Societal Versus Healthcare Perspectives on the Cost Effectiveness of Ocrelizumab for Treatment of Primary Progressive Multiple Sclerosis in Aotearoa New Zealand.

Key Points for Decision Makers

Ocrelizumab is the only disease-modifying therapy currently available to treat primary progressive multiple sclerosis (PPMS). It was publicly funded for this indication in New Zealand (NZ) in 2023. The list price is $NZ37,384 per patient per annum. Based on NZ Treasury criteria, the government could be justified in paying up to 56% of the list price ($NZ20,935 [$US11,828]) per person per annum for treatment of PPMS with ocrelizumab, from a societal perspective, or 10% of list price ($NZ3,738 [$US2,112]) from a healthcare perspective (2024 NZ dollars). The 5.6-fold difference is due to inclusion of direct non-medical and indirect costs, particularly lost wages and productivity. This study illustrates the importance of taking direct non-medical and indirect costs into account when evaluating new pharmaceuticals for chronic illnesses that require significant resources outside of usual public healthcare budgets.

Introduction

Multiple sclerosis (MS) is a chronic, progressive, neurodegenerative autoimmune disease affecting the central nervous system, which is characterised by inflammation, demyelination and degeneration of neuronal axons. Mostly diagnosed in people from ages 20 to 50 years, the global prevalence of MS has been increasing [ 1 ], and MS is the commonest cause of progressive disability in the Western world. The Atlas of MS estimated the global prevalence of MS at 2.8 per million people in 2020 [ 2 ]. Identification of treatments that might significantly impact long-term disability outcomes is likely to reduce costs to payers and improve quality of life [ 3 ].

Most individuals with MS (about 85%) present with relapsing-remitting MS where episodes of neurological dysfunction are followed by partial or full recovery over time; however, disability almost invariably accumulates to secondary progressive MS [ 4 ]. In contrast, primary progressive multiple sclerosis (PPMS) is typically characterised by progressive decline from onset, with occasional temporary plateaus or minor improvement [ 5 ]. Partial paralysis, especially of the lower limbs, and impaired upper limb function are common presentations [ 6 ]. A New Zealand study reported that 272 of 459 participants (59%) with progressive onset MS were female, in contrast to 1856 of 2386 (78%) with relapsing onset MS [ 7 ].

Aotearoa New Zealand (NZ) is a high-risk country for MS. In 2021, an estimated 4130 persons in NZ (population 5.11 million) had MS [ 8 ]. The incidence and prevalence of multiple sclerosis have a clear latitudinal gradient, with relatively high rates in Te Wai Pounamu (the South Island of NZ) [ 9 ] and in Australia (being more prevalent in its southern-most states [ 10 ]). A study of 1727 New Zealanders with MS reported that the mean age at diagnosis was slightly less than 40 years, with wide confidence intervals [ 11 ]. For 459 individuals with progressive onset MS, the median survival age at birth was 9 years lower than that of the general population, with more than double the overall mortality risk compared with the general population [ 7 ].

Long-term MS-related disability severity is typically characterised using the Kurtzke Expanded Disability Status Scale (EDSS), with three main disability milestones: EDSS4 (limited walking but without aid), EDSS6 (walking with unilateral aid) and EDSS7 (wheelchair dependent), with escalating costs. PPMS is usually diagnosed between EDSS states EDSS2 to EDSS5. Caregivers of patients with MS have been reported to experience high levels of distress and reduced health-related quality of life [ 12 ]. Annual costs per individual for an international cohort including NZ patients for mild, moderate and severe disability of the person with MS were $US1123, $US6643 and $US15,855, respectively [ 13 ].

MS is costly to healthcare and to society, particularly as MS-related disability severity increases; hence, delaying severe disability is likely to reduce annual MS-related healthcare payer and societal costs. Early intervention with high-efficacy disease-modifying therapies may represent the best opportunity to delay disability progression. Most individuals in younger age groups in NZ who have MS are likely to be employed and are likely to reduce their degree of employment as disability progresses, with a consequent loss of family income and tax income to the government [ 11 ].

Ocrelizumab (OCR), a recombinant humanised monoclonal antibody that selectively depletes CD20-expressing B cells, is the only disease-modifying treatment (DMT) shown to slow disability progression in individuals with PPMS [ 14 – 16 ]. The main evidence for the efficacy of OCR administered to patients with PPMS comes from the ORATORIO clinical trial, which was an international, multi-centre, double-blind, randomised, placebo-controlled, phase III trial conducted at 182 study locations including New Zealand [ 15 ]. Patients were randomly assigned in a 2:1 ratio to receive 600 mg of OCR by intravenous infusion or matching placebo every 24 weeks for at least 120 weeks and until a prespecified number of confirmed disability progression events had occurred. ORATORIO included patients with MS who were 18–55 years of age at screening and who had an EDSS state of 3.0–6.5. Pre-specified analyses on the intention-to-treat (ITT) population showed that OCR reduced the risk of 12-week confirmed disability progression (i.e. the increase in EDSS that was present for at least 12 weeks) (disability progression hazard ratio [HR] 0.76; 95% confidence interval [CI] 0.59–0.98) and had a similar effect on the risk of 24-week confirmed disability progression [ 15 , 17 ].

OCR has been approved for use in patients with PPMS by the US Federal Drug Administration [ 18 ] and recommended by the UK National Institute for Health and Care Excellence (NICE) [ 19 ] and the European Medicines Agency [ 20 ]. It has been funded for RRMS for several years by New Zealand’s Pharmaceutical Management Agency (Pharmac) and it was funded for PPMS in October 2023 while this study was in progress [ 21 ].

Treatment with OCR requires additional nursing and specialist time, and outpatient clinic chairs for infusions. Treatment is also likely to free up some health system resources by delaying the disability progression of PPMS—meaning that patients spend more time in less severe EDSS states that require fewer health system and other resources. This is also likely to improve employment outcomes [ 11 ].

Economic evaluations designed to inform public funding of pharmaceuticals and medical devices commonly include only direct medical costs and therefore take a healthcare perspective. Such studies could take the form of a cost-effectiveness analysis and/or a form of cost–benefit analysis (CBA) using net monetary benefit (NMB). Both CBA and CEA can take a societal perspective by including both healthcare and direct non-medical costs such as informal caregiving, specialised equipment, home/car modifications and patient transport, and also indirect costs such as loss of actual or potential income [ 22 , 23 ]. Pharmac limits its economic analyses to a healthcare perspective [ 24 ], although many countries take a societal perspective [ 25 ] as recommended by the Second Panel on Cost Effectiveness in Health and Medicine [ 26 ]. Inclusion of non-medical and indirect costs and benefits is likely to affect the ranking of pharmaceuticals for funding under a capped budget (see the Discussion section).

The purpose of this study was to estimate the lifetime cost effectiveness of treatment of persons with PPMS with OCR versus best supportive care (BSC) from both healthcare and societal perspectives, using both cost–utility analysis and cost–benefit analysis. The analysis plan was to develop a Markov model, populated from Australian costs of MS adapted to NZ; NZ costs of administration of OCR; and European health state utilities to estimate the potential cost effectiveness of treatment with OCR in NZ.

Methods

The Economic Model

The present analysis models the ITT cohort in the ORATORIO clinical trial, consistent with Pharmac’s prescribing criteria [ 27 ]. The Markov model has two structurally identical arms. For the treatment arm, each forward transition probability in the transition matrix was converted to a rate, which was then multiplied by the mean hazard ratio from the ITT analysis of ORATORIO (0.76) [ 15 , 28 ]. The product was then converted back to a probability [ 28 ]. The transition matrix in the BSC arm remained unchanged. The model has ten EDSS health states, plus death. It was developed using commercial software (TreeAge ® Pro, 2024). The model considers a cohort of individuals living with PPMS who move progressively from diagnosis to EDSS9 or death from any cause (Fig. 1 ). The start point of the model is initiation of therapy with OCR or the equivalent time in the structurally cloned control arm (‘best supportive care’), and the time horizon of the model is the lifetime of the cohort (terminating at age 95 years). A third arm (not shown) represents the age-matched life expectancy in the general population (59% female [ 7 ]). The health state utility scale is set at zero for death and anchored at 1.0 for full health. Negative values (worse than death) are allowed.

Simplified schematic of the Markov model, showing progression of cohorts of individuals between EDSS health states. For illustrative purposes, only five of ten disability states are shown. EDSS states 1–4 are clones of EDSS0. EDSS Expanded Disability Status Scale, EDSS_initial distribution of starting EDSS states in ORATORIO [ 15 ], OCR ocrelizumab, pDropout annual probability of withdrawal from therapy with OCR, pDth annual probability of all-cause death by age, PPMS primary progressive multiple sclerosis

In NZ, 62% of individuals with MS were between 20 and 40 years of age at the onset of MS [ 7 ] and the mean age at trial entry in ORATORIO was 45 years [ 15 ]. In the base-case analysis, a simulated cohort of patients enters the model at 40 years of age, and sensitivity analysis explores the impact of the age at commencement of treatment on the incremental cost effectiveness ratio (ICER). Each Markov cycle is 12 months to correspond to available annual transition probabilities. A standard half-cycle correction was applied, and all costs and benefits were discounted to net present value at 3.5%, as required by Pharmac [ 24 ], and (in a sensitivity analysis) also 5%, as suggested in guidance by NZ Treasury [ 29 ]. Costs and benefits of treatment with OCR were applied to EDSS states 2.0–6.0. The model includes introduction of a biosimilar within 4 years of 2024 to correspond to patent life in Europe and the USA, because NZ is a party to international treaties on patent expiry and entry of a biosimilar in NZ is expected to follow these timelines. Based on Pharmac experience, we assumed that the biosimilar will be priced at 70% less than the rebated price of OCR while on patent (range 50% to 90%).

Annual state transition probabilities represent the risk of an individual with PPMS progressing from a less severe to a more severe EDSS state, or (uncommonly) the reverse, within 12 months. The Markov model utilises annual state transition probabilities for PPMS patients diagnosed at up to 50 years of age to correspond to most NZ patients [ 11 ]. These were obtained from a large ongoing international neurological database (MSBase [ 30 ]), from which transition probabilities for PPMS were reported by NICE (table 46 on page 105 of committee papers in the single technology appraisal document for OCR in PPMS [ 31 ]). The NICE review of a manufacturer’s ten-state model of OCR in PPMS has been described elsewhere [ 17 ]. Eighty percent of patients represented in the MSBase dataset came from Canada, Spain, Italy, the Netherlands and Australia.

Mortality and Model Calibration

All forms of MS are progressive and potentially fatal. In NZ, the standardised mortality ratio (SMR) for 459 individuals with progressive onset MS was reported as 2.2 (95% CI 2.0–2.5). Life expectancy from birth was 9 years less than the sex-matched general population [ 7 ]. This information was used to calibrate the model.

The age-specific probability of death from all causes at each level of EDSS was estimated by multiplying the age-specific population annual mortality, adjusted to 59% female [ 7 ], by the EDSS-specific SMR and converted to a probability [ 28 ]. Because the SMR increases steeply with EDSS [ 32 ], the EDSS-specific SMR was estimated using the exponential function M = exp ( - a x E D S S ) , where the parameter α was set to 0.14 to give a life expectancy from birth 9 years less than the sex-adjusted NZ population average for PPMS [ 7 ]. It is assumed that mortality for individuals with no disability is equal to that of the general population.

Quality of Life

There are multiple sources for health state utilities (HSUs) for MS [ 15 , 33 – 38 ]. A study of 254 New Zealanders with all phenotypes of MS, using the EQ-5D-5L instrument, reported mean HSU for mild disability of 0.80, moderate disability of 0.57 and severe disability of 0.14 [ 36 ]. For a group of 1577 participants in the Australian MS Longitudinal Study (AMSLS), disability-specific HSU values were similar for different MS onset types. We therefore assumed that EDSS-specific HSU values determined across all phenotypes combined would apply to PPMS. Because the NZ and Australian studies contained only three (NZ) or four (Australia) levels of disability, we utilised European HSUs in our base-case analysis because they included all 10 disability states and were obtained from almost 17,000 participants with a range of phenotypes of MS, from 16 countries [ 37 ]. In this study EDSS9 is negative, indicating that the participants in this study rated this level of disability to be worse than death. Scores from EDSS0 to EDSS9 were as follows: 0.91, 0.81, 0.73, 0.65, 0.61, 0.57, 0.53, 0.38, 0.13, − 0.21. When caregiver disutility was included, the scores were 0.91, 0.81, 0.73, 0.64, 0.60, 0.55, 0.50, 0.33, 0.02, −0.35. Further information is available in the electronic supplementary material [ESM] of this paper (see Table 7).

Costs

Our analyses utilised either (1) direct healthcare costs or (2) societal costs, which include direct healthcare costs plus direct non-medical costs such as carer support and house/car modifications plus indirect costs defined as lost productivity caused by absenteeism and/or presenteeism (reduced activity while at work) or unemployment. In a large European study across 16 countries, healthcare resource use was driven more by system organisation than by medical need [ 37 ]. Because detailed costs of MS by EDSS state are not available for NZ, we assumed that disability-specific resource consumption for persons with MS in NZ (which we define as ‘best supportive care’ or BSC) is similar to that reported in a subset of the AMSLS [ 33 ] with 488 individuals. Just 32 of 488 (6.6%) of these individuals had PPMS. More details can be found in the ESM (Table 6) and the original publication [ 33 ].

AMSLS costs were obtained from cost diaries which captured detailed information on various cost categories relating to MS. Costs were broken down by MS-related disability severity based on disability cut points (no disability = EDSS0, mild disability = EDSS1–EDSS3.5, moderate disability = EDSS4–EDSS6, severe disability = EDSS6.5–EDSS9.5). Key cost categories comprise direct (personal, community/government, nursing home and equivalent care, informal care) and indirect costs (lost wages and productivity including early retirement). In addition, there is a breakdown of total direct costs by key categories. More details can be found in the original publication [ 33 ] and in the ESM of this paper. The corresponding Australian costs were adjusted to 2024 NZD using purchasing power parities (PPP) and the NZ consumer price index (CPI) (PPP in 2017 = 0.9685; NZ CPI in 2017–2024 = 1.2746) [ 39 ]. NZ$1.00 = $US0.5752 as at December 18, 2024.

Because there are ten EDSS states (plus death) but only four levels of cost by disability level are available from the AMSLS, intermediate costs were obtained by linear interpolation, and we assumed conservatively that costs for EDSS9 were the same as for EDSS8. At all levels of disability, indirect costs comprised between one-third and two-thirds of total costs, whereas direct medical costs comprised less than one-third (Fig. 2 ).

Estimated annual costs (2024 NZ dollars) per individual with all phenotypes of multiple sclerosis based on the Australian Multiple Sclerosis Longitudinal Study [ 33 ]. EDSS Kurtzke Expanded Disability Status Scale

In NZ, OCR is administered as two infusions of 300-mg vials in the first 6-month cycle, spaced 2 weeks apart, followed by two 300-mg vials administered as a single infusion every 6 months [ 15 ]. Each infusion is accompanied by other pharmaceuticals (Table 1 ). Vial sharing is unlikely and was not included in our analysis. The listed price of OCR is $NZ9346 per vial ($NZ37,384 per annum) [ 27 ].

Table 1 Input parameters for the multistate Markov model Item Value Comment Age at diagnosis 40 years (base case) EDSS distribution at start ITT distribution in ORATORIO (base case) or EDSS3, EDSS4 or EDSS5 ORATORIO [ 15 ] List price of OCR per vial $9346 Pharmac [ 27 ] Vials per 6-monthly dose of OCR 2 Confidential rebate on list price 10–90% Assumed potential range Annual cost of monitoring OCR $610.52 Pharmac Cost Resource Manual [ 40 ] Cost of administration of OCR (y1) $2012.15 Cost of administration of OCR (y2+) $1341.43 Cost of IV methylprednisolone (y1) a $102.30 Cost of IV methylprednisolone (y2+) a $68.20 Time from 2023 to biosimilar entry 4 years (range 3–5) Based on patent expiry in 2028 in Europe and 2029 in the USA (Roche Products Ltd) Biosimilar discount off rebated price of OCR 70% (range 50–90) Based on Pharmac experience Efficacy of OCR (hazard ratio) 0.76 (0.59–0.98) ORATORIO [ 15 ] Withdrawal rate in y1 and y2 0.054 ORATORIO follow up (Roche Products Ltd) Withdrawal rate in y2+ 0.02 (range 0.01–0.05) Assumed Annual discount rate 3.5% (range 0–5) For both costs and benefits PSA cost sensitivity multiplier Mean 1.0, SD 0.2 Gamma distribution ( n  = 5000) PSA health state utility multiplier Mean 1.0, SD 0.05 Normal distribution ( n  = 5000) PSA hazard ratio Mean 0.76, SD 0.17 Normal distribution ( n  = 5000) PSA transition probability multiplier Mean 1.0, SD 0.05 Normal distribution ( n  = 5000) Costs are expressed in 2024 NZ dollars EDSS Expanded Disability Status Scale, IV intravenous, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], PSA probabilistic sensitivity analysis, SD standard deviation a Prophylaxis with analgesic or antipyretics plus antihistamine is also recommended but the cost is minor

Input parameters for the multistate Markov model

Costs are expressed in 2024 NZ dollars

EDSS Expanded Disability Status Scale, IV intravenous, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], PSA probabilistic sensitivity analysis, SD standard deviation

a Prophylaxis with analgesic or antipyretics plus antihistamine is also recommended but the cost is minor

Because the cost is subject to a confidential rebate, the acquisition cost was varied over a representative range of 10% to 90% of the list price (90% to 10% rebate) in each analysis. Resources required for monitoring and administration of disease management therapies were applied to OCR as recommended in the package insert and costed according to Pharmac’s standardised cost resource manual [ 40 ].

Treatment with Ocrelizumab

Based on findings for the ITT primary endpoint in the ORATORIO clinical trial (12-week confirmed disability progression), the impact of OCR was estimated by multiplying each forward transition probability in EDSS states 2.0–6.0 (converted to a rate) by the mean hazard rato (0.76) reported in ORATORIO [ 15 ], following the NICE Economic Review Group [ 31 ]. This was then converted back to a probability [ 28 ]. Our model assumes that treatment with OCR is withdrawn in EDSS states above 6.0, as required by Pharmac’s funding criteria [ 27 ]. The BSC transition matrix remained unchanged.

Discontinuations in the OCR arm were included as a branch in the OCR arm of the model that takes those patients who withdraw to the BSC arm. Discontinuations in the BSC arm remain in that arm because (in practice) supportive therapy would continue. During the duration of evaluation in ORATORIO (at least 120 weeks), discontinuations due to adverse events were experienced by 4.1% of patients in the OCR arm and 3.3% in the placebo arm [ 15 ]. However, overall discontinuation rates for the treatment arm in ORATORIO were 10.7% for the first 2 years (Roche, data on file). We therefore assumed an annual discontinuation rate for patients treated with OCR of 5.4% for each of the first 2 years of therapy and 2% for subsequent years. The discontinuation rate beyond year 2 was varied from 1 year to 5 years in a sensitivity analysis.

In the ORATORIO clinical trial, there was no statistically significant difference between OCR and placebo groups in the rates of serious adverse events and infections, and discontinuation rates due to serious adverse events were similar in those groups [ 15 ]. Long-term clinical information up to 10 years post-marketing (to November 2022) showed that adverse event rates of OCR in 13 clinical trials remained stable in the long term [ 41 ]. Given this information, the cost and disutility due to adverse events for patients who continue therapy with OCR were not included in the model.

Cost–Utility Analysis

The ICER was defined as the lifetime cost per lifetime gain in quality-adjusted life years (QALYs) and the mean ICER is presented in Tables 2, 3, 4.

Probabilistic Sensitivity Analysis (PSA)

First, the 10×10 BSC transition matrix was deconstructed into 100 individual transition probabilities. Each forward transition probability (from EDSS0 to EDSS6) was converted to a rate then multiplied by a normal distribution (mean 1.0) and also by the hazard ratio from ORATORIO [ 15 ] which was also given a normal distribution (base case mean 0.76 [ 15 ]). The product was then converted back to a probability [ 28 ]. Each EDSS-specific HSU was multiplied by a normal distribution (mean 1.0) and all EDSS costs were multiplied by a gamma distribution [ 28 ] (see Table 1 ). Mean values from these distributions were used in all analyses. For the PSA, random pairs of values of net lifetime costs and utility gain in QALYs were extracted sequentially 5000 times from their respective distributions and the ICER was computed for each pair. Using larger numbers of repetitions did not change the results appreciably.

Net Monetary Benefit

The underpinning theory of cost–benefit analysis is that the outcome measure must include the effects of an intervention on everyone in society [ 42 ]. New Zealand Treasury has developed a CBA toolkit ‘CBAx’ for considering a wide range of impacts across time and multiple dimensions [ 43 , 44 ]. Health benefits are converted into monetary form using a standardised conversion factor. On this metric, a product or service is potentially fundable if its ICER is less than the monetary value of a QALY. A QALY in Treasury’s CBAx ‘Impacts’ database is valued at $NZ43,313 based on Pharmac’s historical investments, or $NZ120,200 based on the societal value of a statistical life ($NZ9.83m) [ 43 , 44 ]. These figures represent the healthcare and societal bounds of the Treasury’s guidance on publicly acceptable values of a QALY for cost-effective funding.

For new pharmaceuticals, the supplier provides a list price and the funder (Pharmac) negotiates a confidential rebate. Therefore, model outputs were determined over a wide range of acquisition costs. Research methods and analyses were aligned with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement [ 45 ] .

Results

Cost–Utility Analyses

At the three acquisition costs shown, OCR is substantially more cost effective from a societal than a healthcare perspective. It is more cost effective when therapy is initiated at a younger than older age, suggesting the value of early diagnosis and treatment (Table 2 ). In the base case at 50% of list price (viz. 50% rebate), therapy with OCR is initiated at age 40 years and the initial ITT distribution of EDSS levels and hazard ratio of 0.76 in ORATORIO are applied [ 15 ]. At 50% of list price and age at commencement of 40 years, the healthcare ICER is 28% higher than that from a societal perspective. Savings to the healthcare payer are not achievable, even at zero acquisition cost.

Table 2 Incremental lifetime costs, QALY gains and cost-effectiveness ratios (2024 NZ dollars per QALY) for ocrelizumab by study perspective and age at commencement of therapy, at three potential acquisition costs and zero cost Perspective Percent of list price ($37,384) 0% 10% 50% 90% Cost ICER Cost ICER Cost ICER Cost ICER Age at commencement of therapy 20 y (1.190 QALYs) Healthcare $15,714 $13,205 $43,166 $36,274 $152,976 $128,551 $262,787 $220,829 Societal a − $28,159 − $23,663 − $707 − $594 $109,104 $91,684 $218,914 $183,961 Age at commencement of therapy 30 y (1.093 QALYs) Healthcare $16,126 $14,754 $42,510 $38,893 $148,046 $135,449 $253,583 $232,006 Societal a −$22,139 −$20,255  $4245  $3884 $109,781 $100,440 $215,317 $196,996 Age at commencement of therapy 40 y (0.960 QALYs) Healthcare $16,492 $17,179 $41,363 $43,086 $140,843 $146,711 $240,324 $250,337 Societal a −$14,501 −$15,105 $10,369 $10,801 $109,850 $114,427 $209,330 $218,052 Age at commencement of therapy 50 y (0.788 QALYs) Healthcare $16,484 $20,919 $39,309 $49,884 $130,606 $165,744 $221,904 $281,604 Societal a − $6221 − $7895 $16,603 $21,070 $107,901 $136,930 $199,199 $252,790 ORATORIO ITT cohort [ 15 ]; annual discount rate 3.5%; age at start 40y; hazard ratio 0.76 Costs are expressed in 2024 NZ dollars ICER incremental cost effectiveness ratio, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], pppa per person per annum, QALYs quality-adjusted life years a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses

Incremental lifetime costs, QALY gains and cost-effectiveness ratios (2024 NZ dollars per QALY) for ocrelizumab by study perspective and age at commencement of therapy, at three potential acquisition costs and zero cost

ORATORIO ITT cohort [ 15 ]; annual discount rate 3.5%; age at start 40y; hazard ratio 0.76

Costs are expressed in 2024 NZ dollars

ICER incremental cost effectiveness ratio, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], pppa per person per annum, QALYs quality-adjusted life years

a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses

At any given drug acquisition cost, treatment with OCR is more cost effective when initiated earlier in the progression of disability (viz, at lower EDSS states), especially from a societal perspective (Table 3 ). For example, from a societal perspective for patients whose therapy with OCR commences at 40 years of age, at 50% list price, the ICER is 37% higher if MS is diagnosed and treatment is initiated at EDSS6 compared with EDSS2. The difference is probably due to higher indirect costs at more severe levels of disability, because the difference is much smaller (11%) from a healthcare perspective.

Table 3 Dependence of the incremental cost-effectiveness ratio on starting EDSS, study perspective and acquisition cost of ocrelizumab EDSS at start of therapy QALYs gained Incremental cost-effectiveness ratio Societal perspective a Healthcare perspective Percent of list price 10% 50% 90% 10% 50% 90% ORATORIO b 0.960 $10,801 $114,427 $218,052 $43,086 $146,711 $250,337 EDSS2 1.120 −$4266 $93,601 $191,469 $40,275 $138,143 $236,011 EDSS3 1.066 $1929 $102,451 $202,973 $41,388 $141,910 $242,432 EDSS4 1.008 $8926 $112,290 $215,654 $42,635 $145,999 $249,362 EDSS5 0.944 $14,762 $118,704 $222,645 $42,740 $146,682 $250,624 EDSS6 0.833 $20,778 $128,146 $235,514 $45,755 $153,123 $260,491 Starting age 40 y; hazard ratio 0.76; annual discount rate 3.5%; list price of OCR $9346 per vial (4 vials pppa) Costs are expressed in 2024 NZ dollars EDSS Expanded Disability Status Scale, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], pppa per person per annum, QALYs quality-adjusted life years a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses b ORATORIO initial distribution of EDSS states (ITT analysis): EDSS2 = 0.003; EDSS3 = 0.2485; EDSS4 = 0.2783; EDSS5 = 0.1889; EDSS6 = 0.2783 [ 15 ]

Dependence of the incremental cost-effectiveness ratio on starting EDSS, study perspective and acquisition cost of ocrelizumab

Starting age 40 y; hazard ratio 0.76; annual discount rate 3.5%; list price of OCR $9346 per vial (4 vials pppa)

Costs are expressed in 2024 NZ dollars

EDSS Expanded Disability Status Scale, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], pppa per person per annum, QALYs quality-adjusted life years

a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses

b ORATORIO initial distribution of EDSS states (ITT analysis): EDSS2 = 0.003; EDSS3 = 0.2485; EDSS4 = 0.2783; EDSS5 = 0.1889; EDSS6 = 0.2783 [ 15 ]

There is a linear relationship between the mean ICER and the acquisition cost of OCR (Fig. 3 ). The ICER is equal to Treasury’s suggested societal WTP threshold ($NZ120,200) when it costs 56% of the list price (viz. $NZ20,935 per annum) or Treasury’s inferred healthcare threshold ($NZ43,313) when it costs 10% of the list price ($NZ3,738 per annum). Based on this metric, a 5.6-fold higher investment in OCR can be justified from a societal perspective compared with a healthcare perspective. Alternatively, an acquisition cost of 37.9% of list price ($NZ14,169) could be justified if the criterion of gross domestic profit (GDP) per capita ($NZ83,011) is used as a societal funding threshold, as it is in many other countries (see the Discussion section). Because of the cost of administration and monitoring of OCR, despite a net savings to society there would be a cost to the healthcare payer, even at zero acquisition cost of OCR (see also Table 2 ; a positive ICER implies a net cost).

The relationship between the lifetime incremental cost-effectiveness ratio (ICER) and annual acquisition cost of ocrelizumab, from both a healthcare and a societal perspective (hazard ratio 0.76, annual discount rate 3.5%). GDP gross domestic product, QALY quality-adjusted life year

Sensitivity Analyses

One-Way Sensitivity Analyses

Several key findings emerge from a series of one-way sensitivity analyses (Table 4 ). At the high end of efficacy (95% CI of the hazard ratio in ORATORIO), at 50% of the listed price the societal ICER is $NZ41,277; and from a healthcare perspective at the same price the ICER is almost 90% higher at $NZ77,064.

Table 4 One-way sensitivity analyses of the incremental cost-effectiveness ratio Societal perspective a Healthcare perspective Percent of list price Percent of list price 10% 50% 90% 10% 50% 90% Hazard ratio (HR) 0.76 (base case) b $10,801 $114,427 $218,052 $43,086 $146,711 $250,337 0.59 (high efficacy) −$31,443 $41,277 $99,842 $18,499 $77,064 $135,629 0.98 (low efficacy) >$300,000 > $300,000 >$300,000 > $300,000 >$300,000 > $300,000 Cost of best supportive care (3.5% discount rate, HR 0.76) Low cost (−30%) $24,860 $128,485 $232,111 $47,459 $151,084 $254,710 High cost (+30%) −$3,257 $100,369 $203,994 $38,712 $142,338 $245,963 Discontinuation rate after 2y (base case 2%) Low (1%) $12,222 $110,587 $208,951 $39,841 $138,205 $236,570 High (5%) $12,021 $136,142 $260,263 $55,109 $179,231 $303,352 Time to biosimilar entry (discount rate 3.5%, base case 4 years, HR 0.76) c 3 years $8590 $103,369 $198,149 $40,874 $135,653 $230,433 5 years $12,870 $124,769 $236,669 $45,254 $157,053 $268,953 Cost of biosimilar (base case 70% discount, HR 0.76) 50% price discount d $20,004 $160,440 $300,876 $52,288 $192,724 $333,160 90% price discount d $1599 $68,414 $135,229 $33,883 $100,698 $167,513 Health state utility source (base case Europe [ 37 ] e UK [ 38 ] $11,895 $126,013 $240,130 $47,448 $161,565 $275,683 SW England [ 66 ] $11,116 $117,757 $224,398 $44,339 $150,981 $257,622 Other parameters Efficacy declines by 20% after 5 years $20,449 $141,554 $262,660 $52,444 $173,549 $294,655 Include carer disutility f $9502 $100,663 $191,825 $37,903 $129,064 $220,225 Discount rate 5% $13,310 $134,180 $255,059 $49,697 $170,577 $291,456 Discount rate 0% $10,189 $81,437 $152,686 $31,295 $102,544 $173,792 Costs are expressed in 2024 NZ dollars a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses b Base case: hazard ratio 0.76 [ 15 ]; annual discount rate 3.5%; price discount on biosimilar agent 70%; time to patent expiry 4 years; discontinuation rate after 5 years 2% c Patent expiry is 2028 in Europe and 2029 in the USA (Roche Products Ltd) d Discounted off the rebated list price of ocrelizumab e See Table 7 in the electronic supplementary material f Carer disutility from a NICE technical assessment of Alzheimer’s disease adapted to multiple sclerosis[ 19 ]

One-way sensitivity analyses of the incremental cost-effectiveness ratio

Costs are expressed in 2024 NZ dollars

a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses

b Base case: hazard ratio 0.76 [ 15 ]; annual discount rate 3.5%; price discount on biosimilar agent 70%; time to patent expiry 4 years; discontinuation rate after 5 years 2%

c Patent expiry is 2028 in Europe and 2029 in the USA (Roche Products Ltd)

d Discounted off the rebated list price of ocrelizumab

e See Table 7 in the electronic supplementary material

f Carer disutility from a NICE technical assessment of Alzheimer’s disease adapted to multiple sclerosis[ 19 ]

Importantly, the cost effectiveness of therapy with OCR depends strongly upon the cost and timing of a biosimilar agent that is very likely to enter the market at patent expiry. The ICER is moderately sensitive to the cost of management of MS from a societal perspective, but not from a healthcare perspective, because of inclusion or exclusion of direct non-medical and indirect costs, respectively. From both perspectives, the results are sensitive to the efficacy and cost of OCR, age at initiation of therapy, the time of entry of a biosimilar pharmaceutical and the discount rate. Notably, if the efficacy of treatment wanes by 20% after 5 years, the ICER increases by 24% at 50% list price from a societal perspective or 18% from a healthcare perspective. The NICE reference case includes caregiver disutility in technology appraisals [ 46 ]. The ICER declines by 12% when estimated caregiver disutility is included in the analysis.

Probabilistic Sensitivity Analyses

Figure 4 shows the incremental cost-effectiveness scatter plots on the cost-effectiveness plane, from societal and healthcare perspectives, with WTP derived from NZ Treasury recommendations (see Methods section). Notably, the discounted lifetime incremental cost is much higher from a societal than a healthcare perspective, although the QALY gain is the same and the distributions in the scatterplots look similar.

Incremental cost-effectiveness scatter plots from societal and healthcare perspectives: note the different vertical scales and willingness to pay (WTP). a Societal perspective (50% list price); b healthcare perspective (10% list price). List price: $NZ37,384 per patient per annum. Monte Carlo analysis, 5000 iterations. Red = ICER > WTP. Green = ICER < WTP. QALY quality-adjusted life year

Cost-effectiveness acceptability curves (Fig. 5 ) illustrate the probability of cost effectiveness from a societal or a healthcare perspective at the acquisition cost (percent of list price) at which there is 50% probability of cost effectiveness (see Sect. 3.1 ). Both graphs cross the 50% probability line at the NZ Treasury WTP values ($NZ120,200 from a societal perspective and $NZ43,313 from a healthcare perspective).

Cost-effectiveness acceptability curves at two levels of drug acquisition cost (percent list price), from societal and healthcare perspectives. List price $NZ37,384 per patient per annum. Monte Carlo analysis, 5000 iterations. X axis, 2024 NZ dollars

Net Monetary Benefit (NMB)

Pharmac does not have a willingness to pay (WTP) funding threshold but it does have ‘factors for consideration’ that include costs to society, although these are not quantified in its economic analyses [ 47 , 48 ]. One alternative metric is NMB, which is a summary statistic that represents the value of an intervention in monetary terms. NMB scales both health outcomes and use of resources to costs, so that comparisons can be made without defined ICERs [ 49 ].

(where WTP = willingness-to-pay threshold; C = discounted lifetime cost)

Positive values for NMB mean that therapy with OCR generates savings over the lifetime of the cohort. Using NMB as a decision metric, if the government would be willing to pay $NZ40,000 per QALY gained (achieving 25 QALYs per million dollars invested in healthcare), at 10% of list price ($NZ3738), there would be a monetary gain to society (positive NMB) but a cost to the healthcare sector (Table 5 ). At 5.8% of list price ($NZ2168), there would be savings to society but a cost to the healthcare payer (not shown).

Table 5 Net monetary benefit of therapy with ocrelizumab (in 2023 NZD) over a range of hypothetical WTP thresholds and OCR prices, from both healthcare and societal perspectives Percent of list price 10% 50% 90% Healthcare Societal a Healthcare Societal a Healthcare Societal a QALYs 0.96 Incremental cost $41,363 $10,369 $140,843 $109,850 $240,324 $209,330 ICER $43,086 $10,801 $146,711 $114,427 $250,337 $218,052 WTP threshold Net monetary benefit $20,000 −$22,163 $8,831 −$121,643 −$90,650 −$221,124 −$190,130 $40,000 −$2,963 $28,031 −$102,443 −$71,450 −$201,924 −$170,930 $60,000 $16,237 $47,231 −$83,243 −$52,250 −$182,724 −$151,730 $80,000 $35,437 $66,431 −$64,043 −$33,050 −$163,524 −$132,530 $100,000 $54,637 $85,631 −$44,843 −$13,850 −$144,324 −$113,330 $120,000 $73,837 $104,831 −$25,643 $5,350 −$125,124 −$94,130 Hazard ratio 0.76; annual discount rate 3.5%; list price of OCR = $NZ37,384 per annum; therapy commences at 40 years of age; ORATORIO ITT distribution of initial EDSS states Positive values are shown in italics EDSS Expanded Disability Status Scale, ICER incremental cost-effectiveness ratio, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], QALYs quality-adjusted life years, WTP willingness to pay a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses

Net monetary benefit of therapy with ocrelizumab (in 2023 NZD) over a range of hypothetical WTP thresholds and OCR prices, from both healthcare and societal perspectives

Hazard ratio 0.76; annual discount rate 3.5%; list price of OCR = $NZ37,384 per annum; therapy commences at 40 years of age; ORATORIO ITT distribution of initial EDSS states

Positive values are shown in italics

EDSS Expanded Disability Status Scale, ICER incremental cost-effectiveness ratio, ITT intent-to-treat, OCR ocrelizumab, ORATORIO pivotal clinical trial of OCR [ 15 ], QALYs quality-adjusted life years, WTP willingness to pay

a Healthcare costs plus direct non-medical costs including special equipment, house/car modifications and informal care plus productivity losses

Discussion

Our study is the first cost-effectiveness analysis of treating people living with PPMS with OCR versus BSC, from both healthcare and societal perspectives, using both cost–utility analysis and cost–benefit analysis. Our model was developed for the funding situation in New Zealand, but the methodology has relevance to other countries with similar healthcare systems and funding criteria for publicly funded pharmaceuticals. Based on NZ Treasury criteria, OCR can be cost effective from both healthcare payer and societal perspectives, but at different acquisition costs; and at 3.25% of list price or lower it can provide overall cost savings to society but a cost to the healthcare funder(s). Societal costs in our model include direct non-medical costs and loss of income through lost productivity, early retirement, absenteeism and presenteeism. Since employment status and quality of life are linked to disability level, protection against disability progression by OCR will lead to a greater likelihood of continuing employment and improved quality of life [ 50 ].

Cost effectiveness improves when younger individuals are treated and when individuals are treated earlier in their disease course. Importantly, the clinical imperative is increasingly shifting to earlier diagnosis and treatment with high efficacy DMTs [ 51 , 52 ]. Unsurprisingly, the cost effectiveness of OCR is very sensitive to its acquisition cost, from either a healthcare or societal perspective. Treatment with OCR is more cost effective when the cost of supportive care is high, and also with high cost and/or rapid market entry of a biosimilar agent. It is less sensitive to the discount rate. Savings to the health funder(s) are not achievable, even at zero drug acquisition cost, because of the cost of bi-annual intravenous administration and monitoring. Possible waning of efficacy after 5 years reduces the cost effectiveness of treatment with OCR substantially.

The multistate Markov model approach has been utilised by others to predict economic outcomes for the more prevalent forms of MS [ 53 – 58 ]. At the time of writing (September 2024), there is a published Portuguese cost–utility analysis of OCR in PPMS which takes a societal perspective [ 59 ] and one US study which takes a healthcare payer perspective [ 60 ]. There is also a NICE evidence group review of an unpublished pharmaceutical company model, based on a healthcare perspective [ 17 ]. The Portuguese study predicted that the discounted lifetime QALY gain is lower than our model predicts (0.78 vs 0.96), probably because of a higher discount rate (5% vs 3.5%). Also, the stated acquisition cost of OCR in Portugal (21,784 Euro per annum) is much higher than in our base case at 50% list price ($NZ18,692) and the ICER therefore is also much higher (81,052 Euro vs $NZ82,588) [1 Euro = $NZ1.80 in October 2024]. The US study reported a still lower QALY gain (0.58 vs 0.96), possibly because it utilised very dated (1997) insurance data on the relationship between EDSS and mortality. The ICER in this study was much higher than in our study ($US264,760 vs $NZ139,986), presumably because of the low QALY gain and higher acquisition cost for OCR ($US65,000 per annum). It is not possible to compare the company study evaluated by NICE with our study because the acquisition cost of OCR was not revealed.

Perspectives and Funding Criteria

The goal of taking a societal rather than healthcare perspective is to maximise social welfare from a flexible rather than a siloed healthcare budget. In this context, impacts on society that fall outside the healthcare sector are also included in the decision-making process. Most guidelines from well-established funding agencies recommend inclusion of a societal perspective, which is used in many countries [ 23 , 25 ]. It is also recommended in most other areas of NZ government like transport policy and environmental policy. Failure to utilise a societal perspective in decision-making risks allocative inefficiency between government departments during budget bids, hence the recommendations of the NZ Treasury to utilise a societal perspective.

Pharmac bases its funding decisions on a range of criteria including the outcome of a cost–utility analysis, a budget impact analysis and other factors [ 24 ]. Because its internal evaluations are confidential, it is not possible to determine the relative influence of these decision inputs or to compare its cost utility analysis with ours. Unlike reimbursement agencies in other jurisdictions, Pharmac does not have a WTP funding threshold [ 47 ], however it does have ‘factors for consideration’ which include costs to society, although these are not quantified in its economic analyses [ 47 , 48 ].

There is an ongoing debate about the choice of WTP cost-effectiveness thresholds that is crucial in determining the value of healthcare interventions and an authority’s willingness to pay. The World Health Organization suggests that the WTP threshold per QALY could be in the range of 1–3 GDP per capita [ 61 ]; however, 97% of 174 countries studied recently had a WTP threshold < 1 GDP per capita. The estimated WTP threshold for Australia, Canada, Germany and the UK ranged from $US38,000–$US50,000 [ 62 ]. In our analysis, from a societal perspective the ICER for OCR is equal to the NZ GDP per capita ($NZ83,011 in 2024) at 37.9% of list price ($NZ14,169). Alternatively, by NZ Treasury criteria, Pharmac can be justified in paying up to $NZ20,935 per person per annum for treatment of PPMS with OCR, when considering all the lifetime costs and benefits at net present value. However, when considering only those costs for healthcare, investment in OCR could be justified only up to $NZ3738 per person per annum.

Pharmac has historically made positive funding decisions over a broad range of ICER levels including those reported in this study. This is evidenced in part by the investment in OCR for PPMS being announced while this study was in progress. Pharmac now operates with an independent appropriation from government and so is obligated to follow the guidelines set out by New Zealand Treasury in budget bids for additional funds. For this, and to ensure comparability between government departments on investment options, it would seem sensible for Pharmac to report NMB according to the value of a QALY set by New Zealand Treasury. We have shown NMB over a range of potential thresholds and drug acquisition costs, as guidance for price negotiations. The methodology is applicable for other novel pharmaceuticals and in other countries.

If Pharmac’s health technology analyses of new pharmaceuticals had been undertaken from a societal perspective, as in China, Denmark, France, Norway, Portugal, Spain, Sweden, Taiwan, Thailand and The Netherlands [ 25 ], that would change the ranking of new pharmaceuticals that are approved and waitlisted for funding from Pharmac’s capped budget. In particular, it would favour those new medicines that reduce disability and/or suffering adequately to allow some patients to return to work and/or to work more productively. This would apply to many chronic illnesses including advanced cancer, MS, many other neurological diseases and other rare diseases.

Additionally, the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) recommends an even broader societal perspective to incorporate new and novel elements of value in health economic evaluation including family spillovers, the value of hope, scientific spillovers and broader productivity impacts. Notably, ISPOR recommends the adoption of this even broader societal perspective to address the limitations of conventional cost-effectiveness analysis [ 63 , 64 ]. Some health technology assessment bodies are now considering these broader elements [ 64 ].

Strengths and Limitations

The main strength of our analysis is that it is built on a validated set of transition probabilities from an ongoing international database (MSBase) and comprehensive and detailed sets of both direct and indirect costs and HSUs. Also, the economic model was calibrated to the mortality of PPMS in New Zealand [ 7 ], which provides face validity (Sect. 2.2 above). The calibration process also necessitated developing a method for estimating the relationship between SMR and EDSS. EDSS-specific SMR values were slightly lower than those reported in 1992 for patients attending MS clinics [ 65 ], probably reflecting better management today, even prior to DMTs. This gives confidence in our calibration method. No external validation was attempted, however a structurally identical model using the same transition probability matrix has been examined in detail by the Economics Review Group of NICE, and treatment of PPMS with OCR was subsequently recommended by NICE [ 31 ].

Costs

Because detailed New Zealand costs by EDSS level are not available, our analysis utilised a subset of the AMSLS comprising 488 individuals who recorded their costs in diaries. Real-world costs are not complicated by the requirements of a clinical trial such as regular reporting and monitoring, but they could be overestimated by recording of events or situations that were not directly associated with MS or underestimated by omission of relevant costs. We suggest that the AMSLS cost data provide a robust representation of Australian costs, given that a detailed cost diary was completed by participants over a 6-month period with instructions regarding the recording of costs directly related to MS. Furthermore, any inaccuracy in the cost of illness is likely to have only minor impact on our results because our sensitivity analysis shows that the ICER is only moderately sensitive to the cost of illness. A more important question is how well the Australian resource consumption and costs relate to New Zealand. Also, because the Australian study utilised only four levels of disability severity (none, mild [0 < EDSS < 4]; moderate; severe [EDSS > 6]), it was necessary to interpolate between them. These questions invite further research. A more comprehensive study could be considered for New Zealand using the integrated data infrastructure (IDI), which links health databases with other databases through unique identifiers.

Total annual costs per individual for a small subset of individuals with PPMS in the Australian cost study ( N  = 39) were similar to those in the parent study ( N  = 448) although informal care costs comprised a much smaller proportion of the total cost of PPMS than for all MS combined, and the proportion of indirect costs through loss of productivity was much larger, especially at high levels of disability [ 33 ]. This does not affect our analysis materially because it is independent of the proportions of direct non-medical costs versus indirect costs. Some Australian costs (such as hospitalisation) are likely to be higher than in New Zealand, and the healthcare system has a larger component of private funding, mostly through health insurance. However, our analysis did not attempt to differentiate public from private funding.

For comparison, Thompson and colleagues (2017) reported UK costs of MS in 2015 taken from a UK subset of a large study throughout Europe. In contrast to the Australian study (and therefore ours), the cost rose exponentially with the severity of disability, and informal care accounted for over one-third of the cost of the most severe disability category. This study, like ours, reported that income lost through early retirement was substantial throughout the range of EDSS [ 38 ].

Health State Utilities

This study utilised European HSUs [ 37 ] because New Zealand HSUs are sparse. It was assumed that EDSS-specific HSUs measured across all phenotypes of MS could be applied to PPMS, based on the AMSLS [ 34 ], although they could arguably be slightly lower for PPMS [ 33 , 66 ]. Carers’ quality of life can be greatly impaired relative to the general population [ 67 , 68 ]. Carer HSUs were derived from a population of individuals providing care for persons with Alzheimer’s disease and adjusted to reflect the time spent providing care for people with multiple sclerosis [ 19 ]. Because HSU data for MS carers have not been validated to our knowledge, we included this factor only in a sensitivity analysis.

Transition Probabilities and Efficacy

The transition probabilities used in our model were derived from a retrospective cohort study of the large, well validated ongoing international database, MSBase [ 30 ], with transition probabilities appropriate to PPMS as reported by NICE [ 31 ]. Such transition probabilities are subject to uncertain variability which was included in our probabilistic sensitivity analysis.

In order to include efficacy in a Markov model, transition probabilities are commonly multiplied by relative risk reported in a clinical trial; however, like other studies using a ‘time to event’ analysis, ORATORIO reported a hazard ratio, which provides an estimate of the ratio of the hazard rates between the experimental group and a control group over the entire study period [ 69 ].

Our analysis follows an analysis of a structurally similar manufacturers’ cost effectiveness model of OCR in PPMS by the Economic Review Group of NICE, which utilised hazard ratio rather than relative risk [ 31 ]. Like NICE, we assumed that the hazard ratio remains constant beyond the time horizon of the clinical trial; however, we applied both a one-way sensitivity analysis in which the hazard ratio increases after 5 years (i.e. the efficacy wanes) and also a probability distribution to the hazard rate.

Model Scope and Uncertainty

OCR was funded in New Zealand in October 2023 for treatment of PPMS from EDSS0 to EDSS6.5 [ 27 ]. Our model covered the range of EDSS0 to EDSS6.0, but EDSS6.5 was not included because transition probabilities for EDSS6.5 are not available. There is no evidence that the treatment effect (hazard ratio) is the same for all EDSS transitions, but that is a limitation of the clinical trial which flows through to our analysis and similar studies.

Our study did not include the cost of adverse effects of therapy with OCR. A large post-marketing study of OCR reported that continuous administration of OCR for up to 7 years in clinical trials, as well as its broader use for > 3 years in the real-world setting, was associated with a ‘favourable and manageable safety profile’ without emerging safety concerns in a heterogeneous MS population [ 70 ]. As of March 2024, more than 350,000 patients with RMS or PPMS have been treated with OCR globally. There have been 16 confirmed cases of progressive multifocal leukoencephalopathy (PML) in patients with MS, of which 12 were carry-over cases from prior disease-modifying therapies (Roche Products—Data on File, June 5, 2024). Therefore, the impact of including the cost of managing PML in the analysis is likely to be minimal, although some case series have reported increased rates of hospitalisation with infection in older, more disabled MS patients [ 71 ]. Weber and colleagues, in an uncontrolled cohort study of OCR to December 2020, reported that in Germany the most common and serious adverse events were infections and infestations, with serious infection rates of 1.5 events per 100 patient years for patients with PPMS [ 72 ]. It is difficult to cost these convincingly, given the uncontrolled study and limited detail on the types of infection.

Our findings are sensitive to the efficacy and acquisition cost of OCR and its efficacy over time; healthcare and other costs; age at the initation of treatment; the cost and timing of a potential biosimilar pharmaceutical; and the discount rate. They are moderately sensitive to the level of disability (EDSS) when treatment begins. Because Pharmac does not have an explicit funding threshold [ 47 ], the ICER is not a strong input to any funding decision, which is more likely to be driven by the budget impact of the new therapy. However, it is an important input for New Zealand and more important in Australia and other countries with a WTP threshold. The age at treatment has a modest impact on the ICER but will have little bearing on a funding decision, because it would be unethical to fund the therapy for some age groups but not others. There is evidence that treatment with some DMTs including OCR becomes less effective in older patients [ 73 ] although there is insufficient information for OCR to model this convincingly. Further research is required.

The efficacy of OCR could wane over time, although in an extension study of ORATORIO, compared with patients switching from placebo, earlier and continuous OCR treatment provided sustained benefits on measures of disease progression over 6.5 study years of follow-up [ 74 ]. Longer-term clinical information is required to establish whether treatment effects wane over time, and to determine long-term adverse effects of therapy. Either of these would reduce the cost effectiveness of therapy to an unknown extent. Table 4 also shows that treatment would become less cost effective if it is not maintained beyond 2 years. Finally, the delay before a biosimilar agent is introduced is an important modifier of the ICER and the budget impact, and this should be taken into consideration by any funding agency . When funding for OCR is considered in other countries, the analysis could benefit from utilising detailed local costs. When the results of extension studies are available, alternative models could be developed to predict cost effectiveness more accurately over time. With only minor modification, the same model could be useful for biosimilar agents.

Conclusions

Compared with best supportive care, ocrelizumab is much more cost effective from a societal than a healthcare perspective, therefore a funding decision is likely to depend critically upon the study perspective. This study illustrates the importance of taking direct non-medical and indirect costs into account when evaluating new pharmaceuticals for chronic illnesses that require significant resources outside of healthcare budgets.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 24 KB)

Acknowledgements

The authors thank Professors Ingrid van der Mie, Andrew Palmer, Bruce Taylor and colleagues at the Menzies Institute, University of Tasmania, for use of their published information on the cost of multiple sclerosis in Australia. The authors are grateful to the MSBase Registry and Investigators for use of their published information on the natural history of PPMS available from the UK NICE technology appraisal for ocrelizumab. The authors thank Andrew Munzer and Belinda Orme at TreeAge Software LLC for their advice.

Declarations

Conflict of interest

This study was funded by a grant from Roche Products (New Zealand) Ltd to Multiple Sclerosis New Zealand (MSNZ). Roche NZ reviewed an early draft of the manuscript but neither MSNZ nor Roche had any role in the analysis or interpretation of results or the decision to submit the manuscript for publication. CS is employed as a health economist by F. Hoffmann-La Roche Ltd (Roche) and owns shares in the company. None of the other authors have any relevant financial interest. RM is an Editorial Board Member of PharmacoEconomics. He was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions.

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions. RM was funded by a grant from Roche Products (NZ) Ltd to Multiple Sclerosis NZ. He has received conference support from Roche Products (NZ) Ltd. CS is an employee of Roche Singapore Pte and was previously an employee of Pharmac. JC has received funding from Multiple Sclerosis Australia. JM has received honoraria from Roche Products (NZ) Ltd.

Author contributions

Model development by RM and CS; review of cost data by JC; analyses by RM and CS; review of clinical information by JM; manuscript development and final approval by all authors.

Data availability

The Markov model can be obtained from the lead author on request

References

Associated Data

Supplementary Materials

Supplementary file1 (DOCX 24 KB)

Data Availability Statement

The Markov model can be obtained from the lead author on request