Clinical Therapeutics/Volume ], Number ], 2015
Cost-effectiveness Analysis of Apixaban against Warfarin for Stroke Prevention in Patients with Nonvalvular Atrial Fibrillation in Japan Isao Kamae, MD, PhD1; Yoichiro Hashimoto, MD2; Yukihiro Koretsune, MD, PhD3; Norio Tanahashi, MD4; Tatsunori Murata, MS5; Hemant Phatak, PhD6; Larry Z. Liu, MD, PhD7; Ann C. Tang, PhD8; Peter Feng Wang, MD, PhD9; and Ken Okumura, MD, PhD10 1
The University of Tokyo, Tokyo, Japan; 2Kumamoto City Hospital, Kumamoto, Japan; 3National Hospital Organization, Osaka National Hospital, Osaka, Japan; 4Saitama International Medical Center, Saitama, Japan; 5Crecon Medical Assessment Inc, Tokyo, Japan; 6US Health Economics and Outcomes Research, EMD Serono, Rockland, MA; 7Global Health Economics and Outcomes Research and Evidence, Pfizer Inc, and Weill Medical College of Cornell University, New York, New York, USA; 8Health Economics and Outcomes Research, Bristol-Myers K.K., Tokyo, Japan; 9Worldwide Health Economics and Outcomes Research, Bristol Myers Squibb, Princeton, New Jersey; and 10Department of Cardiology, Hirosaki University Graduate School of Medicine, Aomori, Japan
ABSTRACT Purpose: The aim of this study was to evaluate the cost-effectiveness of apixaban compared with to warfarin, current standard of care, for stroke prevention in patients with nonvalvular atrial fibrillation (NVAF) in Japan. Methods: A previously published lifetime Markov model was adapted to evaluate the cost-effectiveness of apixaban compared with warfarin in patients with NVAF in Japan. In the same model, the costs associated with each clinical event and background mortality were replaced with Japanese data. Whenever available, some of the utility parameters were derived from Japanese published literature. Lifetime horizon was selected to evaluate the value of the treatment benefit (stroke prevention) against potential risks (such as major bleedings) among patients with NVAF. Direct medical cost, long-term care cost, and qualityadjusted life years (QALYs) were calculated from the payers’ perspective. Findings: Compared with warfarin, treatment with apixaban was estimated to increase life expectancy by 0.231 year or 0.240 QALYs while treatment cost increased by ¥511,692 (US $5117 at an exchange rate of US $1 ¼ ¥100). The incremental cost-effectiveness ratio was ¥2,135,743 per QALY (US $21,357 per QALY). On the basis of the results of the probabilistic
] 2015
sensitivity analysis, when the willingness-to-pay threshold was set at approximately Z¥2,250,000 (US $22,500) per QALY, the probability of apixaban being cost-effective was Z50%. Assuming a willingness-to-pay threshold of ¥5,000,000 (US $50,000) and ¥6,700,000 (US $67,000) in Japan, the probability of apixaban being cost-effective was 85% and 91%, respectively. Conclusion: Although most participants in the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial used for the efficacy data of apixaban in the model were non-Japanese patients, the impact of the limitations on our results was considered small, and our results were deemed robust because of the additional effect in Japanese patients compared with that in the global population according to the subanalysis of Japanese patients in the trial. Therefore, based on an adaptation of a published Markov model, apixaban is a cost-effective alternative to warfarin in Japan for stroke prevention among patients with NVAF.
Accepted for publication October 5, 2015. http://dx.doi.org/10.1016/j.clinthera.2015.10.007 0149-2918/$ - see front matter & 2015 Elsevier HS Journals, Inc. All rights reserved.
1
Clinical Therapeutics (Clin Ther. 2015;]:]]]–]]]) & 2015 Elsevier HS Journals, Inc. All rights reserved. Key words: apixaban, cost-effectiveness, nonvalvular atrial fibrillation, Japan.
INTRODUCTION
Atrial fibrillation (AF) affects 0.6% to 1.6% of the general population and up to 14% of patients in cardiovascular clinics in Japan.1,2 AF increases the risk of stroke by nearly 5-fold.3 Warfarin has been used to prevent the occurrence of stroke, a complication of AF. However, monitoring of the prothrombin time–international normalized ratio is necessary during warfarin use, and it is not easy to maintain the dose of the drug within the desired therapeutic range.4 Use of warfarin can be associated with hemorrhagic adverse effects, including severe and life-threatening events, such as intracerebral hemorrhage. In addition, warfarin is complicated by challenges such as drug-drug interactions and restriction of foods, including vitamin K. Recent guidelines from the European Society of Cardiology recommend the use of novel anticoagulants (NOACs) as alternatives to warfarin or antiplatelet therapy in most AF patients who require stroke prevention.5 Apixaban, an orally active factor Xa inhibitor, has been studied in a Phase III global randomized clinical trial (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation [ARISTOTLE]) versus dose-adjusted warfarin in patients with nonvalvular atrial fibrillation (NVAF). The study found that the incidence of stroke and systemic embolism decreased significantly and the frequency of major bleeding also decreased significantly in the apixaban 5 mg BID group compared with the warfarin control group.6 NOACs, such as dabigatran, rivaroxaban, and apixaban, are currently available for clinical use for the prevention of stroke in patients with NVAF in Japan. According to the 2011 Estimates of National Medical Care Expenditure report, the medical cost of cerebrovascular diseases, including ischemic and hemorrhagic stroke, was ¥1.8 trillion (US $18 billion at an exchange rate of US $1 ¼¥100), which accounted for approximately 6% of the overall medical cost (¥28 trillion [US $280 billion]).7 Prevention of stroke is important not only clinically but also from the perspective of health economics.
2
High drug acquisition cost of NOACs versus warfarin in Japan warrants such analyses to be conducted from the Japanese payer perspective to inform selection of optimal anticoagulation therapy for NVAF patients. Warfarin is currently still considered the standard of care, although NOACs are becoming more widely used in Japan.8 In this study, we evaluated the costeffectiveness of apixaban compared with warfarin based on Japanese data.
METHODS Model In this study, a previously published model for health economic evaluation9,10 (Figure 1) was adapted to evaluate the cost-effectiveness of apixaban compared with warfarin in Japan. The model was a Markov model developed using Microsoft Excel, and a 6-week cycle length was chosen to capture the possibilities of events related to AF that occur within a short period. The detailed description of the model can be found in the publications by Dorian et al.9 In brief, all patients started in the NVAF state, and certain portions of the cohort were assigned during each model cycle to one of the modeled 17 health states. The health states comprised NVAF, ischemic stroke (mild, moderate, severe), hemorrhagic stroke (mild, moderate, severe), recurrent ischemic stroke (mild, moderate, severe), recurrent hemorrhagic stroke (mild, moderate, severe), systemic embolism, gastrointestinal bleeding and myocardial infarction (MI), and death. Intracranial hemorrhage (ICH) that is not hemorrhagic stroke (other ICH), major bleeding other than gastrointestinal bleeding, clinically relevant nonmajor bleeding, and other cardiovascular hospitalization were also considered as events. Patients who experience these events were assumed to stay in their initially assigned anticoagulation state, and events were modeled as transient states. These assignments were based on the probability of experiencing an event that in turn was derived from secondary analyses of trial data9 (Tables I–IV). Some events altered subsequent risks of modeled events, including death (eg, patients with ischemic stroke would have higher probability of death or patients experiencing MI would have greater likelihood of acute mortality). In addition, stroke severity was considered. Stroke severity was evaluated by the modified Rankin Scale (mRS) as mild (mRS scores
Volume ] Number ]
I. Kamae et al.
NVAF
No Change
Systemic Embolism
Mild IS
NVAF Nonfatal
Fatal
Systemic embolism
Death
Moderate IS Mild Nonfatal Ischemic Hemorrhagic
Severe IS
Moderate
Mild IS Moderate IS
Severe Fatal
Severe IS Death Mild
Warfarin
Dabigatran 110 mg
Dabigatran 150 mg
Rivaroxaban
M
Hemorrhagic Stroke
Mild HS
Mild HS
Moderate
Moderate HS
Severe Severe HS
M
ICH Moderate HS
M
Nonfatal
Fatal
Death
NVAF
Tmt Interruption (6wk )
M
Other ICH
Nonfatal Discontinue AC Fatal
M
NVAF NVAF subsequent ASA
Death
Severe HS Apixaban
Bleeding M
Major Bleed
Nonfatal
Discontinue AC Fatal
Systemic embolism CRNM Bleed
Stay on current AC
Death
Nonfatal Myocardial infarction Fatal Other Deaths Other AC Discontinuation
NVAF subsequent ASA
NVAF
Myocardial infarction
NVAF subsequent ASA
NVAF
Myocardial infarction
Death
Death NVAF subsequent ASA
Death
Figure 1. Markov model. The model uses a previously published 6-week cycle model.9,10 All patients start with the nonvalvular atrial fibrillation (NVAF) state, and certain portions of the cohort are assigned during each model cycle to 1 of the modeled 17 health states. The health states comprise ischemic stroke (IS) (mild, moderate, severe), hemorrhagic stroke (HS) (mild, moderate, severe), recurrent IS (mild, moderate, severe), recurrent HS (mild, moderate, severe), systemic embolism, gastrointestinal bleeding and myocardial infarction (MI), or death. Other intracranial hemorrhage (ICH), major bleeding other than gastrointestinal bleeding, clinically relevant nonmajor (CRNM) bleeding, and other cardiovascular hospitalization are also considered events, but patients are assumed to stay in their initially assigned anticoagulation states and the events are modeled as transient states. Some events altered the subsequent risks of modeled events, including death. AC = anticoagulant; ASA = aspirin. M indicates Markov process, and triangles indicate that a patient transitions to the following state in the next cycle.
0–2), moderate (mRS scores 3–4), and severe (mRS score 5; an mRS score of 6 is death). Recurrence was also considered for stroke. For treatment after each event, patients were assumed to continue the first-line treatment (apixaban or warfarin) after ischemic stroke, systemic embolism, MI, clinically relevant nonmajor bleeding, and other cardiovascular hospitalization. However, for ICH (including hemorrhagic stroke), gastrointestinal bleeding, and other major bleeding, patients were assumed to discontinue or suspend first-line treatment for 6 weeks. Patients who discontinued first-line treatment were assumed
] 2015
to start second-line aspirin treatment. In essence, model structure was built to reflect real-life treatment paradigms and consequences of treatment decision.
Localization In the adaptation analysis, we used parameters such as cost of each event and some of the utility parameters and background mortality using the Japanese data. The cost of the acute phase for each event were obtained from Japanese claims data for patients with AF (excluding AF associated with rheumatic mitral valve disease and mechanical valve) using the
3
Clinical Therapeutics
Table I. Distributions of the cTTR and CHADS2 scores used in the model. Parameter cTTR, % o52.38 52.38–66.02 66.02–76.51 Z76.51 CHADS2 0 1 2 3 4 5 6
Setting, %
Source
25.00 25.00 25.00 25.00
Dorian et al,9 2014
0.60 33.40 35.80 18.00 8.70 3.10 0.40
Dorian et al,9 2014
CHADS2 ¼ congestive heart failure, hypertension, age of 75 years, diabetes mellitus, stroke; cTTR ¼ center’s median time in therapeutic range.
Medical Data Vision Co Ltd (MDV) database. The database provided claims data from 131 hospitals (as of April 2013) using the Diagnosis Procedure Combination (DPC) system for medical service claims (21% of general hospitals but 55% of general beds in Japan is under the DPC system in 201415). DPC is a reimbursement system for hospitals to claim a treatment cost comprehensively estimated for one hospitalization per patient. The actual costs per event from admission until discharge (including death; ie, mRS score of 6) with a name of each event in the record of “disease name with the most health resources invested” in DPC data were used to estimate costs associated with the acute phase of each event. The definitions of the target population and events are given in Table V. The costs of chronic stroke were obtained from the study by Hattori et al, which included not only direct medical cost but also the long-term care cost.16 The cardiogenic brain embolism data derived from Stroke Data Bank 2009 were used for distribution of the mRS scores for each stroke severity in Japanese patients.17 The cost of medication therapy and that of percutaneous coronary intervention for the
4
chronic phase of MI are quoted from the report of Tanihata et al.18 The percentage of patients receiving each treatment was obtained from a study using claims data analysis in Japan by Evans et al.19 Because the research by Tanihata et al was conducted in 2001, the cost derived from the study was adjusted to 2014 rates based on the revised total medical treatment fees thereafter (Tables VI and VII). Japanese utility data were used only for chronicphase stroke, for which Japanese evidence is available16 (Table VIII). Background mortality was set based on the 2012 Abridged Life Table according to age and sex in Japan.20
Analyses Lifetime horizon was chosen to evaluate the treatment value of the preventive effect against events such as stroke in NVAF patients. The direct medical cost, long-term care cost, and qualityadjusted life-years (QALYs) were calculated from the payers’ perspective. The number of events avoided per 1,000 patients with NVAF during the lifetime horizon was also evaluated. An annual discount of 2% was applied to both the cost and effectiveness in accordance with a Japanese guideline for health economic analysis.21 In this study, the cost-effectiveness of apixaban compared with warfarin was estimated in base case analysis. One-way sensitivity analysis for each parameter was conducted to confirm the impact of each parameter on the analysis results with the 95% CI for each parameter. In addition, probabilistic sensitivity analysis by Monte-Carlo simulation with 2000 iterations was performed, and parameter uncertainty was assessed by the cost-effectiveness acceptability curve, which plotted the probability that apixaban is cost-effective in willingness to pay (WTP) on the horizontal axis.
RESULTS Base Case Analysis The simulation results in NVAF patients treated with apixaban versus warfarin were projected to reduce the number of strokes by 16 per 1000 patients (2 for ischemic stroke and 14 for hemorrhagic stroke) and the number of bleeding episodes other than hemorrhagic stroke onset by 72 per 1000 patients. The expected life-years accrued during lifetime treatment per patient were 10.285 years for the warfarin
Volume ] Number ]
I. Kamae et al. Table II. Relative event risks for each cTTR used in the model.* Relative Event Risk cTTR
Apixaban
Warfarin
cTTR for stroke, % o52.38 0.92 52.38–66.02 1 66.02–76.51 0.69 Z76.51 0.56 cTTR for ICH, % o52.38 0.58 52.38–66.02 1 66.02–76.51 0.69 Z76.51 0.36 cTTR for gastrointestinal bleeding or other major bleeding, % o52.38 0.72 52.38–66.02 1 66.02–76.51 1.69 Z76.51 1.77 cTTR for CRNM bleeding, % o52.38 0.71 52.38–66.02 1 66.02–76.51 1.25 Z76.51 1.7
Source
1.54 1 0.84 0.72
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
1.05 1 0.68 0.78
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
0.84 1 1.13 1.37
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
0.99 1 1.26 1.27
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
CRNM ¼ clinically relevant nonmajor bleeding; cTTR ¼ center’s median time in therapeutic range; ICH ¼ intracranial hemorrhage. * From the parameter applied in the model used in this analysis, settings for relative risk according to the cTTR of each event are given. Event risk was set as the relative risk in each cTTR value based on a cTTR of 52.38% to 66.02%.
group and 10.516 years for the apixaban group. Therefore, apixaban was expected to extend 0.231 life-year and improve quality of life of patients, which generated 0.240 QALYs gained. As with many new technologies, innovation posts additional costs. The clinical improvement from apixaban compared with warfarin was expected to incur some additional costs to Japanese payers. During a lifetime horizon, total cost per patient was ¥5,144,019 (US $51,440) for warfarin and ¥5,655,712 (US $56,557) for apixaban, resulting in incremental cost of ¥511,692 (US $5117). The incremental cost-effectiveness ratio (ICER) of treatment with apixaban versus warfarin was estimated to be ¥2,135,743 (US $21,357) per QALY gained (Table IX).
] 2015
Sensitivity Analysis The results of the 1-way sensitivity analysis identified the following factors as having a large impact on the ICER: the risk of ischemic and unspecified stroke associated with the use of apixaban and the risk of ICH associated with the use of warfarin (Figure 2). The range of ICER in sensitivity analysis of these parameters were ¥987,444 to ¥5,116,598 per QALY (US $9,874 to $51,166 per QALY) for risk of ischemic and unspecified stroke for apixaban and ¥832,334 to ¥3,819,575 per QALY (US $8323 to 38,196 per QALY) for risk of ICH for warfarin. In the probabilistic sensitivity analysis, the cost-effectiveness plane (scatterplot) suggested apixaban was more effective and more costly than warfarin (Figure 3).
5
Clinical Therapeutics
Table III. Clinical event rates for each parameter in the model. Parameter
Apixaban
Warfarin
Event risk Ischemic stroke CHADS2 score of 0–1 CHADS2 score of 2 CHADS2 score of Z3 ICH Gastrointestinal bleeding or other major bleeding CRNM bleeding MI Systemic embolism Other CV hospitalization Withdrawal from treatment
person-years person-years person-years person-years person-years
2.08/100 person-years 0.53/100 person-years 0.09/100 person-years 10.46/100 person-years 13.18/100 person-years
Recurrence rate of stroke Ischemic stroke Hemorrhagic stroke
4.10/100 person-years 3.00/100 person-years
0.52/100 0.95/100 1.53/100 0.33/100 1.79/100
1.09*
Source
2.39* 1.27*
Dorian Dorian Dorian Dorian Dorian
et et et et et
al,9 al,9 al,9 al,9 al,9
2014 2014 2014 2014 2014
1.43* 1.14* 1.11* 1.00* 1.08*
Dorian Dorian Dorian Dorian Dorian
et et et et et
al,9 al,9 al,9 al,9 al,9
2014 2014 2014 2014 2014
Hemorrhagic stroke and gastrointestinal bleeding, % Hemorrhagic stroke out of ICH 77 Gastrointestinal bleeding out of 38 other major bleeding
Mohan et al,11 2009 Mohan et al,11 2009 64 35
Dorian et al,9 2014 Dorian et al,9 2014
Rate of change in treatment after bleeding event, % Hemorrhagic stroke 100.00 Other ICH 56.00 Gastrointestinal bleeding 25.00 Other major bleeding 25.00
Assumption Claassen et al,12 2008 Sorensen et al,13 2009 Sorensen et al,13 2009
Mortality of each event, % Systemic embolism MI (male) MI (female) Other ICH
9.40 10.80 15.60 13.00
Dorian et al,9 2014 Scarborough et al,14 2010 Scarborough 201014 Dorian et al,9 2014
2.00
Dorian et al,9 2014
Gastrointestinal bleeding or other major bleeding
CHADS2 ¼ congestive heart failure, hypertension, age of 75 years, diabetes mellitus, stroke; CRNM ¼ clinically relevant nonmajor bleeding; CV ¼ cardiovascular; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction. * Hazard ratio based on apixaban use.
The cost-effectiveness acceptability curve revealed that when the WTP was approximately Z¥2,250,000 (US $22,500) per QALY; the probability of apixaban being cost-effectiveness was Z50% (Figure 4).
6
DISCUSSION In this analysis, the cost-effectiveness of apixaban compared with warfarin in Japan was estimated by using Japanese data applied into a previously
Volume ] Number ]
I. Kamae et al.
Table IV. Distributions of the severity after a stroke. Distribution, % Parameter
Mild Moderate Severe Death (mRS Score, 0–2) (mRS Score, 3–4) (mRS Score, 5) (mRS Score, 6)
After ischemic stroke Apixaban Warfarin After hemorrhagic stroke Apixaban Warfarin
Source
53 45
21 30
8 10
18 15
Dorian et al,9 2014
23 20
32 15
10 12
35 53
Dorian et al,9 2014
mRS ¼ modified Rankin Scale.
published model.9,10 Compared with warfarin, treatment with apixaban earned life-years gained of 0.231 and QALYs gained of 0.240. The original study by Dorian et al reported that discounted life-years and QALYs gain were much the same (mainly due to the mortality from the most frequent event of ischemic stroke in the apixaban arm, which was slightly higher than that in warfarin arm [18% vs 15%]). Furthermore, the utility values of severe stroke in the Japanese model were significantly lower than those in the original model, which increased the gain in QALYs from stroke severity improvement by apixaban. These findings could be attributed as the cause of our results that the gain in QALYs was higher than the gain in life-years. Higher gain in QALYs by treatment with apixaban compared with gain in life-years suggests that apixaban has a greater impact on the improvement of patient’s quality of life than on his or her extension in life-years. The ICER for treatment with apixaban versus warfarin was ¥2,135,743 per QALY (US $21,357 per QALY). Although no specific ICER threshold has been set in Japan, according to Ohkusa et al and Shiroiwa et al, the WTP to gain 1 QALY in Japan is ¥6,700,000 per QALY (US $67,000 per QALY) and ¥5,000,000 (US $50,000 per QALY) per QALY, respectively.22,23 On the basis of the reported estimates of the WTP, apixaban is considered cost-effective by a payer’s perspective. Other than the payer’s perspective, whether this extra cost of apixaban for a marginal benefit from a new treatment is acceptable from a patient’s
] 2015
perspective may also be in question. A study by Hori et al reported that patients were willing to pay ¥819 (US $8.2) per day for an alternative drug that does not need monitoring or dietary restrictions, such as warfarin.24 The daily cost of apixaban is approximately ¥550 (US $5.5) in Japan, out of which the amount that a patient copays is ¥165 (US $1.7) (based on a 30% copayment requirement). Thus, a Japanese patient’s copayment amount of apixaban is lower than how much he or she is willing to pay. In the cost-effective analysis, we conducted a oneway sensitivity analysis that identified the risk of ischemic and unspecified stroke associated with the use of apixaban and the risk of ICH associated with the use of warfarin to largely affect the results. We also found that, at the higher at the higher end of the CI, the probability of apixaban being cost-effective with a WTP of ¥5,000,000 yen per QALY (US $50,000 per QALY) and ¥6,700,000 per QALY (US $67,000 per QALY) was 85% and 91%, respectively. Consistent with the pivotal ARISTOTLE trial,6 our analysis had warfarin only as the comparator. Studies from other countries have reported that apixaban was cost-effective when compared with other NOACs.10,25,26,27,28 These studies were based on the results estimated by indirect treatment comparison among the ARISTOTLE trial (apixaban 5 mg BID vs warfarin), the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial (dabigatran 110 mg BID vs dabigatran 150 mg BID vs warfarin), and the Rivaroxaban Once Daily Oral Direct Factor
7
Clinical Therapeutics
Table V. Definitions of target population and events for analysis from claims data. Target population
Definition of event Ischemic stroke Hemorrhagic stroke Other ICH Gastrointestinal bleeding
Patient with standard disease name, such as atrial fibrillation or chronic atrial fibrillation, nonvalvular atrial fibrillation, and nonvalvular paroxysmal atrial fibrillation. Atrial fibrillation with the following conditions were excluded*: rheumatic atrial fibrillation (ICD code I05–I09 [chronic rheumatic disease]) and mechanical valvular atrial fibrillation (T820 [mechanical complication of prosthetic cardiac valve]) ICD code I63 (cerebral infarction) ICD code I60 (subarachnoid hemorrhage) and I61 (intracerebral hemorrhage)
ICD code I62 (other nontraumatic intracranial hemorrhage) ICD code K250 (gastric ulcer, acute and with bleeding) K252 (gastric ulcer, acute and with bleeding and perforation) K260 (duodenal ulcer, acute and with bleeding) K262 (duodenal ulcer, acute and with bleeding and perforation) K270 (peptic ulcer at unknown site, acute and with bleeding) K272 (peptic ulcer at unknown site, acute and with bleeding and perforation) K280 (gastrojejunal ulcer, acute and with bleeding) K282 (gastrojejunal ulcer, acute and with bleeding and perforation) K290 (acute hemorrhagic gastritis) K920 (hematemesis) K921 (melena) K922 (gastrointestinal hemorrhage, details unknown) MI ICD code I21 (acute myocardial infarction) I23 (secondary complication of acute myocardial infarction) Systemic embolism ICD code I26 (as pulmonary embolism), I74 (embolism of artery and thrombosis), I81 (portal vein thrombosis), I82 (other embolism of vein and thrombosis) Other major All disease names including “bleeding” in the MDV data (excluding above intracranial bleeding or hemorrhage and gastrointestinal hemorrhage) CRNM bleeding Other Below ICD codes in MDV data (excluding above ischemic stroke, intracranial cardiovascular hemorrhage and systemic embolism) hospitalization I20–I25 (ischemic heart disease) I26–I28 (pulmonary heart disease and pulmonary circulation disease) I60–I69 (cerebrovascular disease) I70–I79 (artery, arteriole and capillary disease) I80–I89 (vein, lymphatic vessel and lymph node disease, not categorized in other definition) G45 (transient ischemic attack and related syndrome)
CRNM ¼ clinically relevant nonmajor bleeding; ICD ¼ International Classification of Diseases; ICH ¼ intracranial hemorrhage; MDV ¼ Medical Data Vision Co Ltd; MI ¼ myocardial infarction; mRS ¼ modified Rankin Scale. * Patients under dialysis or cancer (these patients presents a great impact on the total medical cost).
8
Volume ] Number ]
I. Kamae et al.
Table VI. Antithrombotic drug costs.
Apixaban Aspirin Warfarin
Tablet Size, mg
Cost per Tablet, ¥
Mean Daily Dose, mg
No. of Tablets per Day
Mean Daily Cost, ¥
5 100 1
272.8 10.0 10.0
10 100 5
2 1 5
545.6 10.0 50.0
Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) trial (rivaroxaban 20 mg once daily vs warfarin).6,29,30 The RE-LY trial and ROCKET-AF trial have different patient backgrounds and study designs (the ROCKET-AF trial targets a high-risk population, the RE-LY trial is open label, but the ROCKET-AF and ARISTOTLE trials are double-blind comparative trials), and the study dose of rivaroxaban in the ROCKET-AF trial is different from the Japanese approved dose (15 mg once daily). In addition, because all these NOACs have the same prices, it would be likely to have no incremental cost between these NOACs in the Japanese setting of clinical practice. Hence, inclusion of the other NOACs was not adapted to comply with the standard approach of cost-effectiveness analysis. As with all cost studies, our study presents several limitations worth mentioning. First, we did not use Japan-specific utility data except for chronic phase stroke because of a lack of utility data for some event states.16 In addition, as mentioned above, the utility values of severe stroke in the Japanese model were significantly lower than those in the original model, but the reason for the difference is unknown because of a lack of data, such as a country comparative study using the EQ-5D for stroke. Sakthong et al reported that the difference in utility values of type 2 diabetes mellitus was observed between the preference weights in the United Kingdom, United States, and Japan, revealing higher scores in the United States than in the United Kingdom and Japan.31 The Japanese researchers Shiroiwa et al also reported that there was a difference between the utility scores for the general population in Japan and the other countries.32 Those findings, however, come from general or mild health condition of patients. It is conjectured that, specific to countries, the utility scores differ in severe
] 2015
conditions more largely than in mild ones. Therefore, the issue of discrepancy between the Japanese and original utility values in the case of severe stroke is still left for further investigations. However, in our sensitivity analysis, we saw only a small impact resulting from a change in the utility parameters. The highest ICER in the sensitivity analysis on utility parameters was ¥2,763,056 per QALY (US $27,631 per QALY) when the utility decrement of warfarin parameter was set to the least favorable for apixaban, that is, with no utility decrement for warfarin (warfarin disutility of 0). We therefore believe that using utility data from overseas would not significantly reverse our conclusion. Second, the claims data provided by the MDV, which was used as the data source of the acute-phase cost for each event, were based on the actual treatment data from 131 hospitals. The MDV database represents only a proportion (9%) of the 1496 hospitals that have adopted the DPC system in Japan. On the other hand, the patient compositions listed by the major diagnostic categories in the MDV data are similar to the 2012 Discharged Patient Survey by the Ministry of Health, Labour and Welfare.15 Therefore, it is reasonable to assume that the MDV claims data are representative of the nationwide composition of hospitalized patients and related costs.33 Moreover, we found similarities in the literature that support our results. On estimating the acute-phase cost of stroke and MI, Toyonaga et al counted the cost of stroke in 171 patients admitted to several laborer hospitals in 2005 and reported an acute-phase cost of approximately ¥1,300,000 (US $13,000) for ischemic stroke, ¥1,500,000 (US $15,000) for intracerebral hemorrhage, and ¥3,800,000 (US $38,000) for subarachnoid hemorrhage.33 These costs were similar to our results of ¥1,401,931 (US $14,019) for acute-phase ischemic stroke and ¥1,697,332 (US $16,973) for
9
Parameter Monitoring visit (warfarin only) Routine care visit Mild ischemic stroke acute care (per episode) Mild ischemic stroke maintenance care (per month) Moderate ischemic stroke acute care (per episode) Moderate ischemic stroke chronic care (per month) Severe ischemic stroke acute care (per episode) Severe ischemic stroke chronic care (per month) Fatal ischemic stroke (per episode) Mild hemorrhagic stroke acute care (per episode) Mild hemorrhagic stroke chronic care (per month) Moderate hemorrhagic stroke acute care (per episode) Moderate hemorrhagic stroke chronic care (per month) Severe hemorrhagic stroke acute care (per episode) Severe hemorrhagic stroke chronic care (per month) Fatal hemorrhagic stroke (per episode) Systemic embolism acute care (per episode)
Volume ] Number ]
Other ICH (excluding hemorrhagic stroke) Gastrointestinal bleeding Other major bleeding or CRNM bleeding MI acute care (per episode) MI chronic care (per month) Other CV hospitalization
Cost, Mean (95% CI) ¥
1,024,406 146,911 1,699,225 335,129 1,817,620 383,444 983,728 1,448,758 157,117 2,131,870 336,541 2,085,764 383,444 902,966
598,077 620,341 672,665 2,156,290 75,036 1,109,681
Distribution
SE, ¥
2120 720 (961,815–1,088,941) (138,299–155,780) (1,596,349–1,805,267) (319,427–351,203) (1,698,097–1,941,151) (361,995–405,502) (870,539–1,103,735) (1,140,846–1,792,900) (147,610–166,917) (1,775,254–2,520,640) (320,626–352,837) (1,576,334–2,665,429) (361,995–405,502) (615,738–1,244,312) 1,123,451
γ γ γ γ γ γ γ γ γ γ γ γ γ γ γ
32,434 4460 53,301 8107 62,011 11,100 59,509 166,536 4926 190,286 8218 278,269 11,100 160,798 83,844
(515,908–686,230) (556,760–687,311) (567,485–786,656) (1,940,641–2,383,138) (71,981–78,153) (1,050,043–1,170,944)
γ γ γ γ γ γ
43,470 33,313 55,946 112,911 1575 30,845
CRNM ¼ clinically relevant nonmajor; CV ¼ cardiovascular; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction.
Source Medical treatment fee Medical treatment fee Claims data Hattori 201216 Claims data Hattori et al,16 2012 Claims data Hattori et al,16 2012 Claims data Claims data Hattori et al,16 2012 Claims data Hattori et al,16 2012 Claims data Hattori et al,16 2012 Claims data Tanihata et al,18 2006 Evans et al,19 2007 Claims data Claims data Claims data Claims data Claims data Claims data
Clinical Therapeutics
10 Table VII. Cost parameters.
I. Kamae et al.
Table VIII. Utility parameters. Parameter Utility value Mild ischemic stroke Moderate ischemic stroke Severe ischemic stroke Mild hemorrhagic stroke Moderate hemorrhagic stroke Severe hemorrhagic stroke Systemic embolism Utility decrement Other ICH
Mean (95% CI)
Distribution
SE
(0.758–0.809) (0.426–0.460) (0.070–0.117) (0.746–0.795) (0.414–0.448) (0.070–0.117) (0.589–0.664)
β β β β β β β
0.013 0.009 0.012 0.013 0.009 0.012 0.019
Hattori Hattori Hattori Hattori Hattori Hattori Hattori
0.151 (0.081–0.238)
β
0.040
Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014
0.784 0.443 0.092 0.771 0.431 0.092 0.627
Gastrointestinal bleeding or other major bleeding CRNM bleeding
0.151 (0.081–0.238)
β
0.040
0.058 (0.029–0.096)
β
0.017
MI
0.610 (0.572–0.647)
β
0.019
Other CV hospitalization
0.128 (0.081–0.182)
β
0.026
Utility decrement due to anticoagulation treatment Aspirin 0.002 (0.000–0.040)
β
0.000
Aspirin (second line)
0.002 (0.000–0.040)
β
0.000
Warfarin
0.013 (0.000–0.080)
β
0.000
Source et et et et et et et
al,16 al,16 al,16 al,16 al,16 al,16 al,16
2012 2012 2012 2012 2012 2012 2012
Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014
CRNM ¼ clinically relevant nonmajor bleeding; CV ¼ cardiovascular; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction.
hemorrhagic stroke (intracerebral hemorrhage: ¥1,599,116 [US $15,991]; subarachnoid hemorrhage: ¥2,614,010 [US $26,140]). Evans et al also reported the cost for the acute phase of MI based on claims data analysis, including 1748 admissions for MI to several hospitals, as approximately ¥2,200,000 (US $22,000),19 which is almost the same as the cost for acute-phase MI of ¥2,156,290 (US $21,563) calculated in our analysis. Despite the limitations we discussed, the impact of these limitations on our results was considered small,
] 2015
and our results were deemed robust because of the additional effect in Japanese patients compared with that in the global population. In conclusion, based on an adaptation of a published Markov model, apixaban was a cost-effective alternative to warfarin in Japanese NVAF patients by lowering risks of clinical events, such as stroke and bleeding and extending QALYs. Further research extrapolating the cost-effectiveness of apixaban versus warfarin with more representative and larger Japanspecific real-world data or trial-based clinical data,
11
Clinical Therapeutics
Table IX. Results of base case analysis.
No. of events (per 1000 patients) Stroke and systemic embolism* Major bleeding, including hemorrhagic stroke* CRNM bleeding MI Other CV hospitalization Other treatment discontinuation Outcomes (per patient) Life-years QALYs Cost, ¥ ICER, ¥ (per QALYs) Discounted by 2% (base case) Undiscounted Discounted by 5%
Warfarin
Apixaban
Difference
358 263 354 94 1250 651
341 220 312 92 1283 656
17 43 42 2 33 5
10.285 7.23 5,144,019
10.516 7.469 5,655,712
0.231 0.24 511,692 2,135,743 1,966,615 2,393,753
CRNMB ¼ clinically relevant nonmajor; CV ¼ cardiovascular; ICER ¼ incremental cost-effectiveness ratio; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction; QALYs: quality-adjusted life-years. * Hemorrhagic strokes are considered in both categories.
2,135,743 Risk of ischemic and unspecified stroke for apixaban (Rate/100 PYs) (0.56 - 1.52)
5,116,598
987,444
Risk of intracranial hemorrhage for warfarin (Rate/100 PYs) (0.35 - 1.87)
832,334
Risk of cardiovascular hospitalization for apixaban (Rate/100 PYs) (5.98 - 16.17)
3,819,575
1,181,399
Risk of other treatment discontinuations for warfarin (Rate/100 PYs) (8.23 - 22.27)
3,350,447
1,307,317
Risk of other treatment discontinuations for apixaban (Rate/100 PYs) (7.53 - 20.38)
3,386,241 3,534,145
1,523,386
Risk of cardiovascular hospitalization for warfarin (Rate/100 PYs) (5.98 - 16.17)
1,030,914
Utility decrement: Warfarin (0.00 - 0.08)
984,165
Risk of ischemic and unspecified stroke for warfarin (Rate/100 PYs) (0.87 - 1.33)
3,010,564 2,763,056
1,426,808
Risk of intracranial hemorrhage for apixaban (Rate/100 PYs) (0.19 - 0.51)
2,964,105
1,705,000
Rate of death comparator trial period (2.71 - 4.04)
2,878,554
1,749,743
Rate of death apixaban trial period (2.50 - 3.72)
1,798,718
Risk of MI for apixaban (Rate/100 PYs) (0.30 - 0.82)
1,765,876
Mean age for males (63% - 77%)
2,872,675 2,883,384 2,762,285 2,439,805
1,816,145 0
1,000,000
2,000,000
3,000,000 4,000,000 ICER (¥ per QALY)
5,000,000
6,000,000
Figure 2. Results of 1-way sensitivity analysis (Tornado diagram). The analysis identified the following factors as having a large impact on the cost-effectiveness of apixaban: the risk of ischemic and unspecified stroke associated with the use of apixaban and the risk of intracranial hemorrhage (ICH) associated with the use of warfarin. ICER = incremental cost-effectiveness ratio; MI = myocardial infarction; PYs = patient-years; QALYs: quality-adjusted life-years. 12
Volume ] Number ]
I. Kamae et al.
2,000,000
Incremental Costs (¥)
1,500,000
1,000,000
500,000
0 -0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
-500,000
-1,000,000
Incremental QALYs
Figure 3. Results of probabilistic sensitivity analysis (scatterplot). QALYs = quality-adjusted life-years.
Apixaban Warfarin Probability for Being the Most Cost-effective Treatment Choice (%)
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
Willingness to Pay (Million ¥ per QALY)
Figure 4. Results of probabilistic sensitivity analysis (cost-effectiveness acceptability curve). The costeffectiveness acceptability curve revealed that when the willingness to pay was approximately Z¥2,250,000 (US $22,500) per quality-adjusted life-year QALY, the probability of apixaban being cost-effectiveness was Z50%.
] 2015
13
Clinical Therapeutics resource utilization, and utility data whenever available would be warranted.
ACKNOWLEDGMENT This research has been funded exclusively by BristolMyers K.K. and Pfizer Japan Inc. Tatsunori Murata is an employee of Crecon Medical Assessment Inc who received funding from Bristol-Myers K.K. and Pfizer Japan Inc for writing of the manuscript. Drs. Kamae, Hashimoto, Koretsune, Tanahashi, Phatak, Liu, Wang and Okuyama provided interpretation of this research and revised the manuscript. Mr. Murata performed all analyses and drafted the manuscript. Dr. Tang drafted concept of this research and revised the manuscript. All authors gave final approval of the manuscript.
CONFLICTS OF INTEREST Dr Kamae has received lecture fees from Bristol-Myers K.K. and a consultant fee from Crecon Medical Assessment Inc. Dr Hashimoto has received lecture fees from Bristol-Myers K.K., Pfizer Japan Inc, Bayer Yakuhin Ltd, Daiichi-Sankyo Company Ltd, and Nippon Boehringer Ingelheim Co Ltd. Dr Koretsune has received lecture fees from Nippon Boehringer Ingelheim Co Ltd, Daiichi Sankyo Company Ltd, Bayer Yakuhin Ltd, and Bristol Myers K.K. and has received research grants from Nippon Boehringer Ingelheim Co Ltd and Daiichi Sankyo Company Ltd. Dr Tanahashi has received lecture fees from Bayer Yakuhin Ltd, Sanofi, Mitsubishi Tanabe Pharma, Nippon Boehringer Ingelheim Co Ltd, and BristolMyers K.K. and has received research grants from Bayer Yakuhin Ltd, Sanofi, Astrazeneca, Nippon Boehringer Ingelheim Co Ltd, Mitsubishi Tanabe Pharma, Takeda, Daiichi-Sankyo Company Ltd, and Bristol-Myers K.K. Dr Okumura has received lecture fees from Nippon Boehringer Ingelheim Co Ltd, Bayer Healthcare, Bristol-Myers K.K., Pfizer Japan Inc, and Eisai and has received research grants from Nippon Boehringer Ingelheim Co Ltd, Bayer Healthcare, and Daiichi-Sankyo Company Ltd. Mr Murata is an employee of Crecon Medical Assessment Inc and a consultant for Bristol-Myers K.K. Drs Wang and Liu are paid employees of Bristol-Myers Squibb or Pfizer Inc and/or own respective company’s stock. At the time of this research, Dr Phatak was an employee of Bristol-Myers Squibb and owned company stocks. Dr Tang was an employee of Bristol-Myers K.K. during the work. The authors have indicated that
14
they have no other conflicts of interest regarding the content of this article.
REFERENCES 1. Lip GY, Brechin CM, Lane DA. The global burden of atrial fibrillation and stroke: a systematic review of epidemiology of atrial fibrillation in regions outside North America and Europe. Chest. 2012;142:1489–1498. 2. Tomita F, Kohya T, Sakurai M, et al, Hokkaido Atrial Fibrillation Study Group. Prevalence and clinical characteristics of patients with atrial fibrillation: analysis of 20,000 cases in Japan. Jpn Circ J. 2000;64:653–658. 3. Wolf PA, Dawber TR, Thomas HE Jr, Kannel WB. Epidemiologic assessment of chronic atrial fibrillation and risk of stroke: the Framingham study. Neurology. 1978;28:973–977. 4. Hayashi T, Hayashi T, Hayashi T. Why most of patients on warfarin have not been anticoagulated to the therapeutic range in Japan? Prog Med. 2009;29:1599–1605. 5. Camm AJ1, Lip GY, De Caterina R, Savelieva I, et al, ESC Committee for Practice Guidelines (CPG). 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J. 2012;33:2719–2747. 6. Granger CB, Alexander JH, McMurray JJ, et al, ARISTOTLE Committees and Investigators. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981–992. 7. Ministry of Health, Labour and Welfare. 2011 Estimates of National Medical Care Expenditure (http://www.mhlw. go.jp/toukei/saikin/hw/k-iryohi/11/index.html). Last accessed at 2014. 8. Yhiroyuki Y. Anticoagulation therapy for atrial fibrillation –from warfarin to new oral anticoagulants. Prog Med. 2013;33:975–982. 9. Dorian P, Kongnakorn T, Phatak H, et al. Costeffectiveness of apixaban vs. current standard of care for stroke prevention in patients with atrial fibrillation. Eur Heart J. 2014;35:1897–1906. 10. Lip GY, Kongnakorn T, Phatak H, et al. Cost-effectiveness of apixaban versus other new oral anticoagulants for stroke prevention in atrial fibrillation. Clin Ther. 2014;36:192–210. 11. Mohan KM, Crichton SL, Grieve AP, et al. Frequency and predictors for the risk of stroke recurrence up to 10 years after stroke: the South London Stroke Register. J Neurol Neurosurg Psychiatry. 2009;80:1012–1018. 12. Claassen DO, Kazemi N, Zubkov AY, et al. Restarting anticoagulation therapy after warfarin-associated intracerebral hemorrhage. Arch Neurol. 2008;65:1313–1318. 13. Sorensen SV, Dewilde S, Singer DE, et al. Cost-effectiveness of warfarin: trial versus “real-world” stroke prevention in atrial fibrillation. Am Heart J. 2009;157:1064–1073.
Volume ] Number ]
I. Kamae et al. 14. Scarborough P, Bhatnagar P, Wickramasinghe K, et al.British Health Foundation Health Promotion research group. Coronary Heart Disease statistics 2010 edition. Department of Public Health University of Oxford, 2010:157. 15. 2012 Discharged patient survey ( http://www.mhlw. go.jp/stf/ shingi/0000023522.html). Last accessed at 2014. 16. Hattori N, Hirayama T, Katayama Y. Medical care for chronic-phase stroke in Japan. Neurol Med Chir (Tokyo). 2012;52:175–180. 17. Kobayashi S, Ogushi Y. Stroke Data Bank 2009. Tokyo: Nakayama Shoten, 2009:34–35. 18. Tanihata S, Nishigaki K, Kawasaki M, et al. Outcomes of patients with stable low-risk coronary artery disease receiving medical- and PCI-preceding therapies in Japan: J-SAP study 1-1. Circ J. 2006;70: 365–369. 19. Evans E, Imanaka Y, Sekimoto M, et al. Risk adjusted resource utilization for AMI patients treated in Japanese hospitals. Health Econ. 2007;16:347–359. 20. Ministry of Health, Labour and Welfare. Abridged Life Table (http://www. mhlw.go.jp/toukei/saikin/hw/life/ life12/). Last accessed at 25 Aug 2014. 21. Guideline for analysis method in health economic evaluation ver. 1.0. Mar 29, 2013 (http://hta. umin.jp/guideline_j.pdf). Last accessed at 25 Aug 2014. 22. Ohkusa Y, Sugawara T. Research for Willingness to Pay for One QALY Gain. J Health Care Soc. 2006;16:157–165. 23. Shiroiwa T, Sung YK, Fukuda T, et al. International survey on willingness-to-pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health Econ. 2010;19:422–437. 24. Hori M, Koretsune Y, Yasaka M, et al. Health economic evaluation of Dabigatran Etexilate for prevention of stroke in patients with
] 2015
25.
26.
27.
28.
nonvalvular atrial fibrillation. Pharma Medica. 2011;29:151–164. Lanitis T, Cotté FE, Gaudin AF, et al. Stroke prevention in patients with atrial fibrillation in France: comparative cost-effectiveness of new oral anticoagulants (apixaban, dabigatran, and rivaroxaban), warfarin, and aspirin. J Med Econ. 2014;17:587–598. Pink J, Pirmohamed M, Lane S, Hughes DA. Cost-effectiveness of pharmacogenetics-guided warfarin therapy vs. alternative anticoagulation in atrial fibrillation. Clin Pharmacol Ther. 2014;95:199–207. Coyle D, Coyle K, Cameron C, et al. Cost-effectiveness of new oral anticoagulants compared with warfarin in preventing stroke and other cardiovascular events in patients with atrial fibrillation. Value Health. 2013;16:498–506. Harrington AR, Armstrong EP, Nolan PE Jr, Malone DC. Costeffectiveness of apixaban, dabigatran, rivaroxaban, and warfarin for stroke prevention in atrial fibrillation. Stroke. 2013;44:1676–1681.
29. Connolly SJ, Ezekowitz MD, Yusuf S, et al, RE-LY Steering Committee and Investigators. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009; 361:1139–1151. 30. Patel MR, Mahaffey KW, Garg J, et al, ROCKET AF Investigators. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365: 883–891. 31. Sakthong P, Charoenvisuthiwongs R, Shabunthom R. A comparison of EQ-5D index scores using the UK, US, and Japan preference weights in a Thai sample with type 2 diabetes. Health Qual Life Outcomes. 2008;6:71. 32. Shiroiwa T, Fukuda T, Ikeda S, et al. Japanese population norms for preference-based measures: EQ5D-3L, EQ-5D-5L, and SF-6D. Qual Life Res. 2015. 33. Toyonaga T. Factors affecting the cost for medical care and rehabilitation in stroke patients. Jpn J Occup Med Trauma. 2006;54: 175–182.
Address correspondence to: Tatsunori Murata, MS, Crecon Medical Assessment Inc, 12-15 Shibuya 2-chome, Shibuya-ku, Tokyo, Japan. E-mail: [email protected]
15
Cost-effectiveness Analysis of Apixaban against Warfarin for Stroke Prevention in Patients with Nonvalvular Atrial Fibrillation in Japan Isao Kamae, MD, PhD1; Yoichiro Hashimoto, MD2; Yukihiro Koretsune, MD, PhD3; Norio Tanahashi, MD4; Tatsunori Murata, MS5; Hemant Phatak, PhD6; Larry Z. Liu, MD, PhD7; Ann C. Tang, PhD8; Peter Feng Wang, MD, PhD9; and Ken Okumura, MD, PhD10 1
The University of Tokyo, Tokyo, Japan; 2Kumamoto City Hospital, Kumamoto, Japan; 3National Hospital Organization, Osaka National Hospital, Osaka, Japan; 4Saitama International Medical Center, Saitama, Japan; 5Crecon Medical Assessment Inc, Tokyo, Japan; 6US Health Economics and Outcomes Research, EMD Serono, Rockland, MA; 7Global Health Economics and Outcomes Research and Evidence, Pfizer Inc, and Weill Medical College of Cornell University, New York, New York, USA; 8Health Economics and Outcomes Research, Bristol-Myers K.K., Tokyo, Japan; 9Worldwide Health Economics and Outcomes Research, Bristol Myers Squibb, Princeton, New Jersey; and 10Department of Cardiology, Hirosaki University Graduate School of Medicine, Aomori, Japan
ABSTRACT Purpose: The aim of this study was to evaluate the cost-effectiveness of apixaban compared with to warfarin, current standard of care, for stroke prevention in patients with nonvalvular atrial fibrillation (NVAF) in Japan. Methods: A previously published lifetime Markov model was adapted to evaluate the cost-effectiveness of apixaban compared with warfarin in patients with NVAF in Japan. In the same model, the costs associated with each clinical event and background mortality were replaced with Japanese data. Whenever available, some of the utility parameters were derived from Japanese published literature. Lifetime horizon was selected to evaluate the value of the treatment benefit (stroke prevention) against potential risks (such as major bleedings) among patients with NVAF. Direct medical cost, long-term care cost, and qualityadjusted life years (QALYs) were calculated from the payers’ perspective. Findings: Compared with warfarin, treatment with apixaban was estimated to increase life expectancy by 0.231 year or 0.240 QALYs while treatment cost increased by ¥511,692 (US $5117 at an exchange rate of US $1 ¼ ¥100). The incremental cost-effectiveness ratio was ¥2,135,743 per QALY (US $21,357 per QALY). On the basis of the results of the probabilistic
] 2015
sensitivity analysis, when the willingness-to-pay threshold was set at approximately Z¥2,250,000 (US $22,500) per QALY, the probability of apixaban being cost-effective was Z50%. Assuming a willingness-to-pay threshold of ¥5,000,000 (US $50,000) and ¥6,700,000 (US $67,000) in Japan, the probability of apixaban being cost-effective was 85% and 91%, respectively. Conclusion: Although most participants in the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial used for the efficacy data of apixaban in the model were non-Japanese patients, the impact of the limitations on our results was considered small, and our results were deemed robust because of the additional effect in Japanese patients compared with that in the global population according to the subanalysis of Japanese patients in the trial. Therefore, based on an adaptation of a published Markov model, apixaban is a cost-effective alternative to warfarin in Japan for stroke prevention among patients with NVAF.
Accepted for publication October 5, 2015. http://dx.doi.org/10.1016/j.clinthera.2015.10.007 0149-2918/$ - see front matter & 2015 Elsevier HS Journals, Inc. All rights reserved.
1
Clinical Therapeutics (Clin Ther. 2015;]:]]]–]]]) & 2015 Elsevier HS Journals, Inc. All rights reserved. Key words: apixaban, cost-effectiveness, nonvalvular atrial fibrillation, Japan.
INTRODUCTION
Atrial fibrillation (AF) affects 0.6% to 1.6% of the general population and up to 14% of patients in cardiovascular clinics in Japan.1,2 AF increases the risk of stroke by nearly 5-fold.3 Warfarin has been used to prevent the occurrence of stroke, a complication of AF. However, monitoring of the prothrombin time–international normalized ratio is necessary during warfarin use, and it is not easy to maintain the dose of the drug within the desired therapeutic range.4 Use of warfarin can be associated with hemorrhagic adverse effects, including severe and life-threatening events, such as intracerebral hemorrhage. In addition, warfarin is complicated by challenges such as drug-drug interactions and restriction of foods, including vitamin K. Recent guidelines from the European Society of Cardiology recommend the use of novel anticoagulants (NOACs) as alternatives to warfarin or antiplatelet therapy in most AF patients who require stroke prevention.5 Apixaban, an orally active factor Xa inhibitor, has been studied in a Phase III global randomized clinical trial (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation [ARISTOTLE]) versus dose-adjusted warfarin in patients with nonvalvular atrial fibrillation (NVAF). The study found that the incidence of stroke and systemic embolism decreased significantly and the frequency of major bleeding also decreased significantly in the apixaban 5 mg BID group compared with the warfarin control group.6 NOACs, such as dabigatran, rivaroxaban, and apixaban, are currently available for clinical use for the prevention of stroke in patients with NVAF in Japan. According to the 2011 Estimates of National Medical Care Expenditure report, the medical cost of cerebrovascular diseases, including ischemic and hemorrhagic stroke, was ¥1.8 trillion (US $18 billion at an exchange rate of US $1 ¼¥100), which accounted for approximately 6% of the overall medical cost (¥28 trillion [US $280 billion]).7 Prevention of stroke is important not only clinically but also from the perspective of health economics.
2
High drug acquisition cost of NOACs versus warfarin in Japan warrants such analyses to be conducted from the Japanese payer perspective to inform selection of optimal anticoagulation therapy for NVAF patients. Warfarin is currently still considered the standard of care, although NOACs are becoming more widely used in Japan.8 In this study, we evaluated the costeffectiveness of apixaban compared with warfarin based on Japanese data.
METHODS Model In this study, a previously published model for health economic evaluation9,10 (Figure 1) was adapted to evaluate the cost-effectiveness of apixaban compared with warfarin in Japan. The model was a Markov model developed using Microsoft Excel, and a 6-week cycle length was chosen to capture the possibilities of events related to AF that occur within a short period. The detailed description of the model can be found in the publications by Dorian et al.9 In brief, all patients started in the NVAF state, and certain portions of the cohort were assigned during each model cycle to one of the modeled 17 health states. The health states comprised NVAF, ischemic stroke (mild, moderate, severe), hemorrhagic stroke (mild, moderate, severe), recurrent ischemic stroke (mild, moderate, severe), recurrent hemorrhagic stroke (mild, moderate, severe), systemic embolism, gastrointestinal bleeding and myocardial infarction (MI), and death. Intracranial hemorrhage (ICH) that is not hemorrhagic stroke (other ICH), major bleeding other than gastrointestinal bleeding, clinically relevant nonmajor bleeding, and other cardiovascular hospitalization were also considered as events. Patients who experience these events were assumed to stay in their initially assigned anticoagulation state, and events were modeled as transient states. These assignments were based on the probability of experiencing an event that in turn was derived from secondary analyses of trial data9 (Tables I–IV). Some events altered subsequent risks of modeled events, including death (eg, patients with ischemic stroke would have higher probability of death or patients experiencing MI would have greater likelihood of acute mortality). In addition, stroke severity was considered. Stroke severity was evaluated by the modified Rankin Scale (mRS) as mild (mRS scores
Volume ] Number ]
I. Kamae et al.
NVAF
No Change
Systemic Embolism
Mild IS
NVAF Nonfatal
Fatal
Systemic embolism
Death
Moderate IS Mild Nonfatal Ischemic Hemorrhagic
Severe IS
Moderate
Mild IS Moderate IS
Severe Fatal
Severe IS Death Mild
Warfarin
Dabigatran 110 mg
Dabigatran 150 mg
Rivaroxaban
M
Hemorrhagic Stroke
Mild HS
Mild HS
Moderate
Moderate HS
Severe Severe HS
M
ICH Moderate HS
M
Nonfatal
Fatal
Death
NVAF
Tmt Interruption (6wk )
M
Other ICH
Nonfatal Discontinue AC Fatal
M
NVAF NVAF subsequent ASA
Death
Severe HS Apixaban
Bleeding M
Major Bleed
Nonfatal
Discontinue AC Fatal
Systemic embolism CRNM Bleed
Stay on current AC
Death
Nonfatal Myocardial infarction Fatal Other Deaths Other AC Discontinuation
NVAF subsequent ASA
NVAF
Myocardial infarction
NVAF subsequent ASA
NVAF
Myocardial infarction
Death
Death NVAF subsequent ASA
Death
Figure 1. Markov model. The model uses a previously published 6-week cycle model.9,10 All patients start with the nonvalvular atrial fibrillation (NVAF) state, and certain portions of the cohort are assigned during each model cycle to 1 of the modeled 17 health states. The health states comprise ischemic stroke (IS) (mild, moderate, severe), hemorrhagic stroke (HS) (mild, moderate, severe), recurrent IS (mild, moderate, severe), recurrent HS (mild, moderate, severe), systemic embolism, gastrointestinal bleeding and myocardial infarction (MI), or death. Other intracranial hemorrhage (ICH), major bleeding other than gastrointestinal bleeding, clinically relevant nonmajor (CRNM) bleeding, and other cardiovascular hospitalization are also considered events, but patients are assumed to stay in their initially assigned anticoagulation states and the events are modeled as transient states. Some events altered the subsequent risks of modeled events, including death. AC = anticoagulant; ASA = aspirin. M indicates Markov process, and triangles indicate that a patient transitions to the following state in the next cycle.
0–2), moderate (mRS scores 3–4), and severe (mRS score 5; an mRS score of 6 is death). Recurrence was also considered for stroke. For treatment after each event, patients were assumed to continue the first-line treatment (apixaban or warfarin) after ischemic stroke, systemic embolism, MI, clinically relevant nonmajor bleeding, and other cardiovascular hospitalization. However, for ICH (including hemorrhagic stroke), gastrointestinal bleeding, and other major bleeding, patients were assumed to discontinue or suspend first-line treatment for 6 weeks. Patients who discontinued first-line treatment were assumed
] 2015
to start second-line aspirin treatment. In essence, model structure was built to reflect real-life treatment paradigms and consequences of treatment decision.
Localization In the adaptation analysis, we used parameters such as cost of each event and some of the utility parameters and background mortality using the Japanese data. The cost of the acute phase for each event were obtained from Japanese claims data for patients with AF (excluding AF associated with rheumatic mitral valve disease and mechanical valve) using the
3
Clinical Therapeutics
Table I. Distributions of the cTTR and CHADS2 scores used in the model. Parameter cTTR, % o52.38 52.38–66.02 66.02–76.51 Z76.51 CHADS2 0 1 2 3 4 5 6
Setting, %
Source
25.00 25.00 25.00 25.00
Dorian et al,9 2014
0.60 33.40 35.80 18.00 8.70 3.10 0.40
Dorian et al,9 2014
CHADS2 ¼ congestive heart failure, hypertension, age of 75 years, diabetes mellitus, stroke; cTTR ¼ center’s median time in therapeutic range.
Medical Data Vision Co Ltd (MDV) database. The database provided claims data from 131 hospitals (as of April 2013) using the Diagnosis Procedure Combination (DPC) system for medical service claims (21% of general hospitals but 55% of general beds in Japan is under the DPC system in 201415). DPC is a reimbursement system for hospitals to claim a treatment cost comprehensively estimated for one hospitalization per patient. The actual costs per event from admission until discharge (including death; ie, mRS score of 6) with a name of each event in the record of “disease name with the most health resources invested” in DPC data were used to estimate costs associated with the acute phase of each event. The definitions of the target population and events are given in Table V. The costs of chronic stroke were obtained from the study by Hattori et al, which included not only direct medical cost but also the long-term care cost.16 The cardiogenic brain embolism data derived from Stroke Data Bank 2009 were used for distribution of the mRS scores for each stroke severity in Japanese patients.17 The cost of medication therapy and that of percutaneous coronary intervention for the
4
chronic phase of MI are quoted from the report of Tanihata et al.18 The percentage of patients receiving each treatment was obtained from a study using claims data analysis in Japan by Evans et al.19 Because the research by Tanihata et al was conducted in 2001, the cost derived from the study was adjusted to 2014 rates based on the revised total medical treatment fees thereafter (Tables VI and VII). Japanese utility data were used only for chronicphase stroke, for which Japanese evidence is available16 (Table VIII). Background mortality was set based on the 2012 Abridged Life Table according to age and sex in Japan.20
Analyses Lifetime horizon was chosen to evaluate the treatment value of the preventive effect against events such as stroke in NVAF patients. The direct medical cost, long-term care cost, and qualityadjusted life-years (QALYs) were calculated from the payers’ perspective. The number of events avoided per 1,000 patients with NVAF during the lifetime horizon was also evaluated. An annual discount of 2% was applied to both the cost and effectiveness in accordance with a Japanese guideline for health economic analysis.21 In this study, the cost-effectiveness of apixaban compared with warfarin was estimated in base case analysis. One-way sensitivity analysis for each parameter was conducted to confirm the impact of each parameter on the analysis results with the 95% CI for each parameter. In addition, probabilistic sensitivity analysis by Monte-Carlo simulation with 2000 iterations was performed, and parameter uncertainty was assessed by the cost-effectiveness acceptability curve, which plotted the probability that apixaban is cost-effective in willingness to pay (WTP) on the horizontal axis.
RESULTS Base Case Analysis The simulation results in NVAF patients treated with apixaban versus warfarin were projected to reduce the number of strokes by 16 per 1000 patients (2 for ischemic stroke and 14 for hemorrhagic stroke) and the number of bleeding episodes other than hemorrhagic stroke onset by 72 per 1000 patients. The expected life-years accrued during lifetime treatment per patient were 10.285 years for the warfarin
Volume ] Number ]
I. Kamae et al. Table II. Relative event risks for each cTTR used in the model.* Relative Event Risk cTTR
Apixaban
Warfarin
cTTR for stroke, % o52.38 0.92 52.38–66.02 1 66.02–76.51 0.69 Z76.51 0.56 cTTR for ICH, % o52.38 0.58 52.38–66.02 1 66.02–76.51 0.69 Z76.51 0.36 cTTR for gastrointestinal bleeding or other major bleeding, % o52.38 0.72 52.38–66.02 1 66.02–76.51 1.69 Z76.51 1.77 cTTR for CRNM bleeding, % o52.38 0.71 52.38–66.02 1 66.02–76.51 1.25 Z76.51 1.7
Source
1.54 1 0.84 0.72
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
1.05 1 0.68 0.78
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
0.84 1 1.13 1.37
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
0.99 1 1.26 1.27
Dorian Dorian Dorian Dorian
et et et et
al,9 al,9 al,9 al,9
2014 2014 2014 2014
CRNM ¼ clinically relevant nonmajor bleeding; cTTR ¼ center’s median time in therapeutic range; ICH ¼ intracranial hemorrhage. * From the parameter applied in the model used in this analysis, settings for relative risk according to the cTTR of each event are given. Event risk was set as the relative risk in each cTTR value based on a cTTR of 52.38% to 66.02%.
group and 10.516 years for the apixaban group. Therefore, apixaban was expected to extend 0.231 life-year and improve quality of life of patients, which generated 0.240 QALYs gained. As with many new technologies, innovation posts additional costs. The clinical improvement from apixaban compared with warfarin was expected to incur some additional costs to Japanese payers. During a lifetime horizon, total cost per patient was ¥5,144,019 (US $51,440) for warfarin and ¥5,655,712 (US $56,557) for apixaban, resulting in incremental cost of ¥511,692 (US $5117). The incremental cost-effectiveness ratio (ICER) of treatment with apixaban versus warfarin was estimated to be ¥2,135,743 (US $21,357) per QALY gained (Table IX).
] 2015
Sensitivity Analysis The results of the 1-way sensitivity analysis identified the following factors as having a large impact on the ICER: the risk of ischemic and unspecified stroke associated with the use of apixaban and the risk of ICH associated with the use of warfarin (Figure 2). The range of ICER in sensitivity analysis of these parameters were ¥987,444 to ¥5,116,598 per QALY (US $9,874 to $51,166 per QALY) for risk of ischemic and unspecified stroke for apixaban and ¥832,334 to ¥3,819,575 per QALY (US $8323 to 38,196 per QALY) for risk of ICH for warfarin. In the probabilistic sensitivity analysis, the cost-effectiveness plane (scatterplot) suggested apixaban was more effective and more costly than warfarin (Figure 3).
5
Clinical Therapeutics
Table III. Clinical event rates for each parameter in the model. Parameter
Apixaban
Warfarin
Event risk Ischemic stroke CHADS2 score of 0–1 CHADS2 score of 2 CHADS2 score of Z3 ICH Gastrointestinal bleeding or other major bleeding CRNM bleeding MI Systemic embolism Other CV hospitalization Withdrawal from treatment
person-years person-years person-years person-years person-years
2.08/100 person-years 0.53/100 person-years 0.09/100 person-years 10.46/100 person-years 13.18/100 person-years
Recurrence rate of stroke Ischemic stroke Hemorrhagic stroke
4.10/100 person-years 3.00/100 person-years
0.52/100 0.95/100 1.53/100 0.33/100 1.79/100
1.09*
Source
2.39* 1.27*
Dorian Dorian Dorian Dorian Dorian
et et et et et
al,9 al,9 al,9 al,9 al,9
2014 2014 2014 2014 2014
1.43* 1.14* 1.11* 1.00* 1.08*
Dorian Dorian Dorian Dorian Dorian
et et et et et
al,9 al,9 al,9 al,9 al,9
2014 2014 2014 2014 2014
Hemorrhagic stroke and gastrointestinal bleeding, % Hemorrhagic stroke out of ICH 77 Gastrointestinal bleeding out of 38 other major bleeding
Mohan et al,11 2009 Mohan et al,11 2009 64 35
Dorian et al,9 2014 Dorian et al,9 2014
Rate of change in treatment after bleeding event, % Hemorrhagic stroke 100.00 Other ICH 56.00 Gastrointestinal bleeding 25.00 Other major bleeding 25.00
Assumption Claassen et al,12 2008 Sorensen et al,13 2009 Sorensen et al,13 2009
Mortality of each event, % Systemic embolism MI (male) MI (female) Other ICH
9.40 10.80 15.60 13.00
Dorian et al,9 2014 Scarborough et al,14 2010 Scarborough 201014 Dorian et al,9 2014
2.00
Dorian et al,9 2014
Gastrointestinal bleeding or other major bleeding
CHADS2 ¼ congestive heart failure, hypertension, age of 75 years, diabetes mellitus, stroke; CRNM ¼ clinically relevant nonmajor bleeding; CV ¼ cardiovascular; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction. * Hazard ratio based on apixaban use.
The cost-effectiveness acceptability curve revealed that when the WTP was approximately Z¥2,250,000 (US $22,500) per QALY; the probability of apixaban being cost-effectiveness was Z50% (Figure 4).
6
DISCUSSION In this analysis, the cost-effectiveness of apixaban compared with warfarin in Japan was estimated by using Japanese data applied into a previously
Volume ] Number ]
I. Kamae et al.
Table IV. Distributions of the severity after a stroke. Distribution, % Parameter
Mild Moderate Severe Death (mRS Score, 0–2) (mRS Score, 3–4) (mRS Score, 5) (mRS Score, 6)
After ischemic stroke Apixaban Warfarin After hemorrhagic stroke Apixaban Warfarin
Source
53 45
21 30
8 10
18 15
Dorian et al,9 2014
23 20
32 15
10 12
35 53
Dorian et al,9 2014
mRS ¼ modified Rankin Scale.
published model.9,10 Compared with warfarin, treatment with apixaban earned life-years gained of 0.231 and QALYs gained of 0.240. The original study by Dorian et al reported that discounted life-years and QALYs gain were much the same (mainly due to the mortality from the most frequent event of ischemic stroke in the apixaban arm, which was slightly higher than that in warfarin arm [18% vs 15%]). Furthermore, the utility values of severe stroke in the Japanese model were significantly lower than those in the original model, which increased the gain in QALYs from stroke severity improvement by apixaban. These findings could be attributed as the cause of our results that the gain in QALYs was higher than the gain in life-years. Higher gain in QALYs by treatment with apixaban compared with gain in life-years suggests that apixaban has a greater impact on the improvement of patient’s quality of life than on his or her extension in life-years. The ICER for treatment with apixaban versus warfarin was ¥2,135,743 per QALY (US $21,357 per QALY). Although no specific ICER threshold has been set in Japan, according to Ohkusa et al and Shiroiwa et al, the WTP to gain 1 QALY in Japan is ¥6,700,000 per QALY (US $67,000 per QALY) and ¥5,000,000 (US $50,000 per QALY) per QALY, respectively.22,23 On the basis of the reported estimates of the WTP, apixaban is considered cost-effective by a payer’s perspective. Other than the payer’s perspective, whether this extra cost of apixaban for a marginal benefit from a new treatment is acceptable from a patient’s
] 2015
perspective may also be in question. A study by Hori et al reported that patients were willing to pay ¥819 (US $8.2) per day for an alternative drug that does not need monitoring or dietary restrictions, such as warfarin.24 The daily cost of apixaban is approximately ¥550 (US $5.5) in Japan, out of which the amount that a patient copays is ¥165 (US $1.7) (based on a 30% copayment requirement). Thus, a Japanese patient’s copayment amount of apixaban is lower than how much he or she is willing to pay. In the cost-effective analysis, we conducted a oneway sensitivity analysis that identified the risk of ischemic and unspecified stroke associated with the use of apixaban and the risk of ICH associated with the use of warfarin to largely affect the results. We also found that, at the higher at the higher end of the CI, the probability of apixaban being cost-effective with a WTP of ¥5,000,000 yen per QALY (US $50,000 per QALY) and ¥6,700,000 per QALY (US $67,000 per QALY) was 85% and 91%, respectively. Consistent with the pivotal ARISTOTLE trial,6 our analysis had warfarin only as the comparator. Studies from other countries have reported that apixaban was cost-effective when compared with other NOACs.10,25,26,27,28 These studies were based on the results estimated by indirect treatment comparison among the ARISTOTLE trial (apixaban 5 mg BID vs warfarin), the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial (dabigatran 110 mg BID vs dabigatran 150 mg BID vs warfarin), and the Rivaroxaban Once Daily Oral Direct Factor
7
Clinical Therapeutics
Table V. Definitions of target population and events for analysis from claims data. Target population
Definition of event Ischemic stroke Hemorrhagic stroke Other ICH Gastrointestinal bleeding
Patient with standard disease name, such as atrial fibrillation or chronic atrial fibrillation, nonvalvular atrial fibrillation, and nonvalvular paroxysmal atrial fibrillation. Atrial fibrillation with the following conditions were excluded*: rheumatic atrial fibrillation (ICD code I05–I09 [chronic rheumatic disease]) and mechanical valvular atrial fibrillation (T820 [mechanical complication of prosthetic cardiac valve]) ICD code I63 (cerebral infarction) ICD code I60 (subarachnoid hemorrhage) and I61 (intracerebral hemorrhage)
ICD code I62 (other nontraumatic intracranial hemorrhage) ICD code K250 (gastric ulcer, acute and with bleeding) K252 (gastric ulcer, acute and with bleeding and perforation) K260 (duodenal ulcer, acute and with bleeding) K262 (duodenal ulcer, acute and with bleeding and perforation) K270 (peptic ulcer at unknown site, acute and with bleeding) K272 (peptic ulcer at unknown site, acute and with bleeding and perforation) K280 (gastrojejunal ulcer, acute and with bleeding) K282 (gastrojejunal ulcer, acute and with bleeding and perforation) K290 (acute hemorrhagic gastritis) K920 (hematemesis) K921 (melena) K922 (gastrointestinal hemorrhage, details unknown) MI ICD code I21 (acute myocardial infarction) I23 (secondary complication of acute myocardial infarction) Systemic embolism ICD code I26 (as pulmonary embolism), I74 (embolism of artery and thrombosis), I81 (portal vein thrombosis), I82 (other embolism of vein and thrombosis) Other major All disease names including “bleeding” in the MDV data (excluding above intracranial bleeding or hemorrhage and gastrointestinal hemorrhage) CRNM bleeding Other Below ICD codes in MDV data (excluding above ischemic stroke, intracranial cardiovascular hemorrhage and systemic embolism) hospitalization I20–I25 (ischemic heart disease) I26–I28 (pulmonary heart disease and pulmonary circulation disease) I60–I69 (cerebrovascular disease) I70–I79 (artery, arteriole and capillary disease) I80–I89 (vein, lymphatic vessel and lymph node disease, not categorized in other definition) G45 (transient ischemic attack and related syndrome)
CRNM ¼ clinically relevant nonmajor bleeding; ICD ¼ International Classification of Diseases; ICH ¼ intracranial hemorrhage; MDV ¼ Medical Data Vision Co Ltd; MI ¼ myocardial infarction; mRS ¼ modified Rankin Scale. * Patients under dialysis or cancer (these patients presents a great impact on the total medical cost).
8
Volume ] Number ]
I. Kamae et al.
Table VI. Antithrombotic drug costs.
Apixaban Aspirin Warfarin
Tablet Size, mg
Cost per Tablet, ¥
Mean Daily Dose, mg
No. of Tablets per Day
Mean Daily Cost, ¥
5 100 1
272.8 10.0 10.0
10 100 5
2 1 5
545.6 10.0 50.0
Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) trial (rivaroxaban 20 mg once daily vs warfarin).6,29,30 The RE-LY trial and ROCKET-AF trial have different patient backgrounds and study designs (the ROCKET-AF trial targets a high-risk population, the RE-LY trial is open label, but the ROCKET-AF and ARISTOTLE trials are double-blind comparative trials), and the study dose of rivaroxaban in the ROCKET-AF trial is different from the Japanese approved dose (15 mg once daily). In addition, because all these NOACs have the same prices, it would be likely to have no incremental cost between these NOACs in the Japanese setting of clinical practice. Hence, inclusion of the other NOACs was not adapted to comply with the standard approach of cost-effectiveness analysis. As with all cost studies, our study presents several limitations worth mentioning. First, we did not use Japan-specific utility data except for chronic phase stroke because of a lack of utility data for some event states.16 In addition, as mentioned above, the utility values of severe stroke in the Japanese model were significantly lower than those in the original model, but the reason for the difference is unknown because of a lack of data, such as a country comparative study using the EQ-5D for stroke. Sakthong et al reported that the difference in utility values of type 2 diabetes mellitus was observed between the preference weights in the United Kingdom, United States, and Japan, revealing higher scores in the United States than in the United Kingdom and Japan.31 The Japanese researchers Shiroiwa et al also reported that there was a difference between the utility scores for the general population in Japan and the other countries.32 Those findings, however, come from general or mild health condition of patients. It is conjectured that, specific to countries, the utility scores differ in severe
] 2015
conditions more largely than in mild ones. Therefore, the issue of discrepancy between the Japanese and original utility values in the case of severe stroke is still left for further investigations. However, in our sensitivity analysis, we saw only a small impact resulting from a change in the utility parameters. The highest ICER in the sensitivity analysis on utility parameters was ¥2,763,056 per QALY (US $27,631 per QALY) when the utility decrement of warfarin parameter was set to the least favorable for apixaban, that is, with no utility decrement for warfarin (warfarin disutility of 0). We therefore believe that using utility data from overseas would not significantly reverse our conclusion. Second, the claims data provided by the MDV, which was used as the data source of the acute-phase cost for each event, were based on the actual treatment data from 131 hospitals. The MDV database represents only a proportion (9%) of the 1496 hospitals that have adopted the DPC system in Japan. On the other hand, the patient compositions listed by the major diagnostic categories in the MDV data are similar to the 2012 Discharged Patient Survey by the Ministry of Health, Labour and Welfare.15 Therefore, it is reasonable to assume that the MDV claims data are representative of the nationwide composition of hospitalized patients and related costs.33 Moreover, we found similarities in the literature that support our results. On estimating the acute-phase cost of stroke and MI, Toyonaga et al counted the cost of stroke in 171 patients admitted to several laborer hospitals in 2005 and reported an acute-phase cost of approximately ¥1,300,000 (US $13,000) for ischemic stroke, ¥1,500,000 (US $15,000) for intracerebral hemorrhage, and ¥3,800,000 (US $38,000) for subarachnoid hemorrhage.33 These costs were similar to our results of ¥1,401,931 (US $14,019) for acute-phase ischemic stroke and ¥1,697,332 (US $16,973) for
9
Parameter Monitoring visit (warfarin only) Routine care visit Mild ischemic stroke acute care (per episode) Mild ischemic stroke maintenance care (per month) Moderate ischemic stroke acute care (per episode) Moderate ischemic stroke chronic care (per month) Severe ischemic stroke acute care (per episode) Severe ischemic stroke chronic care (per month) Fatal ischemic stroke (per episode) Mild hemorrhagic stroke acute care (per episode) Mild hemorrhagic stroke chronic care (per month) Moderate hemorrhagic stroke acute care (per episode) Moderate hemorrhagic stroke chronic care (per month) Severe hemorrhagic stroke acute care (per episode) Severe hemorrhagic stroke chronic care (per month) Fatal hemorrhagic stroke (per episode) Systemic embolism acute care (per episode)
Volume ] Number ]
Other ICH (excluding hemorrhagic stroke) Gastrointestinal bleeding Other major bleeding or CRNM bleeding MI acute care (per episode) MI chronic care (per month) Other CV hospitalization
Cost, Mean (95% CI) ¥
1,024,406 146,911 1,699,225 335,129 1,817,620 383,444 983,728 1,448,758 157,117 2,131,870 336,541 2,085,764 383,444 902,966
598,077 620,341 672,665 2,156,290 75,036 1,109,681
Distribution
SE, ¥
2120 720 (961,815–1,088,941) (138,299–155,780) (1,596,349–1,805,267) (319,427–351,203) (1,698,097–1,941,151) (361,995–405,502) (870,539–1,103,735) (1,140,846–1,792,900) (147,610–166,917) (1,775,254–2,520,640) (320,626–352,837) (1,576,334–2,665,429) (361,995–405,502) (615,738–1,244,312) 1,123,451
γ γ γ γ γ γ γ γ γ γ γ γ γ γ γ
32,434 4460 53,301 8107 62,011 11,100 59,509 166,536 4926 190,286 8218 278,269 11,100 160,798 83,844
(515,908–686,230) (556,760–687,311) (567,485–786,656) (1,940,641–2,383,138) (71,981–78,153) (1,050,043–1,170,944)
γ γ γ γ γ γ
43,470 33,313 55,946 112,911 1575 30,845
CRNM ¼ clinically relevant nonmajor; CV ¼ cardiovascular; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction.
Source Medical treatment fee Medical treatment fee Claims data Hattori 201216 Claims data Hattori et al,16 2012 Claims data Hattori et al,16 2012 Claims data Claims data Hattori et al,16 2012 Claims data Hattori et al,16 2012 Claims data Hattori et al,16 2012 Claims data Tanihata et al,18 2006 Evans et al,19 2007 Claims data Claims data Claims data Claims data Claims data Claims data
Clinical Therapeutics
10 Table VII. Cost parameters.
I. Kamae et al.
Table VIII. Utility parameters. Parameter Utility value Mild ischemic stroke Moderate ischemic stroke Severe ischemic stroke Mild hemorrhagic stroke Moderate hemorrhagic stroke Severe hemorrhagic stroke Systemic embolism Utility decrement Other ICH
Mean (95% CI)
Distribution
SE
(0.758–0.809) (0.426–0.460) (0.070–0.117) (0.746–0.795) (0.414–0.448) (0.070–0.117) (0.589–0.664)
β β β β β β β
0.013 0.009 0.012 0.013 0.009 0.012 0.019
Hattori Hattori Hattori Hattori Hattori Hattori Hattori
0.151 (0.081–0.238)
β
0.040
Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014
0.784 0.443 0.092 0.771 0.431 0.092 0.627
Gastrointestinal bleeding or other major bleeding CRNM bleeding
0.151 (0.081–0.238)
β
0.040
0.058 (0.029–0.096)
β
0.017
MI
0.610 (0.572–0.647)
β
0.019
Other CV hospitalization
0.128 (0.081–0.182)
β
0.026
Utility decrement due to anticoagulation treatment Aspirin 0.002 (0.000–0.040)
β
0.000
Aspirin (second line)
0.002 (0.000–0.040)
β
0.000
Warfarin
0.013 (0.000–0.080)
β
0.000
Source et et et et et et et
al,16 al,16 al,16 al,16 al,16 al,16 al,16
2012 2012 2012 2012 2012 2012 2012
Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014 Dorian et al,9 2014 Lip et al,10 2014
CRNM ¼ clinically relevant nonmajor bleeding; CV ¼ cardiovascular; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction.
hemorrhagic stroke (intracerebral hemorrhage: ¥1,599,116 [US $15,991]; subarachnoid hemorrhage: ¥2,614,010 [US $26,140]). Evans et al also reported the cost for the acute phase of MI based on claims data analysis, including 1748 admissions for MI to several hospitals, as approximately ¥2,200,000 (US $22,000),19 which is almost the same as the cost for acute-phase MI of ¥2,156,290 (US $21,563) calculated in our analysis. Despite the limitations we discussed, the impact of these limitations on our results was considered small,
] 2015
and our results were deemed robust because of the additional effect in Japanese patients compared with that in the global population. In conclusion, based on an adaptation of a published Markov model, apixaban was a cost-effective alternative to warfarin in Japanese NVAF patients by lowering risks of clinical events, such as stroke and bleeding and extending QALYs. Further research extrapolating the cost-effectiveness of apixaban versus warfarin with more representative and larger Japanspecific real-world data or trial-based clinical data,
11
Clinical Therapeutics
Table IX. Results of base case analysis.
No. of events (per 1000 patients) Stroke and systemic embolism* Major bleeding, including hemorrhagic stroke* CRNM bleeding MI Other CV hospitalization Other treatment discontinuation Outcomes (per patient) Life-years QALYs Cost, ¥ ICER, ¥ (per QALYs) Discounted by 2% (base case) Undiscounted Discounted by 5%
Warfarin
Apixaban
Difference
358 263 354 94 1250 651
341 220 312 92 1283 656
17 43 42 2 33 5
10.285 7.23 5,144,019
10.516 7.469 5,655,712
0.231 0.24 511,692 2,135,743 1,966,615 2,393,753
CRNMB ¼ clinically relevant nonmajor; CV ¼ cardiovascular; ICER ¼ incremental cost-effectiveness ratio; ICH ¼ intracranial hemorrhage; MI ¼ myocardial infarction; QALYs: quality-adjusted life-years. * Hemorrhagic strokes are considered in both categories.
2,135,743 Risk of ischemic and unspecified stroke for apixaban (Rate/100 PYs) (0.56 - 1.52)
5,116,598
987,444
Risk of intracranial hemorrhage for warfarin (Rate/100 PYs) (0.35 - 1.87)
832,334
Risk of cardiovascular hospitalization for apixaban (Rate/100 PYs) (5.98 - 16.17)
3,819,575
1,181,399
Risk of other treatment discontinuations for warfarin (Rate/100 PYs) (8.23 - 22.27)
3,350,447
1,307,317
Risk of other treatment discontinuations for apixaban (Rate/100 PYs) (7.53 - 20.38)
3,386,241 3,534,145
1,523,386
Risk of cardiovascular hospitalization for warfarin (Rate/100 PYs) (5.98 - 16.17)
1,030,914
Utility decrement: Warfarin (0.00 - 0.08)
984,165
Risk of ischemic and unspecified stroke for warfarin (Rate/100 PYs) (0.87 - 1.33)
3,010,564 2,763,056
1,426,808
Risk of intracranial hemorrhage for apixaban (Rate/100 PYs) (0.19 - 0.51)
2,964,105
1,705,000
Rate of death comparator trial period (2.71 - 4.04)
2,878,554
1,749,743
Rate of death apixaban trial period (2.50 - 3.72)
1,798,718
Risk of MI for apixaban (Rate/100 PYs) (0.30 - 0.82)
1,765,876
Mean age for males (63% - 77%)
2,872,675 2,883,384 2,762,285 2,439,805
1,816,145 0
1,000,000
2,000,000
3,000,000 4,000,000 ICER (¥ per QALY)
5,000,000
6,000,000
Figure 2. Results of 1-way sensitivity analysis (Tornado diagram). The analysis identified the following factors as having a large impact on the cost-effectiveness of apixaban: the risk of ischemic and unspecified stroke associated with the use of apixaban and the risk of intracranial hemorrhage (ICH) associated with the use of warfarin. ICER = incremental cost-effectiveness ratio; MI = myocardial infarction; PYs = patient-years; QALYs: quality-adjusted life-years. 12
Volume ] Number ]
I. Kamae et al.
2,000,000
Incremental Costs (¥)
1,500,000
1,000,000
500,000
0 -0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
-500,000
-1,000,000
Incremental QALYs
Figure 3. Results of probabilistic sensitivity analysis (scatterplot). QALYs = quality-adjusted life-years.
Apixaban Warfarin Probability for Being the Most Cost-effective Treatment Choice (%)
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
22.5
Willingness to Pay (Million ¥ per QALY)
Figure 4. Results of probabilistic sensitivity analysis (cost-effectiveness acceptability curve). The costeffectiveness acceptability curve revealed that when the willingness to pay was approximately Z¥2,250,000 (US $22,500) per quality-adjusted life-year QALY, the probability of apixaban being cost-effectiveness was Z50%.
] 2015
13
Clinical Therapeutics resource utilization, and utility data whenever available would be warranted.
ACKNOWLEDGMENT This research has been funded exclusively by BristolMyers K.K. and Pfizer Japan Inc. Tatsunori Murata is an employee of Crecon Medical Assessment Inc who received funding from Bristol-Myers K.K. and Pfizer Japan Inc for writing of the manuscript. Drs. Kamae, Hashimoto, Koretsune, Tanahashi, Phatak, Liu, Wang and Okuyama provided interpretation of this research and revised the manuscript. Mr. Murata performed all analyses and drafted the manuscript. Dr. Tang drafted concept of this research and revised the manuscript. All authors gave final approval of the manuscript.
CONFLICTS OF INTEREST Dr Kamae has received lecture fees from Bristol-Myers K.K. and a consultant fee from Crecon Medical Assessment Inc. Dr Hashimoto has received lecture fees from Bristol-Myers K.K., Pfizer Japan Inc, Bayer Yakuhin Ltd, Daiichi-Sankyo Company Ltd, and Nippon Boehringer Ingelheim Co Ltd. Dr Koretsune has received lecture fees from Nippon Boehringer Ingelheim Co Ltd, Daiichi Sankyo Company Ltd, Bayer Yakuhin Ltd, and Bristol Myers K.K. and has received research grants from Nippon Boehringer Ingelheim Co Ltd and Daiichi Sankyo Company Ltd. Dr Tanahashi has received lecture fees from Bayer Yakuhin Ltd, Sanofi, Mitsubishi Tanabe Pharma, Nippon Boehringer Ingelheim Co Ltd, and BristolMyers K.K. and has received research grants from Bayer Yakuhin Ltd, Sanofi, Astrazeneca, Nippon Boehringer Ingelheim Co Ltd, Mitsubishi Tanabe Pharma, Takeda, Daiichi-Sankyo Company Ltd, and Bristol-Myers K.K. Dr Okumura has received lecture fees from Nippon Boehringer Ingelheim Co Ltd, Bayer Healthcare, Bristol-Myers K.K., Pfizer Japan Inc, and Eisai and has received research grants from Nippon Boehringer Ingelheim Co Ltd, Bayer Healthcare, and Daiichi-Sankyo Company Ltd. Mr Murata is an employee of Crecon Medical Assessment Inc and a consultant for Bristol-Myers K.K. Drs Wang and Liu are paid employees of Bristol-Myers Squibb or Pfizer Inc and/or own respective company’s stock. At the time of this research, Dr Phatak was an employee of Bristol-Myers Squibb and owned company stocks. Dr Tang was an employee of Bristol-Myers K.K. during the work. The authors have indicated that
14
they have no other conflicts of interest regarding the content of this article.
REFERENCES 1. Lip GY, Brechin CM, Lane DA. The global burden of atrial fibrillation and stroke: a systematic review of epidemiology of atrial fibrillation in regions outside North America and Europe. Chest. 2012;142:1489–1498. 2. Tomita F, Kohya T, Sakurai M, et al, Hokkaido Atrial Fibrillation Study Group. Prevalence and clinical characteristics of patients with atrial fibrillation: analysis of 20,000 cases in Japan. Jpn Circ J. 2000;64:653–658. 3. Wolf PA, Dawber TR, Thomas HE Jr, Kannel WB. Epidemiologic assessment of chronic atrial fibrillation and risk of stroke: the Framingham study. Neurology. 1978;28:973–977. 4. Hayashi T, Hayashi T, Hayashi T. Why most of patients on warfarin have not been anticoagulated to the therapeutic range in Japan? Prog Med. 2009;29:1599–1605. 5. Camm AJ1, Lip GY, De Caterina R, Savelieva I, et al, ESC Committee for Practice Guidelines (CPG). 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J. 2012;33:2719–2747. 6. Granger CB, Alexander JH, McMurray JJ, et al, ARISTOTLE Committees and Investigators. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981–992. 7. Ministry of Health, Labour and Welfare. 2011 Estimates of National Medical Care Expenditure (http://www.mhlw. go.jp/toukei/saikin/hw/k-iryohi/11/index.html). Last accessed at 2014. 8. Yhiroyuki Y. Anticoagulation therapy for atrial fibrillation –from warfarin to new oral anticoagulants. Prog Med. 2013;33:975–982. 9. Dorian P, Kongnakorn T, Phatak H, et al. Costeffectiveness of apixaban vs. current standard of care for stroke prevention in patients with atrial fibrillation. Eur Heart J. 2014;35:1897–1906. 10. Lip GY, Kongnakorn T, Phatak H, et al. Cost-effectiveness of apixaban versus other new oral anticoagulants for stroke prevention in atrial fibrillation. Clin Ther. 2014;36:192–210. 11. Mohan KM, Crichton SL, Grieve AP, et al. Frequency and predictors for the risk of stroke recurrence up to 10 years after stroke: the South London Stroke Register. J Neurol Neurosurg Psychiatry. 2009;80:1012–1018. 12. Claassen DO, Kazemi N, Zubkov AY, et al. Restarting anticoagulation therapy after warfarin-associated intracerebral hemorrhage. Arch Neurol. 2008;65:1313–1318. 13. Sorensen SV, Dewilde S, Singer DE, et al. Cost-effectiveness of warfarin: trial versus “real-world” stroke prevention in atrial fibrillation. Am Heart J. 2009;157:1064–1073.
Volume ] Number ]
I. Kamae et al. 14. Scarborough P, Bhatnagar P, Wickramasinghe K, et al.British Health Foundation Health Promotion research group. Coronary Heart Disease statistics 2010 edition. Department of Public Health University of Oxford, 2010:157. 15. 2012 Discharged patient survey ( http://www.mhlw. go.jp/stf/ shingi/0000023522.html). Last accessed at 2014. 16. Hattori N, Hirayama T, Katayama Y. Medical care for chronic-phase stroke in Japan. Neurol Med Chir (Tokyo). 2012;52:175–180. 17. Kobayashi S, Ogushi Y. Stroke Data Bank 2009. Tokyo: Nakayama Shoten, 2009:34–35. 18. Tanihata S, Nishigaki K, Kawasaki M, et al. Outcomes of patients with stable low-risk coronary artery disease receiving medical- and PCI-preceding therapies in Japan: J-SAP study 1-1. Circ J. 2006;70: 365–369. 19. Evans E, Imanaka Y, Sekimoto M, et al. Risk adjusted resource utilization for AMI patients treated in Japanese hospitals. Health Econ. 2007;16:347–359. 20. Ministry of Health, Labour and Welfare. Abridged Life Table (http://www. mhlw.go.jp/toukei/saikin/hw/life/ life12/). Last accessed at 25 Aug 2014. 21. Guideline for analysis method in health economic evaluation ver. 1.0. Mar 29, 2013 (http://hta. umin.jp/guideline_j.pdf). Last accessed at 25 Aug 2014. 22. Ohkusa Y, Sugawara T. Research for Willingness to Pay for One QALY Gain. J Health Care Soc. 2006;16:157–165. 23. Shiroiwa T, Sung YK, Fukuda T, et al. International survey on willingness-to-pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health Econ. 2010;19:422–437. 24. Hori M, Koretsune Y, Yasaka M, et al. Health economic evaluation of Dabigatran Etexilate for prevention of stroke in patients with
] 2015
25.
26.
27.
28.
nonvalvular atrial fibrillation. Pharma Medica. 2011;29:151–164. Lanitis T, Cotté FE, Gaudin AF, et al. Stroke prevention in patients with atrial fibrillation in France: comparative cost-effectiveness of new oral anticoagulants (apixaban, dabigatran, and rivaroxaban), warfarin, and aspirin. J Med Econ. 2014;17:587–598. Pink J, Pirmohamed M, Lane S, Hughes DA. Cost-effectiveness of pharmacogenetics-guided warfarin therapy vs. alternative anticoagulation in atrial fibrillation. Clin Pharmacol Ther. 2014;95:199–207. Coyle D, Coyle K, Cameron C, et al. Cost-effectiveness of new oral anticoagulants compared with warfarin in preventing stroke and other cardiovascular events in patients with atrial fibrillation. Value Health. 2013;16:498–506. Harrington AR, Armstrong EP, Nolan PE Jr, Malone DC. Costeffectiveness of apixaban, dabigatran, rivaroxaban, and warfarin for stroke prevention in atrial fibrillation. Stroke. 2013;44:1676–1681.
29. Connolly SJ, Ezekowitz MD, Yusuf S, et al, RE-LY Steering Committee and Investigators. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009; 361:1139–1151. 30. Patel MR, Mahaffey KW, Garg J, et al, ROCKET AF Investigators. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365: 883–891. 31. Sakthong P, Charoenvisuthiwongs R, Shabunthom R. A comparison of EQ-5D index scores using the UK, US, and Japan preference weights in a Thai sample with type 2 diabetes. Health Qual Life Outcomes. 2008;6:71. 32. Shiroiwa T, Fukuda T, Ikeda S, et al. Japanese population norms for preference-based measures: EQ5D-3L, EQ-5D-5L, and SF-6D. Qual Life Res. 2015. 33. Toyonaga T. Factors affecting the cost for medical care and rehabilitation in stroke patients. Jpn J Occup Med Trauma. 2006;54: 175–182.
Address correspondence to: Tatsunori Murata, MS, Crecon Medical Assessment Inc, 12-15 Shibuya 2-chome, Shibuya-ku, Tokyo, Japan. E-mail: [email protected]
15
