Atrial fibrillation: understanding the serious consequences and learning about the causes.

Abstract

The results reported by Kaarisalo et al.1 in this issue fit into the context of an increasingly rich literature that has described the etiology of cerebrovascular accident (CVA). Only 15 years ago most of the available epidemiological evidence seemed to point to hypertension as the dominant risk factor for stroke. As elegant and persuasive clinical trials2,3 confirmed the observational epidemiology on blood pressure as a risk factor for CVA, more detailed, larger, and longer term prospective studies began to suggest a much more interesting and complicated epidemiology, particularly in older people. The age demographics of the United States, the disabling consequences, and medical resource utilization that result from CVA make this story not only interesting but compelling. Some suggestions that the incidence of CVA may no longer be decreasing4,5 add a tinge of urgency to the need to understand more about stroke epidemiology, in general, and the often clinically silent precursors, in particular. The understanding of the epidemiology and etiology of cerebrovascular disease took a major step forward in 1983 when a paper entitled “Manifestations of coronary disease predisposing to stroke. The Framingham Study” was published.6 That report identified coronary heart disease, or ischemic heart disease, as a major precursor to stroke. It also found that hypertension, cardiac failure, atrial fibrillation, and left ventricular hypertrophy augmented the risk of stroke associated with ischemic heart disease. These findings were published at a time when the epidemiology of stroke was in its infancy. Understanding the etiology of CVA was possible only after adequate diagnostic criteria for fatal and non-fatal CVA were established and applied uniformly across study cohorts. The Framingham Study was one of only a few studies in a position to enhance the understanding of CVA etiology because of their complete enumeration of stroke occurrence in their cohort. In retrospect, their findings6 don't seem stunning or even surprising, but they appeared to lead to increased scrutiny of cardiac status as a culprit in stroke etiology. As the Framingham cohort aged, the case against cardiac status became stronger. The paper that focused attention on atrial fibrillation as an important, if not the leading, proximal risk factor for stroke was made possible as the Framingham cohort was followed into advanced age. Most subsequent papers, including the paper by Kaarisalo et al. that appears in this issue,1 confirm the identification of AF as a risk factor, and they do so in a wide array of contexts. The contribution by Kaarisalo et al. is of great relevance in the present era of increased survival after stroke where the dominant risk factor for stroke is prior cardiovascular disease. As Kaarisalo et al. point out, when the Framingham experience was similarly analyzed as a case series, remarkably similar results were found.7 It is also important to acknowledge the sometimes almost glaring dissimilarities of the details across the variety of reports. First, the prevalence of AF differs in the various samples that have been studied. Of course, older samples are particularly susceptible, and may be increasingly susceptible,8,9 to the forces of mortality and morbidity that can heavily impact patient mix and representation. Thus, the 29% prevalence of AF reported by Kaarisalo et al. is not surprising given the average age (82.2 years) and post-CVA status of the study sample. Study prevalence are also too influenced by the array of methods used to define atrial fibrillation. Thus, we are not surprised to find different estimates of the prevalence of AF from different studies of older people.10 There is no question, however, that AF prevalence rises rapidly after age 60 to at least 10% by age 80.11 Second, whereas attributable risk estimates from prospective studies are inextricably tied to the prevalence estimates for atrial fibrillation, the increase of the attributable risk of stroke caused by AF still depends on a strong predictive ability of AF for stroke. Because of the dearth of prospective population-based studies of stroke in older adults, it remains to be determined whether the sizable attributable risks of stroke caused by AF in the Framingham study (36.2% for age 80–89) can be replicated elsewhere.12 However, the 29% prevalence of AF reported by Kaarisalo et al. provides strong circumstantial evidence that the attributable risk is indeed substantial above the age of 75 years. See also p 1297 The critical arguments surrounding the epidemiology of AF also focus on the issues addressed by Kaarisalo et al. about whether AF is an incidental marker for atherosclerosis. A long-standing literature shows patients with rheumatic heart disease with mitral stenosis or cardiomyopathy have a further increase in the presence of AF. As the literature builds on this subject and other subtopics within cardiovascular disease epidemiology, it is crucial to explore and explicate causal pathways for the ultimate, often fatal or disabling, cardiovascular disease outcomes. Arguments made by Furberg et al.,11 showing very low prevalence of AF in subjects with no history of cardiovascular disease or subclinical cardiovascular disease might convince some readers that AF is indeed simply an incidental marker. Kaarisalo et al. argue that since AF is an independent predictor of survival after ischemic stroke, AF can't be just an incidental marker for atherosclerosis. The growing documentation that warfarin therapy is effective in reducing the stroke risk to AF patients,13–15 if it does nothing else, supports the view that AF is not simply an incidental marker but a true contributor and in the causal pathway to stroke. As these trials estimate at least a 70% reduction in stroke risk, and warfarin therapy in congestive heart failure and coronary heart disease is not nearly as effective in preventing stroke, a large part of the reduction of stroke risk is likely to be related to preventing atrial thrombi in the fibrillating atrium. In addition, it has been well documented that stroke may occur with AF, chronic or paroxysmal, even in the absence of heart disease. Further, in the presence of heart failure or coronary disease, stroke occurs two or three times more frequently in patients with AF.16 This is not to say that the critical question of risk/benefit of anticoagulation therapy in the elderly (>age 75) is settled.13,15,17 It also doesn't say that if you have no atherosclerosis or cardiovascular disease you won't have an extremely low risk of atrial fibrillation. Finally, as Kaarisalo suggests, cardiac factors including but not limited to AF deserve attention. Of course, cardiac factors are not the only documented cause of AF. The finding that even subclinical hyperthyroidism,18 rare as it is, is associated with incident AF suggests a role for noncardiac factors. And any thorough discussion of the epidemiology of stroke needs to include more than cursory attention to cigarette smoking as one of the dominant risk factors for stroke.19–21 How cigarette smoking causes stroke, by atherogenesis, by clotting derangement, or both, remains to be determined. Describing how these pathological consequences of cigarette smoking impinge on or modify risks associated with AF will be yet another challenge. It is worth noting that while cigarette smoking appears to weaken as a risk factor in the very old,4 AF remains a substantial risk factor for stroke at any age. Thus, this pattern of risk relationship of AF with its high predictive impact on CVA into very old age sets AF apart from virtually all other cardiovascular disease risk factors. Clearly, the attention needs to focus not only on diminishing the consequences of AF but on understanding its causes as well. Subclinical atrial/valvular disease22,23 is now suggested as a potential risk overlay in patients with cardiovascular disease. While the nature and extent of this noncoronary disease source for AF needs explication, it is also important to pursue the agenda suggested by Kaarisalo et al. They emphasize that “cardiac factors deserve attention” because of findings that recent myocardial infarction is the strongest predictor of death and that cardiac causes of death are the predominant cause of death in the AF group at the acute stage following the index stroke. Finally, the role of an important covariate,22 left ventricular hypertrophy is, at best, ambiguous. Cardiac enlargement is likely one of the set of consequences in the pathological cascade that follows atherosclerosis. Evidence is accumulating, however, that cardiac geometry, if not morphology, plays a special role in determining risk not only of CVA but also of coronary heart disease and congestive heart failure. AF is a strong candidate as an intermediate causal link between cardiac geometry or morphology and stroke. Whether adequate quality and detailed echocardiographic studies of older people will yield useful profiles for the risk of atrial fibrillation remains to be determined.