The manufacturer’s submission to NICE includes:
iii) a review of published economic evaluations of pharmaceutical interventions for the treatment and secondary prevention of VTE.
iv) a report of an economic evaluation undertaken for the NICE STA process. The cost- effectiveness of rivaroxaban is compared to LMWH+VKA for the treatment of PE and prevention of recurrent VTE.
Manufacturer’s review of published economic evaluations
The MS describes a systematic search of the literature which was conducted to identify economic evaluations of rivaroxaban using several health economic databases and medical databases. See Section 3.1.1 of this report for the ERG critique of the search strategy. The review did not identify any studies that evaluated the cost-effectiveness of rivaroxaban specifically.
Depending on the assumed anticoagulation treatment duration, the cost-effectiveness analysis uses either a 13 or 14-state Markov model to estimate the cost-utility of rivaroxaban compared to LMWH+VKA in adults with an acute PE. Results are presented by duration of anticoagulation therapy (3 months/6 months/12 months/lifelong).
A separate cost-minimisation analysis comparing rivaroxaban to LMWH only was undertaken to inform the appraisal of the potential value of rivaroxaban in reducing the monitoring burden for PE patients with active cancer (MS Section 7.9.3 page 269). This lies outside the NICE reference case.
The 14 Markov model states are detailed in MS Section 7.2.2 (MS page 112) and Table 26 (MS page 114). These states describe the management and complications of VTE and include an on-treatment state for the index event; two off-treatment states (off-treatment post index PE and off-treatment post DVT); three recurrent event states (DVT, PE and PE post DVT); three acute bleeding states; and two long-term complication states. Disease progression is not explicitly modelled as DVT and PE are generally acute conditions not classified by severity (MS Section 7.2.5 page 116).
The model has a lifetime horizon of 40 years and a cycle length of 3 months. Costs and outcomes are discounted at 3.5% per annum. The perspective of the model is the UK
NHS/Personal Social Services (PSS) and results are presented as incremental cost per QALY gained (MS Table 27 page 117).
The principal measures of rivaroxaban clinical-effectiveness used in the model are derived from the EINSTEIN-PE trial2. Estimates obtained from EINSTEIN-PE inform both the model
probability of a recurrent VTE whilst on rivaroxaban and the probability of an adverse event (bleed) whilst on rivaroxaban, compared to treatment with LMWH+VKA.
Quality of life in the model is determined by the disease health states. No mapping to utility was undertaken. A systematic literature review was conducted for relevant HRQoL data which yielded six studies (MS Table 42 page 152). Two further studies were used in order to fully populate the model utilities as the six studies identified by the systematic review did not provide all of the utilities required in the model (MS page 151 and Section 7.4.9 page 156).
A systematic review of the literature was undertaken to identify resource and cost data
associated with the treatment of VTE (MS Section 7.5.3 page 164). Twenty publications were included after final screening (MS Table 44 page 166). The model reflects resource use related to initial treatment and ongoing treatment and monitoring costs (MS Section 7.5.5 page 173).
The model dosing data for rivaroxaban match those in the draft SmPC (MS Section 6.10.4 page 102). The dosing data for LMWH (enoxaparin) are based on the UK licensed dose although this is different to the dosing regime which was delivered in the EINSTEIN-PE trial (MS Section 7.5.5 page 174).
Unit costs in the model are taken from the BNF643, Personal Social Services Research Unit (PSSRU) 14, NHS reference costs 2010/1115 and NICE Clinical Guideline 92 on reducing VTE risk in hospital patients.16
One-way deterministic sensitivity analyses were performed for a large number of model parameters including treatment effects and utilities (MS Section 7.6.2 page 193 and Table 52 page 194). Probabilistic sensitivity analyses (PSA) were also carried out (MS Section 7.6.3 page 195).
The MS states that two clinical experts were approached early in model development to provide validation on the initial model structure and parameter values tested in the model (MS Section 7.5.4 page 172 and Section 7.6.1 page 193). The MS notes that the expert comments were taken into account during the finalisation of the model structure, and that parameter values were refined following the literature review and results from EINSTEIN-PE. The model was also validated by comparison of its outcomes with those of EINSTEIN-PE (MS Section 7.7.1 page 196).
CEA Results
Results of the base-case economic evaluation are presented separately according to four treatment durations: 3, 6, or 12 months, or lifelong (Table 9). To be consistent with the MS, and to maintain clarity, Table 9 is presented in the non-standard results format adopted in the MS. At all treatment durations except lifelong, rivaroxaban dominates LMWH+VKA, i.e. rivaroxaban is cheaper and more effective than LMWH+VKA. For lifelong treatment the incremental cost per QALY gained is £13,252 (Table 9). The manufacturer states that there is a greater discounted
life expectancy and quality-adjusted life expectancy with rivaroxaban compared to LMWH+VKA, irrespective of treatment duration (MS page 252).
Results of the PSA indicate that at a threshold willingness to pay (WTP) of £20,000 the
probability that rivaroxaban is cost-effective compared to LMWH+VKA is 99.9%, 95.9%, 93.7% and 59.1% for 3 months, 6 months, 12 months and lifelong treatment respectively (MS Table 85 page 258).
Table 9 Base case cost-effectiveness results by treatment duration. (Reproduced from MS Tables 81-84.) Technologies Total costs (£) Total LY Total QALY s Incr. costs (£) Incr. LY Incr. QALY s ICER (£) (QALYs) 3 Months of Treatment Rivaroxaban 4,511 14.571 11.940 - - - - LMWH+VKA 4,907 14.546 11.912 396 -0.025 -0.027 Dominateda 6 Months of Treatment Rivaroxaban 4,546 14.630 11.992 - - - - LMWH+VKA 4,759 14.622 11.979 213 -0.008 -0.013 Dominateda 12 Months of Treatment Rivaroxaban 4,881 14.683 12.035 - - - - LMWH+VKA 5,015 14.671 12.015 133 -0.011 -0.020 Dominateda Lifelong Treatment LMWH+VKA 9,493 15.303 12.375 - - - - Rivaroxaban 10,868 15.333 12.479 1,375 0.030 0.104 13,252 a
Dominated treatment (LMWH+VKA) is less effective and more expensive than rivaroxaban
The cost-minimisation analysis indicates that over a six month period rivaroxaban is associated with a cost saving of £903.39 compared to LMWH (dalteparin) for treatment of PE in active cancer patients (MS Table 90 page 269). However the ERG considers that this analysis is speculative because, as acknowledged in the MS (MS Section 7.9.3 page 269), there are few data and no robust analyses to support the assumed equivalence of rivaroxaban and dalteparin in treatment of PE.