An economic model of adverse events and costs for oral anticoagulants used for atrial fibrillation

ABSTRACT

Objective: To construct a semi-Markov model to compare health outcomes and medical costs associated with warfarin and a second anticoagulant over 1- and 5-year periods.

Design: We posited a hypothetical cohort of 10 000 identical 70-year-old patients with atrial fibrillation. We posited 20 scenarios for events that included four possibilities for ischemic strokes (mild, moderate, severe, death) and 16 possibilities for hemorrhages. The model allowed for four levels of International Normalized Ratio. Event rates were based on outcomes in clinical trials and observational studies. Costs were estimated from the perspective of the third-party payer.

Results: The greatest cost-generating events were virtually the same for the two drugs and included severe stroke ($1 758 548 for 1 year for both drugs), moderate stroke ($380 355 for 1 year for both drugs), and severe lower gastrointestinal (GI) hemorrhage ($193 804 for 1 year for warfarin and $193 474 for second drug). The least costly events for both drugs were mild intracranial or intracerebral hemorrhage ($7584 for warfarin and $4314 for second drug) and fatal upper GI hemorrhage ($16 781 and $16 752). Total costs for adverse events over 5 years were similar: $18 330 662 for warfarin and $17 102 847 for the second drug. Fatalities for 5 years were 123 for warfarin and 101 for the non-warfarin drug. Varying assumptions for nursing home care and numbers of ischemic strokes and hemorrhages generated the widest variation in costs.

Limitations: We did not account for out-of-pocket expenses, ‘pain and suffering’ costs, or variation across practice settings.

Conclusions: There was substantial variation in numbers and costs of adverse events across 20 scenarios, and for fatalities between the two drugs, but variation in costs between the two drugs was modest.

Key words: :

• Anticoagulants
• Cost-effectiveness
• Economics
• Warfarin