Physicians' perceptions of the benefits and risks of warfarin for patients with nonvalvular atrial fibrillation.

Numerous randomized controlled trials (RCTs) have conclusively demonstrated that long-term anticoagulation in patients with nonvalvular atrial fibrillation (NVAF) can reduce the risk of stroke by approximately 68% per year, with a low annual risk of major hemorrhage (1.3%). Despite these impressive results, several studies have shown that the use of warfarin in the “real world” for the prevention of ischemic stroke is suboptimal, with only 15%–44% of patients with atrial fibrillation who had no contraindication to the therapy actually receiving warfarin. The reasons for this discrepancy between the RCT evidence and clinical practice patterns are unknown. To test the hypothesis that physicians may underestimate the benefits and overestimate the risks of anticoagulation therapy, we conducted a survey of all cardiologists (n = 58), neurologists (n = 34) and general internists (n = 130) and a random sample of family physicians (n = 243) within the province of Alberta (first mailing in December 1998 and third mailing in June 1999). The response rate was 49% for family physicians, 73% for specialists and 60% overall. The survey assessed knowledge by presenting a hypothetical patient with NVAF typical of those enrolled in the RCTs. Physicians were then asked to estimate the relative risk reduction (RRR) for stroke (if the patient was taking warfarin or if the patient was taking ASA) and the absolute risk of a major hemorrhage (if the patient was taking warfarin) on a continuum ranging from 0% to 100%, expressed as annual rates. Respondents underestimated the RRR with warfarin therapy (mean estimate of 53%) compared with that shown in the RCTs (68%) and overestimated the risk of major hemorrhage (10% absolute annual risk v. 1% respectively) (Table 1). In contrast, the mean RRR of ASA was estimated to be higher (27%) than that documented in RCTs (21%). Interestingly, family physicians estimated ASA to be significantly more efficacious and the risk of major hemorrhage with warfarin to be significantly higher than the specialists had estimated. Because warfarin therapy is typically monitored by family physicians, these misconceptions may affect a large absolute number of patients. The estimation of a higher risk of hemorrhage with warfarin by family physicians (compared with specialists) has been reported previously. This may be due to several factors. First, family physicians monitor warfarin therapy on a long-term basis and, therefore, they are more likely to witness adverse outcomes in clinical practice. As such, there may be a natural tendency to remember these sequelae selectively when the therapeutic benefits are intangible (namely, the avoidance of a thromboembolic event). SecPhysicians’ perceptions of the benefits and risks of warfarin for patients with nonvalvular atrial fibrillation