1 Appraisal Committee's preliminary recommendations

1.1 The Committee is minded not to recommend rivaroxaban for the prevention of stroke and systemic embolism in people with atrial fibrillation.

1.2 For the second Appraisal Committee meeting, the manufacturer of rivaroxaban should provide revised cost-effectiveness analyses comparing rivaroxaban with warfarin as follows:

• The characteristics of the cohort in the model should represent people with atrial fibrillation in the UK. Therefore ideally the baseline risks of events in the patient cohort in the model should be derived from the General Practice Research Database or the UK GP practice-based survey (Gallagher et al. 2008).
• The analyses should use clinical-effectiveness data from the safety-on-treatment population of the ROCKET-AF trial, and use all point estimates from this trial regardless of statistical significance.
• The effect of the low proportion of time in therapeutic range on warfarin in the ROCKET-AF trial should be accounted for by considering subgroup analyses by country or centre.
• The analyses should incorporate a fixed annual warfarin international normalised ratio (INR) monitoring cost of £242 per person.

2 The technology

2.1 Rivaroxaban (Xarelto, Bayer HealthCare) is an anticoagulant that directly inhibits activated factor X (factor Xa). Factor Xa is a key component in the formation of blood clots. In September 2011 it received a positive opinion from the Committee for Medicinal Products for Human Use (CHMP) for the 'prevention of stroke and systemic embolism in adult patients with non valvular atrial fibrillation with one or more risk factors: congestive heart failure, hypertension, age 75 years or older, diabetes mellitus, prior stroke or transient ischaemic attack'.

2.2 According to the summary of product characteristics provided by the manufacturer, approximately 14% of people treated with rivaroxaban in clinical studies experienced adverse reactions. Bleeding and anaemia occurred in approximately 3.3% and 1% of patients respectively. Other common adverse reactions were nausea and an increase in transaminases. The summary of product characteristics states that the risk of bleeding may be increased in certain patient groups, for example those with uncontrolled severe arterial hypertension and/or those taking other treatments that affect haemostasis. For full details of side effects and contraindications, see the summary of product characteristics.

2.3 The provisional cost of rivaroxaban is £2.10 per day and £766.50 annually. Costs may vary in different settings because of negotiated procurement discounts.

3 The manufacturer's submission

The Appraisal Committee (appendix A) considered evidence submitted by the manufacturer of rivaroxaban and a review of this submission by the Evidence Review Group (ERG; appendix B).

3.1 The main clinical effectiveness evidence came from one multicentre, double-blind randomised controlled trial. The ROCKET-AF trial (‘Rivaroxaban once daily oral direct Factor Xa inhibitor compared with vitamin K antagonism for prevention of stroke and embolism trial in atrial fibrillation’) compared rivaroxaban with dose-adjusted warfarin. The manufacturer also compared rivaroxaban with aspirin and dabigatran etexilate (110 mg or 150 mg twice a day) using a network meta-analysis in people for whom anticoagulation therapy is considered suitable. The ROCKET-AF trial was designed as a non-inferiority trial in which a blinded dose of rivaroxaban (20 mg or 15 mg once a day) was compared with open-label warfarin (target INR of 2.0–3.0) for the prevention of stroke and thromboembolic events in people with non-valvular atrial fibrillation at risk of future thromboembolic events. People were randomly allocated to one of the two treatment groups with equal probability (1:1 allocation ratio). The study took place in 45 countries including the UK and a total of 14,264 people were enrolled across the two treatment arms (rivaroxaban n = 7131 and warfarin n = 7133). The duration of the treatment period depended on the time needed to obtain approximately 405 adjudicated primary efficacy end point events in the per-protocol population on treatment. As a result, the time on study treatment varied from patient to patient depending on the time of the patients’ enrolment. The expected study duration was approximately 40 months from first patient enrolled to the occurrence of the last event. The median duration of treatment was 590 days.

3.2 The primary efficacy end point in ROCKET-AF was a composite of stroke (ischaemic and haemorrhagic stroke) and non-central nervous system systemic embolism. The primary safety end point was defined as a composite of major bleeding and clinically relevant non-major bleeding. A margin of 1.46 for the risk ratio (rivaroxaban/warfarin) was used to assess non-inferiority in preventing stroke and non-central nervous system embolism. To show non-inferiority, the upper bound of the confidence interval of the hazard ratio (HR) for rivaroxaban compared with warfarin had to be less than 1.46. Once non-inferiority was demonstrated for the primary outcome, further analyses investigated superiority of rivaroxaban over warfarin.

3.3 More than 50% of people in the trial received treatment for at least 18 months. The median age of study participants was 73 years and 60.3% were male. The majority of the trial population (62.4%) had received prior warfarin therapy and 36.5% had received prior aspirin. Risk of stroke at baseline was classified according to CHADS₂ score, which is used to predict the risk of stroke in people with atrial fibrillation. The trial entry criteria included a history of stroke, transient ischaemic attack or systemic embolism, or a CHADS₂  score of 2 or more. The mean CHADS₂ score was 3.48 for the rivaroxaban group and 3.46 for the warfarin group, and 99.8% of the trial population had a baseline CHADS₂ score of 2 or more. In the warfarin group, the mean Time in Therapeutic Range (TTR), for the INR range of 2.0 to 3.0 was 55% (58% median). Some variability was observed in TTR by region: North America had the highest overall INR control followed by Western Europe, Latin America, Asia Pacific, and Eastern Europe.

3.4 Three analyses were defined in the manufacturer’s submission for the efficacy analysis: the intention-to-treat (ITT) set (all patients randomised), the safety-on-treatment set (all ITT patients who had taken at least one dose of study drug and were followed for events) and the per-protocol set (all ITT patients excluding those who had major pre-defined protocol deviations). The primary non-inferiority analysis of the ROCKET-AF trial was conducted on the per-protocol and the safety-on-treatment population data sets. The superiority analyses were conducted on the safety-on-treatment population data sets. In addition to these analyses, sensitivity analyses were performed to assess non-inferiority and superiority in the ITT population. The primary safety analysis was conducted on the safety-on-treatment population data.

3.5 A number of pre-planned subgroup analyses were conducted. These were by region, prior use of vitamin K antagonists (such as warfarin), and history of stroke, transient ischaemic attack, and non-central nervous system systemic embolism. Other subgroups included prior chronic aspirin use, gender, age, family origin, renal function, body mass index, weight, congestive heart failure, hypertension, diabetes, type of atrial fibrillation, proton pump inhibitor use at baseline, and prior myocardial infarction. Results were summarised by subgroup based on data from the safety-on-treatment and ITT populations.

3.6 The non-inferiority of rivaroxaban compared with warfarin was demonstrated for the primary outcome (composite of stroke and non-central nervous system systemic embolism) in both the per-protocol and safety-on-treatment populations. The results for the per-protocol population were HR 0.79 (95% confidence interval [CI] 0.66 to 0.96) and HR 0.79 (95% CI 0.65 to 0.95) for the safety-on treatment population. Superiority of rivaroxaban over warfarin was also demonstrated in the safety-on-treatment population but was not demonstrated for this outcome in the sensitivity analysis using the ITT population data set (HR 0.88, 95% CI 0.75 to 1.03).

3.7 For the primary safety end point of major or non-major clinically relevant bleeding, the results from the safety-on-treatment population data for ROCKET-AF suggest a comparable safety profile for rivaroxaban and warfarin, with no statistically significant difference between the two treatments (HR 1.03, 95% CI 0.96 to 1.11). Bleeding sites for the primary safety end point differed between treatment groups. Rivaroxaban was more often associated with bleeding at sites throughout the gastrointestinal tract (3.15% versus 2.16% p < 0.001) but intracranial haemorrhage rates were significantly lower with rivaroxaban compared with warfarin (0.5 versus 0.7%, p = 0.02). Following a request from the ERG the manufacturer provided subgroup analyses for the safety-on-treatment and ITT populations in people who had previously used vitamin K antagonists, people who had not previously used vitamin K antagonists, people with a time in therapeutic range below 60%, and those with a time in therapeutic range above 60%. In the safety-on-treatment population superiority of rivaroxaban compared with warfarin was demonstrated for the primary outcome (composite of stroke and non-central nervous system systemic embolism) in people who had not previously used vitamin K antagonists (HR 0.72, 95% CI 0.53 to 0.97) but not in people who had previously used vitamin K antagonists (HR 0.84, 95% CI 0.66 to 1.08).

3.8 The manufacturer undertook a Bayesian network meta-analysis comparison that compared rivaroxaban with warfarin, aspirin, no treatment and dabigatran etexilate. The clinical evidence comparing rivaroxaban with warfarin was taken from the ROCKET-AF trial. Studies used for the other comparators were obtained from a systematic literature search. The manufacturer identified 18 studies in total for inclusion in the network meta-analysis, which included one study comparing rivaroxaban with warfarin, seven comparing aspirin with placebo or control, eight comparing warfarin with aspirin, one comparing a vitamin K antagonist with clopidogrel plus aspirin and one comparing dabigatran etexilate with aspirin (the RE-LY study). The manufacturer reported network meta-analysis results for the outcomes using the ROCKET-AF safety-on-treatment population data set. At the request of the ERG, the manufacturer also provided the results for the outcomes using the ROCKET-AF ITT population dataset. The efficacy estimates from this network meta-analysis were used in the manufacturer’s cost-effectiveness analyses. 

3.9 The ERG undertook an exploratory network meta-analysis comparing rivaroxaban with dabigatran etexilate, aspirin, placebo, and adjusted standard dose warfarin. The ERG included data from 8 of the 18 studies used by the manufacturer in its network meta-analysis. The eight studies included one study comparing dabigatran etexilate with warfarin, one study comparing rivaroxaban with warfarin, three studies comparing aspirin with warfarin and three studies comparing warfarin with placebo. The ERG judged that including only these eight trials would reduce the amount of heterogeneity in the network. Only comparable dosing strategies were included (that is, rivaroxaban 20 mg per day, dabigatran etexilate 150 mg twice a day, aspirin 300 mg per day, and adjusted dose warfarin aiming at a target INR range between 2 and 3). A fixed-effect model was used because of the high degree of homogeneity between the included trials. The efficacy estimates from this network meta-analysis were used in the ERG’s cost-effectiveness analyses. 

3.10 The manufacturer developed a Markov model that compares rivaroxaban (20 mg once a day) with warfarin (adjusted dose warfarin at 4.5 mg once a day, target INR 2.5, range 2.0 to 3.0), aspirin (150 mg once a day) dabigatran etexilate (110–150 mg twice a day) and no treatment. The population in the model is the same as the ROCKET-AF safety-on-treatment population. The model has a lifetime time horizon and a UK NHS perspective.

3.11 The model included the following health states: anticoagulant initiation, stable atrial fibrillation (on or off therapy), minor stroke (on or off therapy), major stroke (on or off therapy), post minor stroke (on therapy), post major stroke (on therapy), minor bleed (on or off therapy), major bleed (on or off therapy), intracranial bleed (on or off therapy), post intracranial bleed (on or off therapy), systemic embolism (on or off therapy), myocardial infarction (on or off therapy), post myocardial infarction (on or off therapy) and death. The ROCKET-AF trial results for the safety-on-treatment population were used to inform the efficacy estimates for rivaroxaban compared with warfarin, rivaroxaban compared with warfarin in people whose atrial fibrillation is poorly controlled on warfarin, and the vitamin K antagonist naive model populations. The characteristics of the population for the analyses of rivaroxaban compared with aspirin, dabigatran etexilate and no treatment were based on the patient characteristics of a UK GP practice-based survey (Gallagher et al. 2008). Efficacy estimates were obtained from the manufacturer’s network meta-analysis.

3.12 The manufacturer classified all model events as either transient or permanent depending on associated long-term costs and consequences. Systemic embolism, minor extracranial bleeds and major extracranial bleeds were assumed to have no lasting clinical or economic consequences and as such were considered transient events in the model. Minor stroke, major stroke, intracranial bleeding and myocardial infarction were considered by the manufacturer to be permanent events, in the sense that they have lasting clinical and economic consequences. Consequently, the manufacturer developed post-event health states to account for the different risks, costs and utilities associated with surviving a permanent event.

3.13 The manufacturer highlighted that increasing age was an important risk factor for ischaemic stroke and systemic embolism, and adjusted the baseline risk of these events to account for patients aging as they move through the model. Risks were calculated using the Framingham risk equations. In the model, a weighted average relative risk (weighted by the proportion of patients in each risk group at initiation) is calculated for each age group and applied to the baseline risk as patients enter that age group. The risks of extracranial bleeding, intracranial bleeding and myocardial infarction were assumed to be independent of time and, therefore, were not adjusted for.

3.14 The baseline risk of each event was adjusted according to the treatment regimen the patient was receiving. Patients may stop their primary therapy and switch to a pre-specified secondary therapy at any time although the risk adjustment applied for the remainder of that cycle is that of the primary therapy. The probabilities of treatment discontinuation for warfarin and rivaroxaban were based on data obtained from the ROCKET-AF trial. The manufacturer assumed that treatment-discontinuation rates for aspirin, dabigatran etexilate and placebo were equivalent to that of rivaroxaban given the similarity of administration between these interventions.

3.15 The health-state utility values and treatment-related utility values in atrial fibrillation were obtained from published sources identified by systematic literature searching. The ROCKET-AF trial did not include a generic measure of health-related quality of life (such as the EQ-5D) that could be used to estimate utilities in the model, The estimates of resources and costs were obtained from NHS reference costs for 2009/10 and systematic literature searching. The manufacturer’s model categorised monitoring costs into the following distinct phases: initiation, maintenance, and re-initiation. The manufacturer’s model calculated the quarterly cost of initiation, maintenance and re-initiation by taking a weighted average of each cost; the cost of the individual phases were weighted by the proportion of people treated in primary and secondary care, as indicated by the manufacturer’s survey. The resulting annual cost estimates for warfarin monitoring were £663 (£448 in primary care and £215 in secondary care) for the first year and £524 (£359 in primary care and £165 in secondary care) for subsequent years.   

3.16 The manufacturer’s base-case analysis of rivaroxaban versus warfarin used only statistically significant data from the ROCKET-AF safety-on-treatment population. The resulting incremental cost-effectiveness ratio (ICER) was £18,883 per quality-adjusted life year (QALY) gained. The manufacturer also presented the results of four subgroup analyses:

• For rivaroxaban versus warfarin in people whose INR is poorly controlled on warfarin, rivaroxaban dominated (was more effective and cost less than) warfarin.
• For rivaroxaban versus warfarin in people naive to warfarin, the ICER was £15,494 per QALY gained.
• For rivaroxaban versus aspirin, the ICER was £2,083 per QALY gained.
• For rivaroxaban versus dabigatran etexilate, rivaroxaban dominated dabigatran etexilate.  

The manufacturer carried out univariate sensitivity analysis on the base case, scenario analyses and subgroup analyses. The main drivers of the model results were consistent across analyses, with the cost of warfarin monitoring in primary care having a major impact on all ROCKET-AF-based analyses. The probabilistic sensitivity analyses indicated that, using the base case, rivaroxaban had a 75% probability of being cost effective at a threshold of £20,000 per QALY gained and an 88% probability at £30,000 per QALY gained. The manufacturer’s scenario analysis of rivaroxaban versus warfarin used all point estimates from the ROCKET-AF safety-on-treatment population regardless of their statistical significance. The resulting incremental cost-effectiveness ratio (ICER) was £8,732 per quality-adjusted life year (QALY) gained.

3.17 The ERG considered a Markov model to be an appropriate choice for modelling the chronic condition of atrial fibrillation. The ERG noted that the manufacturer chose a cycle length of 3 months and that only one event per 3-month cycle was possible because of the nature of the model. The manufacturer acknowledged that, in reality, people may experience more than one event in 3 months, but clinical opinion was that the probability of experiencing more than one event in 3 months would be low. The ERG agreed that assuming one event per model cycle was necessary and reasonable. However, the ERG noted that the manufacturer’s model also suspends the risk of further events in the subsequent model cycle. The ERG considered that this additional suspension of risk was likely to bias the analysis against the more effective treatment because the overall event rate would be lower, and the potential to demonstrate clinical and economic benefits would also be lower.

3.18 The ERG identified the following limitations to the manufacturer’s model’s structural assumptions and parameter sources: not separating out the number of hospital visits needed by people who were within and outside recommended INR control; not adjusting risk of bleeding by age; not adjusting utility by age; the source of myocardial infarction risk for people treated with aspirin; the out of date source of post-myocardial infarction mortality risk; the double counting of re-initiation costs of warfarin monitoring; suspending the risk of further events for the subsequent model cycle after an event; and excluding transient ischaemic attack as a potential event.

3.19 The ERG presented an alternative base case in which, where possible, adjustments were made to account for the limitations identified (see section 3.19). The alternative base-case for rivaroxaban compared with warfarin ICER was £33,758 per QALY gained. Similarly, for warfarin-naive people, after incorporating the ERG’s model adjustments the ICER for rivaroxaban compared with warfarin increased from £15,494 to £29,894 per QALY gained. However, rivaroxaban remained dominant in people whose INR was poorly controlled on warfarin after the ERG’s model adjustments were incorporated. The structure of the manufacturer’s model meant it wasn’t possible to remove risk suspension or add transient ischaemic attack as a potential event. Consequently, the ERG was unable to fully quantify the impact of these limitations on the ICERs. However, the ERG considered that suspending risk and excluding transient ischaemic attack as an event would favour warfarin (that is, the removal of these limitations would decrease the ICER for rivaroxaban compared with warfarin), because warfarin is generally less effective than rivaroxaban (based on the safety-on-treatment population of ROCKET-AF).

3.20 The ERG considered that the manufacturer’s base-case model is driven by the cost of anticoagulation monitoring rather than the differential effectiveness of rivaroxaban and warfarin. The ROCKET-AF trial showed that, for most outcomes, there was no statistically significant difference between rivaroxaban and warfarin. The ERG highlighted that when the cost of anticoagulation monitoring was separated out by INR range the ICER substantially increased from £18,883 per QALY gained to £27,281 per QALY gained. In addition to this, the ERG’s scenario analysis using alternative anticoagulation monitoring costs (discussed by the Committee in the ongoing appraisal of dabigatran etexilate for the prevention of stroke and systemic embolism in atrial fibrillation) increased the ICER to £62,568 per QALY gained.

3.21 The ERG was concerned that the network meta-analysis presented by the manufacturer to compare rivaroxaban with aspirin and dabigatran had high levels of heterogeneity, which were not shown when the ERG conducted its own network meta-analysis restricting the network to the comparators specified in the final scope issued by NICE. When the ERG applied the treatment effects estimated by the ERG’s network meta-analysis to the manufacturer’s model, an ICER of £34,680 per QALY gained was obtained for dabigatran etexilate versus rivaroxaban whereas rivaroxaban had dominated dabigatran etexilate in the manufacturer’s analysis. The ERG applied further adjustments to account for the following limitations: the absence of a post-systemic embolism health state, not adjusting bleeding risk by age, not adjusting utility by age, the out of date source of post-myocardial infarction mortality risk, and assuming equivalent discontinuation rates. This reduced the ICER to £12,701. Exploratory analysis assuming an equivalent ability of rivaroxaban and dabigatran etexilate to prevent myocardial infarction, further decreased the ICER to £3,578 for rivaroxaban compared with dabigatran etexilate.

3.22 The ERG noted the presence of potential biases in the model, with limitations of risk suspension and the absence of transient ischaemic attack and dyspepsia as adverse events. Removing risk suspension is likely to favour dabigatran etexilate (that is, reduce its ICER), whereas including transient ischaemic attack and dyspepsia is likely to increase the ICER for rivaroxaban compared with dabigatran etexilate. Furthermore, the ERG noted that there is a large amount of uncertainty in the model and that the model is highly sensitive to even small changes to the discontinuation rates. Therefore, the ERG considered that the results of the probabilistic sensitivity analysis should be taken into account when considering the ERG’s alternative ICER for dabigatran etexilate versus rivaroxaban. The probabilistic sensitivity analysis indicated that dabigatran etexilate was dominant in 45% of the 1000 runs and dominated in 35% of runs.

3.23 Full details of all the evidence are in the manufacturer’s submission and the ERG report, which are available from www.nice.org.uk/guidance/TAXXX

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 stroke and systemic embolism and the value placed on the benefits of rivaroxaban by people with atrial fibrillation, those who represent them, and clinical specialists. It also took into account the effective use of NHS resources.

4.2 The Committee heard from the clinical specialists and patient experts that the current standard treatment for preventing stroke and systemic embolism in people with atrial fibrillation is warfarin, and that aspirin is used only in people for whom warfarin is unsuitable, because of its lesser efficacy. The Committee also heard that warfarin, although an effective treatment, it is associated with a number of problems. The main concerns for people with atrial fibrillation were fear of having a stroke and anxiety about the difficulty of keeping the INR within the satisfactory therapeutic range. The Committee also heard from the patient expert that taking warfarin adversely affects quality of life for these reasons and also because it restricts food and drink choices. In addition, the need for regular monitoring and dose adjustments is disruptive and inconvenient, and the need for regular GP and hospital visits can have an adverse impact on people’s work, social and family life. The Committee also heard from the clinical specialists that a substantial proportion of people taking warfarin have poorly controlled INR and are often not within the target therapeutic range at any one time. In particular, older people with atrial fibrillation are more likely to have poorly controlled INR because the interaction between warfarin and other medications taken for existing comorbidities affects INR. The clinical specialists explained that people with an INR outside the therapeutic range have an increased risk of stroke or bleeding. The clinical specialists explained that controlling anticoagulation in people on warfarin who need medical procedures or surgery was sometimes difficult and could result in additional days in hospital or procedures being delayed. The clinical specialists also explained that the need for regular monitoring and dose adjustments, occasionally involving complicated regimens such as different doses on alternate days, can cause difficulties with adherence to treatment. The Committee recognised the potential benefits of alternatives such as rivaroxaban for people with atrial fibrillation, including the positive effect on quality of life of removing the restrictions and difficulties associated with taking warfarin.

4.3 The Committee considered the clinical-effectiveness data from the ROCKET-AF trial comparing rivaroxaban with warfarin. It noted that this study was the basis of the clinical-effectiveness evidence in the manufacturer’s submission. The Committee noted that the efficacy analysis in the manufacturer’s submission had been undertaken on three different populations in the ROCKET-AF trial, the ITT set (all randomised patients), the safety-on-treatment set (all ITT patients who had taken at least one dose of study drug and were followed for events) and the per-protocol set (all ITT patients excluding those who have major pre-defined protocol deviations). The Committee noted that the manufacturer had presented data from the safety-on-treatment population for its primary analyses. The Committee heard from the clinical specialists that using a trial ITT population was considered to be the gold standard for estimating clinical effectiveness in a superiority trial but ROCKET-AF was a non-inferiority trial so the primary analysis was different. The clinical specialists were satisfied that the safety-on-treatment population would be representative of people who would be treated with rivaroxaban in clinical practice. The Committee considered that the ITT population included people who had either had no treatment or switched treatment during the trial, and agreed that the estimates derived from the safety-on-treatment population of the ROCKET-AF trial provided an adequate basis for evaluating clinical effectiveness.

4.4 The Committee noted that a key uncertainty highlighted by the ERG was the generalisability of the results of ROCKET-AF to people diagnosed with atrial fibrillation in the NHS. The Committee noted that the mean time in therapeutic range for the INR range of 2.0–3.0 for warfarin was 55% for the safety-on-treatment population in the ROCKET-AF trial. The clinical specialists confirmed this could be considered to be around the lower end of the level of control that would be expected in UK clinical practice, but there is considerable variation between different centres and also between different settings, depending on the patient group. The Committee noted that the ROCKET-AF trial had been undertaken in a number of countries, which did not all achieve similar levels of time in therapeutic range. The majority (66.5%) of the participants had been recruited from centres in Eastern Europe, Latin America and Asia and that in these centres, the proportion of time in therapeutic range was lower than in the centres in North America and western Europe. The Committee was concerned that the effectiveness of warfarin could be underestimated if the proportion of time in therapeutic range was low, and that the UK context might be better reflected by results from centres where the time in therapeutic range in the warfarin arm more closely matched the usual levels in the UK.

4.5 The Committee also noted that in its subgroup analyses of people whose atrial fibrillation was well controlled and poorly controlled by warfarin, the manufacturer had assumed equal clinical benefit of rivaroxaban, independent of warfarin control. The Committee agreed that the manufacturer should provide analyses that took account of the impact of the low proportion of time in therapeutic range on warfarin in the ROCKET-AF trial. It agreed this should be done by doing subgroup analyses by country or centre, and that these analyses should be used to inform estimates of cost effectiveness.

4.6 The Committee also noted that the ROCKET-AF population had a mean CHADS₂ score of 3.47, and that 99.8% of the trial population had a baseline CHADS₂ score of 2 or more, but the scope had specified that the appraisal population would be in people with medium to high risk of stroke. The clinical specialists confirmed that people with a CHADS₂ score of 3 or more would be at high risk of stroke and that this population was typical of people seen in secondary care. However this did not necessarily represent people with atrial fibrillation treated in primary care, who tended to have a lower risk of stroke. The Committee heard that people with atrial fibrillation treated with warfarin in primary care often have a CHADS₂ score of less than 2 and that it is estimated that between 20 and 75% of people with a CHADS₂ score of less than 2 are prescribed warfarin in the UK. The clinical specialists agreed that it was likely that although people with a CHADS₂ score of 2 or more would benefit similarly to those in the ROCKET-AF trial, this cannot be assumed for people with a CHADS₂ score of less than 2. The Committee heard from the manufacturer that rivaroxaban would be indicated for the treatment of atrial fibrillation in people with one or more risk factors for stroke, which equates to a CHADS₂ score of 1 or more. The Committee therefore agreed that the population included in the ROCKET-AF trial did not reflect all the people with atrial fibrillation in the UK eligible for treatment and therefore there was uncertainty about whether the results were generalisable to UK clinical practice.

4.7 The Committee then discussed the indirect clinical-effectiveness evidence for rivaroxaban compared with dabigatran etexilate and aspirin. The Committee noted that the population in the study comparing dabigatran etexilate with warfarin (RE-LY) had a lower risk of stroke (mean CHADS₂ score 2.1) than the population in the ROCKET–AF study (mean CHADS₂ score of 3.47). The Committee noted that the manufacturer’s interpretation of its network meta-analysis was that there was no significant difference between rivaroxaban and dabigatran etexilate for any outcome. The Committee noted the ERG’s concerns about the validity of the manufacturer’s network meta-analysis because of the clinical heterogeneity of the included trials, and the different levels of time in therapeutic range in the warfarin arms of the rivaroxaban and dabigatran etexilate trials. The Committee also noted that both the manufacturer’s and ERG’s network meta-analyses contained wide confidence intervals and therefore the resulting efficacy point estimates were subject to considerable uncertainty. The Committee therefore agreed that the clinical-effectiveness estimates for rivaroxaban compared with dabigatran etexilate obtained from the network meta-analyses were unreliable. The Committee concluded that it would not consider further the clinical effectiveness of rivaroxaban compared with aspirin or dabigatran etexilate. 

4.8 The Committee discussed the safety data from the ROCKET-AF trial. It noted that analysis of the primary safety end point of all major and non-major clinically significant bleeding events showed no significant differences between rivaroxaban and warfarin. There was a significant reduction in the rate of fatal bleeds and intracranial haemorrhage compared with warfarin, but a higher rate of gastrointestinal bleeds. The Committee heard from the patient experts that intracranial bleeds were considered to be a more serious complication than gastrointestinal bleeds in clinical practice, because they were more difficult to treat and often result in permanent disability. The Committee noted the possible uncertainty in these results related to the relatively low proportion of time in therapeutic range of 55% in the warfarin arm of the trial, but concluded that the primary safety end point had shown no statistically significant difference between rivaroxaban and warfarin.

4.9 The Committee considered the cost effectiveness of rivaroxaban compared with warfarin. The Committee noted that both the manufacturer and ERG had identified the costs associated with warfarin INR monitoring as a major factor affecting the cost-effectiveness estimate in the model. The Committee noted that the manufacturer had assumed an average annual anticoagulant monitoring cost of £535 per person. The clinical specialists agreed that the annual cost of anticoagulant monitoring for each person treated with warfarin was likely to be lower than £535 in clinical practice, but a precise estimate could not be given because costs varied considerably between people (for example being higher in those with poor INR control), and between centres. The Committee therefore concluded that, in view of this uncertainty and to make the appraisals consistent with each other, the estimate of annual anticoagulant monitoring that the Committee had previously accepted in the ongoing appraisal of dabigatran etexilate for the prevention of stroke in atrial fibrillation should be used. It agreed that an annual anticoagulant monitoring cost of £242 per person should be included in the model. The Committee therefore did not consider the estimates of cost effectiveness of rivaroxaban compared with warfarin from the current model to be appropriate.

4.10 The Committee noted the concerns about generalisability of the trial to the UK population in the light of the higher CHADS ₂ score in the ROCKET-AF trial compared with the UK population with atrial fibrillation who would be eligible for treatment with rivaroxaban. It agreed that in order to maximise consistency between appraisals, ideally the economic model should have derived baseline risk estimates from the population model cohort used in the ongoing appraisal of dabigatran etexilate for the prevention of stroke in atrial fibrillation, which was based on the General Practice Research Database. The Committee had accepted that the General Practice Research Database has advantages over the UK GP practice-based survey (Gallagher et al. 2008) because the data are more recent and therefore better reflect the current UK atrial fibrillation population. The Committee agreed that if it was not possible for the manufacturer to amend its model in this way, it could derive baseline risk estimates from the UK GP practice-based survey.  

4.11 The Committee discussed which data should be used to inform the cost effectiveness of rivaroxaban compared with warfarin. It noted the base-case ICER of £18,900, was calculated using the direct comparison and only the statistically significant differences demonstrated in the ROCKET-AF trial. It also noted that the ICER using all point estimates from the trial in the safety-on-treatment population was significantly lower at £8,730. The Committee considered that it was reasonable to use point estimate data because this would capture all benefits. Nevertheless, if the much lower warfarin monitoring costs favoured by the Committee were incorporated, this figure would be considerably higher. The Committee requested that the manufacturer provide further figures using the Committee’s preferred estimate of £242 per person.

4.12 The Committee summarised the range of concerns related to the evaluation of the clinical and cost effectiveness of rivaroxaban compared with warfarin using the direct comparison. It accepted the use of the safety-on-treatment trial population and the use of point estimates of efficacy. However, the Committee questioned whether the efficacy shown in the trial would apply to the whole UK population eligible for rivaroxaban treatment. The first source of potential variance was the baseline risk of the population, that is the lower mean CHADS₂ scores in the UK population, which would be expected to have a lower baseline risk compared with the mean CHADS ₂ score of 3.5 in the ROCKET-AF trial. The second area of concern was the relatively low proportion of time in therapeutic range in the warfarin arm of the ROCKET-AF trial compared with expected levels in the UK. In addition to these issues related to the generalisability of the trial results, the Committee considered that the warfarin monitoring costs in the economic model were too high. In conclusion, the Committee was minded not to recommend rivaroxaban, and requested that the manufacturer provide further information.

4.13 The Committee requested that, for the second Appraisal Committee meeting, the manufacturer of rivaroxaban provide further information and revised cost-effectiveness analyses comparing rivaroxaban with warfarin as follows:

• The characteristics of the cohort in the model should represent people with atrial fibrillation in the UK. Therefore ideally the baseline risks of events in the patient cohort in the model should be derived from the General Practice Research Database or alternatively or the UK GP practice-based survey (Gallagher et al. 2008).
• The analyses should use clinical-effectiveness data from the safety-on-treatment population of the ROCKET-AF trial, and use all point estimates from this trial regardless of statistical significance.
• The effect of the low proportion of time in therapeutic range on warfarin in the ROCKET-AF trial should be accounted for by considering subgroup analyses by country or centre.
• The analyses should incorporate a fixed annual warfarin INR monitoring cost of £242.

Summary of Appraisal Committee's key conclusions

TAXXX	Appraisal title: Rivaroxaban for the prevention of stroke and systemic embolism in people with atrial fibrillation		Section
Key conclusion
The Committee is minded not to recommend the rivaroxaban to prevent stroke and systemic embolism in people with atrial fibrillation. The Committee questioned whether the efficacy shown in the ROCKET-AF trial would apply to the whole UK population eligible for rivaroxaban treatment and it considered that the warfarin monitoring costs in the economic model were too high.			1.1 4.8 and 4.9
Current practice
Clinical need of patients, including the availability of alternative treatments		The current standard treatment for the prevention of stroke and systemic embolism in people with atrial fibrillation is warfarin, and that aspirin is used only in people for whom warfarin is unsuitable, because of its lower efficacy. Warfarin is associated with a number of problems such as fear of having a stroke and anxiety about the difficulty of keeping the international normalised ratio (INR) within the satisfactory therapeutic range. Taking warfarin adversely affects quality of life because it restricts food and drink choices. In addition, the need for regular monitoring and dose adjustments causes disruption and inconvenience, and the need for regular GP and hospital visits can have an adverse impact on people’s work, social and family life. A substantial proportion of people taking warfarin have poorly controlled INR and are often not within the target therapeutic range at any one time.	4.2
The technology
Proposed benefits of the technology How innovative is the technology in its potential to make a significant and substantial impact on health-related benefits?		The Committee recognised the potential benefits of alternatives such as rivaroxaban for people with atrial fibrillation, including the positive effect on quality of life of removing the restrictions and difficulties associated with taking warfarin.	4.2
What is the position of the treatment in the pathway of care for the condition?		Rivaroxaban would be used for the prevention of stroke and systemic embolism in people with atrial fibrillation.	4.2
Adverse effects		The Committee noted the possible uncertainty in the results from the ROCKET-AF trial related to the relatively low proportion of time in therapeutic range of 55% in the warfarin arm of the trial, but concluded that the primary safety end point had shown no statistically significant difference between rivaroxaban and warfarin.	4.7
Evidence for clinical effectiveness
Availability, nature and quality of evidence		The main clinical effectiveness evidence came from one multicentre, double-blind randomised controlled trial. The ROCKET-AF trial compared rivaroxaban with dose-adjusted warfarin. The manufacturer also compared rivaroxaban with aspirin and dabigatran etexilate (110 mg or 150 mg twice a day) using a network meta-analysis in people for whom anticoagulation therapy is considered suitable. A number of pre-planned subgroup analyses were conducted, these were by region, prior use of vitamin K antagonists (such as warfarin), and history of prior stroke, transient ischaemic attack, and non-central nervous system systemic embolism. Other subgroups included prior chronic aspirin use, gender, age, family origin, renal function, body mass index, weight, congestive heart failure, hypertension, diabetes, type of atrial fibrillation, proton pump inhibitor use at baseline, and prior myocardial infarction. Following a request from the ERG the manufacturer provided subgroup analyses for the safety-on-treatment and ITT populations in people who had previously used vitamin K antagonists, people who had not previously used vitamin K antagonists, people with a time in therapeutic range below 60%, and those with a time in therapeutic range above 60%.	3.1 3.5 3.7
Relevance to general clinical practice in the NHS		The Committee agreed that the population included in the ROCKET-AF trial did not reflect all the people with atrial fibrillation in the UK eligible for treatment	4.5
Uncertainties generated by the evidence		The Committee agreed that the population included in the ROCKET-AF trial did not reflect all the people with atrial fibrillation in the UK eligible for treatment and therefore there was uncertainty about whether the results were generalisable to UK clinical practice. The Committee agreed that the clinical-effectiveness estimates for rivaroxaban compared with dabigatran etexilate obtained from the network meta-analyses were unreliable.	4.5 4.6
Are there any clinically relevant subgroups for which there is evidence of differential effectiveness?		Superiority of rivaroxaban compared with warfarin was demonstrated for the primary outcome (composite of stroke and non-central nervous system systemic embolism) in people who had not previously used vitamin K antagonists (HR 0.72, 95% CI 0.53 to 0.97) but not in people who had previously used vitamin K antagonists (HR 0.84, 95% CI 0.66 to 1.08).	3.7
Estimate of the size of the clinical effectiveness including strength of supporting evidence		The Committee agreed that the manufacturer should provide analyses that took of account the impact of the low proportion of time in therapeutic range on warfarin in the ROCKET-AF trial. It agreed this should be done by doing subgroup analyses by country or centre.	4.4
Evidence for cost effectiveness
Availability and nature of evidence		The manufacturer developed a Markov model that compares rivaroxaban (20 mg once a day) with warfarin (adjusted dose warfarin at 4.5 mg once a day, target INR 2.5, range 2.0 to 3.0), aspirin (150 mg once a day) dabigatran etexilate (110–150 mg twice a day) and no treatment.	3.10–3.17
Uncertainties around and plausibility of assumptions and inputs in the economic model		The Committee noted that both the manufacturer and ERG had identified the costs associated with warfarin INR monitoring as a major factor affecting the cost-effectiveness estimate in the model. The Committee noted that the manufacturer had assumed an average annual anticoagulant monitoring cost of £535 per person. The Committee agreed that an annual anticoagulant monitoring cost of £242 per person was the most appropriate estimate to be included in the model. The Committee therefore did not consider the estimates of cost effectiveness of rivaroxaban compared with warfarin from the current model to be appropriate.	4.8
Incorporation of health-related quality-of-life benefits and utility values Have any potential significant and substantial health-related benefits been identified that were not included in the economic model, and how have they been considered?		No health-related benefits were identified which were not used in the economic model.
Are there specific groups of people for whom the technology is particularly cost effective?		N/A
What are the key drivers of cost effectiveness?		The ERG highlighted that when the cost of anticoagulation monitoring was separated out  by INR range the ICER substantially increased from £18,883 per QALY gained to £27,281 per QALY gained. In addition to this, the ERG’s scenario analysis using alternative anticoagulation monitoring costs (discussed by the Committee in the ongoing appraisal of dabigatran etexilate for the prevention of stroke and systemic embolism in atrial fibrillation) increased the ICER to £62,568 per QALY gained.	3.21
Most likely cost-effectiveness estimate (given as an ICER)		The Committee did not consider the estimates of cost effectiveness of rivaroxaban compared with warfarin from the current model to be appropriate.	4.8
Additional factors taken into account
Patient access schemes (PPRS)		N/A
End-of-life considerations		N/A
Equalities considerations and social value judgements		No equalities issues were identified.

5 Implementation

5.1 The Secretary of State and the Welsh Assembly Minister for Health and Social Services have issued directions to the NHS in England and Wales on implementing NICE technology appraisal guidance. When a NICE technology appraisal recommends use of a drug or treatment, or other technology, the NHS must usually provide funding and resources for it within 3 months of the guidance being published. If the Department of Health issues a variation to the 3-month funding direction, details will be available on the NICE website. When there is no NICE technology appraisal guidance on a drug, treatment or other technology, decisions on funding should be made locally.

5.2 NICE has developed tools to help organisations put this guidance into practice (listed below). These are available on our website (www.nice.org.uk/guidance/TAXXX). [NICE to amend list as needed at time of publication]

• Slides highlighting key messages for local discussion.
• Costing template and report to estimate the national and local savings and costs associated with implementation.
• Implementation advice on how to put the guidance into practice and national initiatives that support this locally.
• A costing statement explaining the resource impact of this guidance.
• Audit support for monitoring local practice.

6 Related NICE guidance

Published

• The management of atrial fibrillation. NICE clinical guideline 36 (2006).  Available from www.nice.org.uk/guidance/CG36
• Dronedarone for the treatment of non-permanent atrial fibrillation. NICE technology appraisal guidance 197 (2010). Available from www.nice.org.uk/guidance/TA197  
• Thoracoscopic exclusion of the left atrial (with or without surgical ablation) for non-valvular atrial fibrillation for the prevention of thromboembolism. NICE interventional procedure guidance 400 (2011). Available from www.nice.org.uk/guidance/IPG400
• Percutaneous occlusion of the left atrial appendage in non-valvular atrial fibrillation for the prevention of thromboembolism. NICE interventional procedure guidance 349 (2010). Available from www.nice.org.uk/guidance/IPG349

Under development

NICE is developing the following guidance (details available from www.nice.org.uk):

• Dabigatran etexilate for the prevention of stroke and systemic embolism in atrial fibrillation. Expected date of publication December 2011.

7 Proposed date for review of guidance

7.1 NICE proposes that the guidance on this technology is considered for review by the Guidance Executive in October 2014. NICE welcomes comment on this proposed date. The Guidance Executive will decide whether the technology should be reviewed based on information gathered by NICE, and in consultation with consultees and commentators.

Jane Adam
Chair, Appraisal Committee
December 2011

Appendix A: Appraisal Committee members, and NICE project team

A Appraisal Committee members

The Appraisal Committees are standing advisory committees of NICE. Members are appointed for a 3-year term. A list of the Committee members who took part in the discussions for this appraisal appears below. There are four Appraisal Committees, each with a chair and vice chair. Each Appraisal Committee meets once a month, except in December when there are no meetings. Each Committee considers its own list of technologies, and ongoing topics are not moved between Committees.

Committee members are asked to declare any interests in the technology to be appraised. If it is considered there is a conflict of interest, the member is excluded from participating further in that appraisal.

The minutes of each Appraisal Committee meeting, which include the names of the members who attended and their declarations of interests, are posted on the NICE website.

Dr Jane Adam (Chair)
Department of Diagnostic Radiology, St George’s Hospital

Professor A E Ades
Professor of Public Health Science, Department of Community Based Medicine, University of Bristol

Dr Jeremy Braybrooke
Consultant Medical Oncologist, University Hospitals Bristol NHS Foundation Trust

Dr Fiona Duncan
Clinical Nurse Specialist, Anaesthetic Department, Blackpool Victoria Hospital, Blackpool

Professor Jonathan Grigg
Professor of Paediatric Respiratory and Environmental Medicine, Barts and the London School of Medicine and Dentistry, Queen Mary University London

Dr Peter Heywood
Consultant Neurologist, Frenchay Hospital

Dr Sharon Saint Lamont
Head of Quality and Innovation, North East Strategic Health Authority

Dr Ian Lewin
Consultant Endocrinologist, North Devon District Hospital

Dr Louise Longworth
Reader in Health Economics, HERG, Brunel University

Dr Anne McCune
Consultant Hepatologist, University Hospitals Bristol NHS Foundation Trust

Dr Alec Miners
Lecturer in Health Economics, London School of Hygiene and Tropical Medicine

Ms Pamela Rees
Lay Member

Dr Ann Richardson
Lay Member

Dr Paul Robinson
Medical Director, Merck Sharp & Dohme

Mr Mike Spencer
Assistant Director Patient Experience, Cardiff and Vale University Health Board

Mr David Thomson
Lay Member

Dr John Watkins
Clinical Senior Lecturer/Consultant in Public Health Medicine, Cardiff University and National Public Health Service Wales  

Dr Anthony S Wierzbicki
Consultant in Metabolic Medicine/Chemical Pathology, Guy’s and St Thomas’ Hospitals NHS Trust

B NICE project team

Each technology appraisal is assigned to a team consisting of one or more health technology analysts (who act as technical leads for the appraisal), a technical adviser and a project manager.

Helen Tucker
Technical Lead

Nicola Hay
Technical Adviser

Bijal Joshi
Project Manager

 Appendix B: Sources of evidence considered by the Committee

A The Evidence Review Group (ERG) report for this appraisal was prepared by BMJ Technology Assessment Group (BMJ-TAG):

• Edwards SJ, Hamilton V, Nherera L et al. Rivaroxaban for the prevention of stroke and systemic embolism in people with atrial fibrillation: A Single Technology Appraisal. BMJ-TAG, London (October 2011)

B The following organisations accepted the invitation to participate in this appraisal as consultees and commentators. They were invited to comment on the draft scope, the ERG report and the appraisal consultation document (ACD). Organisations listed in I were also invited to make written submissions. Organisations listed in II and III had the opportunity to give their expert views. Organisations listed in I, II and III also have the opportunity to appeal against the final appraisal determination.

I Manufacturer/sponsor:

• Bayer HealthCare

II Professional/specialist and patient/carer groups:

• Anti Coagulation Europe (ACE)
• Arrhythmia Alliance (AFA Affiliated)
• Atrial Fibrillation Association (AFA)
• British Association of Stroke Physicians
• British Cardiovascular Society
• British Heart Foundation
• British Society for Haematology
• Heart Rhythm UK
• Primary Care Cardiovascular Society
• Royal College of Nursing
• Royal College of Pathologists
• Royal College of Physicians

III Other consultees:

• Department of Health
• NHS Berkshire East
• Welsh Government

IV Commentator organisations (did not provide written evidence and without the right of appeal):

• BMJ Technology Assessment Group (BMJ-TAG)
• Boehringer Ingelheim
• Bristol Myers-Squibb
• Commissioning Support Appraisals Service
• Department of Health, Social Services and Public Safety for Northern Ireland
• Health Care Improvement Scotland
• Medicines and Healthcare products Regulatory Agency
• National Institute for Health Research Health Technology Assessment Programme

C The following individuals were selected from clinical specialist and patient expert nominations from the non-manufacturer/sponsor consultees and commentators. They gave their expert personal view on rivaroxaban by attending the initial Committee discussion and providing written evidence to the Committee. They are invited to comment on the ACD.

• Dr Rhona Maclean, Consultant Haematologist, nominated by Royal College of Pathologists – clinical specialist
• Ms Fiona Sayers, Head of Nursing, Cardiology & Acute Service, nominated by Royal College of Nursing – clinical specialist
• Professor John Potter, Professor of Ageing Stroke Medicine, nominated by Bayer HealthCare – clinical specialist
• Ms Diane Eaton, nominated by Anticoagulation Europe – patient expert
• Ms Joanne Jerrome, nominated by Atrial fibrillation Association – patient expert

D  Representatives from the following manufacturer/sponsor attended Committee meetings. They contributed only when asked by the Committee chair to clarify specific issues and comment on factual accuracy.

• Bayer HealthCare