Anticoagulation therapy for patients with non-valvular atrial fibrillation: comparison of decision analytic model recommendations and real-world warfarin prescription use.

Abstract

BACKGROUND:

Anticoagulation in patients with atrial fibrillation (AF) is challenging because stroke-risk reduction must be balanced against increased bleeding risk.

OBJECTIVE:

We developed a decision model integrating both stroke and bleeding risk schemes to guide optimal use of anticoagulation in AF, and compared model recommendations with warfarin use in a real-world database.

METHODS:

A Markov model based on demographics, CHADS(2) (Congestive Heart Failure, Hypertension, Age of 75 years and greater, Diabetes Mellitus and History of Stroke) stroke and ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) bleed risk scores, and anticoagulation treatment effects from clinical trials simulated health state transitions for recently diagnosed AF patients. The model recommended the treatment with greater quality-adjusted life expectancy. Model recommendations were contrasted with actual warfarin use recorded in the Thomson Reuters MarketScan database (N = 64,946).

RESULTS:

74.8% (n = 48,548) of the Marketscan AF cohort had CHADS(2) ≥1, of whom 14.3% had moderate/high (≥4) ATRIA bleeding risk. While the model recommended warfarin for almost all patients with CHADS(2) ≥1 who are at low bleeding risk, it recommended warfarin for fewer patients as bleeding risk increased. Of the 44,611 patients recommended warfarin, 63.4% of patients were considered warfarin exposed (concordant with model recommendation), and of the 20,335 patients recommended aspirin (acetylsalicylic acid), 59.7% received warfarin (discordant with model recommendations). Actual warfarin use decreased modestly with higher stroke risk (p < 0.0001) and with higher bleeding risk (p < 0.0001).

CONCLUSION:

High discordance between actual warfarin use and model recommendations suggests that anticoagulation decisions are not based on systematic evaluation of stroke and bleeding risks. Model-based clinical decision aids may improve oral anticoagulation decisions by more systematically weighing bleed and stroke risk.

• Authors: Julian P Casciano, Daniel E Singer, Winghan Jacqueline Kwong, Eben S Fox, Bradley C Martin
• Journal: American journal of cardiovascular drugs : drugs, devices, and other interventions (Am J Cardiovasc Drugs) Vol. 12 Issue 5 Pg. 313-23 (Oct 01 2012) ISSN: 1179-187X [Electronic] New Zealand
• PMID: 22920047 (Publication Type: Journal Article, Research Support, Non-U.S. Gov't)

Chemical References

• Anticoagulants
• Warfarin

Topics

• Adolescent
• Adult
• Aged
• Aged, 80 and over
• Anticoagulants (adverse effects, therapeutic use)
• Atrial Fibrillation (complications, drug therapy)
• Decision Support Techniques
• Female
• Hemorrhage (chemically induced, prevention & control)
• Humans
• Male
• Markov Chains
• Middle Aged
• Quality-Adjusted Life Years
• Risk Factors
• Stroke (etiology, prevention & control)
• Warfarin (adverse effects, therapeutic use)
• Young Adult