Is dabigatran considered a cost-effective alternative to warfarin treatment: a review of current economic evaluations worldwide

Abstract

Objective:

Dabigatran was the first of a new generation of anticoagulation drugs for the indication of non-valvular atrial fibrillation (AF) to be approved. Evidence show that dabigatran 150 mg twice daily significantly reduces the risk of stroke and systemic embolism (RR = 0.65; p < 0.001) and shows a comparable rate of major bleedings (RR = 0.93; p = 0.32), whereas dabigatran 110 mg twice daily was associated with a comparable rate of stroke and systemic embolism (RR = 0.90; p = 0.30) and a significantly lower rate of major bleedings compared to warfarin treatment (RR = 0.80; p = 0.003). The purpose is to review current economic evaluations of these alternatives for healthcare professionals to include these findings in their decision-making.

Methods:

A systematic literature search identified 43 economic evaluations, of which 10 were included and evaluated according to the Consensus Health Economic Criteria list (CHEC-list) and the Oxford model.

Results:

Six economic evaluations concluded that dabigatran was a cost-effective alternative to warfarin. One evaluation concluded the same except when quality in warfarin treatment was excellent, with a mean time in therapeutic range (TTR) > 73%. Three evaluations concluded that dabigatran was a cost-effective alternative to warfarin in patient sub-groups; TTR ≤ 64%, congestive heart failure, hypertension, age ≥ 75, diabetes mellitus, prior stroke or transient ischemic attack (CHADS2 score) ≥3, or a CHADS2 score = 2 unless international normalized ratio (INR) control was excellent, and with high risk of stroke or in a low-quality warfarin treatment. Dabigatran 110 mg twice daily was in general dominated by dabigatran 150 mg twice daily.

Limitations:

The evaluations were not fully homogeneous, as some did not include loss of productivity, costs of dyspepsia, and annual costs of dabigatran patient management.

Conclusions:

In the majority of the economic evaluations, dabigatran is a cost-effective alternative to warfarin treatment. In some evaluations dabigatran is only cost-effective in sub-groups, such as patients with a low TTR-value in warfarin treatment and a CHADS2 score ≥2.

Keywords::

• Economic evaluation
• Dabigatran
• Warfarin
• Non-valvular atrial fibrillation
• Systematic review
• Anticoagulation

Introduction

Worldwide, non-valvular atrial fibrillation (AF) is the most common cardiac arrhythmia, which affects more than 6% of the population above 80 years, and increases mortality and morbidity rates due to a markedly increased risk of stroke and systemic embolism1,2. Such events will often result in serious consequences for the patients’ quality-of-life and significant health economic consequences3.

For more than five decades the vitamin K antagonist warfarin has been the commonly used anticoagulation treatment for this patient group, despite the fact that it has been criticised for its management disadvantages, such as regular monitoring of international normalized ratio (INR) due to a narrow therapeutic index and interactions with alcohol, food products containing high levels of K-vitamin, and other pharmaceuticals4,5.

In August 2011, the direct thrombin inhibitor dabigatran achieved marketing approval in Europe as the first of a new generation of anticoagulation drugs5. This approval was based on a large multinational phase III clinical trial, The Randomized Evaluation of Long-term Anticoagulant Therapy (RE-LY). This trial concluded that dabigatran 150 mg twice daily (high-dose dabigatran) significantly reduced annual risk of stroke or systemic embolism from 1.71% to 1.11% (95% CI = 0.52–0.81, p < 0.001) compared to dose-adjusted warfarin treatment and dabigatran 110 mg twice daily (low-dose dabigatran) was associated with a reduced annual risk of 1.54% (95% CI = 0.74–1.10, p = 0.30). Furthermore, low-dose dabigatran significantly reduced annual risk of major bleedings from 3.57% to 2.87% (95% CI = 0.70–0.93, p = 0.003) compared to dose-adjusted warfarin treatment and high-dose dabigatran was associated with a reduced annual risk of 3.32% (95% CI = 0.81–1.07, p = 0.32)6.

Generally, dabigatran is priced at a higher level compared to warfarin7. However, it is important to estimate the incremental health gain in proportion to the incremental costs, in order to evaluate whether this new treatment option is cost-effective. The decision-makers at different levels ought to take these health economic perspectives into account when evaluating dabigatran treatment. The purpose of this study is to review all current economic evaluations of dabigatran compared to warfarin treatment in patients with AF, thereby evaluating the cost-effectiveness of dabigatran treatment worldwide.

Methods

Systematic literature search

A systematic literature search was conducted on October 24, 2012 by the authors Louise Justesen Hesselbjerg and Heidi Sjoelund Pedersen. This was completed in order to identify relevant economic evaluations of dabigatran and warfarin treatment in the prevention of stroke and systemic embolism in patients with AF. The literature search was conducted in the following databases: MEDLINE, Embase, Cochrane Library, and NHS Economic Evaluation Database.

Because the RE-LY trial was completed in 2009, an inclusion criterion was set to publications between 2009–2012. Other inclusion criteria were one of the following types of publications: journal articles, multi-centre studies, reviews, systematic reviews, Cochrane reviews, economical evaluations, or comparative studies. Keywords searched for in combination were: cost-effectiveness, atrial fibrillation, dabigatran, and warfarin. The exclusion criterion was economic evaluations of the indication hip and knee replacement surgery.

All identified articles were evaluated by means of the title and abstract. The full-texts of the relevant economic evaluations were then screened and the reference lists checked in order to ensure that no relevant evaluations were disregarded.

Quality assessment

The quality of the relevant economic evaluations was assessed applying the Consensus Health Economic Criteria list (CHEC-list) and their level of evidence was stated according to the Oxford model from the Oxford Centre for Evidence-based Medicine8,9. This quality assessment was completed by Louise Justesen Hesselbjerg and Heidi Sjoelund Pedersen. Items in the CHEC-list were evaluated with the criteria; ‘Yes’ indicating that the item was adequately covered or reported in an appropriate way, or ‘No’ indicating that the item was not met. The 19th item, ‘Are ethical and distributional issues discussed appropriately?’, was irrelevant in this review and therefore excluded8. Evaluations that achieved a higher score than 75% were defined as high-quality evaluations, >50–≤75% were defined as moderate-quality evaluations, and ≤50% were defined as low-quality evaluations10. Evaluations with a low-quality were excluded. Furthermore, each evaluation was assigned a number according to the Oxford model of 1a–1c, 2a–2c, 3a–3b, 4, or 5, with 1a being of highest level of evidence9. Disagreements between the authors assessing the quality were resolved through discussion, which is illustrated in Table 1.

Table 1. Summed quality scores according to the CHEC-list and the level of evidence according to the Oxford model of included economic evaluations of dabigatran and warfarin treatment in the prevention of apoplexia and systemic embolism in patients with AF.

	González-Juanatey et al.20	You et al.12	Davidson et al.17	Kansal et al.13	Langkilde et al.16	Freeman et al.18	Kamel et al.19	Sorensen et al.14	Pink et al.15	Shah and Gage11	Total fulfillment*
Oxford model	2b	2b	2b	2b	2b	2b	2b	2b	2b	2b
CHEC-list
1. Is the study population clearly described?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
2. Are competing alternatives clearly described?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
3. Is a well-defined research question posed in answerable form?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
4. Is the economic study design appropriate to the stated objective?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
5. Is the chosen time horizon appropriate in order to include relevant costs and consequences?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
6. Is the actual perspective chosen appropriately?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
7. Are all important and relevant costs for each alternative identified?	No	No	No	No	No	No	No	No	No	No	0%
8. Are all costs measured appropriately in physical units?	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	90%
9. Are costs valued appropriately?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
10. Are all important and relevant outcomes for each alternative identified?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
11. Are all outcomes measured appropriately?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
12. Are outcomes valued appropriately?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
13. Is an incremental analysis of costs and outcomes of alternatives performed?	Yes	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	90%
14. Are all future costs and outcomes discounted appropriately?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
15. Are all important variables, whose values are uncertain, appropriately subjected to sensitivity analysis?	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	No	80%
16. Do the conclusions follow from the data reported?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	100%
17. Does the study discuss the generalizability of the results to other settings and patient/client groups?	Yes	No	Yes	Yes	Yes	No	Yes	No	No	Yes	60%
18. Does the article indicate that there is no potential conflict of interest of study researcher(s) and funder(s)?	Yes	No	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	80%
Total fulfillment^†	94.4%	77.8%	94.4%	94.4%	88.9%	88.9%	83.3%	88.9%	88.9%	88.9%

AF, Non-valvular atrial fibrillation; CHEC-list, Consensus Health Economic Criteria list.

*Calculated as: Number of fulfilled criteria divided by total number of criteria multiplied by 100.

†Calculated as: Number of economic evaluations fulfilling the criteria in question divided by total number of economic evaluations multiplied by 100.

Results

Systematic literature search

The systematic literature search resulted in 43 hits, of which 14 were duplicates and 19 evaluations were excluded after screening of titles and abstracts. Ten remaining economic evaluations were included after full-text reviewing and the reference lists did not yield any additional relevant literature. Thereby, 10 economic evaluations were included in this review11–20, which is illustrated in Figure 1.

Figure 1. Flowchart of different steps in the systematic literature search.

Quality assessment

On average 89.9% of the criteria were met in the CHEC-list, ranging from 77.8–94.4%. Some of the factors not included were; annual costs of dabigatran patient management, loss of productivity, discussing generalizability, a probabilistic sensitivity analysis (PSA), and indicating a potential conflict of interests. Furthermore, health outcomes were in general obtained from a single clinical trial. As no evaluations obtained a score below 75%, they were all evaluated as high-quality evaluations. All economic evaluations were assigned the level of evidence 2b, which corresponds to ‘An analysis based on clinically sensible costs or alternatives; limited review(s) of the evidence, or single studies; and including multi-way sensitivity analyses’9, which is illustrated in Table 1.

Model characteristics

Of the 10 economic evaluations, three were published in 201114,15,18 and seven were published in 201211–13,16,17,19,20. The study design was in all evaluations based on Markov models with varying hypothetical cohorts of patients with AF. In the evaluations warfarin and dabigatran treatment were defined differently. In some of the Markov models dabigatran treatment was divided into sequentially/adjusted drug doses, changing drug dose from 150 mg to 110 mg twice daily at the age of 8015,18,19. In one of the Markov models warfarin treatment was incorporated as genotype-guided anticoagulation care (genotype-guided AC)12. Furthermore, in some of the Markov models warfarin treatment was incorporated as a ‘real world scenario’ including acetylsalicylic acid, clopidogrel, and no treatment, or as ‘trial like scenario’ including the mean TTR-value from the RE-LY trial11–14,16,17,20. The costs, adverse events, annual discount rates, and time horizons varied between the evaluations and only some of the evaluations15–17 included costs of general practitioners, proton pump inhibitors for treating dyspepsia, outpatient visits, and dabigatran patient management. None of the economic evaluations included all relevant adverse events. Some evaluations did not include outcomes such as death15, dyspepsia13–20, and systemic embolism13,14,16,17,20. The annual discount rate of the evaluations varied (2–5%) as well as the time horizon, which was, e.g., stated as lifetime (maximum life expectancy 100 years), mean remaining life years (e.g., 20 years), or simply as expected lifetime without further explanation. All model characteristics of the economic evaluations are illustrated in Table 2.

Table 2. General characteristics of the included economic evaluations on dabigatran and warfarin treatment in prevention of apoplexia and systemic embolism in patients with AF.

	Freeman et al.18	Pink et al.15	Kamel et al.19	Langkilde et al.16	Davidson et al.17
Year	2011	2011	2012	2012	2012
Country	US	UK	US	Denmark	Sweden
Journal	Annals of Internal Medicine	British Medical Journal	American Heart Association	Journal of Medical Economics	European Heart Journal
Design	Markov decision model Hypothetical cohort of 10,000 patients ≥65 years with non-valular AF, a CHADS2 score ≥1, and no contraindications	Markov model, reflected RE-LY patients Hypothetical cohort of 50,000 patients with a mean CHADS2 score of 2.1	Markov model Hypothetical cohort of 70-year-old patients with non-valvular AF, prior stroke or TIA, and no contraindications Cycles of 30 days	Adapted Markov model from Sorensen et al.16 Hypothetical cohort of 10,000 patients with AF and a mean age of 69 years Cycles of 3 months	Decision analytic model, Markov design Hypothetical cohort of patients >65 years with AF Cycles of 1 year
Compared alternatives	Adjusted-dose warfarin (target INR 2–3) Dabigatran 150 mg, twice daily Dabigatran 110 mg, twice daily	Warfarin Dabigatran, twice daily (150 mg < 80 years; 110 mg ≥ 80 years)	Adjusted-dose warfarin (target INR 2–3) Dabigatran 150 mg, twice daily	Dabigatran, twice daily (150 mg < 80 years; 110 mg ≥ 80 years) Warfarin (5 mg, target INR 2–3) Acetylsalicylic acid No treatment	Dabigatran, twice daily (150 mg < 80 years; 110 mg ≥ 80 years) Warfarin Acetylsalicylic acid No anticoagulation treatment
Time horizon	Lifetime (35 years or until death)	Expected lifetime	20 years or until death	Long-term (remaining lifetime)	Long-term (20 years)
Annual discount rate	3%	3.5%	3%	2%	3%
Health outcomes
Included adverse events	IS TIA MI Haemorrhage Death	Stroke ICH TIA MI Major bleeding Minor bleeding Pulmonary embolism	IS ICH TIA MI MEH Recurrent or combined stroke and/or ICH Mortality	IS ICH TIA AMI Other major haemorrhages Minor bleeds SE HS Death	IS MI Other major bleedings Minor bleeding GIB HS Death SE
Effect	Warfarin: 10.28 QALYs Dabigatran 110 mg: 10.70 QALYs Dabigatran 150 mg: 10.84 QALYs	Warfarin: 6.390 QALYs 10.851 Life-years Dabigatran 150 mg: 6.536 QALYs 11.051 Life-years Dabigatran 110 mg: 6.484 QALYs 10.940 Life-years	Warfarin: 3.91 QALYs Dabigatran: 4.27 QALYs	Warfarin: 8.32 QALYs 11.34 Life-years Dabigatran: 8.59 QALYs 11.61 Life-years	Warfarin: 8.31 QALYs 11.83 Life-years Dabigatran: 8.60 QALYs 12.11 Life-years
Costs
Perspective	Socio-economic	UK National Health service	Socio-economic	N/S*	Socio-economic
Currency	2008 USD($)	2009 GBP(£)	2010 USD($)	2011 Euro(€)	2012 Euro(€)
Included costs	Prices of drugs Adverse events INR-measurement	Prices of drugs Adverse events Costs of general practitioners Costs of district nurses’ visits Price of proton pump inhibitors (dyspepsia) Ward costs Inpatient and Outpatient costs	Prices of drugs Adverse events INR-measurements	Prices of drugs Adverse events INR-measurement Consultation Outpatient visits Dabigatran initiation (consultation and CrCl-test)	Prices of the drugs (including annual visits to primary care doctor) Warfarin monitoring visits Social services Outpatient care
Total costs	Dabigatran 110 mg: $164,576 Dabigatran 150 mg: $168,398 Warfarin: $143,193	Dabigatran 150 mg: £9,850 Dabigatran 110 mg: £10,529 Warfarin: £6,480	N/S*	Dabigatran: €18,752 Warfarin: €16,886	Dabigatran: €27,009 Warfarin: €24,797
	Gonzàlez-Juanatey et al.20	Shah and Gage11	You et al.12	Sorensen et al.14	Kansal et al.13
Year	2012	2012	2012	2011	2012
Country	Spain	US	China	Canada	UK
Journal	Revista Española de Cardiología	Circulation	Plos One	Thrombosis and Haemostasis	Heart
Design	Markov Model Hypothetical cohort of 10.000 patients with AF. Mean age was 69.1 years with initiation of model and had no disabilities Duration of a cycle was 3 months	Markov model Hypothetical cohort of 70-year-old patients with AF, moderate risk of stroke, and no contraindications Duration of a cycle was 1 month Intention-to-treat perspective	Markov model Hypothetical cohort of 65-year-old patients with AF and a CHADS2 score of 2 Duration of a cycle was 1 month	Markov model Hypothetical cohort of patients with AF matched that of RE-LY Duration of a cycle was 3 months	Markov model Two hypothetical cohorts of patients with AF (starting age <80 and ≥80 years) Duration of a cycle was 3 months
Compared alternatives	Scenario 1: Dabigatran, twice daily (150 mg ≥ 80 years; 110 mg < 80 years) Warfarin Scenario 2: Dabigatran, twice daily (150 mg ≥ 80 years; 110 mg < 80 years) Prescribing pattern in clinical practice (60% warfarin, 30% acetylsalicylic acid and 10% not treated)	Dabigatran, twice daily (150 mg ≥ 80 years; 110 mg < 80 years) Warfarin Clopidogrel/acetylsalicylic acid Acetylsalicylic acid No treatment	Dabigatran, twice daily (150 mg ≥ 80 years; 110 mg < 80 years) Warfarin (mean TTR 64%) Genotype-guided AC	Trial-like scenario: Dabigatran (sequential), twice daily Dabigatran, twice daily (150 mg ≥ 80 years) Dabigatran, twice daily (110 mg < 80 years) Warfarin Real-world scenario: Dabigatran (sequential), twice daily Anticoagulation treatment (receiving warfarin, acetylsalicylic acid, or no treated)	Age < 80: Dabigatran, twice daily warfarin (mean TTR 64%) Age ≥ 80: Dabigatran, twice daily warfarin (mean TTR 64%) Scenario analysis: Dabigatran, twice daily acetylsalicylic acid No treatment
Time horizon	Lifetime (maximum life expectancy 100 years)	Maximum of 20 years	Lifetime (maximum 25 years)	Lifetime (maximum life expectancy 100 years)	Lifetime (maximum life expectancy 100 years)
Annual discount rate	Base case 3% Medication costs 7.5%	Base case 3%	Base case 3%	Base case 5%	Base case 3.5%
Health outcomes
Included outcomes	IS ICH TIA AMI Non-gastrointestinal haemorrhage Minor bleeding SE HS EH GIB Death	IS ICH TIA MI Major and minor non-cerebral haemorrhage Dyspepsia Death	IS ICH MI Major bleeding EH Dyspepsia Death	IS ICH TIA AMI Minor bleeding HS SE EH Death	IS TIA ICH AMI Minor bleeding HS SE EH Death
Effect	Scenario 1: Warfarin 8.45 QALYs Dabigatran 8.73 QALYs Scenario 2: Prescribing pattern in clinical practice 8.32 QALYs Dabigatran 8.73 QALYs	Warfarin: 8.40 QALYs Dabigatran 110 mg: 8.54 QALYs Dabigatran 150 mg: 8.65 QALYs Clopidogrel/acetylsalicylic acid: 8.32 QALYs Acetylsalicylic acid: 8.17 QALYs	Dabigatran 110 mg: 10.026 QALYs Dabigatran 150 mg: 10.065 QALYs Warfarin: 9.444 QALYs Genotype-guided AC: 9.554 QALYs	Trial-like scenario: Dabigatran (sequential): 7.29 QALYs Warfarin: 7.08 QALYs Dabigatran 150 mg: 6.86 QALYs Warfarin: 6.68 QALYs Dabigatran 110 mg: 6.82 QALYs Warfarin: 6.68 QALYs Real-world scenario: Dabigatran (sequential): 7.29 QALYs Warfarin: 7.01 QALYs	Age < 80: Warfarin: 7.82 QALYs Dabigatran: 8.06 QALYs Age ≥ 80: Warfarin: 4.04 QALYs Dabigatran: 4.11 QALYs Scenario analysis: Dabigatran: 8.06 QALYs Acetylsalicylic acid: 7.59 QALYs No treatment: 7.12 QALYs
Costs
Perspective	Spanish National Health System	Insurance company (or Medicare)	Healthcare payers	Canadian payer	N/S*
Currency	2010 Euro(€)	2010 USD($)	2012 USD($)	2010 CAD($)	2010 GBP(£)
Types of costs	Prices of drugs Adverse events INR-measurements Follow-up costs	Prices of drugs Adverse events INR-measurements Annual prophylaxis	Prices of drugs Adverse events INR-measurements Test of genotype	Prices of drugs Adverse events INR-measurements Follow-up costs	Prices of drugs INR-monitoring Adverse events Follow-up costs
Included costs	Scenario 1: Warfarin: €10,343 Dabigatran: €15,193 Scenario 2: Prescribing pattern in clinical practice: €9426 Dabigatran: €15,193	Dabigatran 150 mg: $43,700 Dabigatran 110 mg: $44,300 Warfarin: $23,000 Clopidogrel/acetylsalicylic acid: $34,000 Acetylsalicylic acid: $20,000	Dabigatran 110 mg: $102,536 Dabigatran 150 mg: $92,684 Warfarin: $90,481 Genotype-guided AC: $85,627	Trial-like scenario: Dabigatran (sequential): $45,124 Warfarin: $42,946 Dabigatran 150 mg: $41,824 Warfarin: $40,169 Dabigatran 110 mg: $44,379 Warfarin: $40,169 Real-world scenario: Dabigatran (sequential): $45,124 Warfarin: $44,020	Age < 80: Dabigatran: £19,645 Warfarin: £18.474 Age ≥ 80: Warfarin: £9,919 Dabigatran: £10,424 Scenario analysis: Dabigatran: £19,645 Acetylsalicylic acid: £18,561 No treatment: £20,475

AF, Non-valvular atrial fibrillation; AMI, Acute myocardial infarction; CHADS2 score, Congestive heart failure, hypertension, age >75, diabetes mellitus, and two points for the factor prior stroke or transient ischemic attack; CrCl, Creatine clearance; TTR, Mean time in the therapeutic range; EH, Extracranial haemorrhage; Genotype-guided AC, Genotype-guided anticoagulation care; GIB, Gastrointestinal bleedings; HS, Haemorrhagic stroke; ICH, Intracranial haemorrhage; INR, International normalized ratio; IS, Ischaemic stroke; MEH, Major extracranial haemorrhage; mg, milligrams; MI, Myocardial infarction; QALYs, Quality-adjusted life years; RE-LY, The Randomized Evaluation of Long-term Anticoagulant Therapy trial; SE, Systemic embolism; TIA, Transient ischemic attack.

*Not stated.

All evaluations stated health outcomes in quality-adjusted life years (QALYs). The willingness-to-pay (WTP) threshold for a QALY varied significantly between the countries, in which the economic evaluations were published, from £20,000 in the UK15 to €50,000 in Sweden17. In all evaluations one-way sensitivity analyses were applied, of which five additionally applied multi-way sensitivity analyses11,12,15,16,18. In eight of the evaluations a PSA was conducted12–15,17–20. The incremental cost-effectiveness ratio (ICER), the WTP thresholds, and choice of sensitivity analysis of each economic evaluation are illustrated in Table 3.

Table 3. Results of the included economic evaluations of dabigatran and warfarin treatment in the prevention of apoplexia and systemic embolism in patients with AF.

	Freeman et al.18	Pink et al.15	Kamel et al.19	Langkilde et al.16	Davidson et al.17
Cost-effectiveness
WTP threshold	$50,000 per QALY gained	£20,000–30,000 per QALY gained	$50,000 per QALY gained	€30,000 per QALY gained	€50,000 per QALY gained
ICER	Dabigatran 110 mg: $51,229 per QALY gained Dabigatran 150 mg: $45,372 per QALY gained	Dabigatran 150 mg: £23,082 per QALY gained Dabigatran 110 mg: £43,074 per QALY gained Ad hoc sub-group analysis: £24,340 per QALY gained Full RE-LY data: £27,940 per QALY gained	Dabigatran: $25,000 per QALY gained	Dabigatran: €6950 per QALY gained	Dabigatran: €7699 per Life-year gained €7742 per QALY gained
Sensitivity analysis
One-way	Key parameters influencing cost-effectiveness: Drug cost (dabigatran 110 mg not cost-effective >$9.36 per day. Dabigatran 150 mg not cost-effective >$13.70 per day) Risk of IS and ICH for dabigatran and warfarin Age (ICER decreased with age) Utility of dabigatran and warfarin Costs after ICH Utility after MI	Dabigatran 150 mg was cost-effective at the highest WTP threshold for all sub-groups except in very well-regulated patients (TTR ≥ 65.5%)	Key parameters influencing cost-effectiveness: Monthly costs of stroke and/or ICH (if >$5000 dabigatran dominated) Starting age of the cohort (if ≤81 years dabigatran dominated) Difference in RR of stroke between Dabigatran and warfarin (if ≥0.92 dabigatran dominated) Costs of dabigatran (if <$320 dabigatran dominated) Average CTTR (if CTTR <73% dabigatran dominated) Utility of mild IS	Key parameters influencing cost-effectiveness: Annual costs of INR-measurements (if >€744 dabigatran dominated) Dabigatran is cost-effective to warfarin irrespective of the range in quality of warfarin treatment	Key parameters influencing cost-effectiveness: Compared to well-controlled warfarin treatment the ICER increased to €12,449 per QALY gained, still below WTP threshold CHADS2 scores = 3–6 generated a better cost-effectiveness of dabigatran than lower scores If only transitions within the RE-LY time frame were included, the ICER increased to €10,287 per gained QALY, still below WTP threshold
Multi-way	Dabigatran 150 mg was the preferred choice unless patients had a low risk of stroke and a high risk of IH	Dabigatran 110 mg exceeded the highest WTP threshold in all cases	–	Higher levels of stroke risk were associated with lower incremental costs of dabigatran and higher QALY gained The ICER stayed below €18,000 per QALY gained in all cases	–
Probabilistic	10,000 first-order Monte Carlo simulations: Dabigatran 150 mg was cost-effective in 53% of cases compared to warfarin Dabigatran 110 mg was cost-effective in 30% of cases compared to warfarin	2,000 first-order Monte Carlo simulations: Warfarin had the highest probability of being cost-effective at threshold ≤£24,400 per QALY gained dabigatran 150 mg was most cost-effective in 68% of cases (at a CHADS2 score of ≥3)	10,000 first-order Monte Carlo simulations: Dabigatran was cost-effective in 57% of cases	–	1000 first-order Monte Carlo simulations: In all cases, the ICER was <€50,000 per QALY gained
Conclusions	Dabigatran was likely to be a cost-effective alternative to warfarin in the US, especially in patients with increased risk for IS and IH and highly dependent on pricing Dabigatran 110 mg was more cost-effective than dabigatran 150 mg in patients with low risk of IS	Dabigatran 150 mg was cost-effective at the lowest WTP threshold for patients with a low-quality INR-control or a high risk of stroke in the UK. Dabigatran offered a positive benefit-to-harm ratio when compared with warfarin	Dabigatran was likely to be a cost-effective alternative to warfarin in the US, but might not apply with excellent INR-control	Dabigatran represented a cost-effective alternative to warfarin in Denmark	Dabigatran was expected as a cost-effective treatment in Sweden, as the ICER was below the WTP threshold in all cases
Limitations	Did not include genotype-guided warfarin treatment or patient self-testing approaches, which might be a more cost-effective alternative Costs of dabigatran patient management were not included Productivity losses were not estimated	Had no access to data of individual patients Costs of dabigatran patient management were not included Productivity losses were not estimated Data of acetylsalicylic acid came from another study A 2-year trial was extrapolated to life-time Assumptions of utility decrements for AF patients were made Widespread of dabigatran might affect the quality of treatment in clinics	Costs of dabigatran patient management were not included Productivity losses were not estimated	Productivity losses were not estimated Did not include a PSA taking all variable uncertainties into account Limitations in the extrapolation to Danish settings Assumptions were made about dabigatran treatment, e.g., healthcare contact and treatment adherence High rate of AF patients treated despite recommendations not included	Productivity losses were not estimated Other complications were assumed to only affect costs and quality-of-life in the year in which they occurred. Transitions between CHADS2 scores occurred only according to age and stroke and no switch in choice of drug could occur (simplifications)
	Gonzàlez-Juanatey et al.20	Shah and Gage11	You et al.12	Sorensen et al.14	Kansal et al.13
Cost-effectiveness
WTP threshold	€30,000 per QALY gained	$50,000 per QALY gained	$50,000 per QALY gained	$30,000 per QALY gained	£20,000–£30,000 per QALY gained
ICER	Scenario 1: €17,581 per QALY gained Scenario 2: €14,118 per QALY gained	Warfarin: Dabigatran 150 mg $86,000 per QALY gained Dabigatran 110 mg $150,000 per QALY gained Clopidogrel/acetylsalicylic acid dominated Acetylsalicylic acid: Dabigatran 150 mg $50,000 per QALY gained Dabigatran 110 mg $66,000 per QALY gained Warfarin $12,500 per QALY gained Clopidogrel/acetylsalicylic acid $99,000 per QALY gained	Dabigatran 110 mg was by dominated Dabigatran 150 mg Dabigatran 150 mg $13,810 per QALY gained compared to Genotype-guided AC Warfarin was dominated by Genotype-guided AC	Trial-like scenario: Dabigatran (sequential): $10,440 per QALY gained Dabigatran 150 mg $9041 per QALY gained Dabigatran 110 mg $29,994 per QALY gained Real-world scenario: Dabigatran (sequential): $3962 per QALY gained	Before age of 80: Dabigatran compared to warfarin: £4831 per QALY gained At age 80 years and above: Dabigatran compared to warfarin: £7090 per QALY gained Scenario analysis: Dabigatran compared to acetylsalicylic acid: £3457 per QALY gained Dabigatran compared to no treatment: Dominated
Sensitivity analysis
One-way	Key parameters influencing cost-effectiveness: Time horizon Discount rate Patients ≥80 years (dabigatran 110 mg) Mean TTR Analysis perspective Degree of INR-control Reduction of risk of stroke Cost for INR-measurements Decrease in degree of disability when treated with dabigatran Scenario 1: The ICER ranged from €14,651–€57,719 per QALY gained Scenario 2: The ICER ranged from €11,519–€52,160 per QALY gained	Key parameters influencing cost-effectiveness: IS and major bleeds (if CHADS2 score = 2 and annual risk of haemorrhage >6% dabigatran 150 mg dominated. If CHADS2 score ≥3 dabigatran 150 mg dominated) Cost of dabigatran (if <$1800 per year then dabigatran 150 mg dominated)	Key parameters influencing cost-effectiveness: Mean TTR in Genotype-guided AC (dabigatran dominated Genotype-guided AC unless when TTR >98%) Utility of warfarin and dabigatran (dabigatran dominated warfarin unless the utility of warfarin ≥0.99 or dabigatran ≤0.95) Difference in risk of stroke in dabigatran 150 mg and warfarin treatment (ICER ranged between $10,000–$38,000) Risk of stroke with warfarin in INR target range	Key parameters influencing cost-effectiveness: Degree of INR-control (varying the fraction of patients in INR-range from 40–80% then the ICER increased from $1716 to $23,263 per QALY gained) Difference in RR and rate of long-term disability of IS in dabigatran and warfarin treatment (ranging RR in dabigatran 110 mg and 150 mg between 0.51 and 0.58 to 1.33 and 1.03, respectively, then the ICER was between $4186–$30,521 per QALY gained) Cost of long-term follow-up care for patients with disabilities Time horizon (a 10-year time horizon increased the ICER to $23,555 per QALY gained) Varying incremental clinical costs and parameters from 20–30% resulted in an ICER ranging from $13,130–$30,521 per QALY gained	Key parameters influencing cost-effectiveness: Degree of INR-control (dabigatran dominated unless TTR was 91–97% (<80 years) and 80–83% (≥80 years)) Costs of INR-measurements Difference in RR and overall rates of IS, ICH, and HS between dabigatran and warfarin treatment (using the upper limit of 95% CI for the RR of IS, ICH, or HS increased the calculated ICERS to £13,353, £10,013, and £8,420 per QALY gained) Long-term cost of follow-up care on disability Time horizon (a 10-year time horizon increased the ICER to £11,898 per QALY gained)
Multi-way	–	If TTR was <57.1% and a CHADS2 score ≥2 then dabigatran 150 mg dominated	If warfarin utility was between 0.95–1.00 and TTR 77–98% then Genotype-guided AC dominated If utility of warfarin was the same as or higher than dabigatran then Genotype-guided AC dominated	–	–
Probabilistic	10,000 Monte Carlo simulations: Scenario 1: Dabigatran was cost-effective in 96.4% of cases compared to warfarin Scenario 2: Dabigatran was cost-effective in 99.9% of cases compared to prescribing pattern in clinical practice Sub-group analysis: Dabigatran was cost-effective in 99.4% of cases compared to warfarin with poor INR-control Dabigatran was cost-effective in 99.3% of cases compared to warfarin with good INR-control	–	10,000 Monte Carlo simulations: Dabigatran was cost-effective in 51.6% of cases, whereas Genotype-guided AC was in 46.2% of cases warfarin was cost-effective in 0.6% of cases and dabigatran 110 mg in 1.6% of cases	10,000 Monte Carlo simulations: Trial-like scenario: Dabigatran (sequential) was cost-effective in 82% of cases compared to warfarin Dabigatran 150 mg was cost-effective in 81% of cases compared to warfarin Dabigatran 110 mg was cost-effective in 42% of cases compared to warfarin Real-world scenario: Dabigatran (sequential) was cost-effective in 99% of cases compared to warfarin	5000 Monte Carlo simulations: At the low WTP threshold dabigatran was cost-effective in 98% of cases compared to warfarin in patient <80 years and 100% compared to acetylsalicylic acid as well as no treatment At the low WTP threshold dabigatran were cost-effective in 63% of cases compared to warfarin in patient ≥80 years
Conclusions	Dabigatran was a cost-effective alternative to warfarin in Spain	Dabigatran 150 mg was a cost-effective alternative to warfarin only in patients with a CHADS2 score of three or higher or in patients with a CHADS2 score of two as well as a poor INR-control	Dabigatran was the most cost-effective alternative compared to warfarin if TTR ≤ 64% If warfarin was the same as or higher than dabigatran utility and TTR was >77% then Genotype-guided AC cost-effective	Dabigatran was a highly cost-effective alternative in all comparisons conducted	Dabigatran was likely to be cost-effective in all comparisons conducted and recommended as a first-line drug
Limitations	Costs of dabigatran patient management were not included Productivity losses were not estimated Data on utilities extrapolated from UK settings The first line drug in Spain is acenocoumarol not warfarin	Costs of dabigatran patient management were not included Productivity losses were not estimated Warfarin administration in RE-LY was not blinded Did not include a PSA taking all variable uncertainties into account	Costs of dabigatran patient management were not included Productivity losses were not estimated RE-LY based on a 2-year period whereas the Markov model was based on lifetime	Costs of dabigatran patient management were not included Productivity losses were not estimated Did not consider rehabilitation Discontinuation rates were based on RE-LY RR were used over the whole time horizon	Costs of dabigatran patient management were not included Productivity losses were not estimated Limited data available of costs of long-term disability Assuming that the decision to treat with anticoagulation is life long

AF, Atrial Fibrillation; CHADS2 score, Congestive heart failure, hypertension, age >75, diabetes mellitus, and two points for the factor prior stroke or transient ischemic attack; TTR, mean time in the therapeutic range; Genotype-guided AC, Genotype-guided anticoagulation care; HS, Haemorrhagic stroke; ICER, Incremental cost-effectiveness ratio; ICH, Intracranial haemorrhage; INR, International normalized ratio; IS, Ischaemic stroke; mg, milligrams; MI, Myocardial infarction; QALYs, Quality-adjusted life years; RE-LY, The Randomized Evaluation of Long-term Anticoagulant Therapy trial; RR, relative risk; WTP, Willingness-to-pay.

Results on the economic evaluations

In all of the economic evaluations dabigatran was associated with higher cost than warfarin treatment but provided higher amounts of QALYs11–20. This was also the case when dabigatran was compared to acetylsalicylic acid13. The real cost drivers and creators of health gain were degree of INR-control13–17,20, cost of recurrent stroke and/or intracranial haemorrhage (ICH)11,16,19, cost of bleedings11,16, CHADS2 scores17, reduction of stroke risk20, haemorrhage rates11, cost associated with INR-measurements13,14,20, difference in relative risk of stroke between dabigatran and warfarin13,14,17,19, cost of long-term follow-up care for patients with disabilities13,14, overall utility rates of IS, ICH, and HS of dabigatran in proportion to warfarin treatment13,15,18,19, age18,19, TTR19, cost of dabigatran and warfarin11,18, utility of dabigatran and warfarin12,18, costs after ICH18, death15, average TTR in genotype-guided AC12, risk of stroke with dabigatran 150 mg vs warfarin12, and stroke rate11,12. Of other key parameters affecting the ICERs were time horizon13,14,20, discount rates16,17, and analysis perspective20. Langkilde et al.16 calculated the ICER of high-dose dabigatran as below the Danish WTP threshold of €30,000 per QALY gained, which thereby was considered a cost-effective alternative to warfarin. The higher the level of stroke risk the more cost-effective dabigatran was compared to warfarin treatment. The calculations were in general evaluated as robust to all changes in parameters16. Davidson et al.17 calculated the ICER of dabigatran as far below the Swedish WTP threshold of €50,000 per QALY gained, thereby cost-effective compared to warfarin treatment. When compared to high-quality warfarin treatment the ICER increased, but was still far below WTP threshold. When taking all parameter uncertainties into account dabigatran was cost-effective in all cases, but the cost-effectiveness was largest at high risk of stroke, with a congestive heart failure, hypertension, age >75, diabetes mellitus, and prior stroke or transient ischemic attack score (CHADS2 score) = 3–617. González-Juanatey et al.20 calculated the ICER of dabigatran compared to warfarin and the ICER of dabigatran treatment compared to prescribing pattern in clinical practice as below the Spanish WTP threshold of €30,000 per QALY gained. Considering all parameter uncertainties dabigatran was cost-effective in 96.4% and 99.9% of cases compared to warfarin and prescribing pattern in clinical practice, respectively20. Sorensen et al.14 calculated the ICER of dabigatran treatment compared to trial-like scenario as below the Canadian WTP threshold of $30,000 per QALY gained and the ICER of dabigatran treatment compared to real-world scenario below the same WTP threshold. Including all parameter uncertainties dabigatran was cost-effective in 82% of cases compared to the trial-like scenario, and in 99% of cases compared to the real-world scenario14. Kansal et al.13 calculated the ICER of dabigatran treatment compared to warfarin treatment as below the English WTP threshold of £20,000–£30,000 per QALY gained at age <80 and age ≥80. Dabigatran was cost-effective in 98% of cases for patients aged <80 compared to warfarin, and in 63% of cases in patients aged ≥8013. Kamel et al.19 showed that high-dose dabigatran was likely to be a cost-effective alternative to warfarin, as the calculated ICER was below the American WTP threshold of $50,000 per QALY gained. Dabigatran was less likely to be cost-effective in patients with a high-qualitywarfarin treatment, with a mean time in therapeutic range (TTR) >73%. Taking all parameter uncertainties into account high-dose dabigatran was cost-effective in 57% of cases19. Freeman et al.18 calculated the ICER of high-dose dabigatran as below the American WTP threshold of $50,000 per QALY gained, thereby cost-effective compared to warfarin treatment. The cost-effectiveness of dabigatran treatment was highly dependent on the pricing of dabigatran. When the price of high-dose dabigatran exceeded $13.70 per day or the price of low-dose dabigatran exceeded $9.36 per day, then dabigatran was no longer cost-effective. As the daily drug price included in the model was $13.00 for high-dose dabigatran and $9.50 for low-dose dabigatran, only a small change in drug price would influence the cost-effectiveness of dabigatran. When taking all parameter uncertainties into account high-dose dabigatran was cost-effective in 53% of cases, and low-dose dabigatran was cost-effective in 30% of cases18. Pink et al.15 calculated the ICER of high-dose dabigatran to £23,082 per QALY gained, which was above the lowest English WTP threshold, but still below 30,000 per QALY gained, ranging from £20,000–30,000 per QALY gained. Thereby, high-dose dabigatran was only considered cost-effective for patients with a high risk of stroke, CHADS2 score ≥3, or in a warfarin treatment, with a TTR ≤65.5%. Combining all parameter uncertainties, warfarin was most probable to be cost-effective at WTP thresholds of £24,400 per QALY gained or lower. Low-dose dabigatran was dominated by high-dose dabigatran and both drug doses were associated with a positive incremental QALY gain compared to warfarin treatment of 86% and 94%, respectively15. Shah and Gage11 calculated the ICER of high-dose dabigatran and low-dose dabigatran compared to warfarin treatment to be well above the American WTP threshold of $50,000 per QALY gained, thereby not cost-effective, but both high-dose and low-dose dabigatran treatment gained higher QALYs than warfarin, with an incremental QALY gain of 0.25 and 0.14, respectively. The calculated ICER of high-dose and low-dose dabigatran compared to acetylsalicylic acid were neither below the American WTP threshold of $50,000 per QALY gained, and thereby not cost-effective, but both high-dose and low-dose dabigatran gained higher QALYs than acetylsalicylic acid, with an incremental QALY gain of 0.48 and 0.37, respectively. If the cost of dabigatran did not exceed $1800 per year, dabigatran treatment would have been cost-effective in all cases. High-dose dabigatran is cost-effective compared to warfarin in patients with a CHADS2 score ≥2 if annual risk of haemorrhage is >6%, for patients with a CHADS2 score ≥2 if TTR <57.1%, and in patients with a CHADS2 score ≥3 unless TTR >72.6%11. You et al.12 calculated the ICER of high-dose dabigatran as below the Chinese WTP threshold of $50,000 per QALY gained compared to genotype-guided AC, and high-dose dabigatran dominated low-dose dabigatran. High-dose dabigatran was significantly more costly (p < 0.001), mean difference $10,416, and gained significantly higher QALYs (p < 0.001), mean difference 0.217 QALYs, than genotype-guided AC. Taking all parameter uncertainties into account, high-dose dabigatran was cost-effective in 51.6% of cases compared to genotype-guided AC. The probability of genotype-guided AC to dominate high-dose dabigatran increased at TTR ≥77% and if QALYs gained in both treatment groups were comparable12. All calculated ICERs and results of sensitivity analysis are illustrated in Table 3.

Discussion

Quality assessment

The quality of the economic evaluations was assessed as high, yet some quality items were not met. None of the evaluations included all relevant costs and two evaluations12,19 did not state a potential conflict of interest, which results in a low transparency. Four evaluations12,14,15,18 did not compare their findings to other studies, thus they appeared less generalizable. Two evaluations did not apply a PSA11,16, thereby not estimating the probability of cost-effectiveness when including all parameter uncertainties. All evaluations were assigned the evidence level 2b, which indicates a high level of evidence, thereby this review is a level 2a review. The quality assessment score and estimated level of evidence of each economic evaluation are illustrated in Table 1.

Model characteristics

It was not possible to fully compare the evaluations, as the results were calculated on the basis of varying alternative treatment approaches, and thereby they were not fully homogeneous. It might be more accurate to compare dabigatran treatment to the actual prescribing pattern of warfarin in clinical practice, as the treatment quality in a trial setting might be higher than in clinical practice due to a higher degree of monitoring.

Furthermore, variations in included costs, adverse events, annual discount rates, time horizons, and WTP thresholds contributed to a decrease in homogeneity, which makes it impossible to directly compare the ICERs of the evaluations. The fact that different health outcomes and costs were included in the Markov models might imply that incorrect estimates have been made in some of the evaluations, which would decrease the validity of the evaluations in question. Despite these variations in estimates the difference in gained QALYs between dabigatran and warfarin treatment seemed the same in the evaluations, which is illustrated in Table 1. The explanation might be that most QALY gains were calculated on the basis of the average rates of efficiency and adverse events in the RE-LY trial. Thereby, it can be questioned whether calculating the amount of QALYs on the basis of health rates from each country in question, rather than on the basis of rates from a trial covering 44 countries, would have provided more accurate results of the cost-effectiveness.

The evaluations by Shah and Gage11 and Langkilde et al.16 were the only ones not applying a PSA, which is against the recommendations by the National Institute for Clinical Excellence (NICE)21. Thereby the probability of cost-effectiveness is not calculated, including all parameter uncertainties. Additionally, NICE recommends to include a cost-effectiveness acceptability curve every time the cost-effectiveness of two or more alternatives are compared21, which was not included in the evaluations by Davidson et al.17 and Kamel et al.19. Thereby, it was not possible to estimate probabilities of cost-effectiveness at different WTP thresholds.

Strengths and limitations

The quality and level of evidence of the included evaluations were assessed ensuring a high-quality review. The low homogeneity due to, e.g., different costs and adverse events made it difficult to fully compare the evaluations.

Maybe it should be considered whether the best available treatment alternatives were compared in the RE-LY trial as self-monitored warfarin treatment might be superior to common warfarin treatment. This is relevant to consider in economic evaluations. Furthermore, the up-coming anticoagulation drugs rivaroxaban and apixaban ought to be considered in future research.

Conclusions

Dabigatran is likely to be a cost-effective alternative to warfarin treatment in Denmark, the UK, Sweden, Spain, and the US. In some countries dabigatran is only cost-effective in sub-groups, such as patients in low quality warfarin treatment and high risk group of stroke. Thereby, the healthcare professionals can take this knowledge into account making decisions in clinical practice regarding dabigatran and warfarin treatment.

Transparency

Declaration of funding

The study was supported by Danish Center for Healthcare Improvements, Faculty of Social Sciences and Faculty of Health Sciences, Aalborg University, Denmark.

Declaration of financial/other relationships

MBA is an employee at Boehringer Ingelheim Denmark which holds marketing approval of dabigatran, and he is co-author of one of the economic evaluations included. The other authors declare no conflicts. The study was kept objective by all authors and there were no restrictions placed on the material presented in the paper.

Acknowledgements

There is no non-author assistance to be stated.