Early Recurrent Embolism Associated with Nonvalvular Atrial Fibrillation: A Retrospective Study

Transcription

1 Early Recurrent Embolism Associated with Nonvalvular Atrial Fibrillation: A Retrospective Study ROBERT G. HART, M.D.,* BRUCE M. COULL, M.D.,t AND DENISE HART, M.D.* SUMMARY Nonvalvular atrial fibrillation (NVAF) can be a source of cardiogenic brain embolism. We retrospectively reviewed the clinical features of 56 patients with acute brain infarction and NVAF. Based on clinical criteria, 35 infarcts (63%) were classified as probably embolic, 13 infarcts (23%) as probably nonembolic and 8 infarcts (14%) as of indeterminate pathogenesis. Among the 35 patients with presumed embolic infarcts, 12 patients were immediately anticoagulated without hemorrhagic complications. Of 23 patients who did not receive immediate anticoagulation, three (13%) experienced recurrent embolism (one each to brain, kidney and leg) within 10 days of initial embolism. There were no early recurrent emboli in patients receiving immediate anticoagulation or patients in the nonembolic or indeterminate category. Five additional patients experienced probable brain or systemic emboli within the 11 days prior to the marker stroke event. Including these patients, 20% (8 of 40) of a/invaf patients who were not immediately anticoagulated experienced recurrent embolism within 11 days of the initial embolus. Early recurrent emboli are common in NVAF patients who experience embolic stroke. Stroke, Vol 14, No 5, 1983 PATIENTS WITH NONVALVULAR (nonrheumatc) atrial fibrillation (NVAF) have 5-7 times the risk of stroke compared to age-matched patients without this dysrhythmia. 1 " 4 About 35% of NVAF patients will eventually experience cerebral infarction. 1,3 > 5 The incidence of ischemic stroke associated with NVAF is about 5% per year, approaching the stroke risk following transient ischemic attack. 1,5 This increased risk of NVAF-associated stroke affects 5% of the elderly. 6,7 Prevention and management of stroke in NVAF patients is a common, important clinical problem. The etiology of cerebral ischemia associated with AF is often assumed to be embolism of left atrial thrombi. However, nonvalvular AF is usually related to ischemic or other heart disease, most often in elderly patients with risk factors for cerebrovascular atherosclerosis. Thromboembolism from coexistent left ventricular disease or cerebrovascular disease may account for many strokes in NVAF patients. Thus, in some portion of patients, NVAF is only a marker of other cardiac and cerebrovascular diseases which cause ischemic stroke. The relative prevalence of these mechanisms in AF patients with stroke is uncertain and often difficult to define in individual patients. 8 It is perhaps not so surprising that patients with chronic NVAF embolize, but that they do so intermittently. Often embolic episodes are separated by years, in other instances by only hours. 5 The mechanisms responsible for formation and intermittent embolization of left atrial clot in these patients are uncertain. Between 2-15% of NVAF patients with embolic stroke have been reported to experience early recurrent emboli within two weeks of the initial embolic From the *Department of Medicine, University of Texas Health Sciences Center, San Antonio, Texas 78284, and tthe Department of Neurology, Oregon Health Sciences University, Portland, Oregon Address correspondence to: Robert G. Hart, M.D., Department of Medicine (Neurology), University of Texas Health Sciences Center, 7703 Floyd Curl Drive, San Antonio, Texas Received February 24, 1983; revision accepted June 13, event. 5, 10 The role of immediate anticoagulation to prevent early recurrence of embolic brain infarction is unclear, because of the fear of creating hemorrhagic infarction from initially pale infarction. 8 Given the frequency and importance of NVAF-associated stroke in clinical practice, the natural history has received relatively scant attention. ' 4 ' 5-9 We retrospectively review the clinical features of 56 patients with NVAF and cerebral infarction to estimate the fraction of NVAF-associated stroke that is embolic and to determine the early recurrence rate following initial embolism. Methods Patients were identified by cross-indexing discharge diagnostic codes for atrial fibrillation (ICD ) and cerebral infarction and/or embolism (ICD ) for the years for inpatients at two referral-based hospitals. Each chart was reviewed in detail by one of the authors. Patients with clinical or echocardiographic evidence of valvular heart disease (rheumatic heart disease, prosthetic valve, mitral valve prolapse) were excluded. Embolic versus Nonembolic Stroke Classification of the cause of stroke as cardiogenic embolism versus cerebrovascular disease is imperfect, even arbitrary, on clinical grounds. Because of the age and concomitent hypertension of these patients, it cannot be assumed that all or even most strokes associated with nonvalvular atrial fibrillation are on the basis of cardiogenic embolism. Clinical criteria were employed in an attempt to separate NVAF patients into embolic versus nonembolic mechanisms. This algorithm is based primarily on the use of abrupt onset of maximal neurologic deficit in an awake patient to characterize cardiogenic embolism. Although non-sudden onset of embolic stroke can occur in a minority of patients with cardiogenic embolism, progressive or stuttering deficits are far more likely to be on the basis

2 of mechanisms other than cardiogenic embolism and, in the absence of other data, such strokes were classified as probably nonembolic. The presence of bilateral infarctions was also used to indicate cardiogenic embolism, but only if the clinical and CT features were not those of lacunar infarctions, which are commonly bilateral. Group A. Embolic Stroke Patients 1. absence of ipsilateral TIA or carotid bruit. 2. presence of one or more of the following: a. abrupt onset while awake of maximal neurological deficit. b. absence of stenotic atherosclerosis of cerebrovasculature by angiography or reliable noninvasive studies. c. history of other embolic episodes (usually systemic) prior to the marker event. d. age less than 50 years without diabetes or hypertension. e. evidence of embolic stroke at autopsy. f. bilateral, nonlacunar infarctions by CT. 3. if chronically hypertensive, nonlacunar infarction by CT. Group B. Nonembolic Stroke Patients 1. presence of one or more of the following: a. ipsilateral bruit or TIA. b. hypertension combined with clinical and CT evidence of lacunar infarction. c. nonsudden onset of deficit while awake. d. evidence of stenotic atherosclerosis by angiography or reliable noninvasive testing. e. prior ipsilateral carotid endarterectomy. f. nonembolic stroke at autopsy. Group C. Indeterminate 1. onset of deficit while asleep with no progression after waking. 2. no information available about the onset of deficit. 3. absence of items allowing classification in Groups A and B. It is appreciated that this may not be an entirely reliable classification. It does include, however, those factors most often considered at the bedside in assessing the likelihood of cardiogenic embolism. The subsequent occurrence of early recurrent embolism was not used in classification. Type and Etiology of Atrial Fibrillation Atrial fibrillation is often associated with other cardiac disease (myocardial infarction, decompensated heart failure) which may result in cerebral embolism independently of left atrial thrombi. Because of potential differences in rate of early recurrent embolism associated with underlying heart disease, patients were divided into three subtypes based on clinical data: I. AF attributed to ischemic-hypertensive cardiovascular disease but without transmural MI, decompensated congestive failure or other recognized sources of embolism. Chronic, persistent "idiopathic" AF in patients over 50 were included. II. AF associated with ischemic heart disease with transmural MI, ventricular aneurysm and/or congestive heart failure. III. AF associated with other specific, nonischemic heart disease (eg, thyrotoxicosis, pre-excitation syndrome) or idiopathic AF in patients under 50. Atrial fibrillation was further subdivided into constant versus intermittent. If NVAF was observed to spontaneously convert to other rhythms within one year before or after stroke, it was classified as intermittent. Stroke was estimated using clinical criteria based on neurologic deficit and CT data, as described elsewhere. 8 Eighteen patients received immediate anticoagulation with heparin or Coumadin in a nonconsecutive, nonrandomized manner at the discretion of their attending physicians. Results Group A Embolic Stroke Thirty-five strokes (63%) were categorized as probably embolic (Group A, Tables 1 and 2). Most patients had idiopathic-hypertensive NVAF (Type I, 71%) without other cardiac embolic sources. NVAF was constant in 18 patients and intermittent in 17 patients. The location of stroke was in the carotid distribution in 30 patients (85%), in the brainstem in 3 patients (9%), and in the posterior cerebral artery distribution in 2 patients (6%). Echocardiography was done in 22 patients (63%) and revealed no evidence of valvular disease. Patients who did not undergo echocardiography TABLE 1 Demographic Features Non- Indeter- Embolic embolic minate (Group A) (Group B) (Group C) Total Number Male/Female 21/14 8/5 2/6 31/25 Age (average) Associated heart disease* Type I 71% 77% 50% 70% Type II 20% 15% 38% 21% Type III 9% 8% 12% 9% Pattern of AF* constant 51% 46% 75% 54% intermittent 49% 54% 25% 46% Stroke * large 37% 31% 50% 37.5% medium 40% 31% 38% 37.5% small 23% 38% 12% 25% Mortality 9% 15% 0% 9% Chronic hypertension 43% 54% 75% 50% Echocardiography! 63% 62% 75% 64% *See methods for classification scheme, AF is atrial fibrillation. tah patients with early recurrent embolism underwent echocardiography, excluding subclinical valvular disease.

3 TABLE 2 Atrial Fibrillation and Embolic Stroke (Group A) # Age/Sex ' Type of AF Stroke therapy Early reembolism A-l ' 53M I, constant M heparin no A-2 64F I, intermittent S heparin no A-3 60M I, intermittent S heparin no Left atrial Site of second embolus A-4 60M I, constant M heparin no* iliac, 5 days prior to CVA A-5 67F I, intermittent L ' Coumadin no 48 A-6 69M I, intermittent S Coumadin no 43 A-7 81F I, intermittent L Coumadin no* iliac, 11 days prior to CVA A-8 71M I, constant L APT no 48 A-9 71M I, constant M APT no 40 A-10 68F I, intermittent L APT no* 45 renal, 6 hrs prior to CVA All 78M I, constant S APT no 37 A-12 58M I, constant M APT no* 46 mesenteric, 11 days prior to CVA A-13 79M I, intermittent L no A-14 80M I, constant L yes-4th iliac, 4th day after CVA A-15 79F I, constant L no A-16 77F I, constant S no A-17 87M I, intermittent M no A-18 8 IF I, constant L no A-19 81M I, constant L no A-20 37F I, intermittent M yes-4th stroke, 4th day after CVA A-21 67F I, constant M no A-22 55M I, intermittent L yes-9th renal, 9th day after CVA A-23 82F I, constant M no 45 A-24 82M I, intermittent M no 46 A-25 77M I, intermittent M no 38 A-26 64F II, intermittent M heparin no* renal, 4 days prior to CVA A-27 66M II, constant M heparin no A-28 65M II, constant S heparin no A-29 74F II, constant L Coumadin no A-30 61M II, constant S APT no 52 A-31 71F II, intermittent L APT no 47 A-32 78M II, intermittent M no A-33 32M III, constant S Coumadin no 49 A-34 74F III, intermittent L no A-35 23M III, intermittent M no *Embolism immediately prior to marker event, see text. APT = antiplatelet therapy, M = medium, L = large, S = small, CVA = brain infarction. were elderly (over age 75) with longstanding atrial fibrillation and no ausculatory evidence of valvular disease. Left atrial by M-mode echocardiography averaged 45 mm (N = 13). Cerebral arteriography was done in seven patients and was normal (5 patients) or demonstrated isolated middle cerebral artery occlusion (2 patients). There were seven patients in Group A with remotely occurring embolic events. Assessment of the reliability of embolic versus nonembolic classification was attempted using cerebral angiographic and autopsy data from patients who were initially classified by clinical criteria. Arteriography in 10 patients did not change the diagnostic category; autopsy in three patients changed the diagnostic category in one patient. Among the 23 patients in Group A who were not immediately anticoagulated, there were three episodes of early recurrent embolism (table 2, fig. 1) within 2 weeks. anticoagulation in 12 patients resulted in no hemorrhagic complications. In five additional patients, probable systemic embolism immediately preceded (within two weeks) embolic stroke. Including these events, early recurrent embolic events occurred in eight of 25 (32%) of NVAF patients who were not immediately anticoagulated at the time of initial brain Or systemic embolism.

4 t Anticoagulation EMBOLIC STROKE PATIENTS (Group A, N = 35) 5 immediately prior embolic events Antiplatelet Agents No Therapy N - 12 N = 7 N - 16 \ \ \ «Early Recurrence «Early Recurrence 3(19%) Early Recurrences Anticoagulation All NVAF Patients (Groups A, B&C, N = 56) 5 immediately prior embolic events Antiplatelet Agents No Therapy N = 18 N = 13 N = 25 «Early Recurrence «Early Recurrence 3(12%) Early Recurrences FIGURE 1. Early Recurrent Embolism in Nonvalvular Atrial Fibrillation (NVAF): Maximum and Minimum Values. A. If clinically embolic strokes (Group A) are considered, 19% of untreated patients and 13% of nonanticoagulated patients experienced early recurrence. B. If all NVAF patients with stroke (embolic and nonembolic) are considered, 12% of untreated patients and 8% of nonanticoagulated patients experienced early recurrences. Patients were nonrandomly selected for different treatments in this retrospective study. Group B (Nonembolic) and C (Indeterminate) Demographic features of Group B patients (N = 13) and Group C patients (N = 8) were similar to Group A (table 1). Nonsudden onset of deficit, usually stuttering or progressing over hours, was the, usual reason for classification in Group B (table 3A). Brainstem strokes occurred in 4 patients (31%) in Group B Left atrial averaged 47 mm (N = 5) in Group B patients. No patient in Group B or C experienced early recurrent embolism: Early Recurrent Embolism: All NVAF Patients Given the uncertainties of clinical categorization into embolic versus nonembolic stroke, minimum values for early recurrent embolism are obtained by considering all patients. Recurrent embolism after initial stroke occurred in 3 patients 8% of those who were not immediately anticoagulated. Including the five patients with probable ernbolic events shortly before the brain embolism that lead to study inclusion, 8 of 40 (20%) of all NVAF patients with initial systemic or cerebral embolism who are not immediately anticoagulated re-embolized within two weeks. In these 8 patients, the initial embolus was systemic in 5 patients and Cerebral in 3 patients; the second embolus was cerebral in 6 patients and systemic in 2 patients. In patients with more than one embolus, 56% were to the brain. TABLE 3A Atrial Fibrillation and Nonembolic Stroke (Group B)* #. Age/Sex Type of AF Stroke therapy Early reembolism LA. Size Reason Nonembolic B-l 70M I, intermittent L heparin no onset over 36 hrs B-2 69M I, intermittent S heparin no onset over 30 hrs B-3 72M I, constant S heparin no 43 onset over 72 hrs B-4 72M I, constant M Coumadin no onset over 24 hrs B-5 74F I, constant S APT no 48 carotid stenosis B-6 72M I, intermittent S no thrombotic at autopsy B-7 64F I, intermittent M no 52 carotid stenosis B-8 92F I, constant L no recent ipsilateral TIA B-9 67F I, intermittent M no in bruit, recent TIA B-10 69M I, intermittent M no bruit, onset over 24hr B-l 1 83M II, constant L no onset over 72 hrs B-12 74F II, intermittent L no bruit, nocturnal onset B-13 62M III, constant S heparin no 53 stuttered over 4 hrs TABLE 3B Atrial Fibrillation and Stroke: Uncertain Pathogenesis (Group C) C-l 77F I, constant L heparin no onset unknown C-2 74F I, constant L APT no onset unknown C-3 65M I, constant M no onset unknown C-4 57M I, constant M no 37 nocturnal onset C-5 83F II, constant M APT no onset unknown C-6 87F II, intermittent L no 40 onset unknown C-7 79F II, intermittent L APT no 50 onset unknown C-8 87F III, constant S APT no 50 lacunar syndrome *APT is antiplatelet therapy, L.A. is left atrial in millimeters by M-mode echocardiography.

5 TABLE 4. Nonvalvular Atrial Fibrillation and Multiple Embolism Age/Sex. Type Left atrial 1st embolus (duration AF) Interval therapy 2nd embolus (interval) A-9 71M I, constant 40mm CVA (> 9 yrs) APT CVA (1.5 yr) & CVA (2.0 yr) A-10 68F I, intermittent 45mm CVA (6 yrs) APT renal (2.9 yr) A-16 77F I, constant LAE CVA (10 yrs) 0 CVA (1.0 yr) A-33 32M III, constant 49mm CVA (1 yr) 0 CVA (1.8 yr) A-13 79M I, constant - Leg (> 8 yrs) 0 CVA (1.2 yr) A-8 71M I, constant 48mm CVA (3 yrs) APT CVA (4.1 yr) A-ll 78M I, constant 37mm CVA (< 1 yr) 0 CVA (1.6) & CVA (4.5) A-23 82F I, constant 45mm CVA (?) 0 multiple CVA at CT A-28 65M II, constant LAE CVA 0-5 yr) Coumadin spleen (autopsy) *Does not include patients with early recurrent emboli, left atrial enlargement. CVA is brain infarction, APT is antiplatelet therapy, LAE is Longterm Risk of Re-Embolism Follow-up information from subsequent hospitalization are available for 22 patients in Group A and 5 patients in Groups B and C. These data would be clearly biased in favor of recurrent embolism (leading to rehospitalization). and not reflect the natural history Of late recurring emboli (table 4). However, the effect of long-term treatment modalities in this group is of interest. Although involving very small numbers, there were therapeutic trends (not statistically significant). Of Group A patients, there were no recurrent emboli in 8 patients treated with Coumadin who were followed for 11.2 patient-years. Among Group A patients who were treated with antiplatelet agents, there were 3 recurrences (involving 2 patients) in 20.8 patient-years of follow-up. Among 6 patients who received no specific therapy, there were 2 recurrent emboli in 10.3 patient-years. Recurrent embolism occurred, on average, two years after initial stroke. Discussion Ten to 15% of all brain infarctions occur in patients with atrial fibrillation. 611 ' 12,13,14 Distinguishing embolic from nonembolic mechanisms is a common clinical dilemma and has implications for longterm management. In the Harvard Cooperative Stroke Registry < 74% of infarctions in patients with atrial fibrillation were categorized as embolic." Sage and VanUitert 9 reported that 97% of 145 patients with NVAF and stroke had cardiogenic embolism as the likely cause. In our retrospective review, 63% of infarctions associated with NVAF were classified as embolic. The clinical criteria used in the above studies to distinguish embolic from nopembolic infarction are probably imprecise. In our methodology, we used the nonsudden onset of symptoms to categorize patients as probably nonembolic (Group B) in seven instances. Although nonsudden onset Of cardiogenic embolic stroke is well established, a stuttering or fluctuating evolution heralds thrombotic stroke "in the great majority of cases." 15 Epidemiologic data confirm that many strokes associated with NVAF are nonembolic. As NVAF occurs in 3-5% of the elderly 6-7 but 9-25% of all patients with brain infarctions, 6, " 1 2, 1 3, 1 4 it is clear that NVAF is a risk factor for stroke. Estimates from clinical and pathological studies place the NVAF patient at three to seven times the risk of stroke compared to agematched controls. 1 " 1 The stroke incidence in NVAF patients (5-8%/year) 1,5 approaches that following TIA- Even if all of this increased risk is due to cardiogenic embolism (an unlikely prospect), cardiogenic embolism would be likely to account for only 80-85% of strokes in NVAF. Based on these clinical and epidemiological data, it seems likely that at least 20-30% of strokes in NVAF patients are not due to cardiogenic embolism. Because of an average age of 72 years and since about 35% of these patients have large strokes (table lj, many clinicians are reluctant to fully investigate individual NVAF patients with stroke, despite the potential importance of accurate diagnosis in longterm management. Only 17% of our series and 2% of Sage and VanUitert's patients underwent cerebral arteriography. 9 With the advent of low risk diagnostic investigations (digital subtraction angiography, real-time carotid Doppler ultrasound), further studies may well be warranted to define stroke mechanism in these patients. Cardiac computed tomography may occasionally confirm the existence of atrial thrombus (fig. 2). anticoagulation of cardiogenic brain embolism has been controversial and the subject of several recent studies. 8 The risk of early recurrent embolism must be balanced against the risk of hemorrhagic transformation of pale infarction due to anticoagulation. The risk of early recurrent embolism (within two weeks) may depend on the type of cardiac source. Fisher reported an 8% prevalence of early re-embolism but his patient population included a minority with valvular disease and immediate therapy was not specified. 5 Sage and VanUitert reported only 2% of NVAF patients who were survivors of initial hospitalization had recurrence within 3 months, but the early recurrence rate in the 38% who did not survive initial hospitalization was not given. 9 Darling et al. reported a 20% early recurrence rate, but his series antedated echocardiography and many patients undoubtedly had occult valvular disease. 10 There were two patients among 13 NVAF who did not receive immediate anticoagulation

6 FIGURE 2. Atrial thrombus detected by Computed Tomography (CT). 1.5 cm thrombus (arrow) adherent to the wall of the left atrium was demonstrated by cardiac CT after two-dimensional echocardiography showed only left atrial dilatation in a patient with nonvalvular atrial fibrillation and stroke. in the Cerebral Embolism Study Group's data who experienced recurrent embolism within two weeks of initial embolism/ Our retrospective results yield a maximum and minimum figure for early recurrence, depending upon inclusion of embolic events occurring just prior to the marker stroke. Three early recurrent emboli occurred in the 16 patients with probable embolism who received no immediate treatment (figure 1). Five additional emboli occurred within 11 days of the marker stroke in patients who were not immediately anticoagulated at the time of initial embolism. A maximum figure for early recurrent embolism in NVAF patients who were not immediately anticoagulated at the time of initial embolism is 20% (8 of 40). A minimum figure, considering only prospective embolic events from marker stroke in all NVAF patients (Groups A, B and C) who were not immediately anticoagulated is 8% (3 of 38). The efficacy of immediate anticoagulation in prevention of early recurrent embolism is suggested by aggregate data which show no episodes of early reembolism in 29 patients who were immediately anticoagulated. 8 There were no major hemorrhagic complications in these patients, although they were nonconsecutive and nonrandomly selected for immediate anticoagulation at the discretion of their attending physician in this retrospective study. In this series, over half of all patients with NVAF and stroke had constant, not intermittent, NVAF. It is possible, however, that many patients with brief episodes of sinus rhythm were misclassified. The Framingham study reported that 86% (37 of 43) of patients with NVAF and stroke had chronic, persistent atrial fibrillation. 14 Most of our patients had no other cardiac sources of stroke except NVAF (Type 1). All instances of early recurrent embolism occurred in Type 1 patients. Two-dimensional echocardiography in NVAF patients with stroke has shown important, unsuspected abnormalities predisposing to cardiogenic embolism in 10-16% Most of our patients had only M-mode echocardiography. Anecdotally, mitral annulus calcification was detected echocardiographically in three patients (9% of those who underwent echocardiography). Of patients who were over age 65, mitral annulus calcification was present in 14% (3 of 23) of echocardiograms. We conclude that 25-35% of patients with NVAF and stroke have cerebrovascular disease as the cause of brain ischemia. Embolic stroke in NVAF occurs in patients with chronic, persistent atrial fibrillation without other cardiac embolic sources. Early recurrent embolism occurs in about 15% of NVAF patients who are not immediately anticoagulated at the time of the initial embolus. Recent development of safe, noninvasive techniques (e.g., digital angiography, real-time Doppler scanning, cardiac CT) should allow better definition of stroke mechanism in these relatively elderly patients, allowing more specific therapy. References 1. Wolf PA, DawberTR. Thomas HE, Kannel WB: Epidemiologic assesment of chronic atrial fibrillation and risk of stroke: The Framingham Study. Neurology 28: , Fairfax AJ, Lambert CD, Leatham A: Systemic embolism in chronic sinoatrial disorder. N Engl J Med 295: , Hinton RC, Kistler P, Fallon JR et al: Influence of etiology of atrial fibrillation on incidence of systemic embolism. Am J Card 40: , Beer DT, Ghitman B: Embolization from the atria in arteriosclerotic heart disease. JAMA 177: Fisher CM: Reducing risks of cerebral embolism. Geriatrics 34: 59-66, Friedman GD, Loveland DB, Ehrlich SP: Relationship of stroke to other cardiovascular disease. Circulation 38: Kannel WB, Abbott RD, Savage DD, McNamara PM: Epidemiologic features of chronic atrial fibrillation: The Framingham Study. N Engl J Med 306: 1018, Cerebral Embolism Study Group: anticoagulation of embolic stroke. A Randomized Trial. Stroke 14: , Sage II, VanUitert RL: Risk of recurrent stroke with atrial fibrillation: differences between rheumatic and arteriosclerotic heart disease. Stroke 14: , Darling RC, Austin WG, Linton RR: Arterial embolism. Surg Gynecol Obstet 124: , Mohr JP, Caplan LR, Melski JW et al: The Harvard Cooperative Stroke Registry: A prospective registry. Neurology 28: , Lovctt JL, Sandok BA, Giuliani ER, Nasser FN: Two-dimensional echocardiography in patients with focal cerebral ischemia. Ann Int Med 95: 1-4, Greenland P, Knopman DS, Mikell FL, Asinger RW, Anderson DC: Echocardiography in diagnostic assessment of stroke. Ann Int Med 95: 51-53, Bharucha NE, Wolf PA, Kannel WB, McNamara PM: Epidemiological study of cerebral embolism: The Framingham Study. Ann Neurol 10: Fisher CM, Pearlman A: The nonsudden onset of cerebral embolism. Neurology 17: , 1967