Factors influencing physicians' reported use of anticoagulation therapy in nonvalvular atrial fibrillation: a cross-sectional survey

Clin Ther. 2003 Jun;25(6):1750-64. doi: 10.1016/s0149-2918(03)80167-4.


Background: Some elderly patients with nonvalvular atrial fibrillation (NVAF) who might benefit from warfarin therapy do not receive it.

Objective: The goal of this cross-sectional study was to identify physicians' attitudes and beliefs that are associated with their reported use of warfarin in case scenarios.

Methods: A self-administered survey was mailed to a cross-section of general internists randomly selected from a national pool of physicians in the American Medical Association Masterfile. Fourteen clinical vignettes were used, incorporating various comorbid conditions and risk factors for either major bleeding episode or embolic cerebrovascular accident (CVA). The outcome measure was the number of case vignettes for which warfarin was recommended.

Results: A total of 142 completed surveys (33% of 426 eligible respondents; 109 men, and 32 women [1 respondent did not provide gender]; mean [SD] age, 45 [10] years) were received. The median number of case vignettes for which warfarin was recommended was 10 (interquartile range, 8-12). We found no relationship between the perceived benefits of warfarin and its use in the case vignettes. However, the perceived risk for warfarin associated hemorrhage was strongly associated with reported warfarin use (P < 0.001). The physicians in our sample provided estimates of the annual rate of warfarin-associated intracerebral hemorrhage that were >10-fold higher than literature-based estimates, and physicians providing higher risk estimates tended to use warfarin less often. On multivariate logistic regression, physicians who recommended warfarin use in more vignettes were less likely to report anticipated regret of committing an error of omission (ischemic CVA in an untreated NVAF patient) (P < 0.001) or a loss-aversive risk preference (P = 0.027), and had a lower perceived annual risk for hemorrhage with warfarin (P < 0.001). Physician age, sex, primary mechanism of reimbursement, academic appointment, and the NVAF patient volume all were unrelated to warfarin use.

Conclusions: Although the decision to use warfarin in NVAF was not driven by the perceived benefit, the perceived risks strongly affected warfarin use. Response bias is a potential limitation, but our data strongly suggest that physicians' attitudes toward anticipated regret and risk aversion can impact on their treatment recommendations.

Publication types

  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Adult
  • Anticoagulants / therapeutic use*
  • Atrial Fibrillation / drug therapy*
  • Cross-Sectional Studies
  • Data Collection
  • Decision Making
  • Drug Utilization / statistics & numerical data
  • Female
  • Humans
  • Male
  • Practice Patterns, Physicians'*
  • Risk Factors
  • Warfarin / therapeutic use*


  • Anticoagulants
  • Warfarin