Cost-effectiveness of rivaroxaban versus enoxaparin for the prevention of postsurgical venous thromboembolism in Canada

Transcription

1 760 Schattauer 2010 Blood Coagulation, Fibrinolysis and Cellular Haemostasis Cost-effectiveness of rivaroxaban versus enoxaparin for the prevention of postsurgical venous thromboembolism in Canada Alexander Diamantopoulos 1 ; Michael Lees 3 ; Philip S. Wells 4 ; Fiona Forster 2 ; Jaithri Ananthapavan 2 ; Heather McDonald 5 1 Symmetron Limited, London, UK; 2 IMS Health, London, UK; 3 Bayer HealthCare, Uxbridge, UK; 4 Department of Medicine, University of Ottawa, Ottawa Hospital Research Institute, Ottawa, Ontario, Canada; 5 Bayer Inc., Toronto, Ontario, Canada Summary This study aimed to evaluate the cost-effectiveness of prophylaxis with rivaroxaban vs. enoxaparin in the prevention of venous thromboembolism (VTE) after total hip replacement (THR) and total knee replacement (TKR) from the perspective of the Canadian healthcare system. A model was developed that included both acute VTE (represented as a decision tree) and long-term complications (represented as a Markov process with one-year cycles). Transition probabilities were derived from phase III clinical trials comparing rivaroxaban with enoxaparin and published literature. Costs were derived from the Ontario Case Costing Initiative and publicly available sources. Utilities were derived from published literature. The model reported VTE event rates, quality-adjusted life expectancy and direct medical costs over a five-year horizon. Costs are reported in 2007 Canadian Dollars (C$). When rivaroxaban and enoxaparin are compared in patients undergoing THR, rivaroxaban enoxaparin. That is, rivaroxaban is associated with improved health outcomes as measured by increased quality-adjusted life years (QALYs; ) and fewer symptomatic VTE events (0.0061), and also with lower cost (savings of C$300) per patient. Similarly, rivaroxaban enoxaparin in patients undergoing TKR, achieving a gain of QALYs, a reduction of symptomatic venous thromboembolic events and savings of C$129 per patient. Rivaroxaban is a costeffective alternative to enoxaparin for VTE prophylaxis in patients undergoing THR and TKR. Over a five-year horizon, rivaroxaban dominated enoxaparin in the prevention of VTE events in patients undergoing THR and TKR, providing more quality-of-life benefit at a lower cost. Keywords Anticoagulants, surgery, thrombosis Correspondence to: Heather McDonald Health Economics and Outcomes Research, Bayer HealthCare 77 Belfield Road, Toronto Ontarion, M9W 1G6, Canada Tel: Financial support: This study was supported by Bayer Schering Pharma AG. Received: January 28, 2010 Accepted after major revision: June 4, 2010 Prepublished online: August 30, 2010 doi: /th Thromb Haemost 2010; 104: Introduction Venous thromboembolism (VTE) comprising deep-vein thrombosis (DVT) and pulmonary embolism (PE) has a high prevalence in hospitals and in the community (1, 2) and is associated with considerable morbidity and mortality (3). Potential thromboembolic DVT usually originates in the large deep distal veins of the lower limbs and tends to be asymptomatic, unless fractured thrombi metastasise as emboli, or morbid sequelae develop (4). Thrombi which propagate into the proximal veins are more likely to embolise (5) and cause pulmonary obstruction, with risk of death. Long-term complications (LTC) that can arise from VTE include recurrent VTE (6), the post-thrombotic syndrome (PTS) (7) and chronic thromboembolic pulmonary hypertension (8). VTE is the third leading cause of cardiovascular death in the US after myocardial infarction and stroke (9). Major orthopaedic surgery such as total hip replacement (THR) and total knee replacement (TKR) is associated with an increased risk of VTE (10). Estimates indicate that VTE may occur in 40 60% of patients undergoing major orthopaedic surgery with prophylactic therapy (11). In Canada, the number of patients undergoing elective THR and TKR has increased by more than 100% over the last 10 years, with 28,045 THR and 40,701 TKR operations in 2005/6 (12). In response to the risks associated with VTE, the American College of Chest Physicians currently recommends that patients receive prophylaxis for a minimum of 10 days and up to 35 days after THR, and suggest prophylaxis for 35 days after TKR (11). In addition, in the UK the National Institute for Health and Clinical Excellence recommend extended prophylaxis for four weeks after surgery for THR patients with one or more risk factors for VTE (13). The most commonly used method of VTE prophylaxis in Canada is parenterally administered low-molecular-weight heparins (LMWHs) such as enoxaparin and dalteparin (12). Rivaroxaban is a novel, once-daily, orally administered thromboprophylactic agent. It is a direct factor Xa inhibitor that demonstrates activity against both clot-associated and free factor Xa, as well as inhibiting prothrombinase activity and reducing thrombin generation (14, 15). In addition to being orally administered, un- Thrombosis and Haemostasis 104.4/2010

2 Diamantopoulos et al. Cost-effectiveness of rivaroxaban 761 like warfarin rivaroxaban does not require any monitoring during administration. Phase III clinical trials have demonstrated that rivaroxaban has superior efficacy compared with enoxaparin in terms of reducing VTE events, and a similar safety profile, after THR and TKR (16 19). Other new agents are also being investigated for the prevention of thromboembolic events after THR or TKR (20, 21). A cost-effectiveness analysis was conducted in order to evaluate the incremental cost per quality-adjusted life year (QALY) gained of rivaroxaban 10 mg once daily (od) compared with enoxaparin 40 mg od over a five-year time horizon for the prophylaxis of VTE in Canadian patients undergoing THR or TKR from the provincial government payer s perspective. Methods A decision-analytic model was developed to evaluate the cost-effectiveness of rivaroxaban compared with enoxaparin as prophylaxis for VTE in patients undergoing THR and TKR from a Canadian provincial government perspective. The analysis of the THR population compared 35 days of rivaroxaban with 35 days of enoxaparin (based on the REgulation of Coagulation in ORthopaedic surgery to prevent Deep vein thrombosis and pulmonary embolism REC- ORD1 clinical trial) (16), and the TKR population analysis compared 14 days of rivaroxaban with 14 days of enoxaparin (based on the RECORD3 clinical trial) (18). A five-year time horizon was used for the analysis base case and a 5% discount rate (as required in Canada) was applied to costs and benefits occurring beyond year 1. Model structure The cost-effectiveness model extrapolates the observed event rates from the RECORD trials (i.e. during the prophylaxis module in the model) to project VTE events occurring between the end of the RECORD trials up to 90 days after surgery (i.e. the post-prophylaxis module in the model), and recurrent venous thromboembolic events and LTC occurring up to five years after surgery (i.e. the LTC module of the model). Thus the model is divided into three modules: prophylaxis, post-prophylaxis, and LTC. The first two modules are represented with a decision tree (see Fig. 1), and the third module is developed as a Markov process (see Fig. 2). The first module is related to the outcomes in the RECORD clinical trials (i.e. five weeks for THR, two weeks for TKR). The prophylaxis module includes VTE events occurring during index hospitalisation and those occurring after discharge from hospital. The events considered in the prophylaxis module are: major bleeding, asymptomatic VTE, symptomatic DVT, non-fatal PE, and fatal PE. Asymptomatic VTEs are used for the extrapolation process, but do not result in costs or utility decrements in and of themselves. Data suggest that patients are at risk of developing VTE for up to three months after surgery (22). Because the clinical trials report VTE events up to the time of venography, the probability of patients with an asymptomatic event detected by venography who would subsequently develop a symptomatic VTE is not available from clinical trial data. The model therefore extrapolates for the proportion of venographically confirmed asymptomatic events that would become symptomatic between the end of prophylaxis and 90 days after surgery. This is the post-prophylaxis module (23). Rebound VTE were not included in the model as there was no evidence from the RECORD programme, including the follow-up period, that rebound occurred and therefore it was not considered relevant. Three months after surgery, patients progress to the LTC module. The LTC module reflects the risk of PTS and recurrent VTE. During this module, patients who have experienced a VTE are at risk of developing PTS and having a recurrent VTE. As shown in Figure 2, the Markov model contains three health states: no PTS, PTS, and death due to all-cause mortality. For completeness the analysis considers background mortality risk, however, this does not impact the incremental results. The LTC phase is analysed Figure 1: Decision tree. DVT, deep-vein thrombosis; PE, pulmonary embolism; VTE, venous thromboembolism. Schattauer 2010 Thrombosis and Haemostasis 104.4/2010

3 762 Diamantopoulos et al. Cost-effectiveness of rivaroxaban Figure 2: Markov model. Note: recurrent venous thromboembolism is modelled as a transitory event. PTS, post-thrombotic syndrome. using a Markov process with one-year cycles. Half-cycle correction is applied to the model outcomes. The risk of developing LTC (PTS and recurrent VTE) varies depending on time after surgery and is modelled with time-dependent transition probabilities ( Table 1). All individuals who experienced DVT or PE in previous modules (prophylaxis or extrapolation period) are at risk of recurrent VTE and PTS. The incidence of recurrent VTE is modelled as a transitory event rather than a health state; that is, patients incur the VTE cost and disutility regardless of health state membership. Due to the absence of data, the model assumes that all recurrent VTE events are DVTs. Because PTS is a chronic condition, once developed, patients cannot regress back to a no PTS health state. The model assumes that the entire cohort is at risk of death based on background mortality, which is the same for all individuals in the model regardless of the health state, and is based on Canadian life tables (24). Transition probabilities Prophylaxis module For THR, the prophylaxis module of the model was populated with efficacy and safety data from RECORD1, which compared five weeks of prophylaxis with rivaroxaban to five weeks of prophylaxis with enoxaparin. For TKR, the prophylaxis module of the model was populated with data from RECORD3, which compared two weeks of prophylaxis with rivaroxaban to two weeks of prophylaxis with enoxaparin. The model incorporates differences in event rates between comparators if statistically significant differences were observed in the RECORD clinical trials. If statistically significant differences have not been observed, the event rate observed in the RECORD trial for the rivaroxaban arm is used. The rivaroxaban arm was selected as an indication of baseline risk because it produces more conservative results in terms of cost-effectiveness. Sensitivity analysis was performed on the alternative scenario, where the observed events from the enoxaparin arm are used as baseline risk. Because the results of the clinical trials did not show any statistically significant differences in PE events or prophylaxis-related major bleeding between the two comparators, the model assumes parity between the two comparator arms for these event risks. A sensitivity analysis was also conducted using the numerical differences for these event rates from the RECORD trials. Extrapolation from prophylaxis to post-prophylaxis module The proportion of the cohort that developed an asymptomatic event during the RECORD trials (i.e. RECORD1 for THR and RECORD3 for TKR) was assumed to be at risk of a symptomatic event up to 90 days after surgery (22). The extrapolation of asymptomatic to symptomatic events was based on a study published by Quinlan et al. (23). This study compared the incidence of asymptomatic DVT in THR and TKR studies where venography was routinely performed with the incidence of symptomatic VTE in THR and TKR studies where venography was not performed. The authors reported a consistent relationship between symptomatic VTE and asymptomatic DVT for the THR (ratio of 1:5) and TKR (ratio of 1:2) populations. This results in an estimated probability of asymptomatic VTE developing into symptomatic VTE of approximately 20% for the THR and 5% for the TKR model cohorts. The proportion of VTE patients who experience a PE vs. those with a DVT was reported by White et al. (25), and this relationship was combined with the symptomatic incidence reported by Quinlan et al. (23), in order to estimate the stable risk of DVT and PE after the clinical trial module (Table 1). Thrombosis and Haemostasis 104.4/2010 Schattauer 2010

4 Diamantopoulos et al. Cost-effectiveness of rivaroxaban 763 Table 1: Risk of treatment-emergent bleeding events, venous thromboembolic events, and transition probabilities. Variable Prophylaxis Baseline variable value Source THR (RECORD1) TKR (RECORD3) Prophylaxis module VTE Rivaroxaban RECORD1 (THR) (16), Enoxaparin RECORD3 (TKR) (19) Symptomatic VTE Rivaroxaban Enoxaparin Asymptomatic VTE Rivaroxaban Enoxaparin Fatal PE Rivaroxaban Enoxaparin Non-fatal PE Rivaroxaban Enoxaparin Symptomatic DVT Rivaroxaban Enoxaparin Probability of bleeding Rivaroxaban Enoxaparin Post-prophylaxis module Develop symptomatic VTE All treatments Quinlan et al (23) after asymptomatic VTE* Develop symptomatic PE after asymptomatic VTE All treatments Long-term complications module Probability of PTS: Year 1 All treatments Prandoni et al (26); reported Probability of PTS: Year 2 All treatments Probability of PTS: Year 3 5 All treatments Prandoni et al (26); calculated Probability of recurrent VTE: All treatments Kearon et al. (36) calculated Year 1 5 *Quinlan et al (23) report the probability of developing symptomatic VTE after asymptomatic DVT. The model therefore assumes that asymptomatic DVT is equal to asymptomatic VTE (i.e. no patients experience an asymptomatic PE). The ratio of DVT:PE in the post-prophylaxis period is based on White et al (25). DVT, deep-vein thrombosis; PE, pulmonary embolism; PTS, postthrombotic syndrome; THR, total hip replacement; TKR, total knee replacement; VTE, venous thromboembolism. Extrapolation for LTC module The probability of developing PTS was derived from a published epidemiological study in European patients (26). All individuals who experienced a symptomatic DVT in previous modules (prophylaxis or post-prophylaxis) are at risk of developing PTS. This includes the approximate 37% of patients who have suffered a PE after a DVT (8, 27, 28). In this extrapolation it was assumed that the risk of an asymptomatic event becoming symptomatic is the same regardless of the prophylaxis used any differences are driven by the different proportion of patients with an asymptomatic event during the prophylaxis phase of the model. All patients who suffered a symptomatic DVT or PE in the previous modules were also at risk of developing a recurrent VTE. The proportion of patients developing a recurrent VTE was based on the incidence of recurrent VTE in patients with a first episode of VTE provoked by a transient risk factor (approximately 3.2% per year) (29). The model assumes that all recurrent VTE events are DVTs. Utility values Utility values for each health state were obtained from published literature. Utility scores are adjusted to reflect the duration of the corresponding events. The utility values used in this analysis and the source of the values are shown in Table 2. It was assumed that patients who have just undergone THR or TKR surgery had a reduced background utility of and 0.805, respectively, for the first year Schattauer 2010 Thrombosis and Haemostasis 104.4/2010

5 764 Diamantopoulos et al. Cost-effectiveness of rivaroxaban Table 2: Utility values. Reported utility Adjusted for THR Adjusted for TKR Source No VTE event Assumed utility of 1 for perfect health, adjusted for utility after THR and TKR (30, 31, 37) Prophylaxis-related bleeding ( ) ( ) Adjusted for utility after THR and TKR (30, 31, 37) Asymptomatic DVT Assumed to be the same as no VTE Symptomatic DVT ( ) PE ( ) ( ) ( ) Adjusted for utility after THR and TKR (30, 32, 39, 40) Adjusted for utility after THR and TKR (30, 32, 39, 40) PTS (37) Recurrent VTE 0.04* 0.04* Assumed the same as symptomatic DVT, applied for 3 months of the 1-year cycle Long-term utility 1 1 Assumed utility of 1 for perfect health No VTE event Death 0 0 Assumption DVT, deep-vein thrombosis; PE, pulmonary embolism; PTS, post-thrombotic syndrome; THR, total hip replacement; TKR, total knee replacement; VTE, venous thromboembolism *calculation = (0.84 3/12) + (1 9/12). and thereafter had a background utility of 1 (30). The impact of assuming a lower background utility after the 1 st year is assessed in a sensitivity analysis. These values were used to weight the utility of any events (bleeding, DVT, PE, or PTS) (31, 32) occurring during the course of the analysis. For patients who experienced more than one event in the same cycle (year), the worst utility value was applied in that cycle with the exception of recurrent VTE, which was modelled as a disutility for a period of three months of the one-year Markov cycle. Costs Costs related to prophylaxis, VTE treatment, LTC, and bleeding were included in the model ( Table 3). The duration of prophylaxis was based on the relevant RECORD trial. The duration of hospitalisation was obtained from the Canadian Joint Registry Replacement Annual report (12) and was incorporated in calculating the inhospital, post-discharge, and total prophylaxis drug cost per course. Pharmacy mark-up and dispensing fees, based on Ontario allowances, were added to the daily drug cost after discharge. It was conservatively assumed that no costs were incurred for the administration or monitoring of prophylaxis while in hospital. However, it was assumed that 19% of patients receiving an injectable method of prophylaxis would be unable to self-inject and would, therefore, require nurse assistance once discharged from the hospital (33). This is consistent with previous Canadian economic evaluations, which have included nursing homecare visits for up to 39% of patients receiving LMWH therapy (34). Outpatient monitoring costs were also conservatively assumed to be zero for all methods of prophylaxis. Costs incurred in hospital (such as those for inpatient management of VTE and diagnostic testing costs) were obtained from the Ontario Case Costing Initiative (OCCI) (35), an undertaking of the Ontario Ministry of Health and Long-Term Care that collects case cost data for acute inpatient, day surgery, and ambulatory care cases. Patients experiencing a non-fatal prophylaxis-related bleeding event were assigned the OCCI-derived cost of a major bleeding event occurring in hospital in patients who had undergone THR and TKR. Due to small patient numbers, it was not possible to obtain the cost of fatal bleeding specifically in patients undergoing THR or TKR, so the average cost of a fatal bleeding event occurring after admission was used instead. Although this average cost may differ from the cost of fatal bleeding occurring in THR or TKR patients, the proportion of the cohort experiencing a fatal bleeding event is very small, and thus the model is very insensitive to this input. Minor bleeds were assumed not to incur a cost or quality-of-life impact and so were excluded from the model. The cost of treating patients who develop a VTE event during their index hospitalisation (i.e. while still in hospital after surgery) and the associated length of stay were calculated by subtracting the OCCI average cost for patients undergoing THR or TKR from the OCCI cost of patients who have undergone THR or TKR and also had a DVT or PE. Of those patients whose VTE event occurred after discharge, it was assumed, based on published Canadian sources (34), that 10% of DVT patients and 67% of PE patients were re-admitted to hospital for VTE management. The cost and length of stay for treating readmitted DVTs and PEs were obtained from the OCCI using the cost of DVTs and PEs coded as the most responsible diagnosis or the main pre-admission diagnosis for patients who were discharged (35). All patients treated for VTE in hospital were assumed to receive warfarin after discharge and incur some outpatient follow-up costs based on Skedgel et al. (34). Physician fees were not included in the OCCI costs and, therefore, for events occurring Thrombosis and Haemostasis 104.4/2010 Schattauer 2010

6 Diamantopoulos et al. Cost-effectiveness of rivaroxaban 765 Table 3: Cost and resource use. Resource use Cost (C$) Source Prophylaxis-related drug costs Rivaroxaban 10 mg od 8.86 per day* (37) Enoxaparin 40 mg od 8.20 per day (37) Dalteparin 5,000 units per day 9.83 per day (37) Prophylaxis-related administration and monitoring Rivaroxaban None required 0 Assumption Enoxaparin and Homecare visits in outpatient period (19% patients) per visit (33, 39) dalteparin Prophylaxis-related major bleeding Bleeding Non-fatal 6, (35) Fatal 35, (35) VTE diagnosis PE diagnosis 1 ventilation lung scan (50% patients) per test (34-36) 1 spiral CT scan (50% patients) per test DVT diagnosis 1 Doppler ultrasound per test (35, 36) VTE treatment Inpatient treatment Extra cost of DVT during index hospitalisation (7.44 additional days) 2,929 per stay (35) Extra cost of PE during index hospitalisation (7.13 additional days) 7,105 per stay (35) Hospitalisation due to DVT (7 days) 8, per stay (35) Hospitalisation due to PE (7.25 days) 7, per stay (35) First haematologist visit (36) Follow-up haematologist visit per hospital day (36) Outpatient visit (34, 36) Warfarin 0.30 per day* (34, 37) INR tests every 5 days (16.5 tests) per test (34, 40) 1 INR test per month per test (34, 40) (3 months) Outpatient 1 GP visit per visit (34, 36) management 1 specialist consultation per visit (34, 36) 2 specialist follow-ups per visit (34, 36) 2 complete blood counts per test (34, 40) 5 days of LMWH per day (34, 37) 5 home care visits (19% of patients) per visit (34, 39) 90 days warfarin 0.30 per day (34, 37) 3 months INR per month (34, 40) monitoring 17 INR tests per test (34, 40) Long-term complications Recurrent VTE Same as treating DVT after discharge 0 Assumption PTS Diagnosis 1, (41) Treatment (41) *The following adjustments were made to the prophylaxis cost post-discharge: 8% mark-up; C$7 dispensing fee; C$2 co-payment. 90 days warfarin treatment minus duration of hospitalisation. Converted to C$ and inflated to 2007 prices. CT, computerised tomography; DVT, deep-vein thrombosis; GP, general practitioner; INR, international normalised ratio; LMWH, low-molecular-weight heparin; PE, pulmonary embolism; PTS, post-thrombotic syndrome; VTE, venous thromboembolism. Schattauer 2010 Thrombosis and Haemostasis 104.4/2010

7 766 Diamantopoulos et al. Cost-effectiveness of rivaroxaban Table 4: Costs, efficacy, and cost-effectiveness of thromboprophylactic intervention. Total hip replacement Total knee replacement Rivaroxaban Enoxaparin Incremental Rivaroxaban Enoxaparin Incremental Total cost (C$) (Medication + direct costs) Medication costs (C$) Direct medical costs (C$) QALY Symptomatic VTE events Cost per QALY Cost per symptomatic VTE event QALY, quality-adjusted life year; VTE, venous thromboembolism. Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban either during index hospitalisation or after discharge, the cost of a daily physician visit, derived from the Ontario Schedule of Benefits (36), was added for the VTE-related duration of hospital stay. Resource use associated with outpatient management of VTE was based on the Canadian economic analysis reported by Skedgel et al. (34), and updated costs for each of the resource use items were obtained from a number of publicly available sources, including the Ontario Schedule of Benefits and the Ontario Drug Benefit Formulary (Table 3) (36, 37). The baseline analysis was conducted for a five-year timeframe, which allowed the impact of LTC to be incorporated. In order to assess the impact of rivaroxaban in the short term, one-way sensitivity analyses (SA) based on the prophylaxis and post-prophylaxis phases only were performed. A number of one-way sensitivity analyses were also conducted on the baseline assumptions in order to examine the effect of potential variations in clinical practice, costs, and event rates. Parameters tested included outpatient administration costs associated with enoxaparin, discount rates, and event treatment costs. A sensitivity analysis incorporating differences in bleeding and VTE rates that were numerically but not statistically different was also conducted. Because dalteparin is a frequently used LMWH in Canada, a sensitivity analysis comparing rivaroxaban with dalteparin in THR and TKR was also done. This analysis assumed that the efficacy and safety of LMWHs are not different and, therefore, used enoxaparin efficacy and safety inputs for the dalteparin arm. Given that there is not a universally accepted perfect health utility value in Canada, the value of 1 was selected for patients with no longterm VTE complications. A SA considers utility weighting based on a UK study (38). Probabilistic sensitivity analyses (PSAs) were performed to address uncertainty at the parameter level (2,000 samples). The parameters of the model that were sampled include drug administration and monitoring costs, prophylaxis-related bleeding costs, diagnosis and treatment costs, DVT and PE treatment costs, LTC costs, utility values and event probabilities for the prophylaxis, post-prophylaxis, and LTC modules (distributions were generated using 95% confidence intervals or standard deviation, depending upon the parameter). Results Sensitivity analyses Deterministic analysis The base case analyses indicate that, when 35 days of prophylaxis with rivaroxaban and enoxaparin are compared in patients undergoing THR, rivaroxaban enoxaparin. That is, it is associated with greater benefit ( QALYs), fewer symptomatic VTE events (0.0061), and lower cost (savings of C$300) per patient. Similarly, when rivaroxaban and enoxaparin are compared in patients undergoing TKR, rivaroxaban enoxaparin, achieving a gain of QALYs, a reduction of symptomatic VTEevents, and savings of C$129 per patient (see Table 4). One-way sensitivity analysis Sensitivity analyses considering only the acute phase (prophylaxis and post-prophylaxis phases) indicate that rivaroxaban is a costeffective method of VTE prophylaxis when compared with enoxaparin in both a THR and a TKR population: rivaroxaban dominated enoxaparin in both THR and TKR populations, providing more benefit at less cost ( Table 5). In the sensitivity analysis where it is assumed the non-statistically-significant event risk is replaced by the enoxaparin group observed event risk instead of rivaroxaban, the cost-effectiveness Thrombosis and Haemostasis 104.4/2010 Schattauer 2010

8 Diamantopoulos et al. Cost-effectiveness of rivaroxaban 767 results of the THR analysis remain the same. This is because in the THR population only VTE event is statistically significant and since all others are symmetrically reversed the incremental cost and incremental effects are unchanged. In the TKR population symptomatic VTE events are also statistically significant which results in changes to the calculations of symptomatic DVT and asymptomatic VTE events. In this scenario the incremental costsaving is increased to approximately C$379 and the incremental quality of life benefit to QALYs. Rivaroxaban remains the dominant intervention. For patients undergoing TKR, rivaroxaban continued to dominate enoxaparin in all scenarios tested. For patients undergoing THR, rivaroxaban was less costly and more effective than enoxaparin in all but one scenario. When using the results from RECORD2, where patients were administered with rivaroxaban for 35 days and enoxaparin for 14 days, the cost per QALY was approximately C$24,977 which is well below the frequently-referenced cost-effectiveness threshold of C$50,000/QALY. Probabilistic sensitivity analysis The results of the PSAs for THR and TKR are shown via cost-utility analysis (CUA) planes in Figures 3 and 4, respectively. For THR, all of the iterations appear in the lower half of the CUA plane, indicating that 35 days of rivaroxaban is less costly compared with 35 days of enoxaparin. In addition, nearly all of the iterations appear in the lower right quadrant of the CUA plane, indicating that 35 days of rivaroxaban is less costly and more effective compared with 35 days of enoxaparin ( Fig. 3). For TKR, all simulations appear in the bottom right quadrant ( Fig. 4), indicating that 14 days of rivaroxaban is less costly and more effective than 14 days of enoxaparin in terms of preventing VTE events after TKR. Both PSAs and the one-way sensitivity analyses suggest that the finding of dominance for rivaroxaban relative to enoxaparin improved health outcomes at a lower cost is robust. Table 5: One-way sensitivity analysis results. Sensitivity analysis Total hip replacement Incremental cost (C$) Incremental QALYs gained Result 90-day time horizon Rivaroxaban Exclude event treatment and diagnosis costs (bleeding, DVT, PE) Rivaroxaban 0% discount rate for costs and outcomes Rivaroxaban Duration of hospitalisation (+2 days) Rivaroxaban Duration of hospitalisation ( 2 days) Rivaroxaban Accept non-significant efficacy and safety data Rivaroxaban Total knee replacement Incremental cost (C$) Incremental QALYs gained Result Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban 35 days rivaroxaban versus 14 days enoxaparin (RECORD2) $24, NA NA NA Use RECORD4 for TKR (enoxaparin 30 mg bid) NA NA NA Rivaroxaban Use Prandoni et al (18) for risk of recurrent VTE Rivaroxaban 8% using home care for enoxaparin injections Rivaroxaban 39% using home care for enoxaparin injections Rivaroxaban Assuming a background utility of post year 1 as reported in Kind et al 1998 Rivaroxaban vs. dalteparin Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban Rivaroxaban bid, twice daily; DVT, deep-vein thrombosis; NA, not applicable; PE, pulmonary embolism; QALY, quality-adjusted life year; TKR, total knee replacement; VTE, venous thromboembolism. Schattauer 2010 Thrombosis and Haemostasis 104.4/2010

9 768 Diamantopoulos et al. Cost-effectiveness of rivaroxaban Figure 3: Total hip replacement cost-utility analysis plane (quality-adjusted life years). Discussion Rivaroxaban is an orally administered agent that does not require any monitoring during administration for the prevention of VTE. The efficacy and safety of rivaroxaban have been demonstrated in four large phase III randomised controlled trials, where rivaroxaban showed a statistically significant improvement over enoxaparin in the primary endpoint of total VTE (16, 19). Rivaroxaban also showed no statistically significant difference relative to enoxaparin in the occurrence of major bleeding or clinically relevant non-major bleeding. The lack of difference in bleeding events between rivaroxaban and enoxaparin indicate there is no difference in incremental costs associated with these events. The base case cost-effectiveness analyses for rivaroxaban in THR and TKR show that rivaroxaban enoxaparin, providing more benefit at less cost in both populations. In addition to providing increased QALYs relative to enoxaparin, rivaroxaban also leads to fewer symptomatic VTE events than enoxaparin. Rivaroxaban is expected to prevent an additional six symptomatic VTE events relative to enoxaparin per 1,000 THR procedures. For TKR, rivaroxaban is expected to prevent an additional 19 events per 1,000 procedures. The sensitivity of the model results to changes in various parameters was evaluated using one-way deterministic sensitivity analyses. Moreover, uncertainty around model parameters was addressed by PSA. For the deterministic sensitivity analyses, rivaroxaban dominated enoxaparin in all scenarios for the TKR population and in the majority of scenarios for THR. The PSA shows that the results of the economic evaluation are robust. Several potential benefits associated with rivaroxaban thromboprophylaxis, compared with enoxaparin, had not been taken into account in this analysis. First, it had been suggested that the method of administration (oral vs. parenteral) may have some effect on patients utility. However, as no data had been identified to support this hypothesis, disutility associated with injections was not included in this model. Similarly, although no in-hospital drug administration costs had been included in the analysis, enoxaparin injections administered in hospital would likely be performed by a nurse or other healthcare professional, and may therefore be associated with higher professional costs when compared with oral rivaroxaban. Furthermore, in Canada, when LMWHs are administered regular monitoring for heparin-induced thrombocytopenia is recommended, but costs associated with such monitoring had not been included in this analysis. Because enoxaparin is administered as a parenteral injection and rivaroxaban is taken orally, the inclusion of any disutility or additional cost associated with injectable prophylaxis would change the results of this analysis in favour of rivaroxaban. In addition, in Canada, many tertiary care hospitals have anticoagulation clinics to manage discharged patients. Services offered at these clinics include nurse-assisted injection of LMWHs and monitoring of warfarin. The cost of operating these clinics and thus of managing prophylaxis after hospital discharge can be substantial. Because rivaroxaban does not require routine monitoring or assisted administration, it is not anticipated that such clinic visits would be required. In order to remain conservative, the clinic costs for patients receiving LMWHs were not included in the analysis. However, even when administration costs were excluded rivaroxaban was highly cost-effective. Thrombosis and Haemostasis 104.4/2010 Schattauer 2010

10 Diamantopoulos et al. Cost-effectiveness of rivaroxaban 769 Figure 4: Total knee replacement cost-utility analysis plane (quality-adjusted life years). The model structure is based on published recommendations and previously conducted economic evaluations, and, therefore, represents best practice of economic modelling in this area. Moreover, the economic evaluation is based on the clinical trial data and all the cost and resource use assumptions in the model are from publicly available sources. As indicated by the scenario analyses overall results of the model are robust to variations in its inputs. As with any economic model, there are limitations with the approach taken in this analysis. Although the prophylaxis module of the model was based on clinical trial data, the assumptions and values used in the post-prophylaxis and LTC modules were based on the literature. Rebound thromboembolic effects were not included in the model, as the results are driven by the reduced administration requirements and reduced symptomatic and asymptomatic events. Therefore the inclusion of rebound effects is unlikely to have changed the overall conclusions. The parameter of clinically relevant non-major bleeding was excluded from the model because expert advice indicated the likely impact on resource use was very low, as was the impact on health-related quality of life. Furthermore, several different sources were required to ascertain the utility values to populate the model. Although the resource use and cost data were taken from Canadian sources, it was not possible to source Canadian data for all transition probabilities, and utilities used in the model. However, extensive sensitivity analyses (both oneway and probabilistic sensitivity analyses) showed that the key findings of rivaroxaban being a highly cost effective anticoagulant remained unchanged. Conclusion In summary, rivaroxaban is the first oral, direct factor Xa inhibitor available for VTE prophylaxis in patients undergoing THR and What is known about this topic? Before the introduction of novel oral anticoagulants, prophylaxis with low-molecular-weight heparins (LMWHs), such as enoxaparin, was the standard of care after total hip or knee replacement (THR, TKR) in Canada. The cost-effectiveness of enoxaparin in Canada, and elsewhere, has been established in the literature for some years. The only published economic analysis of rivaroxaban in this indication is in the setting of Ireland (McCullagh et al. 2009); it shows rivaroxaban to be more effective and less expensive vs. enoxaparin prophylaxis after THR and TKR (based only on short-term events). What does this paper add? It demonstrates that a decrease in short-term thrombotic events (deep-vein thrombosis and pulmonary embolism) reduces healthcare costs associated with treating these events, as well as the cost associated with longer-term complications of these events (recurrent events and post-thrombotic syndrome). In Canada, the improved health outcomes with rivaroxaban translate into reduced healthcare costs (even when drug prices are included) relative to LMWHs. These findings confirm the results of the Irish analysis (which was based on a different model). Schattauer 2010 Thrombosis and Haemostasis 104.4/2010

11 770 Diamantopoulos et al. Cost-effectiveness of rivaroxaban TKR. It has demonstrated meaningful improvements in efficacy relative to enoxaparin and also has a positive health economic profile, providing more benefit at less cost relative to enoxaparin in THR and TKR patients. The convenience of rivaroxaban, combined with its improved efficacy and economic benefits over enoxaparin, suggests that rivaroxaban is a cost-effective option for VTE prophylaxis in both patient populations. Acknowledgements The authors would like to acknowledge Chris Thomas who provided editorial support with funding from Bayer Schering Pharma AG. Disclosures A. Diamantopoulos is an employee of a private consultancy that provides services to Bayer. M. Lees and H. McDonald are company employees of Bayer Schering Pharma AG. P. Wells has received honoraria from Dade Behring, BioMereiux, Sanofi Aventis, Leo Pharma and Organon for presentations. F. Forster and J. Ananthapavan are paid consultants for Bayer. References 1. Heit JA, Melton LJ, III, Lohse CM, et al. Incidence of venous thromboembolism in hospitalized patients vs community residents. Mayo Clin Proc 2001; 76: Turpie AG, Norris TM. Thromboprophylaxis in medical patients: the role of lowmolecular-weight heparin. Thromb Haemost 2004; 92: Blann AD, Lip GY. Venous thromboembolism. Br Med J 2006; 332: Malone PC, Agutter PS. The aetiology of deep venous thrombosis. QJM 2006; 99: Tapson VF. Acute pulmonary embolism. N Engl J Med 2008; 358: Zhu T, Martinez I, Emmerich J. Venous thromboembolism: risk factors for recurrence. Arterioscler Thromb Vasc Biol 2009; 29: Kahn SR, Ginsberg JS. Relationship between deep venous thrombosis and the postthrombotic syndrome. Arch Intern Med 2004; 164: Pengo V, Lensing AW, Prins MH, et al. Incidence of chronic thromboembolic pulmonary hypertension after pulmonary embolism. N Engl J Med 2004; 350: Jaffer AK. An overview of venous thromboembolism: impact, risks, and issues in prophylaxis. Cleve Clin J Med 2008; 75 (Suppl 3): S3-S Deitelzweig SB, McKean SC, Amin AN, et al. Prevention of venous thromboembolism in the orthopedic surgery patient. Cleve Clin J Med 2008; 75 (Suppl 3): S27-S Geerts WH, Bergqvist D, Pineo GF, et al. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008; 133: 381S-453S. 12. Hip and knee replacements in Canada Annual Report. (Accessed July 31, 2009, at http: //secure.cihi.ca/cihiweb/disppage.jsp?cw_page=ar_30_e.) 13. Venous thromboembolism: reducing the risk of venous thromboembolism (deep vein thrombosis and pulmonary embolism) in inpatients undergoing surgery. NICE clinical guidelines 46. (Accessed March 31, 2009, at http: // 14. Perzborn E, Kubitza D, Misselwitz F. Rivaroxaban. A novel, oral, direct factor Xa inhibitor in clinical development for the prevention and treatment of thromboembolic disorders. Hamostaseologie 2007; 27: Graff J, von Hentig N, Misselwitz F, et al. Effects of the oral, direct Factor Xa inhibitor rivaroxaban on platelet-induced thrombin generation and prothrombinase activity. J Clin Pharmacol 2007; 47: Eriksson BI, Borris LC, Friedman RJ, et al. Rivaroxaban versus enoxaparin for thromboprophylaxis after hip arthroplasty. N Engl J Med 2008; 358: Kakkar AK, Brenner B, Dahl OE, et al. Extended duration rivaroxaban versus short-term enoxaparin for the prevention of venous thromboembolism after total hip arthroplasty: a double-blind, randomised controlled trial. Lancet 2008; 372: Lassen MR, Ageno W, Borris LC, et al. Rivaroxaban versus enoxaparin for thromboprophylaxis after total knee arthroplasty. N Engl J Med 2008; 358: Turpie AGG, Lassen MR, Davidson BL, et al. Rivaroxaban versus enoxaparin for thromboprophylaxis after total knee arthroplasty (RECORD4): a randomised trial. Lancet 2009; 373: Turpie AG, Bauer KA, Davidson BL, et al. A randomized evaluation of betrixaban, an oral factor Xa inhibitor, for prevention of thromboembolic events after total knee replacement (EXPERT). Thromb Haemost 2009; 101: Wolowacz SE, Roskell NS, Plumb JM, Caprini JA, Eriksson BI. Efficacy and safety of dabigatran etexilate for the prevention of venous thromboembolism following total hip or knee arthroplasty. A meta-analysis. Thromb Haemost 2009; 101: Sullivan SD, Kahn SR, Davidson BL, et al. Measuring the outcomes and pharmacoeconomic consequences of venous thromboembolism prophylaxis in major orthopaedic surgery. Pharmacoeconomics 2003; 21: Quinlan DJ, Eikelboom JW, Dahl OE, et al. Association between asymptomatic deep-vein thrombosis detected by venography and symptomatic venous thromboembolism in patients undergoing elective hip or knee surgery. J Thromb Haemost 2007; 5: Life Tables, Canada, Provinces and Territories 2000 to (Accessed May 27, 2010, at http: // 537-XIE&lang =eng.) 25. White RH, Romano PS, Zhou H, et al. Incidence and time course of thromboembolic outcomes following total hip or knee arthroplasty. Arch Intern Med 1998; 158: Prandoni P, Villalta S, Bagatella P, et al. The clinical course of deep-vein thrombosis. Prospective long-term follow-up of 528 symptomatic patients. Haematologica 1997; 82: Prandoni P, Hutten BA, van Dongen CJ, et al. Quality of oral anticoagulant treatment and risk of subsequent recurrent thromboembolism in patients with deep vein thrombosis. J Thromb Haemost 2007; 5: Buller HR, Davidson BL, Decousus H, et al. Subcutaneous fondaparinux versus intravenous unfractionated heparin in the initial treatment of pulmonary embolism. N Engl J Med 2003; 349: Kearon C, Ginsberg JS, Anderson DR, et al. Comparison of 1 month with 3 months of anticoagulation for a first episode of venous thromboembolism associated with a transient risk factor. J Thromb Haemost 2004; 2: Rasanen P, Paavolainen P, Sintonen H, et al. Effectiveness of hip or knee replacement surgery in terms of quality-adjusted life years and costs. Acta Orthop 2007; 78: Lenert LA, Soetikno RM. Automated computer interviews to elicit utilities: potential applications in the treatment of deep venous thrombosis. J Am Med Inform Assoc 1997; 4: Haentjens P, De Groote K, Annemans L. Prolonged enoxaparin therapy to prevent venous thromboembolism after primary hip or knee replacement. A cost-utility analysis. Arch Orthop Trauma Surg 2004; 124: Harrison L, McGinnis J, Crowther M, et al. Assessment of outpatient treatment of deep-vein thrombosis with low-molecular-weight heparin. Arch Intern Med 1998; 158: Skedgel C, Goeree R, Pleasance S, et al. The cost-effectiveness of extended-duration antithrombotic prophylaxis after total hip arthroplasty. J Bone Joint Surg Am 2007; 89: Anonymous. OCCI Acute Inpatient Databases 2005/06 and 2006/ Ministry of Health and Long Term Care. Ontario Schedule of Benefits: Physician services under the Health Insurance Act (November 19, 2007 effective February 1, 2008) Anonymous. Ontario Drug Benefit Formulary Kind P, Dolan P, Gudex C, et al. Variations in population health status: results from a United Kingdom national questionnaire survey. Br Med J 1998; 316: Anonymous. Ontario Association of Community Care Access Centres (OAC- CAC) Medical Services Commission Payment Schedule. (Accessed April 1, 2008, at http: // 41. Caprini JA, Botteman MF, Stephens JM, et al. Economic burden of long-term complications of deep vein thrombosis after total hip replacement surgery in the United States. Value Health 2003; 6: Thrombosis and Haemostasis 104.4/2010 Schattauer 2010