LETTER/REPLIES Continuum of Cognitive Impairment to Stroke Possibly via Atrial Fibrillation Jacob I. Haft, MD In their paper ‘‘Incident Cognitive Impairment Is Elevated in the Stroke Belt: The REGARDS Study,’’1 the authors demonstrate that the incidence of development of cognitive impairment is higher in a population in which the incidence of stroke is higher, suggesting that the mechanism for stroke and cognitive impairment is the same (at least in some patients) and that there may be a continuum between cognitive impairment and stroke, that is, cognitive impairment is an early and mild manifestation and stroke is an extreme manifestation of the same underlying pathophysiologic abnormality. Previous studies, referenced in their paper, have demonstrated a high incidence of the conventional risk factors for stroke in patients with cognitive impairment, although not in the same population group. Were the individual patients who developed cognitive impairment followed longitudinally, and was there an increased incidence of subsequent overt stroke in these patients? We have documented that atrial fibrillation, largely intermittent and undiagnosed, is found in a high incidence in stroke patients (at least 34.9% among ischemic stroke patients with hypertension2) in whom all electrocardiograms (ECGs) in a >14-year ECG and clinical database were reviewed, suggesting that intermittent atrial fibrillation may be the common mechanism whereby the conventional risk factors for stroke cause stroke. Are data available on the incidence of atrial fibrillation in patients in the Stroke Belt compared to the incidence of atrial fibrillation in patients in non-Stroke Belt states, and specifically, are there data on the incidence of atrial fibrillation in the patients in this study who developed cognitive impairment? If there is a higher incidence of atrial fibrillation in these patients, it might be advantageous to study patients with increasing dementia with (1) magnetic resonance imaging to identify those with multiple small silent infarcts and/or (2) echocardiograms to find left atrial enlargement, left ventricular enlargement, or low ejection fraction and to identify patients most likely to benefit from long-term monitoring to diagnose occult intermittent atrial fibrillation, who might then be considered for anticoagulation to prevent overt stroke. Potential Conflicts of Interest Nothing to report. Hackensack University Medical Center, Hackensack, NJ C 2011 American Neurological Association 666 V References 1. Wadley VG, Unverzagt FW, McGuire LC, et al. Incident cognitive impairment is elevated in the Stroke Belt: the REGARDS study. Ann Neurol 2011;70:229–236. 2. Haft JI, Teichholz LE. Echocardiographic and clinical risk factors for atrial fibrillation in hypertensive patients with ischemic stroke. Am J Cardiol 2008;102:1348–1351. DOI: 10.1002/ana.22618 Reply Virginia G. Wadley, PhD,1 Virginia J. Howard, PhD,1 George Howard, DrPH (REGARDS Study Principal Investigator),1 and Elsayed Z. Soliman, MD2 We appreciate the opportunity to respond to Dr Jacob Haft’s queries and good suggestions. We agree with Dr Haft’s perspective regarding the probability of shared pathways for incidence of cognitive impairment and stroke, including atrial fibrillation (AF). Our recent Annals of Neurology publication was part of a larger effort to better understand multiple traditional and novel risk factors for cognitive decline and stroke.1 To this end, an analysis of Framingham stroke risk factors and incident cognitive impairment in the REGARDS cohort is currently in press.2 With respect to Dr Haft’s questions as to whether individuals who developed cognitive impairment were followed longitudinally, and whether there was an increased incidence of subsequent overt stroke in these participants, the answer to both questions is yes. We recently reported that low cognitive performance is a potent predictor of physician-adjudicated incident stroke among REGARDS participants younger than 72 years (ie, ages 45–72 years).3 In the population-based REGARDS study, AF was determined at baseline from study scheduled electrocardiograms (ECGs) recorded in participants’ homes, as well as by selfreported history of a previous physician diagnosis of AF. We have shown in prior work that the prevalence of AF defined by ECG plus self-report (ie, AF detected by both methods) is actually lower in the Stroke Belt than the rest of the nation, although this relationship was attenuated and no longer significant when defining AF by either self-report or ECG.4 Thus, our data do not support regional variations in AF that mirror regional differences in incident cognitive impairment and stroke. Even so, we also have demonstrated that irrespective of detection method, AF is a strong and significant predictor of incident stroke (odd ratios ranging from 1.41 to 1.90 depending on case definition).5 To date we have not collected data for determining the geographic distribution of incident or intermittent AF. We appreciate Dr Haft’s suggestions for future research directions as we continue our efforts to identify modifiable risk factors for stroke and cognitive decline.