We found that among individuals confirmed to have AF by ECG, blacks were about one third as likely to be aware that they had AF as whites in this US national biracial large sample of adult men and women. Because atrial fibrillation is such a powerful risk factor for incident stroke, these finding suggest that lower awareness of atrial fibrillation and reduced likelihood of treatment among blacks may place blacks at higher risk of a stroke event, which in turn could contribute to the higher stroke mortality among blacks. The reasons for the racial discrepancy are not known. Many of the study participants may be undiagnosed, as often AF itself is not symptomatic. Alternatively, these persons may have been diagnosed with the condition, but simply did not remember or understand the condition. As these results are based on self-report, we cannot distinguish these two possibilities.
Alternatively, individuals who are unaware that they have the diagnosis of AF might not have ever been diagnosed with the condition. Whether the participant had or lacked insurance coverage was not an independent predictor of awareness of AF but since the majority of study participants were over age 65 (and are covered by Medicare) only a small percentage did not have any insurance coverage. Further, whether the participant had or lacked access to health care was also not an independent predictor of awareness of AF. Other possible explanations for racial differences in awareness of AF might be differences in either utilization or delivery of health care.
We also found that among those who were aware that they had AF and who had confirmation of the diagnosis of AF, blacks were about one quarter as likely to be treated with warfarin as whites. In striking contrast, risk of stroke as stratified by the CHADS2
score was not a predictor of warfarin use. The fact that risk of future stroke did not significantly alter the likelihood of warfarin use would seem to reflect an evidence-practice gap. Evidence-based guidelines on the use of warfarin in non-valvular AF typically recommend that estimated risk of stroke be part of the decision process regarding long-term anticoagulation.13, 14
One possible reason for a lack of correlation between CHADS2
score and likelihood of treatment with warfarin is that some patients may have had stroke due to intracerebral hemorrhage. Our study collected data on self reported physician diagnosed stroke at baseline. This has been shown to have good sensitivity and specificity. Less is known about the sensitivity and specificity of self-reported stroke subtype (i.e., ischemic stroke versus hemorrhagic stroke). Both because intracerebral hemorrhage is a relatively small contributor to all strokes (about 15%) and because the case fatality of this type of stroke is high, we presume that only a small proportion of strokes at baseline represented intracerebral hemorrhage.
We found that women with documented atrial fibrillation who were also aware of their condition had about a third of the chance as men of being treated with aspirin, a treatment known have efficacy in stroke prophylaxis. The reasons for this relative under- treatment with aspirin are unknown. Others have observed that women are less likely to receive aspirin for secondary prevention of ischemic heart disease. 15, 16
Certain limitations of our study need to be acknowledged. It is possible that awareness of AF was underestimated in this cohort as paroxysmal atrial fibrillation may not have been detected at the time of baseline ECG. The decision to anticoagulate is often complex, involving predictions of benefit from stroke reduction balanced against predictions of hemorrhagic complications. Physicians may be reluctant to anticoagulate patients who have uncontrolled hypertension. Whether this might have been a factor in the likelihood of being treated with warfarin could not be assessed in our study as there was only one baseline measure of blood pressure. While CHADS2 is an example of a well-validated clinical tool for predicting stroke risk, no comparably well validated tool exists for predicting hemorrhage risk. Study subjects were not interviewed regarding whether they were offered warfarin therapy in the past but they refused, i.e., we are not able to distinguish between people who were never offered therapy from informed refusers. Information on contraindications to warfarin therapy was not available so it is not known what proportion were actually warfarin candidates. The interview did not explore whether subjects who reported no active warfarin use had taken and subsequently discontinued warfarin in the past. Finally, our study was not originally designed to test the specific and singular question of racial differences in awareness and treatment of AF. This report is of an exploratory analysis and as such is subject to the limitation common to observational studies of a possible spurious association generated by type I error.
In this large biracial cohort, blacks were less likely to be aware of AF and less likely to be treated with warfarin than whites. These findings are consistent with prior studies demonstrating that blacks are less likely to achieve quality of care goals for stroke risk factors such as glycemic control in diabetes and blood pressure in hypertension.17
Such differences may underlie racial disparities in stroke morbidity and mortality and should lend urgency to focused efforts to improve patient education and medical literacy. The additional finding that CHADS2
score was not a predictor of warfarin use highlights an evidence-practice gap that should prompt further efforts focused on practitioner awareness and education. Future investigations to further understand the determinants of racial disparities in stroke and the impact of improved diagnosis and treatment of atrial fibrillation on race-related stroke outcomes would be of public health importance.