Cost-effectiveness of 3-month paliperidone treatment for chronic schizophrenia in Spain

Abstract

Background: A 3-month long treatment of paliperidone palmitate (PP3M) has been introduced as an option for treating schizophrenia. Its cost-effectiveness in Spain has not been established.

Aims: To compare the costs and effects of PP3M compared with once-monthly paliperidone (PP1M) from the payer perspective in Spain.

Methods: This study used the recently published trial by Savitz et al. as a core model over 1 year. Additional data were derived from the literature. Costs in 2016 Euros were obtained from official lists and utilities from Osborne et al. The authors conducted both cost-utility and cost-effectiveness analyses. For the former, the incremental cost per quality-adjusted life-year (QALY) gained was calculated. For the latter, the outcomes were relapses and hospitalizations avoided. To assure the robustness of the analyses, a series of 1-way and probability sensitivity analyses were conducted.

Results: The expected cost was lower with PP3M (4,780€) compared with PP1M (5,244€). PP3M had the fewest relapses (0.080 vs 0.161), hospitalizations (0.034 v.s 0.065), and emergency room visits (0.045 v.s 0.096) and the most QALYs (0.677 v.s 0.625). In both cost-effectiveness and cost-utility analyses, PP3M dominated PP1M. Sensitivity analyses confirmed base case findings. For the primary analysis (cost-utility), PP3M dominated PP1M in 46.9% of 10,000 simulations and was cost-effective at a threshold of 30,000€/QALY gained.

Conclusions: PP3M dominated PP1M in all analyses and was, therefore, cost-effective for treating chronic relapsing schizophrenia in Spain. For patients who require long-acting therapy, PP3M appears to be a good alternative anti-psychotic treatment.

Keywords:

• Paliperidone 3-month
• Schizophrenia
• Cost-effectiveness
• Spain
• Paliperidone long-acting treatment

Introduction

Schizophrenia is a chronic, incurable disease that afflicts a sizable proportion of people around the world, but rates vary across regions¹. In a comprehensive literature review involving 46 countries, Saha et al.² estimated a point prevalence of 4.6/1,000, with a range of 1.9 (10% quantile) to 10.0% (9% quantile). Worldwide, the estimated number of cases was 21.5 million in 2010¹. Using population statistics from 2010³, there were 6,892 million people, making the global prevalence of schizophrenia 3/1,000 inhabitants. The average annual incidence is 15 per 100,000, with a range of 7.7–43.0 per 100,000^4,5.

In Spain, the estimated prevalence of schizophrenia as reported in 2006 was 3.00/1,000 in males and 2.86/1,000 in females⁶. However, a more recent publication in 2016 estimated rates of 5.88/1,000 in males and 2.20/1,000 in females⁷. The estimated incidence reported in 2006 was eight cases per 100,000 inhabitants per year overall, 8.4 for males and 8.0 for females⁶. The annual incidence reported in 2008 was 13.8 per 100,000 inhabitants, with a male:female ratio of 1.61⁸.

It manifests with both positive and negative symptoms, as well as cognitive, mood, and motor symptoms, whose intensities vary between individuals and over the course of time^5,9. Positive symptoms are florid, such as delusions, hallucinations, and disorganized thinking. Negative symptoms present as deficits in cognitive, affective, and social functioning, as well as passive withdrawal¹⁰.

Schizophrenia adversely affects patients, decreasing their quality-of-life and level of functioning^11,12. As well, their lifespan is shortened; men live 15 years less and women live 12 years less than their counterparts who do not have schizophrenia¹³. Suicide and other violence claims ∼40% of the lives lost^14,15. The largest proportion of people with schizophrenia die from cardiovascular disease (33%), followed by cancer (15%), respiratory problems (13%), and infections (10%)¹⁶. It appears that risk factors for these diseases are increased in schizophrenia. In a recent review of the literature, Azad et al.¹⁴ concluded that rates of death due to cardiovascular disease were 90% higher than in the general population. Another risk factor, smoking, is highly prevalent in schizophrenia. Hughes et al.¹⁷ reported that 88% smoked, while Smith et al. (2014)⁷⁷ found that 69.1% smoked, as compared with 30–33% among non-schizophrenics. Hartz et al.¹⁸ reported that those with schizophrenia had 4.6-times more smoking, 4-times more heavy alcohol use, and 3.5-times more drug abuse than non-schizophrenics. Major problems arise not only from the lifestyle of persons with schizophrenia, but examinations of the care provided to those with cardiovascular disease indicated lower levels were provided. For example, cardiac catheterizations were only 41% compared with the general population¹⁹. Also, they are less likely to be diagnosed and treated than are those without psychiatric comorbidity²⁰.

This disease also creates a large burden on both society and healthcare systems^1,21,22. Particularly affected are the families and caregivers, whose burden is substantial^23–25. The average patient requires 48 hours of care per month²⁶, but patients with more severe symptoms may require 70 hours per week²⁵. The average caregiver incurs 2,258€ in direct costs and 6,667€ in indirect costs, for a total of 8,925€per annum²³. Additionally, caregivers suffer from decreased health, professional, family, and social interactions^23,25. In Spain, direct medical costs for schizophrenia constitute 53% of the total, and informal care costs are the other 47%. Hospitalization comprises 73% of the direct costs (38.5% of the total, including indirect/informal costs); drugs are responsible for 24% of the direct costs (12.8% of total), and outpatient care the remainder (∼3%)²⁶.

An association has been established between resource consumption by individuals with schizophrenia and non-adherence or partial adherence to treatment^27,28. Using electronic monitoring on a group of 102 out-patients in Spain, Acosta et al.²⁹ reported that 42.3% of schizophrenia patients exhibited “non-compliant behaviors”. Those who were non-adherent displayed poor insight, conceptual disorganization, stereotyped thinking, and poor attention. Furthermore, they found that psychiatrists, patients, and relatives all over-estimated rates of adherence to prescribed medications. The reported sensitivity (identifying compliers) was 96%, 100%. and 98% for psychiatrists, patients, and relatives, respectively. However, specificity (identification of non-compliers) was only 22%, 5%, and 5%, respectively. Agreement with the MEMS results was only fair for psychiatrists (kappa = 0.24) and slight for patients (kappa = 0.082), and relatives (kappa = 0.069). Another Spanish group examined the relationship between non-adherence and hospitalizations, as well as their associated costs³⁰. They found that poor adherence to anti-psychotics was associated with increased hospitalizations, resource consumption, and direct healthcare costs.

In an effort to address the problem of partial or non-adherence, depot formulations were developed several decades ago. The first were the phenothiazines, then butyrophenones, which have varying administration intervals³¹. However, the frequent appearance of adverse events, especially movement disorders, prompted a shift away from the typicals to the atypicals, in all forms³². The atypicals have fewer such effects, but can induce weight gain and metabolic problems³³.

Among the atypical anti-psychotics, risperidone microspheres (Risperdal Consta) was the first to be approved for use in Spain, on February 26, 2003³⁴. It must be administered every 2 weeks. In the subsequent years, longer acting therapies (LATs) appeared, such as paliperidone palmitate (Xeplion, Invega Sustenna) in 2011³⁵, which could be administered monthly into either the gluteal or deltoid muscle.

The administration interval for paliperidone has been further increased to 3 months with the introduction of Trevicta (Invega Trinza in the US; Janssen Pharmaceutica, Beerse, Belgium; PP3M)³⁶. These products have the same drug as their active ingredient, but vary in the interval of administration. Gopal et al.³⁷ have demonstrated that the 3-month formulation is equipotent to the 1-month formulation over 3 months when dosed at 3.5-times the 1-month dosage. However, the cost-effectiveness of this new 3-month formulation relative to the 1-month product is presently unknown. Therefore, the aim of this project was to determine their relative cost-effectiveness in persons with chronic schizophrenia in Spain.

Methods

In performing this research, we followed the recommendations for the economic evaluation of health technologies in Spain³⁸. The intended audience consisted of healthcare decision-makers in Spain and psychiatrists managing the pharmacotherapy of Spanish patients with chronic schizophrenia. The analyses took the perspective of the Spanish National Health System (NHS), as per the recommendations.

The primary focus was on PP3M, with PP1M as the comparison drug, as reported in a recent trial³⁹. These two drugs have the same active ingredient, but different durations of action (3 months vs 1 month). For analysis of the data, we used a decision tree model, which is depicted in Figure 1. The analysis was conducted on an Excel spreadsheet, over an analytic time horizon of 1 year. The primary clinical inputs were derived from the trial by Savitz et al.³⁹. This was a double-blind, randomized trial that compared PP3M with PP1M head-to-head over 48 weeks after a 17-week PP1M stabilization phase.

Figure 1. Decision tree used to model the economic analyses.

In order to model the consequences and costs of treatments, additional information was obtained from a variety of sources. Local experts provided inputs on care management. Other information was derived from published literature. For PP3M and PP1M, success rates, switches, and relapses (including hospitalizations and non-hospitalized relapses), and dropout rates were those observed in the Savitz et al.³⁹ trial. To estimate the time to relapse in those who discontinued from the drugs, we used the placebo data from Kim et al.⁴⁰. Those authors calculated relapse rates for three forms of paliperidone (oral, PP1M, and PP3M) drug over time. We used those placebo data to model the course of schizophrenia in patients who discontinued. Since all patients had been stabilized on their respective drugs, it is reasonable to assume that these values are a good proxy for discontinuers, and would provide valid inputs for the model.

Patients who dropped out were considered not to consume resources until they relapsed. To determine relapse rates and times for those patients, we used the data for placebos from Kim et al.⁴⁰. If there was a lack of efficacy, patients would be titrated to the maximum dose. According to the clinical experts, all patients who experienced adverse events were switched to second line treatment, which was aripiprazole long-acting therapy (ARI-LAT) for both PP3M and PP1M, with clinical data derived from Kane et al.⁴¹. In the event of intolerance to or lack of efficacy from the second line drug, patients would be switched to clozapine, as per NICE recommendations⁴². Clinical inputs are presented in Table 1, along with their values and sources for the data.

Table 1. Annual rates and regimens used in the analysis.

Category	Item	PP3M	Source	PP1M	Source	ARI-LAT	Source
Rate	Relapse (overall)	0.077	Savitz et al.^39	0.092	Savitz et al.^39	0.100	Kane et al.^41
	Hospitalized	0.035	Savitz et al.^39	0.043	Savitz et al.^39	0.026	Kane et al.^41
	Non-hospitalized relapse	0.043	Savitz et al.^39	0.049	Savitz et al.^39	0.074	Kane et al.^41
	Switch due to AEs	0.030	Savitz et al.^39	0.025	Savitz et al.^39	0.071	Kane et al.^41
	Dropouts	0.133	Savitz et al.^39	0.154	Savitz et al.^39	0.178	Kane et al.^41
	Utility (Stable/Relapse/Hospitalized)	0.700/0.485/0.270	Osborne et al.^50	0.650/0.469/0.270	Osborne et al.^44	0.650/0.469/0.270	considered equal to PP1M
Drug	Dose (mg)	414.75	Savitz et al.^39	119.11	Savitz et al.^39	396.3	Kane et al.^41
	Stable regimen	every 3 months	Savitz et al.^39	monthly	Savitz et al.^39; EMA^66	every 4 weeks	Kane et al.^41
	Restarting regimen after missing doses	If <4 months, give usual dose; if 4–9 months: stabilize with equivalent dose of PP1M immediately and 1 week later, then PP3M every 3 months	EMA^65; Gopal et al.^37	Same dose immediately and 1 week later, then monthly	EMA^66	400 mg immediately +10–20 mg oral aripiprazole daily ×14 days	Kane et al.^41; EMA^67
Hospital	Duration of stay	15 days acute, 12.5 days in day hospital	Average: Haro et al.^68; Mede and Sarria^69; Peiro et al.^70; Salize et al.^71; Salvador-Carulla et al.^72; clinical experts	15 days acute, 12.5 days day hospital	Average: Haro et al.^68; Mede and Sarria^69; Peiro et al.^70; Salize et al.^71; Salvador-Carulla et al.^72; clinical experts	15 days acute, 12.5 days day hospital	Average: Haro et al.^68; Mede and Sarria^69; Peiro et al.^70; Salize et al.^71; Salvador-Carulla et al.^72; clinical experts

AE, adverse event; ARI-LAT, aripiprazole long-acting treatment; EMA, European Medicines Agency; PP1M, paliperidone once monthly injection; PP3M, paliperidone 3-monthly injection.

We considered only the direct costs from the perspective of the payer. Prices of these resources were obtained from official lists issued by various governmental agencies in Spain. These prices appear in Table 2, and are all expressed in 2016 euros. Estimates of resources costed in previous years were adjusted to 2016 values via the Consumer Price Index for Spain⁴³.

Table 2. Costs of resources used in the analysis.*

Cost centre	Resource	Cost^†	Source
Drugs	Paliperidone 3-month injection (XF)	2.360€/mg	Botplus^73 database
	Paliperidone 3-month injection (PSP)	2.544€/mg	Botplus^73 database
	Paliperidone 1-month injection (XF)	2.751€/mg	Botplus^73 database
	Paliperidone 1-month injection (PSP)	3.317€/mg	Botplus^73 database
	Aripiprazole 1-month injection	0.631€/mg	Botplus^73 database
	Clozapine tablets	0.0064€/mg	Botplus^73 database
Medical care	Psychiatrist	73.90€/visit	Diario Oficial de Galicia^74
	Primary care physician	42.38€/visit	Garcia-Ruiz et al.^75 (2012)
	Psychiatric nurse (per injection)	25.94€	Quintero et al.^76 (2016)
Institutional care	Hospital acute psychiatric unit	293.88€	Quintero et al.^76 (2016)
	Day hospital	190.02€	Oliva-Moreno et al.^26 (2006)
	Day care center	105.20€	Oliva-Moreno et al.^26 (2006)
	Emergency room visit	132.07€	Oliva-Moreno et al.^26 (2006)
Tests	Clozapine level	12.64€	Diario Oficial de Galicia^74
	Blood test to monitor for blood abnormalities associated with clozapine	7.00€	Diario Oficial de Galicia^74

XF, Ex-Factory prices (base case); PSP, Public Selling Price (used in scenario analyses).

* All costs are expressed in 2016 euros.

† Costs for drugs are the average price across all available strengths.

Utilities were derived from the paper by Osborne et al.⁴⁴, who measured preferences using the time trade-off approach. It is the only available paper that measured the utility, based on patient preference, of a 3-month formulation for schizophrenia. They estimated a utility of 0.27 for patients in full relapse; for patients in a stable state, the estimated utilities were 0.70 and 0.65 for PP3M and PP1M, respectively⁴⁴. We assumed that utilities for ARI-LAT were the same as for PP1M, based on the analysis by Tempest et al.⁴⁵, wherein the annual QALY scores were only 1.2% different (0.727 vs 0.717). To determine the utility for non-hospitalized relapses, we used the mid-point between stable disease and relapse, as has been done by Druais et al.⁴⁶. That yielded a ratio of stable:non-hospitalized relapse that was virtually the same (1.38 vs 1.35) as we had determined in previous research⁴⁷.

The analytic outcomes were both cost-effectiveness (i.e. incremental cost per relapse or hospitalization avoided) and cost-utility (i.e. incremental cost per QALY gained). Clinical outcomes were rates of relapse, including those treated as outpatients, those requiring hospitalization, and overall relapses.

To examine the model robustness and stability of results, we conducted a series of sensitivity analyses adjusting inputs using various rates published in other studies. Rates altered included those for relapse, dropout, and switches due to adverse events. Other inputs varied were length of hospital stay (increased to 35 days as reported by Mede Herrero and Saria Santamera⁴⁸), doses of drugs (111.56, as administered in real-world practice, as reported by Gaviria et al.⁴⁹), and utilities (i.e. using PP3M values for all drugs and using previously used values)⁴⁷. Scenario analyses included using the Public Selling Price for drugs instead of the Ex-Factory price, eliminating dropouts, using switch rates from Berwaerts⁵², and using placebo-adjusted relapse rates. One-way (break-even) analyses were done on all probabilities and cost inputs. We also performed a probability sensitivity analysis, varying all inputs within their plausible ranges using standard distributions. We used 10,000 simulations to provide a stable estimate of results.

Results

Primary results are presented in Table 3. PP3M had the lower cost of the two drugs (4,780€ vs 5,244€) and the greatest number of QALYs (0.6765 vs 0.6249); therefore, it dominated PP1M in the cost-utility analysis. PP3M was also associated with a lower number of total relapses (0.081 vs 0.160), as well as patients requiring hospitalization (0.036 vs 0.064), and patients relapsing but treated as outpatients (0.045 vs 0.096). Therefore, it also dominated PP1M in all of the cost-effectiveness analyses.

Table 3. Outcomes from the pharmacoeconomic analysis.

Drug	Expected cost*	Total relapses	Hospital admissions	Emergency room visits	QALYs	Economic outcome^†
PP3M	4,780€	0.081	0.036	0.045	0.6765	Dominant
PP1M	5,244€	0.160	0.064	0.096	0.6249	dominated

* Costs are expressed in 2016 euros.

† For all outcomes listed in this table.

QALY, quality adjusted life-year; PP1M, paliperidone palmitate 1-month injection; PP3M, paliperidone 3-month injection.

In 1-way sensitivity analyses, results were generally resistant to changes in all of the rates tested, including rates of hospitalization. In scenario analyses, PP3M remained dominant over PP1M in all but one case (Table 4). In the worst case scenario (i.e. when PP3M was used in its maximum dose of 414.75 mg (as in the Lanzillo⁵⁰ trial, and PP1M was at its minimum of 75 mg/dose), PP1M would cost less than PP3M. However, PP3M remained cost-effective, with an ICER of 16,752€/QALY, which falls well below the accepted threshold in Spain of 30,000€/QALY.

In probabilistic sensitivity analyses, PP3M was associated with more QALYs than PP1M in 72.8% of simulations, and had a lower cost in 64.3%. PP3M dominated PP1M in 46.9% of the 10,000 iterations, and was dominated in 9.8%. Overall, PP3M was cost-effective in 77.8% of the simulations. Figure 2 displays the results in a scatterplot.

Figure 2. Scatterplot of cost-utility results.

Table 4. Results of scenario analyses.

Parameter changed	Present input	Altered input (source)	PP3M cost	PP1M cost	PP3M QALYs	PP1M QALYs	Result
Base case	Savitz et al.^39	n/a	4,780€	5,244€	0.677	0.625	PP3M Dominant
Dropout rate	Savitz et al.^39	0 dropouts	5,109€	5,427€	0.694	0.643	PP3M Dominant
Switching rate due to intolerance	PP3M = 0.028	PP3M = 0.075 (Berwaerts et al.^52)	4,792€	5,244€	0.676	0.643	PP3M Dominant
Drug doses (mg)	PP3M = 414.75 PP1M = 119.11(Savitz et al.^39)	PP3M = 328.685 (3.5 x PP1M dose) PP1M = 93.91 (Gaviria et al.^49)	4,029€	4,474€	0.677	0.625	PP3M Dominant
		PP3M = 414.75 PP1M = 75 (Summary of Product Characteristics^50)	4,780€	3,905€	0.677	0.625	PP3M cost-effective; ICER =16,752€/QALY
Drug cost	Ex-factory PP3M: 2.360 €/mg PP1M: 2.751 €/mg	Public price PP3M: 2.544€/mg PP1M: 3.317€/mg	5,072€	6,005€	0.677	0.625	PP3M Dominant
Hospital stay	23.5 days (clinical experts)	35 days (Mede Herrero and Saria Santamera^48)	4,852€	5,381€	0.677	0.625	PP3M Dominant
		48 days (Savitz et al.^39)	4,934€	5,536€	0.677	0.625	PP3M Dominant
Utilities	Adjusted via Osborne^50	Unadjusted	4,780€	5,244€	0.868	0.862	PP3M Dominant

ICER, incremental cost-effectiveness ratio; n/a, not applicable; PP1M, paliperidone palmitate monthly injection; PP3M, paliperidone palmitate 3-monthly injection; QALY, quality adjusted life-year.

In this analysis, the cost driver was the cost of the drugs, which comprised 76% and 71% of the total costs for PP3M and PP1M, respectively. Hospitalization accounted for 13% and 15%, respectively, while medical care was responsible for 11% and 14%, respectively.

With respect to adverse events, Savitz et al.³⁹ provided a thorough record of all events recorded during the trial (see Savitz et al., Table 4). Overall, 68% of patients who received PP3M experienced side-effects, as compared with 66% of those on PP1M (p = .67). Serious adverse effects were reported in 5% of those on PP3M, compared with 7% for PP1m (p = .22). Pain at the injection site was noted in 2% with PP3M and 3% for PP1M (p = .77). Considering that PP3M is injected fewer times per year than is PP1M, it would have an advantage over PP1M.

Model validation

This model determined that PP3M was associated with 0.677 QALY, which closely reflects the value of 0.70 estimated by Osborne et al.⁴⁴. For PP1M, the model estimated 0.625 QALY, which was quite similar to the values of 0.611 and 0.667 determined by Druais et al.⁴⁶ and Furiak et al.⁵¹, respectively. Also, the difference between PP3M and PP1M was 0.052 QALY, which corresponds with the value of 0.05 from Osborne et al.⁴⁴. Total relapses were 0.80 for PP3M, which is similar to the rate of 0.88 reported by Berwaerts et al.⁵². For PP1M, the rate of 0.161 is essentially the same as the average of rates of 0.175 from Hough et al.⁵³ and 0.152 from Fu et al.⁵⁴.

Discussion

PP3M is the first formulation that provides effective treatment for schizophrenia over a 3-month period. This product was developed using the same Nanocrystal technology, as was used with PP1M, except that PP3M has a different particle size which causes the drug to be released over a longer period of time^55–57. The half-life ranges from 84–139 days, depending on the dose and injection site⁵⁸. Thus, after discontinuation from steady state, the drug can maintain a therapeutic blood level for ∼5–7 months⁵⁵.

This prolonged half-life serves to prevent missed doses, thereby reducing relapses. Clinically, Savitz et al.³⁹ demonstrated that PP3M did numerically decrease relapses compared with PP1M in a head-to-head trial. In the present research, we found a 50.2% reduction overall (including those due to dropouts and medication switches), but derived from different studies. Thus, the theoretical advantages are reflected in the clinical outcomes.

The 3-month treatment also provides benefits to both the healthcare system and the patient. Since drug administration is required only four times per year, nurses are freed to engage in other activities. Thus, it allows for a more efficient use of resources in an already overburdened system. For patients and their caregivers, less time is required to travel to clinics for drug administration. Also, less frequent administration reduces the pain associated with administration. There is less chance of relapsing while on the 3-month treatment, which results in an increased quality-of-life for patients as well as caregivers. In addition, there is less disruption of daily activities.

As of this writing, there have been three separate research projects involving the cost-effectiveness of PP3M, all of which appeared as posters^59–62. In the first, Benson et al.^61,62 compared the 1-year costs of treating schizophrenia with either PP3M or placebo in the US in 2014. Clinical inputs were based on the randomized trial by Berwaerts et al.⁵². Total costs were comparable in the open label phase, but increased for placebo in the double-blind phase, while it decreased for PP3M. Due to lower resource consumption, total overall costs were lower for PP3M, as were all clinical rates (i.e. relapses, hospitalizations, and emergency room visits). PP3M dominated placebo in that analysis. A limitation of their analysis was that drug costs were not included.

The present analysis used a different RCT to model than that used by Benson et al.⁶¹. We used the head-to-head trial of Savitz et al.³⁹, which was also 1 year long. However, there was another difference in that patients who relapsed were censored in the Benson et al.⁶¹ paper, and resources were not costed after that point. In our analysis, we costed all patients over a full year, including those switched to other drugs and those who dropped out. Most importantly, the calculations by Benson et al.⁶¹ did not include drug costs, while the present analysis did, which explains the very large difference in reported costs.

In Belgium, De Moor et al.⁶⁰ utilized a Markov model with a 5-year time horizon to compare PP3M with PP1M. Clinical inputs were taken from two RCTs, which were Berwaerts et al.⁵² and Hough et al.⁵³. The model was different in that second line therapy consisted of no active treatment, followed (when patients relapsed) by third line based on local prescribing patterns. They concluded that prescribing PP3M rather than PP1M would increase quality-of-life and save money. A similar model was used in France by Arteaga Duarte et al.⁵⁹, which produced similar results. Again, our model differed in structure and duration.

Limitations

The time horizon was 1 year, but the disease lasts indefinitely. It is possible that results could be different over a longer time period. Therefore, a lifetime model would be more appropriate for determining long-term outcomes. However, at present, there are no data available beyond 1 year for PP3M.

Another limitation is that the data were derived from clinical trials, in which patients are closely monitored and adherence to prescribed regimens is assured. The situation could be different in “real world” clinical practice, where a host of other factors come into play, such as concomitant diseases and drugs, and impaired body functions.

These results have been determined within the context of the Spanish healthcare system. Thus, they are dependent upon the approach to managing schizophrenia in that country and their method of financing care. Extrapolation to other countries would require adaptation.

We considered only the direct costs of care that would be paid by the Spanish NHS. With respect to lost productivity, Vázquez-Polo et al.⁶³ determined (1997 estimate) that 20.51% of all persons with schizophrenia were employed either full time or part time, while 76.51% received social assistance. In a more recent publication from Norway, Evensen et al.⁶⁴ estimated that the employment rate among working-age (i.e. 18–66) individuals with schizophrenia was 10.24%. That estimate also included people with all severities of the disease. Another recent paper from Israel that examined patients with multiple hospital admissions reported that a mere 5.8% of patients with schizophrenia were earning minimum wage or above⁶⁵. Thus, the difference in indirect costs between drugs is likely to be very small.

In addition, this analysis did not include the costs of treating adverse events. Reasons were that they comprise a very small amount of the total cost and are very similar between drugs, and due to the short time horizon. For example, a recent economic analysis from the UK examining both PP1M and ARI-LAT (as well as other LATs) included these costs. The total discounted costs of care per patient over 10 years ranged from £157,594–£163,359, while the total costs of treating adverse events for 10 years ranged from £153–£190. That is, the cost of managing AEs ranged from 0.10% (1/10 of 1%) to 0.12%, which is negligible and virtually equal between drugs⁴⁵. Similarly, when summarizing their costs between PP3M and PP1M, De Moor et al.⁶⁰ also concluded that “The largest cost component ws the relapse costs, while adverse events contributed only marginally”.

Conclusions

In the present analysis, PP3M was cost-effective, dominating PP1M in all cost-effectiveness analyses, as well as the cost-utility analysis. Sensitivity analyses confirmed the robustness of the base case results. The major cost driver was the cost of the primary drugs (78% and 74%, respectively), which differs from previous analyses that found hospitalization to be the driver. As a conclusion, PP3M could represent a better treatment option for long term-treatment of chronic schizophrenia patients in the Spanish NHS. A limitation was that the model depended on data derived from randomized controlled trials. We still need observational studies to examine the situation under actual clinical conditions in practice.

Transparency

Declaration of funding

This research was funded by grant sponsor Janssen Pharmaceuticals grant sponsor NV, Beerse, Belgium.

Declaration of financial/other interests

TE and BB received funding for the research and manuscript. IGL, BGMM, FT, and KVI are all employees of Janssen. Peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.