The objective was to assess the cost-effectiveness of dabigatran, compared with warfarin, in patients with atrial fibrillation and a history of stroke or transient ischaemic attack. The authors concluded that dabigatran was likely to be cost-effective for stroke prevention. Limited reporting of the methods and evidence makes it difficult to assess the authors' conclusions.

Cost-utility analysis

The objective was to assess the cost-effectiveness of dabigatran, compared with warfarin, in patients with atrial fibrillation and a history of stroke or transient ischaemic attack.

The dose of warfarin was adjusted until patients achieved an international normalised ratio (measure of blood clotting) of between two and three. This was compared with 150mg dabigatran twice daily.

USA/out-patient secondary care.

Analytical approach:

A Markov decision model was used to assess the costs and outcomes of the two options. The time horizon was 20 years. The authors reported that a societal perspective was adopted.

Effectiveness data:

The clinical and effectiveness data were from published studies. The main measure of effectiveness was the probability of adverse events, such as ischaemic stroke, intracranial haemorrhage, myocardial infarction, and death. These estimates were from a sub-study of the Randomised Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial of dabigatran and warfarin in secondary stroke prevention, and other trials of warfarin therapy for atrial fibrillation.

Monetary benefit and utility valuations:

The utility estimates for no disability on different drugs, and for the adverse events, were from published studies. There were no data available on the utility for patients on dabigatran, so it was assumed to be the same as that of patients on ximelagatran, an older thrombin inhibitor.

Measure of benefit:

The measure of benefit was quality-adjusted life-years (QALYs) gained. As benefits could be generated over 20 years, future benefits were discounted at an annual rate of 3%.

Cost data:

The direct costs were those of dabigatran, office doctor visits, warfarin, anticoagulation management, international normalised ratio tests, hospitalisations for adverse events, and ongoing medical care for adverse events. The costs of dabigatran were based on its wholesale price. The costs of doctor visits, tests and management were from Medicare reimbursements. Hospitalisation costs were from the Healthcare Cost and Utilization Project. The price year was 2010 and costs were discounted at an annual rate of 3%. All costs were reported in US $.

Analysis of uncertainty:

One-way sensitivity analyses were performed by varying all the model inputs over plausible ranges of values. A probabilistic sensitivity analysis was undertaken, using Monte Carlo simulation, with 10,000 iterations.

The QALYs gained were 4.27 with dabigatran, compared with 3.91 with warfarin. The additional average cost for dabigatran, over warfarin, was $9,000 per patient.

Compared with warfarin, dabigatran had an incremental cost-utility ratio of $25,000 per QALY gained.

The one-way sensitivity analyses showed that the results were most sensitive to variations in the costs of recurrent strokes, age, and the relative risk of stroke while on dabigatran. The probabilistic sensitivity analysis showed that at a willingness-to-pay threshold of $50,000 per QALY gained, dabigatran was cost-effective, over warfarin, in 57% of simulations.

The authors concluded that dabigatran was likely to be cost-effective for stroke prevention, unless excellent control could be achieved on warfarin.

Interventions:

The interventions were reported appropriately. Warfarin was a common therapy in this setting for patients with atrial fibrillation, but there might have been other relevant comparators that were not included.

Effectiveness/benefits:

The clinical and effectiveness estimates were from published studies. All the variables used in the model, with the main estimates and their sources, were reported in appendices. No systematic review was reported to identify these sources and it is not possible to determine if all the relevant data were included. It was not clear if the different studies provided different risks of adverse events or if an evidence synthesis was conducted. Various trials were referenced, but in the discussion the authors referred to only the main trial.

Costs:

The authors reported that a societal perspective was adopted, but only the health care costs appear to have been included. A societal perspective would usually include productivity losses and informal care. For a health care system perspective, it appears that all the relevant costs were included. The sources for these costs were adequately reported. The price year, time horizon and discount rate were reported, but it was not clear if any costs had to be adjusted to the price year and how that was achieved.

Analysis and results:

The cost and outcome information was synthesised using a Markov model. The model structure was described and a diagram was given in the online appendices. The impact of uncertainty on the model’s results was appropriately tested in one-way and probabilistic sensitivity analyses. The main limitation of the study, as reported by the authors, was that the effectiveness data were mainly from a sub-study of one trial.

Concluding remarks:

Limited reporting of the methods and evidence makes it difficult to assess the authors' conclusions.

Not stated.