Stroke: choice of warfarin or aspirin was cost-effective in low-risk patients.

Clinical bottom line (level 1b)

Preference-based therapy with aspirin or warfarin was more cost-effective in patients with nonvalvular atrial fibrillation who were at low risk of stroke, than those given warfarin.
Preference-based therapy with aspirin or warfarin was slightly more cost-effective in patients at a medium risk of stroke, than warfarin.
Preference-based therapy with aspirin or warfarin was not cost-effective in patients at high-risk of stroke.

Gage et al: Stroke 1998; 29: 1083-1091

Expires December 2003

The study

Decision analysis model stratified by the number of stroke risk factors (history of stroke, transient ischaemic attack, hypertension, diabetes or heart disease).
Setting: two medical centres, USA

Patients with nonvalvular atrial fibrillation who were at least 50 years old and could read English. Patients were either given warfarin or given the choice of warfarin or aspirin therapy.

Viewpoint: societal

Benefit assessment: Quality-adjusted survival occurring over a 10 year period was assessed.

Resources and costs: Direct costs of prophylactic therapy (including monitoring for warfarin therapy), adverse events (stroke, TIA, haemorrhage and death) and preference elicitation (the provider time needed to elicit and incorporate a patient's preferences). All costs were estimated from a societal perspective in 1994. The cost of eliciting each patient's preference was $50. All future costs and benefits (quality-adjusted life-years gained) were discounted at a rate of 5% per annum.

Sensitivity analysis: Aspirin-for-all therapy, error in preference elicitation, patient's ability to comprehend and complete preference-based therapy, preference-based therapy in older patients (75 years), change in cost of preference elicitation, worst-case scenarios and effect of including a no antithrombotic therapy option.

Patients had a mean age of 70 years, 86% were male and 87% were white.
Patients 60-69 years old and with nonvalvular atrial fibrillation (NVAF) and none of the other stroke risk factors were considered at low risk.
With NVAF and one additional risk factor: medium risk.
With NVAF and two risk factors: high risk.

The evidence

intervention	cost
low risk of stroke and warfarin for all	$9000 (6.70 projected quality-adjusted life-years (QALY) )
low risk of stroke and preference based therapy	$8330 ( 6.75 projected QALYs )
medium risk of stroke with warfarin for all	$10,860 ( 6.60 projected QALYs )
medium risk of stroke with preference based therapy	$10,770 ( 6.62 projected QALYs )
high risk of stroke with warfarin for all	$12,490 ( 6.51 projected QALYs )
high risk of stroke with preference based therapy	$12,600 ( 6.52 projected QALYs )

Effect of sensitivity analysis: Preference-based therapy was still beneficial after all sensitivity analyses except in the worst-case scenario of high-risk patients, where quality adjusted life years were slightly decreased and cost slightly increased.

Citation

Gage BF, Cardinalli AB, Owens DK: Cost-effectiveness of preference-based antithrombotic therapy for patients with nonvalvular atrial fibrillation. Stroke 1998; 29: 1083-1091

Contributor: Clare Wotton and Musab Hayatli, December 1999
Reviewer: Santiago Alvarez Montero

_{Clinical Question.}

_Patient	_{nonvalvular atrial fibrillation}
_{Intervention or Exposure}	_{preference-based antithrombolysis}
_Comparison	_{warfarin-for-all therapy}
_Outcome	_{quality-adjusted survival and cost-effectiveness}