Summary
Background
Atrial fibrillation is the main cause of stroke, but the risk can be reduced, usually with vitamin K antagonists (VKAs) such as warfarin. The RE-LY atrial fibrillation study demonstrated that the rates of stroke and systemic embolism with dabigatran (an oral direct thrombin inhibitor) were similar to or lower than those with warfarin.
Aims
To estimate the cost-effectiveness, from a French payer perspective, of dabigatran (150 or 110 mg bid for patients < or ≥ 80 years, respectively) versus warfarin.
Methods
Cost-effectiveness was modeled using a Markov model in a cohort of 10,000 patients with atrial fibrillation followed over their lifetime. Events accounted for included ischemic stroke, systemic embolism, transient ischemic attack, hemorrhage, myocardial infarction and death. The model patient population matched the RE-LY patients. Dabigatran was compared with “trial-like” warfarin and “real-world” prescribing. Risks of clinical events were obtained from RE-LY. Event and follow-up costs were based on the French national tariff or published literature. Clinical events, QALYs, total costs and incremental cost-effectiveness ratios (ICERs) were calculated.
Results
The ICERs of dabigatran compared with “trial-like” warfarin and “real-world” prescribing were €15,838/QALY and €7473/QALY, respectively. Deterministic and probabilistic sensitivity analyses showed these to be robust to uncertainty and variability in the model parameters. The ICER for dabigatran was below €24,000/QALY or €36,000/QALY in 71% or 92%, respectively, of the simulations when compared with “trial-like” warfarin and 100% and 100%, respectively, when compared with “real-world” prescribing.
Conclusion
This study suggests that the use of dabigatran in French atrial fibrillation patients is cost-effective, according to usually accepted thresholds.
Résumé
Contexte
La fibrillation auriculaire est la principale cause d’accident vasculaire cérébral, mais le risque peut être diminué, principalement avec des antivitamines K (AVK) tels que la warfarin. L’étude RE-LY a démontré que les taux d’AVC et d’embolie systémique avec dabigatran étaient similaires ou inférieurs à ceux observés chez des patients traités par warfarine (AVK).
Objectif
Estimer le coût-efficacité, dans une perspective payeur, de dabigatran (150 mg ou 110 mg deux fois par jour chez les patients âgés de moins et de plus de 80 ans, respectivement) versus warfarin.
Méthodes
Le coût-efficacité a été modélisé à l’aide d’un modèle de Markov appliqué à une cohorte de 10 000 patients présentant une fibrillation jusqu’au décès. Les événements cliniques pris en compte étaient : les AVC ischémiques, les embolies systémiques, les accidents ischémiques transitoires, les hémorragies intra- et extra-crâniennes, les AVC hémorragiques, les infarctus du myocarde ainsi que les décès. La population étudiée présente les mêmes caractéristiques que celles de la population RE-LY. Le dabigatran a été comparé à la warfarin en condition expérimental ( trial-like warfarin) et dans la vraie vie ( real-world prescribing) Les risques d’occurrence des événements cliniques sont ceux de l’essai RE-LY. Les coûts associés aux événements ainsi que les coûts de suivi ont été calculés à partir soit des tarifs nationaux français, soit de la littérature. Le nombre d’événements cliniques, les QALYs, les coûts totaux ainsi que les ratios coût-efficacité incrémentaux ont été calculés.
Résultats
Les ratios coût-efficacité incrémentaux du dabigatran comparativement aux AVK, en condition expérimentale et dans la vraie vie, étaient, respectivement, de 15 838 €/QALY et 7473 €/QALY. Les analyses de sensibilité déterministes et probabilistes montrent que ces ratios sont robustes à l’incertitude et la variabilité des paramètres du modèle. Le ratio coût-efficacité incrémental du dabigatran est en dessous de 24 000 €/QALY et 36 000 €/QALY dans respectivement 71 % et 92 % des simulations en contexte expérimental et dans, respectivement, 100 % et 100 % des simulations dans la vraie vie.
Conclusion
L’utilisation du dabigatran chez des patients français avec une fibrillation auriculaire est coût-effective selon les seuils couramment acceptés.
Background
Atrial fibrillation (AF) is the most common arrhythmia in France. The incidence increases with age, and 10% of people over 80 years of age have the condition . AF is the main cause of stroke, and as strokes in AF patients are particularly severe and disabling, their prevention is of great importance . Vitamin K antagonists (VKAs) and aspirin are presently the main preventive agents . VKAs have been shown to reduce the risk of stroke by 68% compared with aspirin , but have a narrow therapeutic window of anticoagulation and have many interactions with food and other drugs. If overdosed, VKA administration presents an excess risk of hemorrhage, while protection from stroke is diminished in case of underdosing. Thus, VKAs require regular international normalized ratio (INR) monitoring and dose adjustments to be effective and safe.
Dabigatran is an oral direct thrombin inhibitor that provides stable anticoagulation with a fixed dose and no need for anticoagulation monitoring. It has been marketed since 2008 for the prevention of venous thromboembolic events in patients undergoing scheduled surgery for total knee or hip replacement. The RE-LY study demonstrated that dabigatran is effective in the prevention of thromboembolic events in patients with AF. It compared two doses of dabigatran (110 or 150 mg twice daily [bid]) to warfarin (dosage adjusted on INR). It demonstrated that the rates of stroke and systemic embolism were similar (110 mg bid) or lower (150 mg bid) than those observed in patients treated with warfarin. The risk of major hemorrhage was similar (150 mg bid) or lower (110 mg bid). Although both doses of dabigatran have been approved, the higher dose is recommended for patients < 80 years old and the lower dose for patients aged ≥ 80 years .
The aim of the present study was to assess the cost-effectiveness of dabigatran compared with VKAs in patients with AF using a Markov model. Following the recommendations, the model evaluated dabigatran 150 and 110 mg bid doses for patients aged < 80 and ≥ 80 years, respectively.
Methods
Patient population
We considered a cohort of 10,000 patients with AF followed for their remaining lifetime to capture the lifelong consequences of stroke and hemorrhage. The population matched the criteria of the RE-LY trial . Patients had diagnosed AF and at least one of the following characteristics: at least one additional risk factor for stroke or embolism (as defined by the CHADS 2 score risk stratification scheme) or impaired left ventricular ejection fraction. They were eligible for anticoagulation treatment but were not taking concomitant anticoagulation treatment. The CHADS 2 score ranged from 0 to 6 (mean 2.1). Twenty percent of patients had a history of previous stroke or transient ischemic attack. The mean age of the patients was 69 years and 65% were men.
Model structure
The model is an adaptation of Sorensen et al.’s cost-effectiveness model to the French setting. It has been fully described elsewhere . The cost-effectiveness of dabigatran was modeled with a 3-month cycle Markov model. The different events taken into account were: ischemic stroke, intracranial hemorrhage, hemorrhagic stroke (all fatal, independent, moderate disability and totally dependent); systemic embolism, extracranial hemorrhage, acute myocardial infarction (all fatal and non-fatal); transient ischemic attack and death from other causes. At each cycle of the model, patients could experience one of the pre-cited clinical events or remain unchanged in their current health state. A 3-month cycle length was used because it is unlikely that patients would experience more than one major event during any 3-month period. The consequences of the clinical events were disability as defined by the modified Rankin Score (after a stroke) or Glasgow Outcomes Scale (after intracranial hemorrhage), death, reduction in quality of life, changes in risks of future events and treatment status.
Fig. 1 presents a simplification of the Markov model. Severe hemorrhagic events could result in discontinuation of current treatment and a switch to aspirin. Patients could also discontinue for other reasons, clinical or not. A first ever stroke resulted in an increase of the CHADS 2 score of two points. The score remained unchanged after a recurrent stroke. It increased by one point when patients passed the age of 75 years. The risk of extracranial hemorrhage doubled when patients passed 70 years of age .
The model assumed that patients not discontinuing remained adherent to anticoagulation treatment and the relative treatment effect remained constant over time. Patients discontinuing treatment received no further clinical benefit.
Future costs and outcomes were discounted at 4% per annum to convert them to present day equivalents according to the recommendations of the Haute Autorité de santé (HAS; the French National Authority of Health) .
Model scenarios
In RE-LY, 64% of patients under VKAs were at their INR target, whereas in France, published data showed that only 46% are at target . Moreover, two other publications have suggested that patients presenting similar profiles to RE-LY patients with AF received other anticoagulant agent than VKAs or were not treated, leading to less efficacious stroke prevention. For this reason, we modeled two scenarios: a “trial-like” scenario (where patients received dabigatran or warfarin) and a “real-life” scenario (where patients received dabigatran, a VKA, aspirin or no anticoagulant). In the “real-life” scenario, the relative efficacy of dabigatran versus VKAs was the same as in the RE-LY trial and was derived from indirect comparisons with aspirin or no treatment . The percentages of patients taking VKAs, aspirin or untreated were derived from Cohen et al. , while data from Cegedim Strategic Data were used in a sensitivity analysis.
Data sources
Probabilities of events
Annual probabilities of events in the VKA arm and the relative risks (RRs) of dabigatran 110 and 150 mg bid, aspirin and no treatment versus VKAs were taken from the original model . All-cause mortality data adjusted for age were obtained from the French Institute of Health and Medical Research . Baseline characteristics of the patient population in the model matched the CHADS 2 distribution of patients entering into the model was the same as observed in RE-LY. Because fluindione is the most frequently used VKA in France, we assumed that outcomes from RE-LY with warfarin could be transposed.
Utilities
Since no utility values were available in the French context, utility values for each disability level and utility decrements due to clinical events were taken from the published literature . The effective decreases in utility at the time of the event were 0.139 for stroke, 0.120 for systemic embolism, 0.181 for intracranial hemorrhage, 0.103 for transient ischemic attack and 0.125 for myocardial infarction. The level of impairment corresponding to the patient’s functional status was also included: 0.65 for independent patients with stroke history, 0.46 for moderately dependent patients and 0.30 for totally dependent patients.
Resource use and costs
The baseline scenario considered only direct medical expenditures covered by the National Sickness Fund (NSF), thus adopting a health services payer perspective. The NSF covers all healthcare services expenditures (acute hospital care, ambulatory services, drug treatment, rehabilitation and healthcare delivered in “nursing home”-type facilities. The French “départements” (France is administratively divided into 96 “départements”) are responsible for partially covering expenditures related to functional dependency of patients, through payments adjusted on the degree of dependency. Finally, in nursing homes, patients have to pay out-of-pocket accommodation. HAS recommends that all direct expenses by all potential payers are taken into account when performing a cost-effectiveness analysis, therefore, a secondary scenario was considered including payments from the social services of the departments and out-of-pocket money from patients. All costs have been updated to the year 2011 according to the price index of medical services. All costs are summarized in Table 1 .
| Cost (€) | Source | |
|---|---|---|
| Drug costs per day (public prices) | ||
| Dabigatran etexilate | 2.53 | |
| Warfarin (fluindione) | 0.13 | |
| Aspirin | 0.09 | |
| Monitoring costs per year | ||
| Monthly INR monitoring | 134.04 | |
| Event costs per year | ||
| Ischemic stroke | ||
| Fatal (initial hospitalization) | 7108 | PMSI + tariffs 2011 |
| Independent (including initial hospitalization) | 11,740 | PMSI + tariffs 2011 |
| Moderate disability (including initial hospitalization) | 21,123 | PMSI + tariffs 2011 |
| Totally dependent (including initial hospitalization) | 36,690 | PMSI + tariffs 2011 |
| Intracranial hemorrhage or hemorrhagic stroke | PMSI + tariffs 2011 | |
| Fatal (initial hospitalization) | 7240 | PMSI + tariffs 2011 |
| Independent (including initial hospitalization) | 11,872 | PMSI + tariffs 2011 |
| Moderate disability (including initial hospitalization) | 21,255 | PMSI + tariffs 2011 |
| Totally dependent (including initial hospitalization) | 36,822 | PMSI + tariffs 2011 |
| Systemic embolism or transient ischemic attack | 5429 | PMSI + tariffs 2011 |
| Extracranial hemorrhage | 1550 | PMSI + tariffs 2011 |
| Acute myocardial infarction | 4860 | PMSI + tariffs 2011 |
| Follow-up costs per quarter | ||
| Independent | 699 | |
| Moderate disability | 1107 | |
| Totally dependent | 2944 | |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
