overall (aged 18–64 years) and for the 18–35 years age group, specifically.

All-cause healthcare resource utilization and costs were measured during the baseline and follow-up periods. Healthcare utilization was reported by type of service in the following categories: inpatient (IP), outpatient (OP), including emergency room (ER), OP office visits, laboratory tests, other OP, and OP prescriptions. Healthcare costs were based on paid amounts of adjudicated claims, including insurer and health plan payments, as well as patient cost-sharing in the form of copayment, deductible, and coinsurance. Difference in mean per member, per month (PMPM) healthcare costs were reported in the following categories: IP, OP (including ER, office visit, laboratory and other OP), OP prescription, and total costs. All dollar estimates were inflated to 2016 dollars using the Medical Care Component of the Consumer Price Index. To account for the mandatory minimum discount for branded pharmaceutical products in Medicaid where the amount of rebate for each unit of a drug is based on a statutory formula, a 23.1% discount was applied to all prescription drug claims for branded drugs to estimate total medical costs and total outpatient prescription costs during the 12-month follow-up period as a sensitivity analysis^17,18.

Treatment pattern outcomes included duration of continuous exposure to index treatment (number of days with index drug on hand with a gap of no greater than 90 days), proportion of patients with treatment gaps of ≤30 and ≤60 days, number of unique AP agents received (during baseline and follow-up), concomitant AP use other than index treatment (during baseline and follow-up), concomitant psychiatric medication use (during baseline and follow-up (antidepressants, anxiolytics, mood stabilizers, or none)), and polypharmacy (AP and psychiatric). AP polypharmacy was defined as overlapping coverage of two or more unique AP agents for 60 or more consecutive days with no more than a 7-day gap, while psychiatric polypharmacy was defined as overlapping coverage of one or more AP agent with at least one anxiolytic, antidepressant, or mood stabilizer for 60 or more consecutive days with no more than a 7-day gap¹⁴. Adherence (captured as proportion of days covered (PDC)) was measured for index treatment, as well as any AP drug, and patients with PDC ≥80% were considered adherent.

Statistical analysis

Inverse probability of treatment (IPT) weighting was conducted to adjust for potential baseline demographic and clinical characteristics between PP1M and OAA treatment groups without reducing the size of the study sample¹⁹. Weights were calculated based on propensity scores (PS): 1/PS for the PP1M group and 1/(1 – PS) for the OAA group and normalized by dividing each weight by the mean. Multivariable regression models were used to estimate PS while adjusting for age, gender, plan type, race, index year, baseline DCI, presence of baseline comorbidities (substance abuse, cardiovascular disease, hypertension, hepatitis C virus, HIV, obesity), number of unique mental health diagnoses during baseline, baseline healthcare resource utilization (number of patients with an IP admission, ER visit, and OP office visit), baseline total costs, number of unique AP agents received, AP use during baseline (typical and atypical oral, typical and atypical long-acting injectables), concomitant psychiatric medication use (antidepressant, anxiolytic, mood stabilizers, or none) and AP polypharmacy during baseline.

Both IPT-weighted and unweighted descriptive results were presented for the overall study cohorts, as well as young adult patients aged 18–35 years. Categorical variables were presented as the counts and percentages of patients in each category, while continuous variables were summarized as the means, standard deviations, and medians. Standardized differences were used to compare baseline demographic and clinical characteristics to assess the quality of IPT weighting. Variables with standardized differences of ≤10% were considered well-balanced²⁰. Treatment pattern and healthcare resource and utilization outcomes were compared between the PP1M and OAA cohorts using the Pearson Chi-square test for categorical variables and Student’s t-test for continuous variables.

Multivariable modeling was used to compare healthcare resource and utilization outcomes between the PP1M and OAA treatment groups. Total costs, outpatient pharmacy costs, and medical (e.g. inpatient and outpatient) costs during the 12-month follow-up period were compared using weighted ordinary least squares (OLS) regression models, adjusting for treatment group, baseline total medical costs, and number of unique mental health diagnoses during baseline. OLS models were chosen because the distribution of residuals indicated their appropriateness compared to other generalized linear modeling techniques specifying a gamma distribution and log link function. Total rebated costs and rebated outpatient pharmacy cost outcomes were also examined using OLS as part of a sensitivity analysis. Weighted logistic regression models were used to estimate odds ratios (ORs) related to healthcare utilization of IP admissions, ER visits, and OP office visits, adjusting for treatment group, average length of hospital stay, and number of unique mental health diagnoses. A generalized linear model, assuming an underlying negative binomial distribution, and weighted via propensity weights, was used to examine count data regarding healthcare utilization, adjusting for treatment group, baseline average length of hospital stay, and number of unique mental health diagnoses during baseline. The 95% confidence intervals (CIs) for cost outcomes were estimated using a non-parametric bootstrap procedure. All analyses were performed using SAS, version 9.4 (SAS Institute, Inc., Cary, NC).

Results

Demographic and clinical characteristics

A total of 949 PP1M patients and 14,649 OAA patients met the study inclusion criteria (Figure 1). Among these, 439 PP1M patients and 5,811 OAA patients were aged 18–35 years. Following IPT-weighting, the corresponding cohorts consisted of 7,672 PP1M and 7,926 OAA patients overall, and of 3,095 PP1M and 3,155 OAA patients aged 18–35 years. Tables 1 and 2 present selected IPT-weighted baseline demographic and clinical characteristics for the overall cohort and the sub-set of young adult patients.

Figure 1. Patient selection.

Table 1. Selected demographic characteristics at index, overall (aged 18–64 years) and aged 18–35 years.

Demographic characteristic	All patients					Aged 18–35 years
	PP1M patients		OAA patients		Std. Diff.	PP1M patients		OAA patients		Std. Diff.
	(n = 7,672)		(n = 7,926)			(n = 3,095)		(n = 3,155)
	n/Mean	%/SD	n/Mean	%/SD		n/Mean	%/SD	n/Mean	%/SD
Age, (Mean, SD)	40.3	35.7	40.0	9.5	1.4%	26.4	12.8	26.7	3.7	2.8%
Male, n (%)	3746	48.8%	3,873	48.9%	0.1%	1,659	53.6%	1,721	54.5%	1.9%
Race, n (%)
White	2,619	34.1%	2,723	34.4%	0.5%	1,089	35.2%	1,054	33.4%	3.7%
Black	4,024	52.4%	4,155	52.4%	0.1%	1,551	50.1%	1,626	51.5%	2.8%
Hispanic	78	1.0%	101	1.3%	2.3%	35	1.1%	52	1.6%	4.3%
Other	951	12.4%	947	11.9%	1.4%	419	13.5%	423	13.4%	0.4%
Insurance Plan Type, n (%)
Comprehensive/Indemnity	4,637	60.4%	4,842	61.1%	1.4%	1,893	61.2%	1,817	57.6%	7.3%
POS/POS w/cap	844	11.0%	891	11.2%	0.8%	288	9.3%	323	10.2%	3.1%
HMO	2,191	28.6%	2,191	27.6%	2.0%	914	29.5%	1,014	32.2%	5.7%
Other	0	0%	1	0.1%	1.7%	0	0%	1	0.03%	2.6%

Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; Std. Diff., standardized difference; POS, point-of-service; HMO, health maintenance organization.

Table 2. Selected clinical characteristics, 12-month baseline period, overall (aged 18–64 years) and aged 18–35 years.

Clinical characteristic	All patients					Aged 18–35 years
	PP1M patients		OAA patients		Std. Diff.	PP1M patients		OAA patients		Std. Diff.
	(n = 7,672)		(n = 7,926)			(n = 3,095)		(n = 3,155)
	n/Mean	%/SD	n/Mean	%/SD		n/Mean	%/SD	n/Mean	%/SD
Depression Disorder, n (%)	2,663	34.7%	3,355	42.3%	15.7%	1,124	36.3%	1,313	41.6%	10.9%
Tobacco Use, n (%)	2,996	39.0%	3,037	38.3%	1.5%	1,244	40.2%	1,171	37.1%	6.3%
Bipolar Disorder, n (%)	2,296	29.9%	2,397	30.2%	0.7%	1,210	39.1%	1,045	33.1%	12.4%
Substance Abuse, n (%)	1,930	25.2%	2,113	26.7%	3.4%	897	29.0%	892	28.3%	1.5%
Number of Unique Mental Health Diagnoses (Mean, SD)	2.8	6.9	2.7	1.7	1.2%	3.3	8.0	2.8	1.7	7.3%

Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; Std. Diff., standardized difference.

Prior to IPT-weighting (using unadjusted samples), young adult PP1M patients were slightly older than OAA patients (27.4 vs 26.6, standardized difference = 21.0%), more likely to be male (72.2% vs 53.1%, standardized difference = 40.3%), and of Black race (61.3% vs 50.7%, standardized difference = 50.7%), while the overall cohort of PP1M patients was slightly younger than OAA patients (38.5 vs 40.1, standardized difference = 17.4%).

Young adult PP1M patients had a lower number of unique mental health diagnoses (2.1 vs 2.9, standardized difference = 49.4%) and a lower proportion of most mental-health related comorbidities as compared to young adult OAA patients (e.g. anxiety, bipolar disorder, depression disorder, tobacco use). Among the overall cohort, PP1M patients had a lower baseline DCI (0.5 vs 0.9, standardized difference = 35.1%), a lower number of unique mental health diagnoses (2.0 vs 2.8, standardized difference = 51.4%), and a lower proportion of most mental health-related comorbidities as compared to the overall OAA cohort.

Following IPT-weighting, all baseline demographic characteristics and most baseline clinical characteristics were well-balanced between the treatment groups, both overall and for the young adult cohort.

Treatment patterns

IPT-weighted treatment outcomes compared during the 12-month follow-up period are presented in Table 3. Among young adults aged 18–35 years, PP1M patients had a longer duration of continuous exposure to the index treatment as compared to OAA patients (168.2 vs 132.5 days, p = .004). While the PP1M cohort had a lower proportion of patients without a treatment gap of >30 days (55.0% vs 64.8%, p < .001) as compared to OAA patients, they had a higher proportion of patients without a treatment gap of >60 and >90 days. Additionally, a higher proportion of PP1M patients had a PDC ≥80% for their index drug (19.0% vs 17.1%, p = .049) and any AP drug (29.6% vs 23.5%, p < .001) as compared to OAA patients. The use of other AP agents (other than index treatment) was higher among the OAA cohort, and OAA patients were more likely to have concomitant psychiatric medication use as compared to PP1M patients (antidepressant use: 63.7% vs 46.4%, anxiolytic use: 48.2% vs 35.7%, and mood stabilizer use: 42.8% vs 38.1%, all p < .001). The number of unique AP agents received and AP use during baseline was similar between the two groups. PP1M patients were less likely to have AP polypharmacy (5.3% vs 7.3%, p < .001) and psychiatric polypharmacy (35.2% vs 41.9%, p < .001) during follow-up as compared to OAA patients.

Table 3. Treatment patterns, 12-month follow-up period, overall (aged 18–64 years) and aged 18–35 years.

Treatment outcome	All patients					Aged 18–35 years
	PP1M patients		OAA patients		p-value	PP1M patients		OAA patients		p-value
	(n = 7,672)		(n = 7,926)			(n = 3,095)		(n = 3,155)
	n/Mean	%/SD	n/Mean	%/SD		n/Mean	%/SD	n/Mean	%/SD
Duration of continuous exposure to index drug,days (Mean, SD)	176.8	348.6	148.9	92.2	.010	168.2	316.0	132.5	86.9	.004
PDC for index drug (Mean, SD)	0.5	0.9	0.4	0.2	.118	0.5	0.8	0.4	0.2	.161
PDC ≥80% (n, %)	2,048	26.7%	1,769	22.3%	<.001	589	19.0%	540	17.1%	.049
Concomitant medication use, n (%)
Antidepressants	3,910	51.0%	5,334	67.3%	<.001	1,437	46.4%	2,009	63.7%	<.001
Anxiolytics	3,198	41.7%	4,085	51.5%	<.001	1,104	35.7%	1,521	48.2%	<.001
Mood stabilizers	2,844	37.1%	3,444	43.4%	<.001	1,178	38.1%	1,349	42.8%	<.001
None of the above	2,154	28.1%	1,207	15.2%	<.001	976	31.5%	552	17.5%	<.001
Polypharmacy, n (%)
Antipsychotic	690	9.0%	752	9.5%	.292	164	5.3%	232	7.3%	<.001
Psychiatric	3,271	42.6%	3,820	48.2%	<.001	1,089	35.2%	1,323	41.9%	<.001

Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; PDC, proportion of days covered.

Treatment pattern outcomes among the overall cohort were similar to that of the young adult cohort. PP1M patients had a longer duration of continuous exposure to the index drug (176.8 vs 148.9 days, p = .010), a higher proportion of patients with a PDC ≥80% for their index drug (26.7% vs 22.3%, p < .001) and any AP drug (36.8% vs 29.9%, p < .001), and less concomitant AP and psychiatric medication use as compared to OAA patients. While PP1M patients were less likely to have psychiatric polypharmacy (42.6% vs 48.2%, p < .001) than OAA patients, the proportion of PP1M and OAA patients with AP polypharmacy was similar (9.0% vs 9.5%, p = .292).

Healthcare resource utilization

Among young adult patients, the PP1M cohort was 37% less likely to have an all-cause IP admission (OR = 0.63, 95% CI = 0.53–0.74), 33% less likely to have an all-cause ER visit (OR = 0.67, 95% CI = 0.55–0.81), but 27% more likely to have an OP office visit (OR = 1.27, 95% CI = 1.02–1.56) as compared to the OAA cohort. The findings were similar for the overall cohort, where PP1M patients were less likely to have an all-cause IP admission (OR = 0.60, 95% CI = 0.54–0.66) and less likely to have an all-cause ER visit (OR = 0.60, 95% CI = 0.54–0.68) as compared to the overall OAA cohort (Table 4, Figures 2 and 3).

Figure 2. Likelihood of healthcare resource utilization, by resource type, compared between PP1M and OAA patients, overall (aged 18–64 years) and aged 18–35 years. Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; ER, emergency room; CI, confidence interval.

Figure 3. Differences in healthcare resource utilization, by resource type, overall (aged 18–64 years) and aged 18–35 years. Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; IP, inpatient; ER, emergency room; CI, confidence interval.

Table 4. All-cause healthcare utilization, 12-month follow-up period, overall (aged 18–64 years) and aged 18–35 years.

All-cause healthcare utilization	All patients					Aged 18–35 years
	PP1M patients		OAA patients		p-value	PP1M patients		OAA patients		p-value
	(n = 7,672)		(n = 7,926)			(n = 3,095)		(n = 3,155)
	n/Mean	%/SD	n/Mean	%/SD		n/Mean	%/SD	n/Mean	%/SD
Inpatient admissions
Patients with an admission (n, %)	1,968	25.6%	2,691	33.9%	<.0001	902	29.1%	1,042	33.0%	.0008
Number of admissions per patient (Mean, SD)	0.5	3.1	0.8	1.3	.109	0.5	3.0	0.7	1.2	.083
Average length of stay per admission (Mean, SD)	7.3	23.1	6.8	5.5	.030	7.3	22.6	6.9	6.5	.024
Outpatient services
Patients with an ER visit (n, %)	4,194	54.7%	5,214	65.8%	<.0001	1,890	61.1%	2,161	68.5%	<.0001
Number of ER visits per patient (Mean, SD)	2.2	13.9	3.1	4.4	.090	2.4	12.3	3.4	4.6	.109
Patients with an OP office visit (n, %)	6,377	83.1%	6,696	84.5%	.0197	2,453	79.2%	2,562	81.2%	.0522
Number of OP office visits per patient (Mean, SD)	6.8	25.6	7.2	5.7	.020	4.7	13.6	6.0	5.2	.129
Patients with a laboratory test (n, %)	6,116	79.7%	6,477	81.7%	.0015	2,275	73.5%	2,433	77.1%	.0009
Number of laboratory tests per patient (Mean, SD)	16.1	59.6	19.9	20.2	.085	12.7	47.0	16.8	18.5	.117
Patients with other OP services (n, %)	7,636	99.%	7,826	98.7%	<.0001	3,061	98.9%	3,108	98.5%	.1927
Number of other OP services per patient (Mean, SD)	79.1	291.0	74.1	73.4	.024	65.4	218.2	62.1	66.5	.020
Outpatient prescriptions
Patients with an OP prescription (n, %)	7,614	99.2%	7,926	100.0%	<.0001	3,084	99.6%	3,155	100.0%	.0008
Number of OP prescriptions per patient (Mean, SD)	51.4	131.5	57.5	38.8	.063	33.3	76.5	39.4	28.5	.105

Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; ER, emergency room; OP, outpatient.

Young adult PP1M patients had lower mean rates of healthcare utilization as compared to young adult OAA patients in most resource categories. Specifically, PP1M patients aged 18–35 years had less all-cause number of IP admissions (difference in means = –0.043 [–0.597, –0.0264]), number of all-cause ER visits (difference in means = –1.38 [–1.764, –1.005]), number of all-cause OP office visits (difference in means = 1.54 [–2.052, –1.030]), number of other OP services (difference in means = –12.25 [–16.715, –7.788]), and number of all-cause OP laboratory services (difference in means = –6.03 [–7.761, –4.303]) compared with OAA patients. Among the overall cohort, the findings were consistent for IP admissions, ER visits, OP office visits, and OP laboratory services. However, PP1M patients had a longer average length of stay (in days) as compared to OAA patients, both in the young adult (difference in means = 0.65 [0.154, 1.142]) and the overall cohort (difference in means = 0.56 [0.226, 0.894]).

Healthcare costs

Total costs were not significantly different between the PP1M and OAA cohorts, both overall (difference in per member per month (PMPM) mean costs = $225 [–31, 573]) and among young adults (difference in PMPM means = $242 [–62, 469]). Outpatient pharmacy costs were significantly higher for PP1M patients overall (difference in PMPM means: $634 [554, 728]) and among young adults (difference in PMPM means = $606 [552, 699]) as compared to OAA patients, while medical costs were significantly lower among the PP1M cohort overall (difference in PMPM means = –$408 [–652, –79]) and among young adults (difference in PMPM means = –$364 [–673, –184]) (Table 5).

Table 5. Differences in per member, per month (PMPM) mean all-cause healthcare costs, with and without 23.1% rebate, overall (aged 18–64 years) and aged 18–35 years.

Healthcare cost category	Age cohort	Difference in PMPMmean costs (PP1M vs OAA)	Lower 95% CI	Upper 95% CI
Total costs	Overall	$225.42	–$31.34	$573.81
	18–35	$241.82	–$61.56	$469.29
Outpatient prescription costs	Overall	$633.75	$554.47	$728.12
	18–35	$605.72	$551.74	$699.48
Medical costs	Overall	–$408.26	–$652.22	–$79.15
	18–35	–$363.90	–$672.95	–$183.55
Total rebated costs	Overall	$103.11	–$153.69	$408.61
	18–35	$32.21	–$209.31	$321.91
Rebated outpatient prescription costs	Overall	$433.86	$420.22	$556.68
	18–35	$472.33	$419.75	$534.30

Results were generated using ordinary least squares regression models adjusting for baseline total costs, number of unique mental health diagnoses, and study cohort.

Abbreviations. PP1M, once-monthly paliperidone palmitate; OAA, oral atypical antipsychotics; CI, confidence interval.

After applying a 23.1% discount (the amount the Centers for Medicare and Medicaid Services require manufacturers to rebate a drug’s average manufacturer price) to branded prescription drugs as part of the sensitivity analysis, total costs remained similar between PP1M and OAA, while prescription costs were higher among PP1M patients, both overall (difference in PMPM means = $434 [420, 557]) and among young adults (difference in PMPM means = $472 [420, 534]) (Table 5).

Discussion

The aim of this study was to provide real world evidence on the impact of PP1M vs OAA treatment on treatment patterns, healthcare resource utilization, and costs among Medicaid patients with schizophrenia, with a focus on young adults aged 18–35 years. Findings for young adult patients suggest that treatment with PP1M is statistically significantly associated with increased adherence and persistence, lower concomitant AP and psychiatric medication use, and lower polypharmacy relative to treatment with OAAs, during the 12-month follow-up period. These findings are consistent with that of Pilon et al.¹⁴, who showed that PP1M was associated with better adherence and less use of other psychiatric medications as compared to OAAs, among patients with schizophrenia aged 18–25 years. Among the overall cohort, treatment pattern outcomes were similar to that of the young adult cohort, with better adherence and persistence, and lower concomitant medication use among patients treated with PP1M compared to patients treated with OAAs. Non-adherence to schizophrenia treatment can lead to increased severity of symptoms and relapse²¹. While costs for patients with schizophrenia have been estimated to be $1,387 higher per month than for age- and gender-matched individuals from the general US population, non-adherent patients with schizophrenia incur even higher costs and require more healthcare services^22–24.

While both young adult PP1M patients and the overall PP1M cohort had higher outpatient prescription costs than OAA patients, they had lower medical costs and, thus, total costs were similar between the two treatment groups. Schizophrenia-related costs (data not shown) followed similar trends, with no statistically significant difference in total costs between the PP1M and OAA cohorts overall and for young adult patients. After applying a 23.1% rebate as part of a sensitivity analysis, prescription costs remained higher among PP1M patients, while total costs were similar between the two treatment groups. These findings are in line with previous research that showed comparable total costs between PP1M and OAA treatment groups, with significantly lower medical costs^14,25. Additionally, rebates for branded medications under Medicaid could be up to 28%¹⁸, and forecasting has shown that AP expenditures among Medicaid patients could be reduced to less than half of their current levels within 5 years²⁵.

Healthcare resource utilization was lower across all categories of care (IP, ER, OP, laboratory) among the young adult PP1M patients. PP1M patients were also less likely to have an ER visit and less likely to have an IP admission as compared to young adult OAA patients; however, PP1M patients had a longer average length of hospital stay. Given that PP1M patients were less likely to have an IP admission as compared to OAA patients, the higher prescription costs incurred by PP1M patients were offset by the lower medical costs. Xiao et al.²⁴ found that lower medical costs were primarily driven by lower IP utilization among Medicaid beneficiaries. Interestingly, while healthcare resource utilization findings were mostly consistent between the overall PP1M cohort and young adult PP1M patients, reductions in resource use were more pronounced in the younger cohort. Schizophrenia-related healthcare resource utilization also tended to be lower among the PP1M young adult patients (IP, OP, laboratory, other OP) as compared to the OAA cohort, and average length of hospital stay was lower among PP1M patients (7.3 vs 7.9 days, p = .036).

Anderson et al.²⁶ found that PP1M use was a strong predictor of treatment adherence among patients with schizophrenia in community behavioral health organizations. Healthcare resource use, in particular the number of hospitalizations, was another predictor of discontinuation²⁶. Therefore, given the association of PP1M use and increased adherence, as well as reduced healthcare resource utilization found in this study, PP1M may be a superior option to OAAs in the management of schizophrenia.

Limitations

The study sample was limited to patients with Medicaid coverage, and, thus, results of this analysis may not be generalizable to patients with schizophrenia with other types of insurance or without insurance coverage. This study relied on administrative claims data for individuals with Medicaid coverage. Therefore, the potential for misclassification of schizophrenia diagnosis, comorbid conditions, PP1M or OAA use, or study outcomes was present as patients were identified through administrative claims data rather than medical claims. While neither PP1M patients or OAA patients could have exposure to OAAs prior to index medications use, each was required to have at least two prescriptions of the same agent (PP1M or the same OAA agent) within 90 days of each order during the follow-up period, and no previous use of their index agent during the baseline period in order to be included in the study sample. The data contained within the database are subject to data coding limitations and data entry errors. Additionally, medication adherence was based on filled prescriptions. Patients were assumed to have taken the medications as prescribed, but claims-based adherence measures do not account for whether the medication was actually taken. Medicaid benefits for prescription drug coverage vary by state²⁷, thus, the amount reimbursed for branded AP drugs may differ. While our study did not examine outcomes by individual state, the data source included multiple geographically dispersed states. Further, changes to Medicaid prescription drug coverage may affect the results of future studies, but the impact of such changes was beyond the scope of the current study. Finally, systematic differences between the study cohorts that may have accounted for differences in study outcomes were adjusted for during IPT-weighting and multivariate analysis, but adjustment was limited to characteristics that could be measured from administrative claims data.

Conclusions

The findings of this study suggest that Medicaid patients with schizophrenia (aged 18–35 years and overall) treated with PP1M have higher medication adherence and are likely to have fewer hospitalizations as compared to patients treated with OAAs. Overall, patients with schizophrenia treated with PP1M were likely to incur lower rates of healthcare utilization and lower medical costs as compared to patients treated with OAAs. While findings were generally consistent for the overall PP1M cohort and young adult PP1M patients, young adults had a lower rates of healthcare resource utilization than did the overall cohort. While a direct comparison between age groups was not made, these results may suggest a potential impact of PP1M on this age group specifically. PP1M is a viable treatment option for young adult patients with schizophrenia, and may lead to reduced overall healthcare utilization and improved clinical outcomes. Future research is needed to evaluate the long-term effects of initiating PP1M treatment in young adult patients with schizophrenia and to confirm these findings among patients with schizophrenia with other types of insurance.

Transparency

Declaration of financial/other interests

TBA and ACE are employees of Janssen. AV, JM, and AC are employees of IBM Watson Health, which received compensation from Janssen to complete this study. PJ was an employee of IBM Watson Health during the completion of the study. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Previous presentations

A portion of the findings from this study were presented at the International Society of Pharmacoeconomics and Outcomes Research (ISPOR) 23^rd Annual International Meeting on May 19–23, 2018 in Baltimore, MD and at the American Society of Clinical Psychopharmacology (ASCP) Annual Meeting on May 29–June 1, 2018 in Miami, FL.

Acknowledgments

No medical writing assistance in the preparation of this article is to be declared.

Additional information

Funding

This study was funded by Janssen Scientific Affairs, LLC.