Rivaroxaban for treating pulmonary embolism and preventing recurrent venous thromboembolism

4 Consideration of the evidence

4.1 The Appraisal Committee reviewed the data available on the clinical and cost effectiveness of rivaroxaban, having considered evidence on the nature of pulmonary embolism and recurrent thromboembolism and the value placed on the benefits of rivaroxaban by people with the condition, those who represent them and clinical specialists. It also took into account the effective use of NHS resources.

4.2 The Committee discussed the clinical management of pulmonary embolism. It noted the statements received from the clinical specialists, which stated that people with suspected pulmonary embolism are generally treated with immediate parenteral anticoagulation, most commonly with a low molecular weight heparin (LMWH) delivered by subcutaneous injection, and when the diagnosis has been confirmed, an oral vitamin K antagonist such as warfarin. The LMWH is continued for at least 5 days or until the patient's international normalised ratio (INR) has been within the therapeutic range for at least 24 hours, at which point it is stopped. The Committee also heard that a minority of people receive unfractionated heparin or fondaparinux instead of LMWH. People presenting with pulmonary embolism and haemodynamic instability may receive thrombolysis (or undergo embolectomy if thrombolysis is contraindicated) before receiving a vitamin K antagonist. The Committee discussed the manufacturer's decision problem, noting that the manufacturer had excluded fondaparinux as a comparator even though it was specified in the final scope issued by NICE. The Committee noted the comments from the manufacturer and the ERG highlighting that fondaparinux is rarely used in UK clinical practice. The Committee accepted that fondaparinux is rarely used and agreed that it was appropriate to consider only LMWH and a vitamin K antagonist as the comparator as listed in the manufacturer's decision problem.

4.3 The Committee considered the treatment duration with anticoagulation. The Committee was aware that the NICE clinical guideline 144 on Venous thromboembolic diseases recommends 3 months' anticoagulation with a vitamin K antagonist for people with confirmed pulmonary embolism, with treatment continued beyond 3 months for those with permanent risk factors for venous thromboembolism recurrence, taking into account individual risk factors such as bleeding. However, it noted that approximately 95% of people in EINSTEIN‑PE received anticoagulation for 6 months or more. The Committee heard from the clinical specialists that anticoagulation therapy is often started with an expected duration of therapy, but that clinical evaluation is usually carried out at 3 or 6 months, when a decision is made as to whether or not therapy should be continued long term. It also heard that pulmonary embolism is a potentially life-threatening event, and that it would be unusual for people to be treated for less than 6 months. The clinical specialists also explained that people who have had a massive pulmonary embolus, recurrent venous thromboembolism, or are considered to be at significant risk of recurrence would usually receive lifelong anticoagulation. The clinical specialists estimated that overall, as many as 50% of people with pulmonary embolism would subsequently receive lifelong anticoagulation. The Committee accepted that although NICE clinical guideline 144 recommends an initial treatment duration of 3 months, the usual duration of treatment in UK practice was 6 months or more.

4.4 The Committee heard from the patient expert about the perceived benefits of rivaroxaban for treating pulmonary embolism and preventing recurrent venous thromboembolic events over the standard care of LMWH with a vitamin K antagonist (such as warfarin). The patient expert highlighted the disadvantages of warfarin, including the need for regular monitoring of INR, and dose adjustment. Monitoring, which needs regular visits to hospital or GP appointments, can be costly and inconvenient, and means some people may have to take time off work. The patient expert said young people of school or university age may also experience clots and INR monitoring can lead to disruption of education. The patient expert further explained that the burdens of monitoring and dose adjustments with warfarin may also affect carers of people who have had a pulmonary embolus and they may have to make adjustments to their own schedules to enable the person for whom they are caring to monitor their INR. As warfarin has many drug interactions, it may be unsuitable for people with comorbidities, and because of various food interactions, people may need to adjust and monitor their diet and lifestyle. The patient expert said that there are some people whose INR is unstable on warfarin or who are allergic to it, for whom an alternative is needed. The patient expert highlighted that rivaroxaban can be used at the onset of pulmonary embolism in primary or secondary care. Rivaroxaban has the advantage that it avoids injections, regular blood tests and the diet and lifestyle considerations necessary with the combination of heparin and warfarin. The patient expert said that patients are very interested in rivaroxaban and other newer anticoagulants, but noted that a small proportion of patients are concerned about these new agents and find regular monitoring of INR reassuring to confirm that they are adequately anticoagulated. The patient expert expressed the view that rivaroxaban should be made available as an additional treatment option. The Committee accepted the limitations of treatment with LMWH and a vitamin K antagonist and acknowledged the potential benefits of rivaroxaban.

4.5 The Committee considered the clinical management of bleeding resulting from treatment with anticoagulants. The clinical specialists highlighted that there are currently no validated guidelines on how to treat bleeding experienced while taking the new anticoagulants. The clinical specialists stated that people who have a bleed while taking a vitamin K antagonist may receive vitamin K, which takes 8–12 hours to reverse the effect of the vitamin K antagonist. One clinical specialist estimated that approximately 0.1% of people receiving warfarin experience a major bleed and for major bleeds needing rapid reversal, the most effective treatment is prothrombin complex concentrate (PCC) that provides correction for 12 hours. However, PCC would only be considered appropriate for a proportion of people with a major bleed. The Committee noted that the Evidence Review Group (ERG) had suggested that recombinant factor VIIa may be used to reverse a bleed. The clinical specialists agreed that recombinant factor VIIa may be used, but as it has a very short duration of action of approximately 2 hours, and is extremely expensive, they considered that PCC would be used in preference. The Committee further noted that recombinant factor VIIa is not licensed for the reversal of bleeding experienced on the new anticoagulants. It heard from the clinical specialists and the patient expert that there is no established antidote for rivaroxaban but because of its relatively short half-life, some bleeds can be managed simply by discontinuing rivaroxaban. The Committee concluded that there are standard approaches to stop bleeding experienced while on standard vitamin K antagonists such as warfarin, but there is uncertainty about the best approach to reverse bleeding experienced while taking rivaroxaban.

4.6 The Committee considered the generalisability of the EINSTEIN‑PE trial to UK clinical practice. It discussed whether the population in the trial reflects those seen in UK clinical practice. The Committee noted that the mean age of the population in the trial was 58 years. The clinical specialists highlighted that the incidence of deep vein thrombosis and pulmonary embolism increases with age, but EINSTEIN‑PE included few people over 80 years. The clinical specialists considered that the proportion of people in the trial with provoked and unprovoked pulmonary embolism was similar to that observed in clinical practice. The Committee noted the ERG's concern that EINSTEIN‑PE excluded people with severe renal impairment, even though they may be eligible for treatment with rivaroxaban with dose adjustment if needed. The clinical specialists stated that rivaroxaban would probably not be used in routine clinical practice for people with severe renal impairment. The Committee heard from the clinical specialists that in their opinion, the population in EINSTEIN‑PE generally reflected the corresponding UK patient population. The Committee concluded that the baseline characteristics of the population in EINSTEIN‑PE were generalisable to UK clinical practice.

4.7 The Committee discussed the comparator used in EINSTEIN‑PE and its relevance to UK clinical practice. The clinical specialists stated the LMWH used in the trial, enoxaparin, has the largest evidence base of all the LMWHs. The clinical specialists also stated that there were no known differences in the clinical effectiveness of the different available LMWHs. The Committee considered whether the dosage of enoxaparin used in EINSTEIN‑PE is applicable to UK clinical practice. It noted that in EINSTEIN‑PE, the US licensed dose of enoxaparin was used (that is, 1.0 mg/kg twice a day), whereas the dose used in the UK is 1.5 mg/kg once a day. The Committee heard from the clinical specialists that the 2 dosages are similar in terms of efficacy, and although people in EINSTEIN‑PE received a higher overall daily dose of enoxaparin than would be seen in UK clinical practice, this is not expected to have affected the generalisability of the trial to the UK. The Committee accepted that the differences in the dosage did not appear to be clinically significant and was satisfied that the comparators used in the trial represented routine and best practice.

4.8 The Committee considered the time in therapeutic range in the enoxaparin and vitamin K antagonist arm of the trial. It noted that the average time in therapeutic range for people receiving enoxaparin with a vitamin K antagonist was 62.7%. The clinical specialists stated that time in therapeutic range varies, but a range of between 60% and 70% would be expected in UK clinical practice. The Committee therefore concluded that the data from the enoxaparin with a vitamin K antagonist arm in the trial were applicable to routine clinical practice.

4.9 The Committee considered the generalisability of the EINSTEIN‑PE trial to the subgroup of patients who have active cancer. The Committee noted that EINSTEIN‑PE included a small proportion of people with cancer. The clinical specialists stated that people with cancer who experience venous thromboembolism would currently receive extended treatment with a LMWH alone, as evidence has shown that LMWH is more effective than warfarin for this group of people, and has been shown to reduce mortality. The Committee noted that the manufacturer had not presented any clinical evidence for a comparison of LMWH alone with rivaroxaban. The clinical specialists suggested that, without evidence from a direct comparison, it was unlikely that clinicians would offer rivaroxaban as an alternative treatment to LMWH for people with cancer. The Committee agreed that the comparator treatment in EINSTEIN‑PE that included a vitamin K antagonist did not reflect UK clinical practice for people with cancer, and there was no evidence for the relative efficacy of rivaroxaban compared with long-term LMWH, the standard treatment for these patients. The Committee concluded that without direct evidence of the relative efficacy of rivaroxaban compared with LMWH alone, it would be inappropriate to make a recommendation for this group.

4.10 The Committee considered the trial design and the clinical-effectiveness results of the EINSTEIN‑PE trial. It noted that the trial was designed to assess whether rivaroxaban was non-inferior to LMWH with a vitamin K antagonist for preventing recurrent thromboembolism after pulmonary embolism and that the manufacturers had also tested for statistical superiority for the primary efficacy outcome. The Committee noted that for the whole trial population, the rates of recurrent venous thromboembolism were not statistically significantly different in the rivaroxaban and LMWH with a vitamin K antagonist arms in the trial. The Committee concluded that rivaroxaban has acceptable clinical effectiveness compared with low molecular weight heparin and a vitamin K antagonist.

4.11 The Committee discussed the safety data from EINSTEIN‑PE. It noted that there was no statistically significant difference in the composite end point of major bleeding and clinically relevant non-major bleeding between rivaroxaban and LMWH with a vitamin K antagonist, but that the incidence of major bleeds was statistically significantly lower with rivaroxaban. The Committee concluded that rivaroxaban has an acceptable safety profile compared with low molecular weight heparin and a vitamin K antagonist.

4.12 The Committee considered the results from the treatment duration subgroups and noted that groups were based on clinical criteria. It discussed the level of uncertainty around the hazard ratios and noted that the confidence intervals surrounding the hazard ratio for recurrent thromboembolism overlapped. It was aware that the confidence intervals were particularly wide for the 3-month treatment duration subgroup and noted the analysis was based on a small number of people recruited to the group (around 5% of the study population) and on a small number of events in both treatment arms. In addition, the Committee heard from the clinical specialists that they were not aware of any biological reason to expect a differential effect in the 3-month treatment duration subgroup. The Committee therefore concluded that evidence of treatment effect should be based on the whole trial population of EINSTEIN‑PE.

4.13 The Committee considered the issue of long-term or lifelong treatment with rivaroxaban. It noted that the maximum length of treatment in EINSTEIN‑PE was 12 months, but was mindful that it had heard from the clinical specialists that some people will need longer treatment durations in clinical practice (section 4.3). In the absence of long-term data, the Committee considered the plausibility of the effects of rivaroxaban being maintained beyond 12 months. The clinical specialists stated that there is no biological or pharmacological reason why the effects of rivaroxaban should not be maintained over time but noted that there was uncertainty about how people would adhere to treatment with rivaroxaban over the long term. The Committee accepted that there was no biological or pharmacological reason why the effects of rivaroxaban would not be maintained over the long term.

4.14 The Committee noted that the manufacturer had presented 4 base-case scenarios for 3-, 6-, 12-month treatments and a lifelong treatment analysis. It noted that the economic model used clinical-effectiveness data from EINSTEIN‑PE and utility data derived through systematic review. The Committee noted that rivaroxaban dominated treatment with LMWH and a vitamin K antagonist, that is, was less costly and more effective in the manufacturer's deterministic analysis of 3-, 6- and 12-month treatment durations. It noted that for lifelong treatment, the manufacturer's base case incremental cost-effectiveness ratio (ICER) was £13,300 per quality-adjusted life year (QALY) gained but after the corrections made to the model by the ERG, the base case for lifelong treatment was reduced to £7070 per QALY gained. The Committee noted that this figure was calculated using the manufacturer's estimate of INR-monitoring costs.

4.15 The Committee discussed the estimate of the cost of INR monitoring. It heard from the clinical specialists that although there is a trend towards more monitoring being performed in primary care, some people still receive monitoring in secondary care. The clinical specialists also stated that there is huge variation in the frequency of INR monitoring for people receiving a vitamin K antagonist; some people may need weekly visits, particularly at the beginning of therapy, and some people may only need INR monitoring twice a year. The clinical specialists estimated that, as a guide, on the basis of 1 audit, INR monitoring once every 5–6 weeks might be a reasonable average estimate. The patient expert also highlighted that there is interest from patients in self-monitoring of INR, but there is variation in the availability of INR-home monitoring machines. The Committee concluded that there is considerable variability and uncertainty surrounding service provision and the frequency of INR monitoring that makes determining an accurate cost of INR monitoring problematic. It noted that the ERG had assumed fewer INR-monitoring visits than the manufacturer, which the ERG confirmed resulted in lower first year monitoring costs of between £304 and £379. The application of these scenarios to lifelong treatment increased the ICER from £7070 per QALY gained in the base case to £17,900 and £22,900 per QALY gained respectively. The Committee considered how INR-monitoring costs had been estimated in previous appraisals. In Rivaroxaban for the treatment of deep vein thrombosis and prevention of recurrent deep vein thrombosis and pulmonary embolism (NICE technology appraisal guidance 261) the Committee had considered that after deep venous thrombosis, an INR-monitoring frequency of 6 visits in the first quarter and 3 in each subsequent quarter, resulting in first-year costs of £320, to be reasonable. The Committee took into account the clinical specialists' estimates of average INR-monitoring frequency (that is, every 5–6 weeks) and also the deliberations during technology appraisal 261 and concluded that the ERG's scenarios were reasonable estimations of the impact of INR monitoring on the cost effectiveness of rivaroxaban, and that the manufacturer's estimate of frequency of monitoring visits was too high.

4.16 The Committee discussed the health-related quality of life data used in the economic model. It noted that health-related quality of life had not been measured in EINSTEIN‑PE and that the manufacturer had obtained the utility values used in its model through systematic review. The Committee was aware that the ERG had considered the utility values used in the economic model to be generally appropriate; however, the Committee considered that some of the studies that the manufacturer had used to obtain the utility values were too small and did not meet the reference case outlined in NICE's Guide to the methods of technology appraisal. The Committee expressed its disappointment that the manufacturer had not followed the reference case to derive the utility values. In particular, it noted that only 1 study fully met NICE's reference case criteria (that is, the source of data for measurement of health-related quality of life was reported directly by patients or carers, the source of preference data for valuation of changes in health-related quality of life was a representative sample of the public, and the preferred measure of health-related quality of life, the EQ-5D, had been used). The Committee noted that similar utility values had been used by the manufacturer in its economic model for NICE technology appraisal guidance 261 and it also noted that in the scenario analysis carried out by the ERG in which the utility value for the intracranial bleed state was increased from 0.33 to 0.55, there was only a minimal effect on the ICER. The Committee concluded that although the health-related quality of life studies selected by the manufacturer to obtain utility values for its economic model did not meet the NICE reference case, the cost-effectiveness estimates did not appear to be sensitive to the utility values used.

4.17 The Committee considered the 2 additional scenario analyses performed by the ERG for lifelong treatment that addressed the impact of an increase in the risk of a thromboembolic event or a bleeding event for rivaroxaban after 12 months. The Committee noted that the worst-case ICER for rivaroxaban compared with LMWH and a vitamin K antagonist was £14,100 per QALY gained when the hazard ratio for a recurrent venous thromboembolism was increased from 1.12 to 2, and £14,200 per QALY gained when the hazard ratio for a major bleed was increased from 0.49 to 0.79 (the upper limit of the 95% confidence interval for the major bleed hazard ratio seen in EINSTEIN‑PE). The Committee noted that the ERG had also carried out a multiple assumption scenario that resulted in an ICER of £35,900. This included the assumption that both the effectiveness of rivaroxaban in preventing venous thromboembolic events decreased and the risk of bleeds increased relative to conventional therapy after 12 months compared with the base case. The Committee heard from the clinical specialists that if an anticoagulant was less effective in preventing venous thromboembolic events, then it would be expected to have a lower rate of bleeds. The clinical specialists stated that a scenario in which both the risk of venous thromboembolism and bleeding increased was clinically implausible. The Committee concluded that the ERG's multiple assumption scenario was not plausible and the resulting ICER was not appropriate or relevant for this appraisal.

4.18 The Committee considered the 3 scenario analyses undertaken by the ERG in which the costs of reversing a major bleed had been incorporated and had been assessed in the 12-month and lifelong treatment analyses. The Committee noted that in these scenarios, the ERG had assumed that all people who had a major bleed would receive either treatment with PCC or recombinant factor VIIa; however, it was mindful of the testimony from the clinical specialists that only a proportion of people who had a major bleed would receive these treatments and PCC would preferentially be used over recombinant factor VIIa (see section 4.5). The Committee noted that recombinant factor VIIa is a particularly expensive drug and that the manufacturer did not consider the costs used by the ERG for recombinant factor VIIa to be relevant to patients in the decision problem. The Committee stated that consideration of bleeding reversal costs was relevant but that the ERG had presented extreme scenarios because it assumed that all people who had a major bleed would receive PCC or recombinant factor VIIa. The Committee agreed that there was too much uncertainty surrounding the treatment of bleeds experienced while on anticoagulants to reliably estimate the impact of treatment costs for reversing bleeding on the cost-effectiveness estimates. The Committee therefore concluded that the ICERs presented by the ERG for the 12-month analysis (worst-case scenario £23,400 per QALY gained) and the lifelong analysis (worst-case scenario £44,000) were not relevant for this appraisal.

4.19 The Committee considered that in all scenarios assessed for the 3-, 6- and 12-month treatment durations, rivaroxaban either continued to dominate, or the ICER compared with LMWH and a vitamin K antagonist could be considered a cost-effective use of NHS resources. The Committee concluded that rivaroxaban was cost effective for treating pulmonary embolism for 3, 6 or 12 months.

4.20 For lifelong treatment, the Committee considered the ERG's assumptions about the frequency of INR monitoring to be valid and concluded that the most plausible ICER for lifelong treatment with rivaroxaban compared with lifelong treatment with a vitamin K antagonist after initial treatment with a LMWH was between £17,900 and £22,900 per QALY gained. The Committee concluded that rivaroxaban is a cost-effective treatment option for the lifelong treatment of pulmonary embolism and prevention of recurrent thromboembolism for people in whom long-term treatment is indicated.