Healthcare utilization and costs of veterans health administration patients with schizophrenia treated with paliperidone palmitate long-acting injection or oral atypical antipsychotics

Abstract

Objective:

This study aimed to compare real world healthcare costs and resource utilization between patients with schizophrenia treated with paliperidone palmitate long-acting injection (PP) and oral atypical antipsychotics (OAT).

Methods:

Patients (18–64 years) were selected from the Veterans Health Administration dataset (1 July 2007–31 May 2012). Patients with 2+ claims for PP or 2+ claims for the same OAT comprised the two study cohorts with the first prescription date designated as the index date. Participation in the VA healthcare system for 24 months pre- and 12 months post-index, schizophrenia diagnosis (International Classification of Disease 9th Revision Clinical Modification [ICD-9-CM] code 295.1x–6x, 295.8x–9x) and ≥1 claim for an antipsychotic medication during the baseline period were required. Propensity scores and Mahalanobis metric distances with calipers were used to create two matched cohorts. All-cause healthcare utilization and costs for the 12-month follow-up period were compared between matched cohorts.

Results:

The matching process produced two cohorts of 335 patients with similar baseline characteristics. During the 12-month follow-up period, patients in the PP cohort had lower mean inpatient costs (18,560 vs $31,505, p = 0.002), lower frequency of hospitalization (34% vs 53%, p < 0.001) and fewer average inpatient days (13.24 vs 24.18, p = 0.002) vs matched OAT patients. While mean pharmacy costs were higher for the PP cohort ($10,063 vs $4167, p < 0.001), mean total healthcare costs were not significantly different ($45,529 vs $52,569, p = 0.128).

Conclusion:

VA patients, diagnosed with schizophrenia and treated with PP, had lower inpatient costs and admission rates compared to a matched cohort of OAT patients. Total healthcare costs were not significantly different.

Keywords::

• Schizophrenia
• Paliperidone palmitate
• Long-acting antipsychotics
• Oral atypical antipsychotics
• Hospitalization
• Healthcare cost

Introduction

In 2011, schizophrenia affected ∼0.3–0.7% of the population, 24 million people worldwide1. With direct healthcare costs estimated at $22.7 billion in the US alone2, schizophrenia is among the most costly of all mental illnesses3. It is a chronic, complex, and debilitating mental illness which impacts patients, families, and the wider healthcare and social welfare system2. Patients with schizophrenia suffer from a variety of symptoms, including hallucinations, delusions, thought disorder, and cognitive impairment, which may result in difficulties achieving everyday tasks and interacting with others4.

Due to the complex nature of schizophrenia and treatment options from medications to behavioral therapy, schizophrenia can be challenging to manage in clinical practice1. Relapse associated with this debilitating illness has a major impact on the patient and the cost of care5. Not only does relapse correlate with increased healthcare costs, inpatient costs disproportionally accelerate as the number of relapses increase6, making relapse prevention a key goal in managing schizophrenic patient populations7.

The US Department of Health and Human Services estimated that, in 2012, only 61% of Medicaid patients with schizophrenia continuously refilled their antipsychotic prescriptions, underscoring the potential opportunity to reduce relapse through improved adherence8. One population that is significantly impacted by mental health conditions, including schizophrenia, is US veterans9. The VA operates the nation’s largest integrated healthcare system, with more than 1700 hospitals, clinics, community living centers, counseling centers, and other facilities10. In a retrospective database analysis, from October 2006 and September 2011, almost 60,000 incident cases of schizophrenia were diagnosed in the VA system, with an average yearly healthcare cost of $31,29711. Compared to US veterans with other illnesses, veterans with schizophrenia were found to occupy more hospital beds at any time12. The VA has implemented mental health intensive case management (MHICM) programs across the country to promote independent living, reduce the need for inpatient care, and decrease treatment costs. Information that can help improve the care of these individuals and reduce their economic impact on the VA system is of national importance13.

Paliperidone palmitate long-acting injection (PP) was approved in July 2009 by the US Food and Drug Administration (FDA) and is indicated for the treatment of schizophrenia in adults14. Unlike oral antipsychotic therapy, which requires patients to take their medication daily, PP is an injectable antipsychotic medication administered monthly by a healthcare professional.

Among the oral antipsychotics, the atypical category represents the majority of prescriptions for treating schizophrenia15. The hypothesis for a positive effect of PP on costs is supported by earlier studies of long-acting injectable antipsychotics16–18 and a post-hoc analysis of hospitalization data from an open-label extension trial of PP vs placebo19. The published health economic literature on PP is comprised primarily of economic models assessing cost-effectiveness or cost-utility20–22. The lack of comparative real-world economic evidence between PP and oral antipsychotics is due in part to the relatively short period of time that PP has been available23.

While oral antipsychotics and injectable antipsychotics are both indicated for the treatment of schizophrenia, formulary and policy decision-makers need evidence of real world economic outcomes associated with these therapies. Based on a data set from the Veteran’s Health Administration, this study evaluated economic outcomes among the VA population of patients with schizophrenia treated with PP compared to patients treated with OAT in a real-world setting. The purpose of this study is to compare healthcare costs and resource utilization from a payer perspective between schizophrenic patients treated with PP and OAT in a real world setting.

Methods

This retrospective study analyzed medical and pharmacy service records from the Veterans Health Administration (VHA) Medical SAS dataset from July 1, 2007 through May 31, 2012. This time frame was chosen due to the FDA approval of PP in July 2009 and the 24-month baseline period requirement. Study data were extracted from VHA Medical Inpatient and Outpatient data-sets and from VHA Decision Support System (DSS) data-sets. The VHA DSS is a longitudinal, relational database combining the clinical data and cost data needed to integrate expenses, workload, and patient utilization and allocate costs to the healthcare goods and services provided to VHA patients25.

Since the core study proposed herein did not involve the collection, use, or transmittal of individual identifiable data, Institutional Review Board (IRB) approval to conduct this study was not required as per VHA’s classification of the de-identified data for release. De-identified patient IDs were created by VHA and used to link health records in different settings within VHA. These de-identified IDs cannot be linked to any other ID in the outside database.

Study sample

The study included patients diagnosed with schizophrenia with at least two injections of PP (within 60 days of each other) or who received at least two prescriptions of the same OAT medication (within 60 days of each other) from July 1, 2009 through May 31, 2011. The first pharmacy claim date for PP or OAT was designated as the index date.

Patients were required to have a medical claim with a diagnosis code for schizophrenia (International Classification of Disease 9th Revision Clinical Modification [ICD-9-CM] code: 295.1x–295.6x, 295.8x–295.9x). Patients with claims showing schizoaffective disorder (ICD-9-CM 295.7x) were only included if they also had one of the ICD-9-CM codes listed as inclusion criteria.

To be included in the study, patients were required to meet the following criteria: aged ≥18 and <65 on the index date; continuous health plan enrollment for 24 months prior to the index date (baseline period) and for 12 months after the index date (follow-up period); at least one pharmacy claim for an oral or injectable antipsychotic in the baseline period; no injectable antipsychotic use between the first two claims for the index therapy, and no claims for PP in the 24-month baseline period. Individuals in the OAT cohort were also required to have no claims for their index OAT product for 6 months prior to the index date and no claims for PP in the 12-month follow-up period. These criteria defined a study population that had been exposed to baseline antipsychotic treatment, but had not been treated with PP prior to index.

Patients were assigned to one of two mutually exclusive cohorts, PP cohort or OAT cohort, depending on their index antipsychotic medication. These two cohorts comprised all patients qualified for the study, and formed the basis from which two matched cohorts were drawn.

Study variables

Baseline characteristics examined for all study patients included demographics, diagnoses, medication use, medical and psychosocial service utilization, and treatment costs. The number of psychiatric medications, including antidepressants, anticholinergics, mood-stabilizing agents, anxiolytics, sleep agents, and stimulants prescribed during the baseline period were calculated. Mental health disorders, individual comorbidities, and the Charlson Comorbidity Index (CCI), a score that evaluates 22 co-morbid conditions were measured during the baseline period.

For the 12-month follow-up period, the economic outcomes of interest included all-cause healthcare costs for inpatient stays, emergency department visits, outpatient services and pharmacy, plus total healthcare costs defined as the sum of all VA healthcare costs per patient. The costs fields used in this analysis represent VA allocated costs taken from each record of service (i.e. claim) and adjusted to 2012 US dollars using the medical care component of the Consumer Price Index. The utilization of inpatient resources was measured in three ways; the percentage of each treatment group experiencing a hospitalization, the number of hospitalizations, and the mean number of inpatient days per patient. Similar measures for emergency department utilization were also defined. Although the focus of the study is on cost outcomes associated with the treatment of schizophrenia, the diagnosis code entered on each individual claim is insufficient to reliably identify whether that claim is clinically related or unrelated to the patient’s schizophrenia. Therefore, the analysis used all observed claims for a patient and did not limit the analysis to claims with schizophrenia diagnosis codes.

Statistical analysis

For all study-qualified patients, variables for the 24-month baseline period and 12-month follow-up period were analyzed descriptively for the two cohorts. Statistical tests used to evaluate the differences between the cohorts included chi-square tests for categorical variables and Student t-tests for the means of continuous variables. The level of significance was set to α = 0.05.

A logistic regression model was fitted to determine the association of baseline variables with treatment cohort membership, designated as 1 for PP or 0 for OAT. Using this model, a propensity score was calculated for each study-qualified patient, representing the likelihood of receiving PP based on the patient’s baseline data. Covariates in the propensity model included demographic and clinical characteristics and pre-index treatment and resource utilization (as detailed above).

To form matched cohorts for comparison, propensity scores of patients from the OAT group were examined to find a match for each PP patient. Matches were selected as the closest propensity score within a Mahalanobis metric caliper set at one quarter of the standard error of the propensity scores26. If no OAT patients were available within the Mahalanobis distance from a given PP patient, the PP patient was not used in the matched cohort analysis. When two matched cohorts were formed, the quality of the match was assessed by descriptive statistics on the baseline variables and visual inspection of the propensity score histograms. Histograms depicting the frequency distribution of propensity scores from 0 to 1 were compared for PP and OAT cohorts comprising the overall study-qualified population and for the matched PP and OAT cohorts. Mean values for 12-month follow-up cost and resource utilization outcomes were compared between the two matched cohorts using student t-tests.

Results

Baseline characteristics of study-qualified patients

A total of 3918 patients were qualified for the study based on the inclusion and exclusion criteria. This study-qualified population was segmented into a cohort of 381 patients meeting the PP cohort criteria, and a cohort of 3537 meeting the OAT cohort criteria (Table 1). Significant differences in baseline characteristics between cohorts were observed for the majority of demographic, clinical, and economic baseline variables. Some notable differences between PP and OAT cohorts’ baseline characteristics, respectively, were the percentage of patients exposed to more than one antipsychotic in the 24 months prior to index (82% vs 43%, p ≤ 0.001), the percentage of patients with an inpatient admission (64% vs 52%, p ≤ 0.001), and mean total costs ($37,899 vs $30,055, p = 0.013).

Table 1. Baseline characteristics for study qualified and propensity score matched cohorts.

	Study qualified cohorts					Matched cohorts
	PP (n = 381)		OAT (n = 3537)			PP (n = 335)		OAT (n = 335)
24-month baseline period	n/Mean	%/SD	n/Mean	%/SD	p Value	n/Mean	%/SD	n/Mean	%/SD	p Value
Age, years	50.16	10.54	52.17	9.14	<0.001	51.25	9.85	51.20	10.32	0.945
Gender
Female	31	8%	266	8%	0.666	24	7%	29	9%	0.474
Male	350	92%	3271	92%	0.666	311	93%	306	91%	0.474
Psychiatric medication
Number of unique antipsychotic medications	2.66	1.30	1.71	1.02	<0.001	2.56	1.23	2.56	1.27	0.975
Baseline comorbidity
Charlson Comorbidity Index	1.21	1.21	1.34	1.31	0.058	1.25	1.25	1.29	1.20	0.659
Other mental health diagnoses	240	63%	2389	68%	0.072	211	63%	233	70%	0.072
Diabetes	95	24.93	895	25.30	0.875	87	26%	89	27%	0.861
Alcohol abuse	88	23%	863	24%	0.573	78	23%	89	27%	0.326
Obesity	139	36%	945	27%	<0.001	112	33%	108	32%	0.742
All-cause healthcare utilization
Any hospitalization	242	64%	1830	52%	<0.001	209	62%	216	64%	0.574
Number of hospitalizations per patient	2.14	3.32	1.55	2.71	0.001	2.08	3.33	2.17	3.00	0.724
Number of inpatient days per patient	30.07	53.70	38.17	100.77	0.012	30.69	56.17	44.13	118.45	0.061
Any ER visit	44	12%	757	21%	<0.001	43	13%	49	15%	0.501
Number of ER visits per patient	0.27	0.96	0.61	2.26	<0.001	0.30	1.01	0.57	2.88	0.108
Baseline all-cause healthcare costs per patient
Inpatient costs	$39,081	$57,566	$32,432	$77,978	0.040	$39,601	$59,805	$39,822	$80,791	0.968
Pharmacy costs	$10,601	$13,991	$4079	$5895	<0.001	$10,281	$14,320	$6401	$7533	<0.001
ER costs	$54	$268	$261	$1031	<0.001	$60	$284	$214	$1125	0.015
Outpatient office visit costs	$35,742	$32,440	$22,915	$26,719	<0.001	$33,879	$31,111	$34,483	$33,067	0.808
Other outpatient costs	$422	$1556	$341	$2787	0.377	$435	$1639	$195	$527	0.011
Total costs (Outpatient + Inpatient + Pharmacy)	$85,900	$73,363	$60,028	$86,699	<0.001	$84,257	$75,611	$81,116	$91,145	0.628

SD, standard deviation. All-cause healthcare utilization was calculated by the number of claims (e.g. hospitalizations, ER visits) in the health record and the proportion of the study population that had a claim. Costs from the health record were adjusted to 2012 US dollars using the medical care component of the Consumer Price Index (CPI).

Propensity score model

The fitting of a logistic regression model for the dependent variable representing PP treatment selection resulted in covariates controlling for demographics (i.e. age, gender, region), schizophrenia sub-type (i.e. paranoid type, co-morbid schizoaffective disorder), and baseline co-morbidities (i.e. post-traumatic stress disorder, substance & alcohol abuse, obesity, Charlson Comorbidity Index). The covariates representing pre-index treatment, measured over the 6 months prior to the index date, were the type of antipsychotic exposure (i.e. OAT only or ≥1 claim for an injectable antipsychotic), the number of unique antipsychotics used, and adherence measured as a percentage of days covered. Factors to control for baseline resource utilization were defined either for the 6 months prior to index (i.e. number of outpatient visits, number of MHICM visits) or the full 24 months prior to index (i.e. number of inpatient admissions, number of emergency department visits) based on the time period considered most likely to have an effect on treatment selection.

Matching

An examination of the propensity scores for the full study-qualified population showed that 60% of the OAT cohort had less than a 1% likelihood of receiving PP, based on their baseline attributes. Figure 1(a) depicts this spike at the low end of propensity scores for the OAT cohort in contrast to the more even dispersion of propensity scores across PP patients.

Figure 1. Propensity score distributions of the study qualified population cohorts and the matched cohorts. (a) Study qualified cohorts. (b) Matched cohorts. PP, paliperidone palmitate long-acting injection; OAT, atypical oral antipsychotic therapy.

Through application of the Mahalanobis caliper matching process, 335 of the 381 PP patients were successfully matched to similar OAT patients. Figure 1(b) provides a visual confirmation of the congruence of propensity distribution between the two matched cohorts. The visible congruence in Figure 1(b) illustrates that the individual probabilities of receiving PP are similarly distributed in the two matched cohorts. The matching process also accounted for many of the potential confounders observed at baseline. As illustrated in Table 1, after the propensity score matching, key characteristics are at the same level for both cohorts.

Few significant differences were found between baseline characteristics of the two matched cohorts. Comorbidities, as measured by the Charlson Comorbidity Index and frequencies of baseline individual comorbidities, were similar between the two groups, although the percentage of OAT cohort with ‘other mental health diagnosis’ was nominally higher than that of the PP cohort (70% vs 63%, p = 0.072). The mean numbers of hospital admissions in the 24 month baseline period for the PP and OAT groups were close (2.08 vs 2.17, p = 0.724). The average number of hospital days for the PP group (30.69) was not statistically different (p = 0.061) to that of the OAT group (44.13). Mean total costs and inpatient costs for the baseline period were not significantly different between the two matched cohorts. Mean baseline pharmacy costs were higher for the matched PP cohort ($10,281 vs $6401, p ≤ 0.001) and mean emergency department costs were lower for the PP cohort ($60 vs $214, p = 0.015), relative to the matched OAT cohort. Additionally, the mean of ‘other outpatient costs’ was lower for the PP cohort ($435 vs $195, p = 0.011) compared to the matched OAT cohort.

Matched cohort outcomes

Economic outcomes were compared between the matched cohorts for the 12 month post-index follow-up period (Table 2, figure 2). Mean total healthcare costs were $45,529 for the PP cohort and $52,569 for the OAT cohort (p = 0.128). While total healthcare costs were not significantly different, significant differences were evident in lower inpatient costs ($18,560 vs $31,505, p = 0.002), lower ER costs ($20 vs $64, p = 0.015), and higher pharmacy costs ($10,063 vs $4167, p < 0.001) for PP patients compared to OAT patients.

Table 2. Utilization and cost outcomes for the matched PP and OAT cohorts (12-month follow-up period).

	PP cohort (n = 335)			OAT cohort (n = 335)
12-month follow-up period	n/Mean	%/SD	Median	n/Mean	%/SD	Median	p Value
Post-index all-cause healthcare utilization
Any hospitalization	113	34%		176	53%		<0.001
Number of hospitalizations per patient	0.81	1.60		1.30	1.87		<0.001
Number of inpatient days per patient	13.24	36.19		24.18	52.05		0.002
Any ER visit	23	7%		33	10%		0.163
Number of ER visits per patient	0.12	0.55		0.28	1.29		0.033
Any outpatient office visit	335	100%		327	98%		0.004
Number of outpatient office visits per patient	46.68	38.12		42.71	39.88		0.189
Post-index all-cause healthcare costs per patient
Inpatient costs	$18,560	$46,761	$0	$31,505	$60,713	$7771	0.002
Pharmacy costs	$10,063	$18,977	$6889	$4167	$4446	$2845	<0.001
ER costs	$20	$112	$0	$64	$312	$0	0.015
Outpatient office costs	$16,723	$15,809	$11,349	$16,556	$20,915	$10,839	0.908
Other outpatient costs	$163	$534	$0	$276	$1125	$0	0.099
Total costs (outpatient + inpatient + pharmacy)	$45,529	$54,696	$28,982	$52,569	$64,391	$30,836	0.128

SD, standard deviation. All-cause healthcare utilization was calculated by the number of claims (e.g. hospitalizations, ER visits, outpatient visits) in the health record and the proportion of the study population that had a claim. Costs from the health record were adjusted to 2012 US dollars using the medical care component of the Consumer Price Index (CPI).

A lower percentage of patients in the PP cohort experienced a hospital admission compared to those in the OAT cohort (34% vs 53%, p < 0.001) and the mean number of admissions was also lower in the PP cohort (0.81 vs 1.30; p < 0.001) (Table 2, figure 3). The average number of inpatient days over the 12-month follow-up for the PP cohort was less than that of the OAT cohort (13.24 vs 24.18, p = 0.002). Some measures of ER and outpatient visit utilization reached statistical significance without obvious clinical meaningfulness.

Discussion

The results of this study have particular relevance to formulary and policy decision-makers, with evidence of the association of PP with lower hospitalization costs. The results are consistent with the findings of previous studies of injectable antipsychotics suggestive of a reductive effect on the costs associated with inpatient care17–19. Supporting the findings of lower inpatient costs, the study also found that PP was associated with lower hospital admission rates and fewer inpatient days than OAT. The lower total healthcare costs for PP treated patients were not significantly different than the costs for OAT patients over the 12-month follow-up period.

Treatment selection bias was addressed by using rigorous statistical matching methodology. There was a stark contrast in patient characteristics between the study-qualified cohorts treated with PP and OAT. For example, 82% of study-qualified PP patients had been treated with multiple antipsychotic medications in the 24-month baseline period, while this was characteristic of less than half (43%) of OAT patients. Similarly, a significantly higher percentage of patients in the overall PP population (28%) had been hospitalized three or more times in the baseline period than had patients in the OAT population (20%).

Data to assess clinical severity of the disease, such as Positive and Negative Syndrome Scale or Brief Psychiatric Rating Scale scores, were not available, but these examples suggest that PP was being used in a more difficult to treat patient population. A visible manifestation of the treatment selection bias can be seen in the histograms of propensity scores for the two study-qualified cohorts (Figure 1a). Calculated from baseline characteristics associated with treatment selection, the two histograms summarize how likely patients in each cohort were to be treated with PP. Distinctly different patterns seen in the two histograms in Figure 1(a) are indicative of significant disparities in the attributes associated with treatment selection. The histogram for the OAT group is concentrated close to zero, representing that most patients in that group have a low predicted likelihood of receiving PP therapy.

Figure 2. Healthcare costs for the matched PP and OAT cohorts. * p < 0.05. PP, paliperidone palmitate long-acting injection; OAT, atypical oral antipsychotic therapy; ER, emergency room.

Figure 3. Healthcare utilization for the matched PP and OAT cohorts. PP, paliperidone palmitate long-acting injection; OAT, atypical oral antipsychotic therapy; ER, emergency room.

Generation of rigorous evidence comparing the economic outcomes of the two treatments was predicated through creating well-matched cohorts for comparison. The incorporation of a Mahalanobis procedure into the matching process resulted in the identification of a strong set of analogs for the PP patients from within the OAT study-qualified cohort. The Mahalanobis procedure was selected because it relied not only on the simple proximity of propensity scores between patients, but also proximity in a distance derived from the covariance matrix to better account for correlation between factors. The matching process identified solid matches for 88% of the study-qualified PP cohort. The matched OAT cohort had statistically similar values for all key attributes at baseline. One attribute that did differ even after matching was the baseline pharmacy cost. At an annual mean cost of $7160, the baseline pharmacy cost for the matched PP cohort was nearly twice that of the OAT pharmacy costs. A deeper investigation revealed that, although the matched cohorts were similar in the rate of baseline exposure to injectable antipsychotics, 34% of the OAT cohort had been treated with generic injectables, while only 9% of the PP cohort used generic injectables in the baseline period. Therefore, the difference in pharmacy cost may represent a difference in the cost of products used, rather than a difference in the level of medication utilization. Despite the successful matching process, cohorts were not identical and the effects of the remaining differences are unknown.

In defining the primary outcomes for this study, the decision was made to focus on all-cause utilization and costs, rather than assign clinical causality to individual medical and pharmacy claims. This definition is aligned with the perspective of real world policy-makers who rely on claims data to make population health decisions. More importantly, the risk of excluding study relevant medical claims based on the absence of a specific diagnosis code presented a strong argument in favor of including all medical and pharmacy claims in the analysis, regardless of diagnosis code.

This analysis may be subject to inherent limitations of the source administrative claims data, such as coding errors or diagnoses entered for administrative processing rather than clinical completeness. Claims collected for administrative purposes do not contain many of the relevant data elements that would ideally have been incorporated into this analysis. It was not possible to view or control for individual differences in disease severity prior to the index date, nor was it possible to evaluate efficacy of treatments in the follow-up period. Many individual patient characteristics, such as living situation or time since first diagnosis, were not available in the data and, therefore, their effect on study outcomes remains unknown.

The matched cohort method provides insight into how outcomes for matched PP patients differed from outcomes for similar patients treated with OAT, but findings cannot be generalized to unmatched PP patients or to the full OAT population. Furthermore, since this analysis was conducted on observations from within the VA, a system with unique care delivery and reimbursement mechanisms, the findings cannot necessarily be generalized to other healthcare delivery systems, such as commercial managed care plans or Medicaid plans. The study perspective was that of the VHA system and does not reflect the costs for any healthcare delivered to these patients outside of the VHA system.

This study adds to the body of evidence on the role of PP in the treatment of schizophrenia, specifically in the VA system, but more research needs to be done in other treatment settings. Although distinct effects on the cost and utilization of inpatient care were observed in the 12-month follow-up period, it would be valuable to analyze longer-term outcomes for patients afflicted with this lifelong condition.

Conclusion

This study used matched cohorts of patients treated with OAT and PP to quantify the effects of PP, an injectable antipsychotic product, in terms of medical resource utilization and cost. Total healthcare costs were similar between the matched cohorts; however, PP treatment was associated with significantly lower hospital admissions, length of stay, and inpatient cost. A treatment with the potential to help more patients remain in community care rather than inpatient care would offer value in terms of patient experience in addition to any economic benefits.

Transparency

Declaration of funding

This study was funded by Janssen Scientific Affairs.

Declaration of financial/other relationships

JP & MD are employees of Janssen Scientific Affairs. OB & LX are employees of STATinMED Research, a paid consultant for Janssen Scientific Affairs.

Acknowledgments

Portions of this manuscript and an initial statistical analysis were presented as posters: (1) Healthcare use and costs of patients with schizophrenia in a VA population treated with paliperidone palmitate or atypical oral antipsychotics. Poster presented at the American Psychiatric Association 65th Institute on Psychiatric Services (IPS), October 10–13, 2013, Philadelphia, PA; and (2) A comparison of healthcare utilization and costs of schizophrenia patients treated with paliperidone palmitate long-acting injection or atypical oral antipsychotics within the Veterans Health Administration. Poster presented at the 26th Annual US Psychiatric and Mental Health Congress, September 30–October 3, 2013, Las Vegas, NV.