The threshold rate of oral atypical anti-psychotic adherence at which paliperidone palmitate is cost saving

Abstract

Objective:

To identify, estimate, and compare ‘real world’ costs and outcomes associated with paliperidone palmitate compared with branded oral atypical anti-psychotics, and to estimate the threshold rate of oral atypical adherence at which paliperidone palmitate is cost saving.

Methods:

Decision analytic modeling techniques developed by Glazer and Ereshefsky have previously been used to estimate the cost-effectiveness of depot haloperidol, LAI risperidone, and, more recently, LAI olanzapine. This study used those same techniques, along with updated comparative published clinical data, to evaluate paliperidone palmitate. Adherence rates were based on strict Medication Event Monitoring System (MEMS) criteria. The evaluation was conducted from the perspective of US healthcare payers.

Results:

Paliperidone palmitate patients had fewer mean annual days of relapse (8.7 days; 6.0 requiring hospitalization, 2.7 not requiring hospitalization vs 17.8 days; 12.4 requiring hospitalization, 5.4 not requiring hospitalization), and lower annual total cost ($20,995) compared to oral atypicals (mean $22,481). Because paliperidone palmitate was both more effective and less costly, it is considered economically dominant. Paliperidone palmitate saved costs when the rate of adherence of oral atypical anti-psychotics was below 44.9% using strict MEMS criteria. Sensitivity analyses showed results were robust to changes in parameter values. For patients receiving 156 mg paliperidone palmitate, the annual incremental cost was $1216 per patient (ICER = $191 per day of relapse averted). Inclusion of generic risperidone (market share 18.6%) also resulted in net incremental cost for paliperidone palmitate ($120; ICER = $13). Limitations of this evaluation include use of simplifying assumptions, data from multiple sources, and generalizability of results.

Conclusions:

Although uptake of LAIs in the US has not been as rapid as elsewhere, many thought leaders emphasize their importance in optimizing outcomes in patients with adherence problems. The findings of this analysis support the cost-effectiveness of paliperidone palmitate in these patients.

Keywords::

• Schizophrenia
• Adherence
• Long-acting injectables
• Paliperidone palmitate
• Atypical anti-psychotics
• Cost-effectiveness

Introduction

Poor adherence to medication treatment can have devastating consequences for patients with mental illness1. Medication adherence is among the most important health-related behaviors in relapse prevention and recovery in schizophrenia2, yet studies show that clinicians do not make systematic efforts to identify and monitor it over time3. Awareness of the need for medication, awareness of social consequences, patient’s perceived trust in the physician, and the severity of negative symptoms all predict attitudes toward medications that in turn predict adherence4.

Because a large percentage of patients with schizophrenia do not consistently take their oral medication, long-acting injectable (LAI) formulations may provide a more effective way to administer medications3,5. LAI anti-psychotics can improve adherence and provide the clinician with reliable information about whether patients are taking their medication as well as the dosage received for purposes of dose adjustments or to guide response to relapse5. United States (US) expert consensus guidelines emphasize the importance of switching patients with schizophrenia to a long-acting anti-psychotic when lack of insight, substance use, persistent symptoms, logistic problems, lack of routines, or lack of family/social support interfere with adherence1.

LAIs remain an under-utilized option in many countries, despite frequent non-adherence with oral medication and subsequent relapse6. The rate of LAI use in the US is particularly low (10–15%), and this rate hardly matches the US estimated rate of non-adherence with oral schizophrenia treatments (<50%)7. US clinical experts emphasize the importance of LAIs in preventing recurrent relapses in schizophrenia3,5–7. Recurrent episodes in schizophrenia have been shown to lead to substantial chronic deterioration8 and reduced responsiveness to anti-psychotics or ‘treatment resistance’9–11. Maintenance anti-psychotic treatment has been shown to be as important for newly diagnosed patients as it is for patients with chronic or ‘revolving door’ schizophrenia12,13.

Existing prospective and retrospective studies comparing the benefits of LAI with those of oral anti-psychotics often do not adequately capture the important advantages of LAIs14 because of (i) protocol-driven patient selection and frequent follow-up requirements in prospective trials15,16; (ii) confounding due to non-randomization16–18; (iii) wide variation and poor definition of outcome assessment19; or (iv) wide variation and poor definition of adherence (i.e., self- and physician report are inaccurate and may under-estimate non-adherence, pill counts and pharmacy records do not provide information about actual dose taking, much less dose timing)20.

In 1996, Glazer and Ereshefsky21 developed a clinical decision analysis model for schizophrenia that took into account adherence rates and associated re-hospitalization to compare LAI (first-generation typical anti-psychotics) with oral treatments. The decision analytic modeling techniques developed by Glazer and Ereshefsky were subsequently employed to ascertain the benefit of two long-acting atypical anti-psychotics, LAI risperidone22,23, and, more recently, LAI olanzapine24,25. Here, we used those same techniques, along with updated, improved comparative clinical and cost data, to assess paliperidone palmitate.

Approved by the US Food and Drug Administration (FDA) for acute and maintenance treatment of schizophrenia, paliperidone palmitate is a once-monthly long-acting atypical anti-psychotic. To our knowledge, an assessment of the cost-effectiveness of paliperidone palmitate has not yet been published in peer-reviewed literature. The objective of this analysis was to evaluate the relative clinical benefits and associated costs of paliperidone palmitate compared with branded oral atypical anti-psychotics, and to estimate the threshold level of oral atypical anti-psychotic adherence at which paliperidone palmitate would be a cost-saving alternate treatment option.

Methods

Model structure and study design

The decision tree was based on a decision analytical model for schizophrenia developed by Glazer and Ereshefsky21. Glazer and Ereshefsky developed a clinical decision analysis model that took into account adherence rates and associated re-hospitalization to compare direct treatment costs associated with alternate neuroleptic strategies for ‘revolving door’ patients with schizophrenia. The anti-psychotic treatments they considered were traditional oral neuroleptics (e.g., haloperidol), depot neuroleptics (e.g., haloperidol decanoate), and atypical oral agents (e.g., risperidone). The premise of the Glazer and Ereshefsky model was based upon the vast literature showing that relapse rates were very dependent upon adherence to therapy in schizophrenia. Improvement of adherence through the use of LAI formulations of anti-psychotics was found to have dramatic impacts on clinical and economic outcomes of patients with schizophrenia.

Long-acting paliperidone palmitate was compared with a ‘blended’ oral atypical anti-psychotic arm to understand how long-acting paliperidone palmitate compared with commonly prescribed branded oral atypical anti-psychotics overall. The choice of a comparator was based on the US Academy of Managed Care Pharmacy (AMCP) Submission Guidelines, which state that the comparator should be the existing standard of care, best available care, usual care, or best supportive care. Only branded aripiprazole (Abilify, Otsuka America Pharmaceutical, Inc., Tokyo, Japan), olanzapine (Zyprexa, Lilly USA, LLC, Indianapolis, IN), paliperidone (Invega, Janssen Pharmaceuticals, Inc., Titusville, NJ), quetiapine (Seroquel, AstraZeneca Pharmaceuticals, LP, Wilmington, DE), and ziprasidone (Geodon, Pfizer, Inc., New York, NY) were included in the base case analysis. Oral generic risperidone was considered in sensitivity analyses. Recently available oral atypical anti-psychotics (i.e., asenapine, iloperidone, and lurasidone) were not included in the current model because their market share remains very low. The weight of impact of each oral anti-psychotic was determined by the market share of each individual drug.

Two inter-dependent clinical outcomes are modeled in the decision tree: patient treatment adherence and patient disorder manifestation or presentation (Figure 1). First, patients will be adherent, partially adherent, or non-adherent with the treatment regimen. Second, and dependent upon the level of adherence, the patient may (i) remain stable or (ii) suffer exacerbation(s) or relapse(s). Both relapses not severe enough to warrant re-hospitalization and relapses requiring hospitalization are possible.

Figure 1.  Decision tree for the evaluation of paliperidone palmitate compared with oral atypical antipsychotics.

A 1-year time horizon was chosen for the model, which was created using Microsoft Excel (Microsoft, Redmond, WA). The economic component of the analysis was undertaken from the perspective of US healthcare payers and included all major direct medical costs associated with the treatment of schizophrenia (hospitalizations, emergency room [ER] visits, outpatient mental health visits, medications for adverse events), anti-psychotic medications, and the cost of administration for injectable medications).

The majority of clinical data and resource use was obtained from the published literature accessed through the electronic PubMed database as of September 19, 2011. When published data were not available, database information or expert clinical opinion were used. Modified Delphi panel techniques, which included three rounds of exchanges to reach consensus, were used with our expert panel17. Details regarding data and respective sources are described in greater detail in the following paragraphs.

The primary effectiveness measure applied was the number of stable days and the number of relapse days (with and without hospitalization) associated with each therapy. Other inputs included medication adherence, adverse event rates, and relapse frequency and duration. The unit of analysis for the cost evaluation was the total average schizophrenia-related healthcare cost per year, including costs of health services use, ancillary drugs, and medications.

Clinical inputs

Medication adherence

A comprehensive systematic literature search was conducted to identify appropriate studies for the decision analytic model. The National Library of Medicine PubMed database was searched using the terms ‘schizophrenia’ and (‘adherence’ or ‘compliance’). Articles were limited to ‘Humans’, ‘Clinical Trial’, and ‘English’. The original search was performed in July 200422,23, so only studies published after this date were included for the current search.

One hundred and eighty-seven studies were identified. A systematic selection took place based on the following a priori inclusion criteria: (i) the study had to report relapse rates according to different mutually exclusive levels of adherence; (ii) the study had to employ an appropriate definition of adherence (i.e., the definition could not be limited to exclusively subjective reporting, and both dose taking and dose timing had to be taken into consideration); (iii) a broad patient population was required (i.e., not limited to first episode or severely ill patients); and (iv) relapses requiring and not requiring hospitalization had to be distinguished and reported. Three studies were closest in meeting all of these inclusion criteria (Diaz et al.26; Olfson et al.27; and Gutierrez Casares et al.30), two of which were identified in the original search and were already employed in the model26,27.

Medication Event Monitoring System (MEMS) technology is considered to be the most sophisticated way of measuring adherence and serves as the gold standard28,29. MEMS medication bottles contain a microelectronic chip that registers the date and time of every bottle opening. Assuming that bottle openings represent medication intake, MEMS provides a detailed profile of the patient’s adherence behavior, including information about dose timing and dose taking. Two of the studies identified in the literature review employed the MEMS technology26,30.

Gutierrez Casares et al.30 conducted a multi-center observational study of 1848 outpatients with schizophrenia or schizoaffective disorders. Patient adherence with the pharmacological treatment was assessed by the use of questionnaires, scales, medication accountability, and the MEMS. Oral atypical anti-psychotic adherence with the MEMS device was defined as consistency between the number of pill bottle openings (regardless of the hour) and the prescribed regimen on at least 75% of the days of the study period, as per Lacro et al.31. Adherence with LAIs was evaluated as the percentage of correct administrations in keeping with indications, and correct administration was considered as a dose administered within 3 days of the scheduled dose. Patients receiving LAIs were considered adherent if they reported an adherence rate ≥80%. Using the MEMS technology definition, 29% of patients on oral medication were found to be adherent compared with 79% of patients on injectable medication (injection counting) (odds ratio = 9.11; 95% confidence interval 6.02–13.77; p < 0.0001). These rates were used as adherence inputs in the model.

The rate of non-adherence for oral atypical anti-psychotics was obtained from a 2000 study by Olfson et al.27. This study utilized a stringent definition of non-adherence, considering only those patients who acknowledged stopping their medication for 1 week or more as non-adherent.

Partial adherence rates were calculated as the percentage of patients who fell into neither this non-adherent group nor the adherent group (i.e., the remaining proportion of patients). The rate of non-adherence was fixed in sensitivity analyses; therefore changes in input parameters shifted the proportion of patients between the ‘adherent’ and ‘partially adherent’ categories. The annual rates of adherence, non-adherence, and partial adherence for each comparator are presented in Table 1.

Table 1.  Default model clinical input parameters.

Parameter	LAI atypicals	Oral atypicals	Reference
Adherence Rates
Adherent	79.0%	29.0%	Gutierrez Casares et al.30
Partially adherent	21.0%	61.9%	Remaining patients
Non-adherent	0.0%	9.1%	Olfson et al.27
Annualized Relapse Rates
Relapses requiring hospitalization
Adherent	14.3%		Diaz et al.26
Partially adherent	58.6%		Olfson et al. 27
Non-adherent	90.1%		Olfson et al.27
Relapses not requiring hospitalization
Adherent	14.3%		Diaz et al.26
Partially adherent	52.2%		Olfson et al.27
Non-adherent	83.8%		Olfson et al.27
Duration and frequency of relapse
Duration of relapse
Requiring hospitalization	11.5 days		HCUP 2009 Data
Not requiring hospitalization	5 days		Expert Panel
Annual frequency of relapse (per patient with relapse)
Requiring hospitalization	2.2 relapses		Nicholl et al.35
Not requiring hospitalization	2.5 relapses		Nicholl et al.35
Adverse event rates	Weight gain	Extrapyramidal symptoms
Aripiprazole	7.0%	8.0%	Assumed same as ziprasidone
Olanzapine	30.0%	7.0%	CATIE Study36
Paliperidone palmitate	9.0%	4.0%	Pandina et al.37
Paliperidone ER	4.5%	9.0%	Assumed same as risperidone
Quetiapine	16.0%	3.0%	CATIE Study36
Ziprasidone	7.0%	8.0%	CATIE Study36
Relative market share of oral atypicals
Aripiprazole	21.1%		Wolters Kluwer Pharma Solutions and Premier38
Olanzapine	24.8%
Paliperidone ER	2.6%
Quetiapine	39.9%
Ziprasidone	11.6%

CATIE, Clinical Anti-psychotic Trials of Intervention Effectiveness; ER, extended release; HCUP, Healthcare Cost and Utilization Project; LAI, long-acting injectable.

Relapse rates

The Diaz et al.26 study also provided information on the rate of relapse requiring hospitalization for adherent patients in the model (14.3% in 1 year). No published studies provided a suitable rate of relapse not requiring hospitalization in adherent patients. The Modified Delphi panel involved in the original modeling studies22,23 noted that relapses not requiring hospitalization are often not measured in studies or reported in the literature, yet they are significant to include. The experts stated that these relapses were ‘at least as likely’ as relapses requiring hospitalization, which is why the rate of relapse requiring hospitalization was used as a proxy and as a conservative estimate in the model. Relapses not requiring hospitalization are not significant cost drivers, but the experts wanted them considered because they felt they had significant clinical burden nonetheless. Therefore, the Diaz et al. study rate of relapse requiring hospitalization for adherent patients was also used as the rate of relapse not requiring hospitalization for adherent patients (14.3%).

Rates of relapse with partial and non-adherence were obtained from Olfson et al.27 in which non-adherence was defined as patients stopping their medication for 1 week or more. The remainder of patients in the Olfson et al.27 study were grouped together and were considered the ‘partially adherent’ group. Their study results showed that 19.8% of partially adherent patients and 43.9% of non-adherent patients had relapses requiring hospitalization in 3 months. Relapses not requiring hospitalization were recorded as ER visits in their study, and results showed that 16.9% of partially adherent patients and 36.6% of non-adherent patients had relapses not requiring hospitalization in 3 months. These 3-month relapse rates were annualized by first calculating a monthly rate of adherence with the following formula:

Depending on the variable of interest, the term ‘survivors’ in the above equation represents the number of non-relapsed patients remaining by the end of the study period. This calculation is not linear but reflects residual changes in the number of patients over time. This equation assumes that rates of relapse stay constant over time. This assumption is consistent with results found in the literature32–34. The annual rate of relapse was then calculated by using the following equation:

The calculated annual rates of relapse requiring hospitalization and relapse not requiring hospitalization are presented in Table 1.

Relapse frequency and duration

Information on the duration of relapse requiring hospitalization was based on the average length of hospital stay for a patient with a principal diagnosis of schizophrenia from nationally representative data from the 2009 Healthcare Cost and Utilization Project (HCUP) database (http://hcupnet.ahrq.gov/) (Table 1). No published or unpublished primary sources of the duration of relapse not requiring hospitalization were found; therefore, this input was estimated by the expert panel.

HCUP data on the number of relapses per patient per year were not available. The annual frequencies of relapses requiring hospitalization and relapses not requiring hospitalization were obtained from a 2010 US burden of illness article by Nicholl et al.35, which reported the mean number of psychiatric hospitalizations and ER visits per patient with schizophrenia per year (Table 1).

Adverse event rates

The economic consequences of two adverse events were considered in this model: treated extrapyramidal symptoms (EPS) and clinically significant weight gain. Other potential adverse events were not included because evidence for event rate differences is not easily interpreted and the economic consequences relating to these events are not well documented41.

For patients treated with oral atypical anti-psychotics, rates of EPS and weight gain were derived from the National Institute of Mental Health’s Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study36. Data regarding the use of anti-cholinergics were used as a proxy measure for incidences of treated EPS. Clinically significant weight gain rates were based on the proportion of patients in CATIE with a ≥7% increase in body weight from baseline, consistent with the FDA’s definition for studies of psychotropic drugs42. Aripiprazole was not included in the CATIE study. Rates of EPS treatment and clinically significant weight gain for patients taking aripiprazole were assumed to be equivalent to those of ziprasidone, because these agents have been found to exhibit similar efficacy and tolerability profiles44.

Rates of EPS (4%) and weight gain (9%) for patients treated with paliperidone palmitate were taken from a clinical trial (Pandina et al.37 and data on file) because observational data were not yet available (Table 1).

Market share data

Wolters Kluwer Pharma Solutions outpatient and Premier Health inpatient market share data for schizophrenia were used to weight the effect of the individual oral atypical anti-psychotics based on current practice38. Weighting of the individual oral atypical was required for parameters that were entered separately for each of the oral atypical anti-psychotics (i.e., rates of adverse events and drug costs). Market share data are presented in Table 1.

Economic inputs

Unit costs

The costs of healthcare services provided are reported in Table 2 and were based on HCUP data, Centers for Medicare and Medicaid Services Medicare Reimbursement Data, and the PharMetrics claims database. All costs were adjusted to 2011 dollars using the Bureau of Labor Statistics Medical Care Services Inflation Rates (www.bls.gov/news.release/cpi.nr0.htm).

Table 2.  Default model economic input parameters.

Resource	Unit	Unit cost	Reference
Inpatient care
Hospitalization	Day	$814.04	HCUP 2009 Data
Day hospital	Day	$145.14	PharMetrics (CPT HOO35, MH Part Hosp Therapy Under 24 hours)
Emergency room	Visit	$438.25	PharMetrics (ICD-9-CM 290.XX-319.XX in the first diagnosis field and at least half of total costs mental health related)
Outpatient care
Physician office visit	Visit	$102.27	Medicare reimbursement for CPT 99214
Mental health clinic visit	Visit	$102.27	Assumed same as for physician visit
Injection administration	Visit	$23.10	Medicare reimbursement for CPT 96372
Home healthcare	Hour	$110.44	PharMetrics (CPT G0154, SN Services Home Healthcare Each 15)
Social/group therapy meetings	Hour	$110.75	PharMetrics (CPT 90853, Psychotherapy Group)
Nutritionist	Visit	$108.96	PharMetrics (CPT 97802, Medical Nutrition Therapy Initial Assessment each 15 min)
Medications	Daily dose	Daily cost
Aripiprazole	18.0 mg	$21.05	WAC aripiprazole (Abilify), Otsuka America Pharmaceutical Inc., WK Source Lx (APLD) Database
Benztropine		$0.25	Red Book
Olanzapine	16.1 mg	$30.56	WAC olanzapine (Zyprexa), Eli Lilly and Company, WK Source Lx (APLD) Database
Paliperidone palmitate	Day 1: 234 mg; day 8: 156 mg, then 117 mg	$30.88	WAC paliperidone palmitate (Invega Sustenna), Janssen LP
Paliperidone ER	7.5 mg	$20.57	WAC paliperidone ER (Invega), Janssen LP, WK Source Lx (APLD) Database
Quetiapine	350.7 mg	$16.57	WAC quetiapine (Seroquel), AstraZeneca Pharmaceuticals LP, WK Source Lx (APLD) Database
Ziprasidone	125.0 mg	$16.88	WAC ziprasidone (Geodon), Pfizer Inc., WK Source Lx (APLD) Database

ER, extended-release; HCUP, Healthcare Cost and Utilization Project; WAC, wholesale acquisition cost; WK, Wolters Kluwer.

Schizophrenia-specific drug utilization data for each oral atypical comparator were obtained from Wolters Kluwer Source Lx (APLD) outpatient database 2010–2011. The distribution of claims for each of the various doses was used to weight the individual dose wholesale acquisition cost (WAC) prices, and obtain an overall average daily cost per patient for schizophrenia39. The daily mean cost of paliperidone palmitate was calculated based on the labeled dosing40. Recommended initiation of paliperidone palmitate is with a dose of 234 mg on treatment day 1 and 156 mg 1 week later, followed by a monthly maintenance dose of 117 mg for the remainder of the year (Table 2). Paliperidone palmitate doses are expressed as milligram equivalents (mg eq) of paliperidone outside of the US (i.e., a 78 mg dose of paliperidone palmitate = 50 mg eq of paliperidone, 117 mg dose = 75 mg eq, 156 mg dose = 100 mg eq, and 234 mg dose = 150 mg eq). Product replacement and patient assistance programs are sometimes available to help mitigate the cost of LAIs in the inpatient and outpatient settings, however such programs were not considered in this model.

Drug costs for the comparators were adjusted based on the proportion of adherent, partially adherent, and non-adherent patients. It was assumed that adherent patients incurred 100% of their medication costs, partially adherent patients incurred 70% of their medication cost, and non-adherent patients incurred 10% of their medication costs. The impact of these parameter estimates was tested in sensitivity analyses.

Resource use estimates

As described above, the duration of hospitalization was obtained from the HCUP database. Expert clinical opinion was used to determine outpatient resource use as no published data presented this information in the format required by the decision tree model (even after conduction of a very thorough literature review). Information was needed on resource use per (i) stable day; (ii) relapse not requiring hospitalization; (iii) relapse requiring hospitalization (excluding the hospital stay); (iv) episode of EPS; and (v) episode of weight gain17,30.

Sensitivity analyses

One-way sensitivity analyses were used to examine the impact of altering key model assumptions and parameter values. In order to test the robustness of the model, input parameter values were varied ±25%, with the exception of the following:

1. The adherence rate of oral atypical anti-psychotics was varied from 19–79%;

2. The rate of EPS with paliperidone palmitate was set at 9% and the rate of weight gain with paliperidone palmitate was set at 30%. Rates of EPS and weight gain from the NICE Clinical Guidance for Atypical Antipsychotics (http://www.nice.org.uk/nicemedia/live/11786/43607/43607.pdf) were also implemented in a sensitivity analysis;

3. The adherence medication cost adjustment rate was removed (i.e., medication costs were calculated as 100% of prescribed dose);

4. The maintenance dose of paliperidone palmitate was changed to 156 mg (i.e., daily drug cost changed to $35.34) to determine the cost-effectiveness of paliperidone palmitate in patients treated with a higher maintenance dose; and

5. Generic risperidone was included in the oral atypical anti-psychotics arm at its current market share (18.6%).

Results

Clinical outcomes

In the base case scenario, it is estimated that, within 12 months, 22.6% of patients administered paliperidone palmitate would experience a relapse requiring hospitalization, compared with 48.6% of patients taking oral atypical anti-psychotics. Similar proportions of patients experienced relapses not requiring hospitalization (22.3% paliperidone palmitate vs 44.1% for oral atypical anti-psychotics). Overall, paliperidone palmitate had a lower mean number of days of relapse compared with oral atypical anti-psychotics over 1 year (8.7 days; 6.0 requiring hospitalization and 2.7 not requiring hospitalization, vs 17.8 days; 12.4 requiring hospitalization and 5.4 not requiring hospitalization, respectively) (Table 3).

Table 3.  Base case clinical and economic consequences for the comparators over 1 year.

	Paliperidone palmitate	Oral atypical anti-psychotics	Difference (paliperidone palmitate – oral atypical anti-psychotics)
Clinical consequences
Relapses requiring hospitalization
% of patients experiencing relapse requiring hospitalization	23.6%	48.6%	25.0% fewer patients with relapse
# of relapses requiring hospitalization per patient, mean	0.52 relapses	1.08 relapses	0.55 relapses averted
# of days of relapse requiring hospitalization per patient	6.0 days of relapse	12.4 days of relapse	6.4 days of relapse averted
Relapses not requiring hospitalization
% of patients experiencing relapse not requiring hospitalization	22.3%	44.1%	21.8% fewer patients with relapse
# of relapses not requiring hospitalization per patient	0.55 relapses	1.08 relapses	0.54 relapses averted
# of days of relapse not requiring hospitalization per patient	2.7 days of relapse	5.4 days of relapse	2.7 days of relapse averted
Economic consequences
Cost of relapses
Cost of relapses requiring hospitalization	$5620	$11,579	($5959)
Cost of relapses not requiring hospitalization	$761	$1509	($747)
Cost of stable days	$3643	$3551	$93
Cost of adverse events
Cost of extrapyramidal symptoms	$9	$13	($4)
Cost of weight gain	$102	$184	($82)
Medication cost	$10,861	$5648	$5213
Total cost	$20,995	$22,481	($1486)

Economic outcomes

Paliperidone palmitate had a lower annual total cost ($20,995) compared with oral atypical anti-psychotics ($22,481) (Table 3), despite having higher medication costs and a higher cost for stable days in the base case scenario. The largest outcomes-related cost component was relapse requiring hospitalization (cost ranged from $5620 per patient receiving paliperidone palmitate to $11,579 per patient receiving oral atypical anti-psychotics), followed by the cost of stable days, and then the cost of relapse not requiring hospitalization. The direct costs associated with the treatment of EPS and weight gain were found to be minimal for all comparators. Medication costs were also a significant component of total costs, ranging from $5648 for oral atypical anti-psychotics to $10,861 for paliperidone palmitate.

Figure 2 illustrates the cost breakdown by healthcare resource category. Hospitalization cost was the largest cost component for oral atypical anti-psychotics and medication cost was the largest cost component for paliperidone palmitate. The costs of outpatient physician and clinic visits were also significant cost categories, followed by costs for social/group therapy, outpatient physician visits, ER visits, home healthcare, day hospital visits, and nutritionist visits.

Figure 2.  Breakdown of overall mean total cost of treatment arms by resource category.

Sensitivity analyses

The relationship between the adherence rate of oral atypical anti-psychotics and total incremental annual per patient cost appears to be a linear one (Figure 3). The threshold MEMS rate of oral atypical adherence at which paliperidone palmitate is cost saving compared with oral atypical anti-psychotics is 44.9% (a difference of 34.1%). At levels of oral atypical adherence above this threshold rate, paliperidone palmitate results in a net incremental cost compared with oral atypical anti-psychotics in the model. At adherence rates below this threshold, paliperidone palmitate results in net cost savings compared to oral atypical anti-psychotics in the model.

Figure 3.  Impact of oral atypical antipsychotic adherence rate (as measured by the Medication Event Monitoring System (MEMS)) on mean total annual incremental cost per patient.

Additional sensitivity analyses showed that results were relatively robust to changes in input parameter values (Table 4). Paliperidone palmitate remained the lower-cost treatment strategy throughout all ±25% variations in input parameter values. The model was most sensitive to the frequency of relapse requiring hospitalization, the duration of relapse not requiring hospitalization, and the cost of hospitalization. The model was not very sensitive to the duration and frequency of relapse not requiring hospitalization and to the rate of adverse events. Removing the medication cost adjustment for adherence increased the cost savings with paliperidone palmitate because a higher proportion of drug costs were included for oral atypical anti-psychotics. For patients who would receive the 156 mg maintenance dose of paliperidone palmitate, there would be an incremental cost of $1216 per patient per year. The incremental cost-effectiveness ratio (ICER) for this dose would be $191 per day of relapse averted. The inclusion of generic risperidone in the model (at a market share of 18.6% at the time of the analysis, daily dose of 2.25 mg and daily cost of $0.19) also resulted in a net incremental cost for paliperidone palmitate ($120 per patient per year; ICER = $13 per day of relapse averted).

Table 4.  Sensitivity analyses on model input parameters.

	Mean incremental cost per patient per year (paliperidone palmitate – oral atypical anti-psychotics)
Base case values
Base case	($1463)
Rate of relapses requiring hospitalization
−25%	($1275)
+25%	($2689)
Rate of relapses not requiring rospitalization
−25%	($1306)
+25%	($1652)
Duration of relapses requiringhospitalization
−25% (8.6 days)	($206)
+25% (14.4 days)	($2766)
Duration of relapses not requiring hospitalization
−25% (3.8 days)	($1386)
+25% (6.2 days)	($1582)
Frequency of relapses requiring hospitalization
−25% (1.7 per year)	($13)
+25% (2.8 per year)	($2959)
Frequency of relapses not requiring  hospitalization
−25% (1.8 per year)	($1306)
+25% (3.1 per year)	($1666)
Rates of adverse events
Weight gain paliperidone palmitate = 30.0%	($1475)
Extrapyramidal symptoms   paliperidone palmitate = 9.0%	($1248)
Weight gain from NICE Clinical Guidance	($1419)
Extrapyramidal symptoms   from NICE Clinical Guidance	($1394)
Cost of Hospitalization
−25% ($611)	($790)
+25% ($1,018)	($2782)
Daily drug cost adjustment for adherence
No adjustment (100% drug cost)	($2839)
Paliperidone palmitate maintenance dose
156 mg per month	$1216 ($191 per day of relapse averted)
Inclusion of generic risperidone
18.6% market share	$120 ($13 per day of relapse averted)

Discussion

Anti-psychotic medication management in patients with schizophrenia is a complex process characterized by prevalent and prolonged rates of poor adherence30. Poor adherence to medication is a modifiable risk factor for sub-optimal outcomes in schizophrenia anti-psychotic treatment41. LAIs, which provide a much less frequent dosing schedule and improved monitoring of adherence behavior, may be an important treatment strategy in reducing the risk of relapse in patients who have difficulty with medication adherence.

The challenge to both payers and providers of healthcare is to maximize the net benefit obtained from healthcare expenditures. Economic models are not intended to replace healthcare providers’ insight and judgment, in that they summarize only a sub-set of information needed to make decisions about resource allocation. However, since they provide details regarding the implications of alternate decisions, they can be valuable during that decision-making process.

Comparative effectiveness research is intended to identify cost-effective medical treatments and, in turn, help curb spending for expensive illnesses such as schizophrenia. This change in healthcare spending trajectory is to be achieved, in part, through the selection of drugs that have been proven effective while costing less than other options. Our analysis found that paliperidone palmitate is a more effective and cost-saving treatment option when the rate of adherence of oral atypical anti-psychotics is below 44.9%. This threshold rate of adherence is based on strict MEMS adherence criteria, which required that patients opened their pill bottle the correct number of times on at least 75% of the days of the study period31. Based on the key study used for adherence rate inputs in our modeling analysis, this rate is well above the actual measured MEMS rate of adherence with oral atypical anti-psychotics (29%)31. In populations of patients where the rate of oral atypical anti-psychotic adherence is below the 44.9% threshold, paliperidone palmitate is projected to be a cost-saving treatment option. It is important to keep in mind that these MEMS adherence criteria are far more stringent than other adherence measures that do not take both dose taking and dose timing into consideration (e.g., medication possession ratios (MPRs)). Therefore, it should not necessarily be assumed that a population with an MPR ≥ 50% for patients taking oral atypical anti-psychotics would not provide an opportunity for paliperidone palmitate to achieve cost savings.

This analysis applied modeling techniques to evaluate the cost-effectiveness of paliperidone palmitate compared to branded oral atypical anti-psychotics. The approach used expressed the clinical and economic findings in terms that are meaningful for target audiences (i.e., costs per relapse days averted vs quality-adjusted life years (QALYs)). Overall, this evaluation was conducted using a limited number of assumptions and relied on a high level of published, peer-reviewed evidence for inputs. Atypical anti-psychotic drug costs inputs obtained from the Wolters Kluwer Source Lx (APLD) Database are very representative of the actual atypical doses for schizophrenia in the US as this database captures 82% of the prescription transactions in the US, including Medicaid, Medicare, Private Payer, and ‘Other Payer’ patients. Extensive one-way sensitivity analyses were performed on the model. The objective was to conduct a straightforward, easily interpretable analysis of the input parameters to which the model was most sensitive (i.e., what were the key cost drivers in the model and how did uncertainty around these parameter estimates affect the key results and conclusions).

The results of paliperidone palmitate being a cost-saving treatment strategy held true when tested through a multitude of sensitivity analyses. The results of the cost-effectiveness model were found to be most sensitive to the frequency, duration, and cost of relapses requiring hospitalization resulting from poor adherence. Regardless of what alternate plausible parameter values were implemented in the model, paliperidone palmitate remained more effective than oral atypical anti-psychotic care. Use of a higher maintenance dose of paliperidone palmitate (156 mg vs 117 mg) and inclusion of generic risperidone in the analysis resulted in a net incremental cost of paliperidone palmitate. Decision-makers may use the calculated ICER values in these two analyses to determine whether they find the additional cost to be worth the improved clinical outcomes (fewer days hospitalized or relapses). Generic oral olanzapine has very recently become available in the US. Initial US prices of generics are often not significantly lower than branded drug prices. Also, the availability of generic olanzapine in the US might prompt some physicians to switch users of other branded anti-psychotics to this highest-priced oral atypical anti-psychotic. Therefore, we would not expect the availability of generic olanzapine to have a significant impact on model results at the present time.

It is difficult to compare the current model results with other cost-effectiveness models of atypical anti-psychotics because of the variability in study design and methodology. This is the first economic evaluation of paliperidone palmitate to our knowledge. Previous economic models evaluating atypical anti-psychotics (i) have included different comparators22–25,42–44; (ii) have evaluated a different patient population45–47; (iii) have extrapolated results over long time horizons47; and (iv) have typically used preference-based end-points (e.g., QALYs) as the primary end-point47, rather than objective clinical measures such as days of relapse averted.

In order to understand the magnitude of the adherence problem, to identify contributing factors, and to assess the effectiveness of interventions, one must be able to accurately assess adherence48. Unrecognized incomplete adherence can result in counterproductive attempts by the physician to titrate the medication dose, switch patients to different medication(s), or add additional medication(s) to treatment regimens. Electronic monitoring systems of pill bottle systems, simplified dosing, supervised medication taking, reminder system, long-acting injectable medications, and psychotherapeutic interventions are strategies that have been shown to be helpful in improving adherence48.

This modeling evaluation has limitations. First, simplifying assumptions are required in decision analytic models. Otherwise, the models become too complex, not transparent, and require too many inputs that become harder to identify from actual data sources as the complexity of the model increases. When data sources are not identifiable, clinical expert opinion must be utilized. A balance in determining reasonable clinical treatment pathways and creating a transparent model based on published evidence is strived for. The simplifications made in outlining potential clinical pathways of patients in a model are determined based on the product evaluated and its attributes, as well as the comparators being assessed.

A second limitation is that the model was built by combining data from multiple sources to identify inputs for effectiveness, resource utilization, and costs. This lack of homogeneity of data sources is a common critique of economic evaluations based on modeling techniques. Decision models, by definition, bring together knowledge from a variety of sources when adequate experimental and/or long-term data are not available. Data obtained from a retrospective claims analysis might have provided a more homogenous representation of actual costs and utilization; however, properly adjusting for covariates in administrative claims analyses is very difficult. A prospective, randomized cost-effectiveness evaluation would be another approach to collecting economic evidence first-hand.

Finally, care should be taken in generalizing results found in this modeling evaluation to other patient populations. The results of this study may not be applicable to all patients with schizophrenia. This analysis was also conducted from the perspective of a US payer and the results of the evaluation would differ from those from a different perspective or in a different country. One-way sensitivity analyses (including threshold analyses) were used to examine the impact of altering key model assumptions and parameter values. However, multivariate analyses were not carried out.

Conclusions

Although uptake of LAIs in the US has not been as rapid as elsewhere, many thought leaders in schizophrenia treatment emphasize their importance in optimizing outcomes in patients with medication adherence problems3,5. Paliperidone palmitate is expected to be associated with reduced relapse rates compared with oral atypical anti-psychotics, largely because of poor adherence among patients receiving oral therapies. Based on results from this cost-effectiveness model, this reduction in relapse rates may also result in lower total healthcare costs. Future studies with paliperidone palmitate are needed to validate these findings in a real-world setting.

Transparency

Declaration of funding

The sponsor of the study was Janssen Scientific Affairs, LLC. The sponsor had involvement in the preparation of this article.

Declaration of financial/other relationships

The primary author is a research consultant hired by Janssen Scientific Affairs, LLC. The remaining authors are employees of Janssen Scientific Affairs, LLC.

Acknowledgments

The authors would like to acknowledge Matt Grzywacz and Amy Eves of ApotheCom for their assistance in editing and formatting the manuscript.