`DOI 10.1007/S402’73 016 0405 0
`
`ORIGINAL RESEARCH ARTICLE
`
`® CrossMark
`
`Cost-Utility Analysis of Lurasidone Versus Aripiprazole in Adults
`with Schizophrenia
`
`Krithika Rajagopalanl - David Trueman2 - Lydia Crowe2 - Daniel Squirrel]2 -
`Antony Loebell
`
`Published online: 11 April 2016
`© The Author(s) 2016. This article is published with open access at Springerlink.com
`
`Abstract
`
`lurasidone, an atypical antipsy-
`Background In 2014,
`chotic, was approved for the treatment of schizophrenia in
`adults. It is an alternative treatment option to aripiprazole,
`and when compared with aripiprazole,
`lurasidone was
`associated with improved symptom reduction and reduced
`risk of weight gain and relapse. We conducted a cost-utility
`analysis of lurasidone versus aripiprazole from the per-
`spective of healthcare services, using Scotland and Wales
`as specific case studies.
`incorporating a
`Methods A 10-year Markov model,
`6-week acute phase and a maintenance phase across three
`health states (discontinuation,
`relapse, death) was con-
`structed. Six-week probabilities of discontinuation and
`adverse events were based on a published independent
`mixed-treatment comparison; long-term risks of relapse and
`discontinuation were from an indirect comparison. Costs
`included drug therapy, relapse, and outpatient, primary and
`residential care. Costs and benefits were discounted at
`
`35 %. Utility estimates were taken from published litera-
`ture, and cost effectiveness was expressed as total 10-year
`incremental costs and quality-adjusted life-years (QALYs).
`Results Lurasidone yielded a cost saving of £3383 and an
`improvement of 0.005 QALYs versus aripiprazole,
`in
`
`Electronic supplementary material The online version of this
`article (doi:10.1007/s40273 016 0405 0) contains supplementary
`material, which is available to authorized users.
`
`[3 Krithika Rajagopalan
`Krithika.Rajagopalan@sunovion.com
`
`Health Economics and Outcomes Research, Sunovion
`Pharmaceuticals Inc., 84 Waterford Drive, Marlborough, MA
`01752. USA
`
`2 DRG Abacus. Bicester, Oxfordshire, UK
`
`Scotland. Deterministic sensitivity analysis demonstrated
`that results were sensitive to relapse rates, while proba-
`bilistic sensitivity analysis suggested that lurasidone had
`the highest expected net benefit at willingness-to-pay
`thresholds of £20,000 30,000 per QALY. The probability
`that lurasidone was a cost-effective treatment strategy was
`approximately 75 % at all willingness-to-pay thresholds,
`with similar results being obtained for the Welsh analysis.
`Conclusions Our analysis suggests that
`lm'asidone would
`provide an effective, cost-saving alternative for the healthcare
`service in the treatment of adult patients with schizophrenia.
`
`Key Points for Decision Makers
`
`Treatment of schizophrenia with atypical
`antipsychotics may be associated with weight gain
`and metabolic side effects.
`
`Lurasidone is a recently approved atypical
`antipsychotic for the treatment of schizophrenia in
`adults in Scotland.
`
`Lurasidone is associated with statistically significant
`improvements in efficacy and was generally well-
`tolerated in clinical studies when compared with
`other common atypical antipsychotics.
`
`Lurasidone is most likely to displace aripiprazole in
`patients with schizophrenia at risk of weight gain
`and/or metabolic disease.
`
`Lurasidone is likely to provide overall savings due to
`lower relapse rates and greater improvements in
`quality of life when compared with aripiprazole.
`
`AAdis
`
`Exhibit 2086
`Slayback v. Sumitomo
`IPR2020-01053
`
`
`
`710
`
`1 Introduction
`
`Schizophrenia is a chronic and disabling mental health
`condition resulting in progressive neurocognitive dys-
`function, leading to alterations in perception,
`thoughts,
`mood and behaviour [1]. It has a lifetime risk of approxi-
`mately 1 % and has a significant health, emotional and
`social impact on the patient, leading to social isolation,
`disability and dependence, unemployment and, in extreme
`cases, imprisonment and homelessness [2, 3]. The condi-
`tion has a significant financial burden; in England, the total
`combined annual cost to society and the public sector was
`estimated to be £19 billion in 2010/11 [4]. The mainstay of
`current treatment for acute schizophrenic episodes, symp-
`tom reduction, and relapse prevention in patients with
`schizophrenia is antipsychotic medication [5]. It is recog-
`nised in numerous national and international guidelines
`that patients with schizophrenia should be treated with
`first- or second-line antipsychotics, and offered clozapine
`after prior failure of two antipsychotics [6 8]. The choice
`of antipsychotics should be based on a combination of
`treatment efficacy, tolerability, and patient and carer pref-
`erences [6, 7]. In the UK, current Scottish Intercollegiate
`Guidelines Network (SIGN) guidelines (131) recommend
`that olanzapine, risperidone or amisulpride should be pre-
`scribed for first-line treatment of patients with acute
`exacerbation or
`recurrence of schizophrenia, and for
`maintenance treatment [9].
`Compounding this debilitating mental condition, comor-
`bidities related to cardiovascular disease and metabolic
`disorders, such as diabetes, hypertension, metabolic syn-
`drome, and obesity, are disproportionately prevalent among
`patients with schizophrenia [10]. Compared with the general
`population, patients with schizophrenia have almost twice
`the risk of metabolic syndrome (40.9 vs. 23.7 %, respec-
`tively) and diabetes (10.3 vs. 5.6 %, respectively) [11, 12],
`as well as an increased risk of cardiovascular disease-related
`mortality, with patients’ life expectancy reduced by an
`average of 15 years [13]. The prevalence of cardiovascular
`risk factors is also disproportionately high among patients
`with schizophrenia, of whom 58 % have dyslipidaemia,
`45 % have hypertension and 15 % have abnormal fasting
`glucose, while 68 % are obese [14].
`Although the presence of some modifiable cardio-
`vascular disease risk factors, such as an increased
`sedentary lifestyle, may be specifically attributable to
`schizophrenia, a number of atypical antipsychotics have
`been associated with an increased risk of weight gain and
`other metabolic abnormalities [15 17]. These adverse
`effects frequently lead to discontinuation and/or cycling
`between different
`therapies
`[18 21]. Schizophrenia
`
`K. Rajagopalan et al.
`
`remains one of the most challenging disorders to treat
`due to a number of factors, including heterogeneity of
`presentation and patient response to treatment, disease-
`related risk of morbidity and mortality, and treatment-
`emergent adverse effects such as weight gain [22, 23].
`For patients who are at risk of, or concerned about,
`weight gain, aripiprazole, haloperidol or amisulpride are
`recommended in SIGN guideline 131. This is supported
`by current National
`Institute for Health and Care
`Excellence (NICE) guidelines (CG178), which recom-
`mend that
`the potential
`risk of
`treatment-emergent
`weight gain should be considered when making treat-
`ment choices [5].
`In January 2014, lurasidone, a new atypical antipsy-
`chotic, obtained marketing authorisation in Europe for the
`treatment of schizophrenia in adults [24]. In the UK,
`lurasidone has received positive recommendations for use
`by the Scottish Medicines Consortium (SMC) in Scotland
`‘‘as an alternative treatment option in patients in whom it is
`important
`to avoid weight gain and metabolic adverse
`effects’’ and by the All Wales Medicines Strategy Group
`(AWMSG) as an option for use in adults aged 18 years and
`over [25, 26]. In five phase II and III, 12-month, double-
`blind, head-to-head studies, lurasidone was associated with
`significant improvements in symptom reduction and mini-
`mal changes in weight, body mass index, and metabolic
`outcomes versus placebo [27 31]. In studies where patients
`switched from a previous atypical antipsychotic to lurasi-
`done,
`lurasidone was associated with improvements in
`weight and lipid levels, and demonstrated a low rate of
`treatment failure and high rate of study completion [32,
`33]. When indirectly compared with other studies that have
`evaluated the efficacy and safety profile of atypical
`antipsychotics, such as aripiprazole, olanzapine, and que-
`tiapine, lurasidone is associated with significant improve-
`ments in terms of weight gain, metabolic outcomes, relapse
`rates, hospitalisations, and rates of all-cause discontinua-
`tion [34 36].
`While the clinical effectiveness of lurasidone in the
`treatment of schizophrenia has been demonstrated, the cost
`effectiveness of lurasidone versus alternative therapies
`remains to be established. We developed a model
`to
`evaluate the cost utility of introducing lurasidone as a
`treatment option for adult patients with schizophrenia from
`the perspective of healthcare services. In this study, we
`focus on Scotland and Wales as specific case studies in
`light of the recent SMC and AWMSG recommendations.
`These case studies compared the cost effectiveness of
`lurasidone versus aripiprazole as lurasidone is likely to
`replace aripiprazole as a treatment option for patients with
`schizophrenia.
`
`2
`
`
`
`Cost Utility Analysis of Lurasidone Versus Aripiprazole in Adults with Schizophrenia
`
`711
`
`2 Methods
`
`2.1 Model Overview
`
`To reflect the chronic nature of the disease, a Markov
`model was constructed in Microsoft Excel (Microsoft
`Corporation, Redmond, WA, USA) (Fig. 1) to estimate the
`effectiveness (relapse, discontinuations, side effects and
`mortality) and costs for adult patients with schizophrenia.
`In line with previous economic evaluations [27, 30, 37],
`this cost-utility model assumes that treatment is initiated in
`a population with acute schizophrenia (acute phase), who
`then continue into a maintenance phase following disease
`stabilisation. In line with other models, a 10-year time
`horizon was used so that longer-term differences between
`treatments could be considered, and a discount rate of
`3.5 % was applied to both costs and benefits [7, 38].
`The model compares two alternative treatment sequen-
`ces. For Scotland, current SIGN guidelines [9] state that
`‘‘clozapine should be offered to service users who have
`treatment-resistant schizophrenia’’, with treatment-resistant
`schizophrenia defined as ‘‘… failure to respond to an
`adequate trial of two different antipsychotics’’. Based on
`this guidance, simplified treatment sequences were con-
`structed. The first strategy consisted of lurasidone, fol-
`lowed by amisulpride, clozapine and, finally, an augmented
`clozapine strategy. The second differed from the first
`therapy in sequence only, which was aripiprazole.
`
`Patients enter the model in an acute phase of relapse
`undergoing trials of antipsychotic agents (‘non-stable/Rx
`trial’ health state). Patients who have not discontinued
`treatment by week 6 are assumed to enter the ‘stable/ad-
`herent’ disease state
`the maintenance phase
`and are
`assumed initially to be on treatment. Those who have
`discontinued treatment at week 6 for any reason are
`assumed to switch therapy at this point and re-enter the
`non-stable/Rx trial health state to continue the process of
`trialling alternative antipsychotic agents. Patients may also
`die from any health state within the model. The 6-week
`endpoint for the acute phase of the model, and ongoing
`cycle length in the Markov model, was chosen to be con-
`sistent with the short-term studies of lurasidone [27, 37].
`Individuals in the ‘stable/adherent’ health state in the
`maintenance phase are further subject to risks of all-cause
`discontinuation, relapse and death. Individuals discontinu-
`ing treatment in the maintenance phase are assumed to
`receive no therapy, and reside in the ‘stable/non-adherent’
`health state until the onset of relapse, at which point they
`enter the ‘relapse’ health state. Relapse is considered to be
`treated either in an inpatient setting or at home, with
`treatment administered via the crisis resolution home
`treatment teams (CRHTTs), and patients who relapse are
`assumed to discontinue current therapy and switch to the
`next therapy in the sequence.
`Reductions in health-related quality of life (HRQoL), as
`well as costs associated with weight gain (defined as a
`
`Fig. 1 Model schematic
`
`3
`
`
`
`712
`
`Table 1 Summary of efficacy
`and safety data used in the
`model
`
`Therapy
`
`Lurasidone
`
`0.77 (0.61, 096)
`
`2.46 (1.55, 3.72)
`
`5.22
`
`0.699c (0.303, 1.244)
`0.723
`
`K. Rajagopalan et al.
`
`Aripiprazole
`
`0.61 (0.51, 0.72)
`
`1.20 (0.73, 1.85)
`
`7.04
`
`1.029d
`0.98
`
`Acute phase model inputs
`All cause discontinuation [OR] (95 % CI)a
`EPS [OR] (95 % CI)a
`Weight gain [%]b
`
`Maintenance phase model inputs
`
`Relapse [HR] (95 % CrI) vs. quetiapine
`
`Discontinuation [HR] vs. quetiapine
`
`CI confidence interval, CrI credible interval, EPS extrapyramidal symptoms, HR hazard ratio, OR odds
`ratio
`a Calculated vs. placebo [34]
`b Probability of weight gain C7 % at week 6 estimated assuming a common standard deviation, assuming
`mean change in weight is normally distributed
`c From study D1050234, as reported Loebel et al. [37]
`d Calculated by indirection comparison with quetiapine [36]
`
`C7 % change in weight), presence of extrapyramidal
`symptoms (EPS) and diabetes, were applied, as experi-
`enced by patients in the model. Weight gain was assumed
`to persist while on treatment; EPS was assumed to persist
`for 3 months from the start of treatment, in line with the
`economic evaluation in NICE CG82, and incurred a one-
`off HRQoL decrement and cost; diabetes
`incidence
`occurred cumulatively over time from any state.
`The main outcome measure of the analysis was the
`incremental cost-effectiveness ratio (ICER) for lurasidone
`versus aripiprazole, reported as cost per quality-adjusted
`life-year (QALY) gained. The electronic model has pre-
`viously been reviewed by economists from UK national
`health technology assessment bodies [26, 39], and all
`clinical data and the model design were validated by an
`independent expert advisory board comprising nine clini-
`cians in the UK.
`
`2.2 Data Used in the Model
`
`2.2.1 Clinical Efficacy
`
`A 2013 independent systematic review and mixed treat-
`ment comparison (MTC) of atypical antipsychotics by
`Leucht et al. [34], including lurasidone and aripiprazole
`versus placebo, was used to inform estimates of short-term
`efficacy (probability of all-cause discontinuation) in the
`acute phase. Since the systematic review and MTC con-
`sidered the relative effectiveness of lurasidone and arip-
`iprazole versus placebo, it was necessary to establish an
`absolute placebo effect in order to estimate absolute effects
`for these therapies [27]. Model data inputs for all-cause
`discontinuation, EPS and weight gain for the acute phase
`are shown in Table 1, and a summary of all model input
`
`data is provided in Online Resource 1. Weight gain was
`considered clinically relevant if the patient experienced
`C7 % change in weight (measured in kilograms) from
`baseline. The independent MTC meta-analysis did not
`report long-term clinical outcomes, and no other compar-
`ative clinical data were available for lurasidone versus
`aripiprazole. Therefore, for the maintenance phase of the
`model, long-term risks of relapse and all-cause discontin-
`uation for lurasidone were taken from a 12-month, ran-
`domised, double-blind, active-controlled study versus
`quetiapine [37]. To inform aripiprazole data, the quetiapine
`arm of the lurasidone trial was then compared with arip-
`iprazole via an adjusted indirect comparison (via the
`Bucher method using olanzapine as the common com-
`parator [40]), with relapse data taken from a 52-week,
`open-label extension to a 26-week comparison of arip-
`iprazole with olanzapine [41], and from a 12-month, open-
`label extension study of quetiapine versus olanzapine [42].
`This approach ensures that the relative effect of aripipra-
`zole versus lurasidone can be calculated by discounting for
`the effect of the common comparator, quetiapine. To
`clarify, the adjusted indirect comparison of aripiprazole
`(A) computed an effect relative to quetiapine (B) by
`comparing aripiprazole (A) versus olanzapine (C) and
`quetiapine (B) versus olanzapine (C). In the absence of a
`common definition of relapse available across studies, all-
`cause hospitalisation was considered a proxy for relapse in
`the estimation of relative effects. We believe it is reason-
`able to consider relative treatment effects for all-cause
`hospitalisation as a proxy for relative treatment effect for
`relapse since hospitalisation is one of the variables mea-
`suring the composite endpoint ‘relapse’ in all clinical trials.
`For example, the definition of relapse provided by Loebel
`et al. [37] is ‘‘… the earliest occurrence of any of the
`
`4
`
`
`
`Cost Utility Analysis of Lurasidone Versus Aripiprazole in Adults with Schizophrenia
`
`713
`
`Table 2 Health state utility values used in the economic model
`
`State
`
`Value
`
`Source
`
`Stable
`
`Relapse
`
`Weight gain
`
`EPS
`
`Diabetes
`
`0.799
`
`NICE CG82 [38]/Lenert et al. [47]
`
`0.670
`0.959 %a
`0.888 %a
`0.150b
`
`Estimated from Briggs et al. [48]
`
`following 3 criteria: (1) worsening of C30% in the PANSS
`total score from Day 42 of the initial acute treatment study
`and a CGI-S C3; (2) re-hospitalisation for worsening of
`psychosis; or (3) emergence of suicidal ideation, homicidal
`ideation and/or risk of harm’’.
`While the cause is unknown, the prevalence of diabetes
`in patients with schizophrenia ranges from 11.3 to 22.3 %,
`and therefore the risk of developing diabetes was included
`in the model [43 45]. To include the effect of diabetes in
`the current analysis, an approach similar to that of the
`NICE CG82 was used. The relative effect of developing
`diabetes was equal to the relative effect of experiencing
`weight gain. Cardiovascular events were not considered
`since including them would potentially lead to double-
`counting of the consequences of diabetes.
`Mortality was based on published life tables of the
`general population, and adjusted to reflect the increased
`risk of mortality in patients with schizophrenia [46].
`In the acute phase, patients cycled through a number of
`treatment regimens until they reached a stable disease state.
`The efficacy and safety data of subsequent
`therapies
`(amisulpride, clozapine, and augmented clozapine) were
`taken from Leucht et al. [34]. Data for augmented cloza-
`pine were assumed to equal the data for clozapine. In the
`absence of data, the risk of relapse and discontinuation
`versus quetiapine were assumed to be equal to quetiapine
`in the maintenance phase; the risk of relapse and discon-
`tinuation were assumed to remain constant
`throughout
`subsequent lines of therapy.
`
`2.3 Health-State Utilities
`
`A systematic review of health state utility values and
`HRQoL evidence in schizophrenia was performed. Elec-
`tronic database searches were undertaken in November
`2012, and conferences were searched between 2010 and
`2012. Of the identified literature, those that were used in
`the development of the NICE guidelines were deemed to be
`the most appropriate to a UK clinical setting and were
`subsequently used to inform model estimates. To consider
`the impact of schizophrenia on patient HRQoL, utility
`scores reported in NICE CG82 and Lenert et al. were
`applied to patients in the stable and relapse health states [7,
`47]. Lenert et al. derived utility weights using a conve-
`nience sample of the general population employing a
`standard gamble approach [47]. Disutilities associated with
`clinically relevant weight gain and EPS (expressed as
`percentage reductions in the utility score for stable disease)
`were taken from the same sources. Disutility for diabetes
`was not presented in NICE CG82; for this adverse event, an
`absolute utility decrement observed between schizophrenia
`with diabetes and stable schizophrenia of 0.15 was
`
`EPS extrapyramidal symptoms, NICE National Institute for Health
`and Care Excellence
`a Percentage
`decrement
`stable schizophrenia
`b Absolute decrement in utility
`
`applied
`
`to
`
`utility
`
`value
`
`for
`
`assumed from the values presented by Briggs et al. [48]
`(Table 2).
`
`2.4 Costs
`
`Cost assumptions were based on those in NICE CG82 [7]
`and were updated with current estimates or adjusted to
`2013/14 costs using the Hospital Pay and Prices Index [49].
`All costs were presented to an advisory board consisting of
`five psychiatrists and four pharmacists, and country-
`specific data were used where available. Costs included
`pharmacological
`therapies, adverse events,
`switching
`therapies, and outpatient, primary and community care
`costs related to general management of care for patients
`with schizophrenia, relapse, and residential care (Table 3).
`Individual costs for outpatient, primary and community
`care costs are reported in Online Resource 2.
`List prices for pharmacological therapies were taken from
`the Monthly Index of Medical Specialities [50]. It was
`estimated that patients with schizophrenia receiving arip-
`iprazole would require a once-daily dose of 15 mg based on
`UK prescribing data [21]. For lurasidone, the assumed once-
`daily dose was 40 or 80 mg, based on data used for the
`World Health Organization Anatomical Therapeutic
`Chemical application (data on file). Adverse event costs
`included those associated with EPS and weight gain.
`Treatment for patients with EPS was based on 100 % of
`patients receiving procyclidine (5 mg/day for 3 months) and
`one psychiatrist outpatient visit, while treatment for weight
`gain consisted of the cost of two general practitioner visits
`and three dietetic outpatient contacts based on 100 and 20 %
`of patients receiving these services, respectively. Outpatient,
`primary and community care costs were all adjusted to
`6-week costs to fit the model cycle length. Cost of relapse
`was the combined cost of acute hospital admissions and
`CRHTT, assuming 30 and 70 % of patients receiving these
`services, respectively, and based on expert clinical opinion
`provided at
`the lurasidone advisory board. The mean
`
`5
`
`
`
`714
`
`K. Rajagopalan et al.
`
`Table 3 Summary of cost inputs used in the model
`
`Source data
`
`Value used (£)
`
`Pharmacological therapies (annual)
`
`Lurasidone
`
`Aripiprazole
`
`Adverse events (weighted per event)
`
`EPS
`
`Weight gain
`
`Switching therapy
`
`List price from MIMS [50] and daily doses based on data
`used for the WHO ATC application (data on file)
`
`List price from MIMS [50] and daily doses based
`on UK prescribing data [51]
`
`MIMS [50] and total direct cost per attendance general psychiatry [52]
`2 9 GP consultation costs [53] and 3 9 dietetic outpatients contact [54]
`
`Consultant psychiatrists
`
`Total direct cost per attendance general psychiatry [52]
`
`Outpatient, primary and community care (per 6 weeks)
`
`Total combined cost of all services
`
`See online resources for detailed costs
`
`1183a
`
`984b
`
`140c
`128d
`
`402e
`
`Stable patient: 647
`
`Relapse patient: 349
`
`Relapse
`
`Acute hospital (per inpatient week)
`
`CRHTT (per case)
`
`Residential care (weighted per 6 weeks)
`Sheltered housingh
`
`Group homei
`Long term hospital carek
`Total cost/cycle in stable health state
`
`Mean length of stay for mental health specialties
`combined with gross cost per inpatient week for general psychiatry [55]
`
`Crisis Resolution Team for adults with mental health problems.
`Average cost per case [53]
`
`Private sector care homes for people with mental health problems.
`Cost per residential week [53]
`NICE CG82, assuming 2006/07 costs [7]j
`NICE CG82, assuming 2007/08 costs [7]j
`
`2807f
`
`29,971g
`
`746
`
`15
`
`51
`
`812
`
`CRHTT crisis resolution home treatment team, EPS extrapyramidal symptoms, GP General Practitioner, NICE National Institute for Health and
`Care Excellence, WHO World Health Organization, ATC Anatomical Therapeutic Chemical, MIMS Monthly Index of Medical Specialities
`a Based on an assumed dose of 40 or 80 mg daily
`b Based on an assumed dose of 15 mg daily
`c Assumes 100 % of patients receive procyclidine 5 mg/day for 3 months [50] and have one psychiatrist outpatient visit
`d Cost of two GP visits and three dietetic outpatient contacts, based on 100 % and 20 % of patients receiving these services, respectively
`e Based on three visits to a consultant psychiatrist
`f Based on 30 % of patients receiving treatment
`g Based on 70 % of patients receiving treatment
`h Based on 18 % of patients in sheltered housing
`i Based on 2 % of patients in group accommodation
`j Costs adjusted to 2013/14 using the Hospital Pay and Price Index
`k Based on 3 % of patients in long term hospital care
`
`duration of treatment for relapse was based on the duration
`reported in a review of crisis resolution home treatment
`services in Scotland [56], and the cost per case from that
`reported by Curtis [53]. Due to regional variations in the
`number of days of treatment, the cost per case was used and
`adjusted from £29,628 to £29,971 for the 2013/14 price year
`[49]. The 6-week cost of residential care was the total
`combined cost of patients being in private accommodation
`(77 %), sheltered housing (18 %), group housing (2 %) and
`long-term hospital care (3 %).
`
`2.5 Sensitivity Analysis
`
`Uncertainty surrounding model inputs was assessed through
`sensitivity analysis. In the univariate sensitivity analysis, all
`model parameters were systematically and independently
`varied over realistic ranges determined by (1) the 95 %
`confidence intervals surrounding the point estimates, or (2)
`sensible ranges of values where there was an absence of
`sampling uncertainty (±25 % of the point estimate). The net
`monetary benefit, assuming a recommended willingness-to-
`
`6
`
`
`
`Cost Utility Analysis of Lurasidone Versus Aripiprazole in Adults with Schizophrenia
`
`715
`
`pay threshold of £20,000 per QALY [57 59], was recorded
`for lurasidone versus aripiprazole at the upper and lower
`parameter values, and was used to plot a Tornado diagram.
`Scenario analyses were also performed in which the values
`of key individual parameters were varied.
`Joint parameter uncertainty was explored through
`probabilistic sensitivity analysis. All parameters were
`assigned distributions and varied jointly over 5000 simu-
`lations. Where possible, correlation between parameters
`was preserved by assuming multivariate normality. Results
`were plotted on a cost-effectiveness plane and cost-effec-
`tiveness acceptability curve.
`
`3 Results
`
`3.2 Univariate Sensitivity Analysis
`
`the model
`Univariate sensitivity analysis revealed that
`parameters with the most impact on the cost effectiveness of
`lurasidone versus aripiprazole were those related to relapse
`rates. A negative net monetary benefit was generated (i.e. an
`ICER[£20,000 per QALY) when the hazard ratio of relapse
`for aripiprazole versus lurasidone was varied to the lower
`limit of the confidence interval. Despite being varied by a
`conservative ±25 % of the point estimate values, HRQoL
`values were not deemed as influencing factors for cost
`effectiveness. Figure 2 illustrates the results of the univariate
`analysis in the form of a Tornado diagram. Similar results
`from univariate sensitivity analyses were observed when the
`model was run using Wales-specific data.
`
`3.1 Base-Case Analysis
`
`3.3 Scenario Analysis
`
`Table 4 presents the results of the base-case analysis of
`lurasidone versus aripiprazole over a 10-year time horizon.
`Lurasidone was associated with an overall cost saving of
`£3383 per patient and a modest increase of 0.005 QALYs,
`meaning that
`it was a dominant strategy versus generic
`aripiprazole (associated with reduced costs and increased
`benefits). Although total drug acquisition costs were £416
`higher per patient treated with lurasidone (due to lower dis-
`continuation modelled in the maintenance phase), these were
`offset by reduced costs of relapse (£3942), switching (£17),
`and adverse events (£50). Similar results were observed when
`data from Wales were used, where there was an overall cost
`saving of £3072 and an increase of 0.005 QALYs.
`
`Results of the scenario analyses are presented in Table 5.
`At willingness-to-pay thresholds of £20,000 and £30,000
`per QALY, lurasidone was a dominant strategy in all sce-
`narios when compared with aripiprazole. The only scenario
`in which lurasidone was not considered cost effective was
`when no difference in relapse rates was assumed; in this
`scenario, lurasidone was associated with increased costs
`and fewer QALYs versus aripiprazole. However, this sce-
`nario should be interpreted with caution since relapses are
`driven largely by discontinuations due to the lack of effi-
`cacy and poor tolerability. It is therefore expected that if
`the relapse rates were equal between lurasidone and arip-
`iprazole, then efficacy and tolerability may also be equal to
`
`Table 4 Base case analysis
`results for lurasidone versus
`aripiprazole (10 year time
`horizon)
`
`Lurasidone
`
`Aripiprazole
`
`Incremental outcomesa
`
`Costs (£)
`
`Drug acquisition
`
`Inpatient relapse
`
`CRHTT relapse
`
`Residential care
`
`Switching
`
`AEs
`
`Outpatient/primary/community care
`
`Total
`
`Outcomes
`
`QALYs
`
`Relapse free days
`
`Life years
`
`Incremental analysis
`
`2195
`
`18,933
`
`47,844
`
`56,093
`
`566
`
`1354
`
`44,344
`
`171,329
`
`6.490
`
`3415
`
`8.284
`
`1779
`
`20,054
`
`50,665
`
`55,962
`
`583
`
`1404
`
`44,264
`
`174,712
`
`6.485
`
`3408
`
`8.284
`
`416
`
`1121
`
`2821
`
`131
`
`17
`
`50
`
`80
`
`3383
`
`0.005
`
`7
`
`0
`
`ICER (lurasidone vs. aripiprazole)
`
`Lurasidone dominant
`
`AEs adverse events, CRHTT crisis resolution home treatment team, ICER incremental cost effectiveness
`ratio, QALYs quality adjusted life years
`a Incremental outcome is equal to lurasidone minus aripiprazole
`
`7
`
`
`
`716
`
`K. Rajagopalan et al.
`
`Fig. 2 Univariate sensitivity analysis for lurasidone versus aripiprazole. NMB net monetary benefit (willingness to pay was £20,000 per quality
`adjusted life year)
`
`or similar between the two drugs. In fact, it is implausible
`to have a scenario in which efficacy and tolerability were
`assumed to be different but have similar relapse rates. Note
`that similar results were observed when the model was run
`using Wales-specific data.
`
`3.4 Probabilistic Sensitivity Analysis
`
`The results from 5000 simulations are presented on a cost-
`effectiveness plane (Fig. 3) and cost-effectiveness accept-
`ability curve (Fig. 4). Lurasidone was associated with the
`highest expected net benefit at all willingness-to-pay
`thresholds. The probability that lurasidone was the cost-ef-
`fective strategy was approximately 75 % at all willingness-
`to-pay thresholds. Similar results were observed when the
`model was run using Wales-specific data, where the proba-
`bility of lurasidone being a cost-effective strategy was
`approximately 70 % at all willingness-to-pay thresholds.
`
`4 Discussion
`
`We evaluated the cost effectiveness of lurasidone versus
`aripiprazole from the perspective of the National Health
`Service (NHS) and personal and social services in Scotland
`and Wales. The SIGN guidelines recommend that arip-
`iprazole
`should
`be
`prescribed
`for
`patients with
`schizophrenia who are at risk of, or concerned about,
`weight gain; aripiprazole has come to represent the most
`widely prescribed treatment for this specific population.
`The economic evaluation suggests that treatment of adult
`patients with schizophrenia in Scotland and Wales with
`
`lurasidone is a cost-effective strategy when compared with
`aripiprazole. In the base-case analysis,
`lurasidone was
`associated with an overall 10-year cost saving of £3383 and
`an increase of 0.005 QALYs per patient. The saving in
`costs was primarily driven by reduced relapse, switching
`and adverse events. When the model was run to evaluate
`the cost utility of lurasidone in Wales, lurasidone was
`associated with an overall 10-year cost saving of £3072 and
`an increase of 0.005 QALYs. Similar to the case for the
`Scottish analysis, the saving in costs was driven by reduced
`relapse, switching, and adverse events.
`Sensitivity analyses suggested that the cost effectiveness
`of lurasidone versus aripiprazole was only sensitive to
`relapse rates due to the high costs associated with relapse
`and all-cause discontinuation. Univariate analysis sug-
`gested that a negative net monetary benefit was generated
`only when relapse rates for lurasidone versus quetiapine
`were varied to the lower limits of the confidence interval.
`In addition, scenario analyses suggested that lurasidone
`was associated with increased costs and fewer QALYs
`when compared with aripiprazole when relapse rates were
`assumed to be equal for the two therapies. However, this
`analysis should be treated with caution. Probabilistic sen-
`sitivity analysis further supported the base-case results,
`demonstrating that lurasidone has a 75 and 70 % proba-
`bility of being cost effective at all willingness-to-pay
`thresholds in Scotland and Wales, respectively.
`There are several strengths to this economic analysis. Due
`to the chronic nature of schizophrenia and the long-term
`effects of the condition, the implementation of a Markov
`model and a 10-year time horizon allows the long-term
`assessment of cost effectiveness, and is in line with the model
`
`8
`
`
`
`Cost Utility Analysis of Lurasidone Versus Aripiprazole in Adults with Schizop