This study aims to estimate the cost-effectiveness of dabigatran 150 mg twice daily versus warfarin for stroke and systemic embolism risk reduction in patients with nonvalvular atrial fibrillation initiating treatment before age 75 (<75), at or after age 75 (≥75), and the overall population (All) from a US Medicare payer perspective. Clinical event rates by age cohort with dabigatran or warfarin for safety-on-treatment and intent-to-treat populations were estimated from Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY). An economic model was adapted using these data to evaluate the impact of starting age on clinical and economic outcomes. Costs were obtained from Medicare payment schedules and utilities from publications. Model outputs included event rates, costs, quality-adjusted life-years, and incremental cost-effectiveness ratios. The RE-LY analysis shows that the <75 cohort has lower rates of all events than the ≥75 cohort; versus warfarin, dabigatran performed better in main efficacy and safety in all age cohorts with the exception of extracranial hemorrhage in the ≥75 cohort. The clinical event costs avoided per patient for dabigatran were $1,100, $135, and $713 for cohorts <75, ≥75, and All, respectively. Extrapolating over a lifetime horizon, the model found that dabigatran resulted in lower rates of stroke and intracranial hemorrhage and higher rates for extracranial hemorrhage versus warfarin for all age cohorts. Lifetime quality-adjusted life-years and costs were higher for dabigatran than warfarin, resulting in incremental cost-effectiveness ratios of $52,773, $65,946, and $56,131 for cohorts <75, ≥75, and All, respectively. In conclusion, dabigatran was cost-effective versus warfarin in US patients with atrial fibrillation regardless of age of treatment initiation.
Atrial fibrillation (AF) is a common arrhythmia and the leading cause of stroke, which is associated with a high humanistic and economic burden. The Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial demonstrated the safety and efficacy of dabigatran versus warfarin for the prevention of stroke and systemic embolism, and in additional analyses, the efficacy and safety by age cohorts in patients with AF at moderate to high risk of stroke. The effects of dabigatran compared with warfarin on stroke or systemic embolism were consistent across age cohorts, but a significant treatment-by-age interaction was seen in major bleeding outcomes. In the dabigatran 150 mg twice daily dose arm, the dose approved by the FDA, patients <75 years had lower rates of intracranial hemorrhage (ICH) and extracranial hemorrhage (ECH) than those treated with warfarin, whereas patients ≥75 years had a lower risk of ICH but higher rates of ECH than warfarin-treated patients. In addition, the overall risk of stroke increases with age, as indicated by the CHADS 2 scoring increasing by a point at age 75. Given the different safety results by age group and the increasing risk of stroke with age, there is a need to understand the cost-effectiveness for each age group. The objective of this study is to investigate the cost-effectiveness of dabigatran etexilate 150 mg twice daily versus warfarin in patients with AF in whom anticoagulation is appropriate, in cohorts initiating treatment before age 75 (<75), those initiating at or after age 75 (≥75), and all RE-LY trial patients (All) from a US payer perspective. The analysis uses efficacy and safety data of the RE-LY trial by age cohorts specifically generated from individual patient data for this analysis.
Methods
A previously published Markov model was adapted to simulate anticoagulation treatment in individual age cohorts and the resulting clinical events in the United States. The model followed patients with nonvalvular AF at risk of relevant clinical events through the natural course of the disease until the end of their lives. Using a US Medicare payer perspective, a lifetime horizon was considered in the base case.
The clinical events included were primary and recurrent ischemic stroke, transient ischemic attack, systemic embolism, acute myocardial infarction, ICH (including hemorrhagic stroke), ECH, and death. Patients could experience no clinical event, 1 of the events, and/or death during any 3-month model cycle. The key consequences of the clinical events were a change in treatment status (switching treatment to aspirin or permanent discontinuation of treatment), a change in functional disability after a stroke or ICH (defined by modified Rankin Score [strokes] or Glasgow Outcomes Scale [ICH]), and/or a reduction in quality of life and death. Patients could also switch treatment for reasons not related to a clinical event. Patients were also subject to mortality from other causes at every cycle using age-, gender-, and event-adjusted all-cause mortality data.
Three patient cohorts were defined: those initiating anticoagulation treatment <75 years, ≥75 years, and all patients (All). Gender distribution and acute myocardial infarction history were assigned at baseline. Each cohort included subcohorts for each year of age at baseline with specific CHADS 2 and previous stroke distribution for that baseline age obtained from a RE-LY analysis by starting age.
Also obtained from the age analysis were the baseline risks of each event while on warfarin and the relative risks of dabigatran versus warfarin. In addition to the bleeding events in the intent-to-treat population reported by Eikelboom et al, the new analysis reported all clinical events for patients in each cohort for both the safety-on-treatment and the intent-to-treat populations.
The safety-on-treatment population data (based only on patients who received at least 1 dose of a study drug and were followed for events from first to last dose plus 6 days, regardless of adherence to the protocol or temporary discontinuations ) were used for the “on-treatment” health states in the model, whereas the treatment efficacy of aspirin and no treatment were applied for patients who switched or discontinued treatment. The use of the safety-on-treatment population for the clinical outcomes avoided double counting of the effects of treatment discontinuation.
The baseline risks of clinical events were based on rates in the warfarin arm for patients aged <75. Relative risks of the age (≥75 vs <75) and treatment (dabigatran vs warfarin) were calculated using age-stratified data from RE-LY. Relative risks of aspirin and no treatment versus warfarin were obtained from a network meta-analysis. Postevent disability was based on analysis of the RE-LY trial. For patients receiving warfarin, 22.1% of ischemic stroke events were fatal; and 4.3%, 19.7%, and 53.9% were completely dependent, moderately dependent, and independent functioning after ischemic stroke events, respectively. This distribution was 25.1%, 1.6%, 15.6%, and 57.7%, respectively, in the dabigatran arm. After ICH events, 51.6% of the events were fatal, 31.8% of the patients were completely dependent, 8.8% were moderately dependent, and 7.8% were independent functioning regardless of current anticoagulation treatment. The proportions of ECH that are gastrointestinal bleeding were obtained from RE-LY and estimated to be 48.7% and 35.7% in the dabigatran and warfarin arms, respectively.
Event and disability costs were calculated using Medicare reimbursement data ( Table 1 ). The acute event costs and long-term follow-up costs were obtained from a database analysis on the long-term costs of stroke and major bleeding events in patients with nonvalvular AF. The study analyzed cost data of nondisabled Medicare patients. In contrast to commercial claims data sources, Medicare has a large number of patients older than 65 years, which fits the model population well. The cost components included in the analysis are inpatient hospital, outpatient hospital, skilled nursing facility, hospice, home health agency, and durable medical equipment. The study captured costs up to 3 years after stroke or bleeding events. Acute event costs were based on costs of the first 3 months postevent. Long-term follow-up costs were based on average costs from the fourth month through 3 years postevent, prorated to fit the model cycles. Stratification of the long-term follow-up costs by disability level was not available from the study by Mercaldi et al. To differentiate the cost consequences of stroke and bleeding events resulting in different disability levels, the unstratified follow-up costs were adjusted to reflect the dependence of costs on disability level according to a population-based study on long-term care cost of patients with AF in the United Kingdom. All costs were adjusted to 2013 values using the medical component of the consumer price index. Acute event costs not available from the study by Mercaldi et al were based on the US diagnosis-related group data. Health-related quality of life was calculated for each health state using published utility data.
| Drug Costs | Per Day | Source ∗ |
|---|---|---|
| Dabigatran 150 mg twice daily | $8.84 | Red Book™ |
| Warfarin | $0.16 | Red Book™ |
| Aspirin | $0.01 | Red Book™ |
| Treatment Monitoring Cost | Annual | Source |
|---|---|---|
| Warfarin | $285.10 | Assume 14 international normalized ratio tests Shah et al Physicians fee and Coding Guide 2012 (CPT 99211) |
| Event Costs | Per Acute Event | Source |
|---|---|---|
| Ischemic stroke, fatal and non-fatal | $22,653.16 | Mercaldi et al 2012 |
| Systemic embolism, fatal and non-fatal | $7,291.83 | Mercaldi et al 2011 |
| Transient ischemic attack | $3,905.03 | Mercaldi et al 2011 |
| Intracranial hemorrhage and hemorrhagic stroke | $34,572.50 | Mercaldi et al 2012 |
| Extracranial hemorrhage (non-brain), fatal and non-fatal | $16,769.55 | Mercaldi et al 2012 |
| Minor bleed | $211.05 | Physicians fee and Coding Guide 2012 (CPT 99215) |
| Acute myocardial infarction, fatal | $5,447.99 | DRG 283, 284, 285 |
| Acute myocardial infarction, non-fatal | $6,560.89 | DRG 280, 281, 282 |
| Follow-up Costs | Per Quarter | Source |
|---|---|---|
| Independent with stroke history | $203.67 | Mercaldi et al 2012; Luengo-Fernandez et al 2012 |
| Moderate disability | $1,764.56 | |
| Dependent disability | $3,746.33 |
∗ Drug costs were obtained January 2014. All the other costs were inflated to 2013 value.
Model outcomes presented in this analysis include the number of clinical events per 100 patient-years, total costs, breakdown by cost type, and quality-adjusted life-years (QALYs). Incremental cost-effectiveness ratios (ICERs) were also computed as incremental cost in the dabigatran arm versus the warfarin arm divided by incremental QALYs. These analyses were conducted for each of the 3 age cohorts. All costs and outcomes were discounted by 3% annually.
The model was validated by comparing the results of the model output over a 2-year time horizon for dabigatran and warfarin with the intent-to-treat results of the RE-LY trial. One-way sensitivity analyses were performed by varying time horizon and discount rates, and clinical inputs such as relative risks (upper and lower 95% confidence interval [CI]), event rates, treatment discontinuation, and event and long-term follow-up costs. Because starting age determines the length of a lifetime analysis, a limited time horizon of 10 years was examined to create a similar follow-up time for the 3 cohorts. Relative risks were varied based on the 95% CI. A scenario with event rates based on the intent-to-treat population and excluding treatment discontinuation was also explored. Scenario analyses were also conducted to investigate the impact of aging and stroke events on ischemic stroke risk.
A probabilistic sensitivity analysis characterized the uncertainty surrounding the cost-effectiveness results. All model parameters were simultaneously varied based on statistical distribution and their 95% CI. Beta distributions were assumed for baseline event probabilities and utility estimates, log-normal distributions for relative risks, and gamma distributions for event and health state costs.
Results
The results of the RE-LY analysis by initiating age cohort for the safety-on-treatment and intent-to-treat populations are presented in Table 2 . Rates of all events were higher in the older cohort at treatment initiation versus the younger cohort. In the intent-to-treat population, the annual risks of ischemic stroke in the warfarin arm increased steadily with the baseline CHADS 2 score in both age cohorts. Dabigatran treatment was associated with a lower risk of stroke across all CHADS 2 scores for both age cohorts. However, because of small sample sizes, no trends in the relative risks of dabigatran versus warfarin were seen in terms of increasing CHADS 2 risk. In the <75 age cohort, the risks of all other events were also lower for dabigatran-treated patients, except a numerical increase in acute myocardial infarction. In the older cohort, dabigatran had lower risk of all events except acute myocardial infarction, minor bleeds, and ECH. The distributions of baseline CHADS 2 scores by age are presented in Figure 1 . In the group of patients between 65 and 75 years, there was a greater proportion of patients in lower CHADS 2 categories than in older patients.
| Variable | Warfarin | Dabigatran 150 mg Twice Daily | ||
|---|---|---|---|---|
| Annual Rate Per 100 Patient-years, Age <75 | Annual Rate Per 100 Patient-years, Age ≥75 | Relative Risk, Age <75 vs. Warfarin | Relative Risk, Age 75≥ vs. Warfarin | |
| Safety-on-treatment | ||||
| Ischemic stroke, CHADS 2 = 0–1 ∗ | 0.61 | 0.82 | 0.75 (0.37, 1.55) | 0.95 (0.27, 3.28) |
| Ischemic stroke, CHADS 2 = 2 ∗ | 0.79 | 1.06 | 0.64 (0.29, 1.40) | 0.58 (0.27, 1.24) |
| Ischemic stroke, CHADS 2 = 3–4 ∗ | 1.51 | 1.36 | 0.64 (0.33, 1.24) | 0.80 (0.41, 1.55) |
| Ischemic stroke, CHADS 2 = 5–6 ∗ | 2.28 | 3.47 | 0.93 (0.13, 6.63) | 0.43 (0.13, 1.43) |
| Systemic embolism ∗ | 0.14 | 0.19 | 0.57 (0.19, 1.69) | 0.40 (0.11, 1.51) |
| Stroke/systemic embolism | 1.31 | 1.87 | 0.56 (0.39, 0.80) | 0.56 (0.38, 0.82) |
| Transient ischemic attack ∗ | 0.71 | 1.10 | 0.80 (0.52, 1.23) | 0.78 (0.50, 1.22) |
| Intracranial hemorrhage (including hemorrhagic stroke) ∗ | 0.60 | 1.03 | 0.34 (0.18, 0.63) | 0.30 (0.16, 0.56) |
| Extracranial hemorrhage ∗ | 2.36 | 3.47 | 0.80 (0.63, 1.02) | 1.50 (1.21, 1.86) |
| Acute myocardial infarction ∗ | 0.49 | 0.74 | 1.40 (0.89, 2.21) | 1.20 (0.74, 1.95) |
| Minor bleeds ∗ | 16.76 | 19.15 | 0.80 (0.74, 0.88) | 1.09 (0.99, 1.20) |
| Disabling or fatal stroke | 0.76 | 1.03 | 0.48 (0.29, 0.78) | 0.54 (0.33, 0.91) |
| Death from any cause | 2.47 | 2.92 | 0.69 (0.54, 0.89) | 1.01 (0.79, 1.31) |
| Intent-to-treat | ||||
| Ischemic stroke, CHADS 2 0–1 | 0.72 | 0.91 | 0.65 (0.34, 1.25) | 0.74 (0.23, 2.43) |
| Ischemic stroke, CHADS 2 2 | 0.77 | 1.29 | 0.77 (0.38, 1.57) | 0.59 (0.31, 1.11) |
| Ischemic stroke, CHADS 2 3–4 | 1.74 | 1.76 | 0.79 (0.45, 1.37) | 0.87 (0.52, 1.46) |
| Ischemic stroke, CHADS 2 5–6 | 2.05 | 3.82 | 0.98 (0.14, 6.96) | 0.90 (0.38, 2.12) |
| Systemic embolism | 0.16 | 0.21 | 0.72 (0.29, 1.78) | 0.49 (0.17, 1.43) |
| Stroke/systemic embolism | 1.42 | 2.15 | 0.63 (0.46, 0.86) | 0.67 (0.49, 0.90) |
| Transient ischemic attack | 0.70 | 1.04 | 0.83 (0.55, 1.25) | 0.90 (0.60, 1.35) |
| Intracranial hemorrhage (including hemorrhagic stroke) | 0.61 | 1.00 | 0.43 (0.25, 0.74) | 0.39 (0.23, 0.67) |
| Extracranial hemorrhage | 2.44 | 3.45 | 0.78 (0.62, 0.97) | 1.39 (1.14, 1.71) |
| Acute myocardial infarction | 0.54 | 0.79 | 1.34 (0.88, 2.03) | 1.20 (0.77, 1.85) |
| Minor bleeds | 15.70 | 17.39 | 0.81 (0.74, 0.88) | 1.06 (0.97, 1.17) |
| Disabling or fatal stroke | 0.80 | 1.32 | 0.59 (0.38, 0.90) | 0.73 (0.50, 1.06) |
| Death from any cause | 3.47 | 5.13 | 0.77 (0.63, 0.92) | 0.99 (0.83, 1.19) |
The validation analysis showed that the event rates predicted in the model largely matched the event rates in the intent-to-treat population from the RE-LY trial, with all model outcomes within 5% of the trial intent-to-treat population except ECH (overestimated for dabigatran by 9%) and transient ischemic attack (overestimated in both arms by 10%). This validation demonstrates that the model replicated the clinical trial with on-treatment clinical inputs based on safety-on-treatment data and simulated effects of treatment switching and discontinuation.
The model calculated overall and specific clinical events per 100 patient-years ( Table 3 ). In all age subgroups, the model found that in comparison with warfarin-treated patients, dabigatran-treated patients experienced fewer instances of ischemic stroke and ICH and more instances of ECH and acute myocardial infarction over the lifetime horizons. The differences in the number of events experienced per 100 patient-years by dabigatran versus warfarin-treated patients were the greatest in the age ≥75 cohort, with the exception of acute myocardial infarction for which the age <75 cohort showed the greatest difference.
| Dabigatran | |||
|---|---|---|---|
| Age <75 | Age ≥75 | All Patients | |
| All events (excluding minor bleed) ∗ | 9.31 | 9.86 | 9.47 |
| Ischemic stroke (overall) | 1.84 | 1.60 | 1.77 |
| Ischemic stroke (disabling and fatal) | 0.88 | 0.74 | 0.84 |
| Intracranial hemorrhage and hemorrhagic stroke | 0.41 | 0.41 | 0.41 |
| Extracranial hemorrhage | 4.55 | 5.45 | 4.80 |
| Acute myocardial infarction | 1.16 | 1.10 | 1.14 |
| Systemic embolism and transient ischemic attack | 1.35 | 1.30 | 1.34 |
| Minor bleed | 20.31 | 21.04 | 20.51 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
