RESUMO
BACKGROUND: Optimal treatment of atrial fibrillation (AF) in the haemodialysis population is uncertain due to the exclusion of this group from randomized trials. The risk-benefit profile for anticoagulation and anti-platelet therapy in haemodialysis differs from the general population due to platelet dysfunction from uraemia, altered pharmacokinetics and increased falls risk. METHODS: This decision analysis used a Markov-state transition model that took a patient perspective over a 5 year timeframe. The Markov model compared life-years gained and quality-adjusted life-years gained (QALY) for three AF treatment strategies: warfarin, aspirin and no treatment. The base case was a 70-year-old man on haemodialysis with non-valvular AF. RESULTS: In the base case, the total health outcomes in life-years and QALY were 2.37 and 1.47 respectively for warfarin, 2.38 and 1.61 respectively for aspirin, and 2.39 and 1.61 respectively for no treatment. Thus, warfarin led to 0.14 fewer QALY or 1.7 fewer months of life lived in full health, compared with either aspirin or no therapy. The finding that warfarin generated the lowest expected QALY was robust to one-way, two-way and probabilistic sensitivity analyses. CONCLUSIONS: Our results suggest that warfarin should not be the default choice for older haemodialysis patients with non-valvular AF as it provides the fewest QALY compared with aspirin or no therapy.