We sought to determine whether thrombi of the left atrium or the left atrial appendage, spontaneous echocardiographic contrast, and size of the left atrial appendage are associated with stroke or embolism. Our hypothesis was that patients with thrombi of the left atrium or the left atrial appendage, spontaneous echocardiographic contrast, and enlargement of the left atrial appendage have a higher rate of stroke or embolism than do patients without these abnormalities. The baseline clinical and echocardiographic characteristics of our study sample have been reported elsewhere [9].

Methods

Top Methods Results Discussion Author & Article Info References

Patient Recruitment

Our study was conducted in Austria and Slovakia. At the beginning of the study, general practitioners and internists from surrounding participating centers were informed about the study and were asked to refer patients who had constant or intermittent atrial fibrillation.

At each patient's first visit, a cardiologist recorded baseline data, associated illnesses, clinical characteristics (definitions of which are listed in Appendix 2), and the cause of atrial fibrillation. A 12-lead electrocardiogram was recorded on 2 separate days. We excluded patients with prosthetic valves and those who were already receiving anticoagulation. Current therapy with aspirin was not an exclusion criterion. Patients who had mitral valve stenosis or had recently had stroke were excluded because anticoagulation was indicated for such conditions. For the definition of recent stroke, we chose a cutoff value of 1 year because a previous study reported that the risk for recurrent stroke was increased in patients who had had an ischemic event in the previous year [10].

Echocardiography

For transthoracic echocardiography, 2.5- or 3.5-MHz probes were used. For M-mode measurements, we applied the recommendations of the American Society of Echocardiography [11]. M-mode measurements were performed on-line and were documented on printouts. Regional wall-motion and valvular abnormalities were recorded. In patients without regional wall-motion abnormalities, left ventricular systolic function was assessed by calculation of left ventricular fractional shortening.

For transesophageal echocardiography, biplane 5-MHz probes were used in 64% of the patients and monoplane 5-MHz probes were used in 36%. After local anesthesia with lidocaine spray was administered, the probe was introduced into the nonsedated patients, who lay on their left side. The entire study was recorded on S-VHS videotapes. The cardiac chambers and valves were investigated by using all obtainable views [12, 13]. Special care was taken to visualize the whole cavity of the left atrium and its appendage from different planes, with the tip of the probe slightly flexed.

The size of the left atrial appendage was measured off-line. A frame that best visualized the left atrial appendage with respect to its orifice, extension, and delineation of borders was selected just before the QRS complex. The length of the left atrial appendage was defined as the distance between the limbus of the left superior pulmonary vein and the apex of the left atrial appendage; the width of the left atrial appendage was defined as the longest distance between the lateral and ventricular left atrial appendage wall, measured in a line perpendicular to the course of the left atrial appendage curvature; and the left atrial appendage area was defined as described elsewhere [14]. For the assessment of intra- and interobserver deviations, left atrial appendage size was measured 1 year after the initial measurement in 20 randomly selected video recordings by the same observer and in 60 randomly selected video recordings by three observers from different centers.

A thrombus of the left atrium or left atrial appendage was considered definite if at least three of the following five criteria were present: clear borders, echogenicity different from that of surrounding structures, independent mobility, longest diameter greater than 15 mm, and evidence of thrombus in more than one imaging plane [15]. A second observer from a different center reviewed video recordings to confirm diagnosis of thrombus of the left atrium or left atrial appendage. A thrombus was considered probable if only two criteria were present or if the observers disagreed.

Spontaneous echocardiographic contrast was diagnosed in the presence of dynamic smoke-like echoes within the left atrium or left atrial appendage with a characteristic swirling motion that was distinct from white noise artifact [16]. When spontaneous echocardiographic contrast was suspected, the gain settings were decreased in a stepwise manner to exclude artifact caused by excessive gain. The compression and gain controls were then adjusted to ensure the best possible visualization. Spontaneous echocardiographic contrast was not quantified. A second observer from a different center checked the diagnosis of spontaneous echocardiographic contrast by reviewing all video recordings. When the observers disagreed about the presence or absence of spontaneous echocardiographic contrast, the decision made at the time of examination was chosen. This was done because spontaneous echocardiographic contrast was not always visible on review of the video recordings.

Outcome Events

Primary events were stroke and embolism that extended to viscera or extremities. Stroke was defined as acute onset of a focal neurologic deficit that lasted more than 24 hours and was consistent with a vascular event. Severity and subtypes of stroke were assessed according to established criteria [17]. Embolism extending to viscera or extremities was defined as a sudden onset of pain that was associated with typical findings on Doppler ultrasonography, angiography, surgery, or autopsy.

Secondary events were the need for anticoagulation during follow-up in patients who were primarily receiving aspirin and death not caused by stroke or embolism. Death not caused by stroke or embolism was classified as cardiac or noncardiac on the basis of information from the treating physician and the patient's family.

Follow-up

Visits took place 3, 6, 12, 18, and 24 months after study entry. After a follow-up period of at least 2 years (ending in February 1994), it was evident that the number of primary outcome events was low. We therefore prolonged the follow-up period and planned further visits, one to be held in 1994 and one to be held in 1995. Patients who were unwilling or unable to attend the visits were contacted by telephone. Medical records were obtained for patients who had been admitted to a hospital. If a patient had died, an autopsy report or death certificate was obtained.

Treatment and Ethical Considerations

Although our study was designed to be observational, we recommended antithrombotic treatment to the patients because of ethical considerations. Because patients with thrombi of the left atrium or left atrial appendage are considered to have a high risk for embolism, we recommended that they receive anticoagulation. The target range for the international normalized ratio was 2.0 to 3.0, and the treating physicians monitored the ratio [5, 18]. We recommended that the remaining patients receive aspirin, 160 mg/d, because preliminary data indicated that aspirin reduces the risk for stroke or embolism in patients with atrial fibrillation [19].

The institutional review board of the participating centers approved the study. Patients were informed about the background and the procedures of the trial, and informed consent was obtained according to the rules of each center.

Statistical Analysis

For baseline data, group comparisons were done by analysis of variance for normally distributed data and by the nonparametric Kruskal-Wallis test for data that were not normally distributed. If appropriate, we used the Scheffe post hoc test. The Wilk-Shapiro test was used to check for normality. Contingency Table methods, including the chi-square test and the Fisher exact test, were used to analyze categorical data. We used the Cox proportional-hazards model to obtain estimates of relative risk and 95% CIs for all univariate and multivariate analyses of time-to-event data. All tests were two-sided. The level of significance assumed was an level of 0.05. We used SAS software (SAS Institute, Cary, North Carolina) for analyses.

Role of Funding Sources

The funding sources had no role in the collection, analysis, or interpretation of the data; did not review the manuscript at any time; and were not involved in submitting the paper for publication.

Results

Top Methods Results Discussion Author & Article Info References

Patients

Patients were recruited between January 1990 and February 1992. Of the 474 consecutive outpatients with constant or intermittent atrial fibrillation who were screened, 409 were included in the study. Sixty-five patients were excluded because of indications for anticoagulation (3 patients with prosthetic valves, 4 with mitral valve stenosis, and 6 with recent stroke), lack of evidence of atrial fibrillation (14 patients), consuming illness (10 patients), the patient's decision not to participate (19 patients), and the inability to perform transesophageal echocardiography (9 patients). At the time of the baseline examination, no included patients were receiving anticoagulant therapy. The clinical characteristics of the 409 included patients (263 men and 146 women) are listed in Table 1. The 149 patients with intermittent atrial fibrillation were younger than the patients with constant atrial fibrillation (mean age, 60 years compared with 64 years; P < 0.001); these two groups did not differ with respect to sex, associated illnesses, and clinical characteristics.

Echocardiography

The baseline echocardiographic results are listed in Table 2. Valvular abnormalities were mild mitral insufficiency in 57 patients, moderate mitral insufficiency in 53 patients, tricuspid insufficiency in 31 patients, severe mitral insufficiency in 27 patients, mild to moderate aortic stenosis in 20 patients, and mild aortic insufficiency in 10 patients. During transesophageal echocardiography, 304 patients had atrial fibrillation and the remaining 105 were in sinus rhythm. The left atrial appendage could not be visualized in 13 patients (3%). The clinical characteristics of these patients did not differ from those of patients whose left atrial appendages were visualized. Monoplane probes were used in 4 of the 13 patients, and biplane probes were used in 9 patients.

Initially, thrombi of the left atrium or left atrial appendage were identified in 20 patients (13 had definite thrombi, and 7 had probable thrombi). For 3 of the 13 definite thrombi, the second observer could not confirm the initial diagnosis. Therefore, these 3 cases were categorized as cases of probable thrombi. The remaining 10 patients with definite thrombi constituted the "thrombus" group (2.5%). Thrombi of the left atrium or left atrial appendage were associated with diabetes (P = 0.02), heart failure (P = 0.01), and low fractional shortening (P = 0.002). The 10 patients with probable thrombi and the 13 patients with a nonvisualized left atrial appendage constituted the "uncertain" group (5.6%). The remaining 376 patients made up the "no thrombus" group (91.9%). The prevalence of thrombi of the left atrium or left atrial appendage did not differ among the participating centers.

Spontaneous echocardiographic contrast was diagnosed by the initial observer in 47 patients (12%). Although the second observer could not confirm this diagnosis in 4 of these cases, the diagnosis was kept. All patients with spontaneous echocardiographic contrast had atrial fibrillation at the time of transesophageal echocardiography. Spontaneous echocardiographic contrast was associated with age (P = 0.01), constant atrial fibrillation (P < 0.001), hypertension (P = 0.004), heart failure (P = 0.04), valvular abnormalities (P = 0.007), and left atrial diameter (P = 0.003). The prevalence of spontaneous echocardiographic contrast was higher in one participating center (which was in a rural area) than in the other centers (28% compared with 9%; P = 0.001). Patients from the rural center had a higher prevalence of heart failure (71% compared with 51%; P = 0.005), a higher percentage of cases categorized as New York Heart Association class III or IV (28% compared with 7%; P = 0.001), a higher percentage of cases categorized as constant rather than intermittent atrial fibrillation (92% compared with 56%; P = 0.001), and larger left atria (49 mm compared with 45 mm; P < 0.001) than did patients from the other centers.

Left atrial appendage size was measured in 396 patients. The length ranged from 18 to 83 mm, the width ranged from 9 to 57 mm, and the area ranged from 1.2 to 16.4 cm². The mean (±SD) differences between the first and second measurement of one observer were as follows: length, 0.4 ± 7.0 mm (range, –24 to 19 mm); width, 0.2 ± 5.2 mm (range, –14 to 18 mm); and area, 0.15 ± 1.84 cm² (range, –4.3 to 4.5 cm²). The deviations in measurement among the three observers did not exceed the deviations assessed for one observer. Left atrial appendage size was associated with constant atrial fibrillation (P = 0.03) and valvular abnormalities (P < 0.001) [9]. Size of the left atrial appendage did not differ among the participating centers.

Follow-up

Follow-up ranged from 1 to 74 months (mean, 58 months). No patient was lost to follow-up. Conversion to sinus rhythm occurred in 57 of the 304 patients who had atrial fibrillation at the time of baseline transesophageal echocardiography. At the final follow-up visit, however, 24 of these 57 patients again had atrial fibrillation.

Stroke or embolism occurred in 50 patients (12%) (Table 3). The rate of stroke or embolism was 3.0% per year and was constant throughout the follow-up period. Secondary events occurred in 91 patients (22%), 38 of whom required anticoagulation (Table 3). The 268 patients without events had seven follow-up visits over 42 to 74 months (median, 56 months). The final follow-up visit was held in the physician's office (n = 217) or was conducted over the telephone (n = 51). The rates of telephone follow-up visits among the thrombus, no thrombus, and uncertain groups were similar. The rate of stroke or embolism did not differ between the rural center and the other centers (risk ratio, 1.5 [95% CI, 0.7 to 3.1]; P > 0.2), but the mortality rate did (risk ratio, 2.1 [CI, 1.2 to 3.7]; P = 0.009).

In the thrombus group. 7 of the 10 patients consented to receive anticoagulation and 3 received aspirin. Aspirin was recommended for patients in the uncertain group. Three patients in this group had stroke or embolism (1 patient whose left atrial appendage was not visualized and 2 patients with probable thrombus of the left atrium or left atrial appendage).

In a univariate analysis, the risk for stroke or embolism differed between the thrombus group and the no thrombus group (Table 2). Although 70% of patients in the thrombus group were receiving anticoagulation, they had a higher rate of stroke or embolism than patients in the no thrombus and uncertain groups (Table 2 and Table 3). Rates of stroke or embolism did not differ between patients with and those without spontaneous echocardiographic contrast (Table 2). The length and width of the left atrial appendage, but not the area, were associated with stroke or embolism (Table 2). Stroke or embolism was associated with age, systolic and diastolic blood pressure, hypertension, and previous stroke. No association was seen between stroke or embolism and constant atrial fibrillation. Patients with lone atrial fibrillation had a lower rate of stroke or embolism than patients with atrial fibrillation caused by other factors (Table 1). Reduced left ventricular fractional shortening and valvular abnormalities were associated with stroke or embolism (Table 2).

In the multivariate analysis (Cox proportional-hazards regression analysis), the following baseline clinical and echocardiographic characteristics were included: sex; age; systolic and diastolic blood pressure; presence of hyperthyroidism, hypertension, diabetes, coronary heart disease, myocardial infarction, heart failure, previous stroke, or constant atrial fibrillation; duration of atrial fibrillation; left atrial diameter; left ventricular fractional shortening; presence of valvular abnormalities, spontaneous echocardiographic contrast, or thrombi of the left atrium or left atrial appendage; and length, width, and area of the left atrial appendage. According to the multivariate Cox model, the following characteristics were associated with stroke or embolism: hypertension (risk ratio, 3.6 [CI, 1.8 to 8.4]; P = 0.001), previous stroke (risk ratio, 3.7 [CI, 1.5 to 7.5]; P = 0.002), and age (risk ratio, 1.1 [CI, 1.0 to 1.1]; P < 0.001). The significance of these three characteristics as risk factors for stroke or embolism remained the same after exclusion of the transesophageal echocardiographic characteristics from the regression analysis. Inclusion of hypertension, age, and previous stroke with presence of thrombus of the left atrium or left atrial appendage in a Cox regression analysis showed that the latter was not a statistically significant risk factor for stroke or embolism (risk ratio, 2.4 [CI, 0.9 to 6.9]; P = 0.09), but the CI suggested the possibility of a clinically significant association. When either the length (risk ratio, 1.1 [CI, 0.9 to 1.5]; P > 0.2) or width (risk ratio, 1.2 [CI, 0.8 to 2.0]; P > 0.2) of the left atrial appendage was included with hypertension, age, and previous stroke in a Cox regression analysis, the CIs of the risk ratio did not exceed 2.0.

Discussion

Top Methods Results Discussion Author & Article Info References

In this observational, multicenter follow-up study of outpatients with atrial fibrillation who had not recently had stroke, thrombi of the left atrium or left atrial appendage and length and width of the left atrial appendage were associated with stroke or embolism in a univariate analysis, whereas spontaneous echocardiographic contrast and enlarged area of the left atrial appendage were not. In a multivariate analysis, age, hypertension, and previous stroke were risk factors for stroke or embolism, as was seen in another study [20]. Among the transesophageal echocardiographic findings, thrombi of the left atrium or left atrial appendage may be risk factors for stroke or embolism.

Because of the low prevalence of thrombi of the left atrium or left atrial appendage in our series (2.5%), it is questionable whether such thrombi can be considered powerful risk factors for stroke or embolism. This low prevalence contrasts with findings from previous studies, which reported a prevalence of 7% to 21% [7, 21-24]. This difference can be explained in part by patient characteristics. Our patients were younger than those in the other studies (mean age, 62 years compared with 64 to 73 years) and had a higher prevalence of intermittent atrial fibrillation (39% compared with 0% to 36%) and lone atrial fibrillation (27% compared with 4% to 25%) [7, 22-24]. In the previous studies, 10% to 41% of patients had recently had embolism, 6% to 8% had mitral valve stenosis, and 1% to 4% had prosthetic valves; in contrast, we excluded patients with these conditions from our study.

An additional reason for the low prevalence of thrombi of the left atrium or left atrial appendage in our study may be the diagnostic criteria that we used [15]. More sensitive diagnostic criteria (that is, including the categorization of "uncertain" in the thrombus group) would have increased the prevalence of thrombi of the left atrium or left atrial appendage to 8.1%. As a result of this reclassification, however, thrombi of the left atrium or left atrial appendage would no longer be associated with stroke or embolism (risk ratio, 1.9 [CI, 0.9 to 4.3]; P = 0.1) on univariate analysis (Table 2). Although multivariate analysis did not identify thrombi of the left atrium or left atrial appendage as statistically significant risk factors for stroke or embolism, the high risk ratio of 2.4 and the 95% CI of 0.9 to 6.9 still provide evidence of an association that might be clinically relevant.

The 11.5% prevalence of spontaneous echocardiographic contrast was lower in our study than previously reported [7, 22-24]. Again, patient characteristics may explain these differences. Spontaneous echocardiographic contrast was associated with age and constant atrial fibrillation; in the other studies, patients were older and constant atrial fibrillation was more prevalent than in our study [7, 22-24]. Differences in the echocardiographic equipment used and in the standardization of technical adjustments and recording techniques may also have contributed to the different prevalence of spontaneous echocardiographic contrast [9].

In our study, spontaneous echocardiographic contrast was not associated with stroke or embolism. Two other prospective studies have examined this issue: Leung and associates' study of 272 patients with constant atrial fibrillation [7] found that spontaneous echocardiographic contrast was related to stroke or embolism, whereas Archer and coworkers' study of 55 patients [8] did not. These conflicting results may be explained in part by differences in sample size, which resulted in a power of 41% in Leung and associates' study and of 1% in Archer and coworkers' study. The difference may also be explained by patient characteristics: Archer and coworkers included only outpatients, and they excluded patients with recent embolism and those with an absolute requirement for anticoagulation. Leung and associates included these patients. The overall mortality rate was 4.0% per year in our patients, 6.4% per year in Archer and coworkers' patients, and 10.5% per year in Leung and associates' patients. We infer from these data that Leung and associates' patients were sicker than our patients and those of Archer and coworkers. In our study, the low rate of stroke or embolism and the low prevalence of spontaneous echocardiographic contrast led to a power of only 14% of the applied statistical tests. It is therefore inappropriate to definitely conclude that there was an absence of an association between spontaneous echocardiographic contrast and stroke or embolism.

Researchers performing anatomic and transesophageal echocardiographic studies have reported an association of enlarged left atrial appendage with atrial fibrillation and thrombi [25, 26]. In our study, length and width of the left atrial appendage were associated with stroke or embolism in univariate analysis (Table 2). These associations, as well as the association of thrombi of the left atrium or left atrial appendage with stroke or embolism, vanished when age, hypertension, and previous stroke were included with the transesophageal echocardiographic characteristics in a multivariate analysis. Age, hypertension, and previous stroke are similar to risk factors for stroke or embolism identified by other studies of patients with atrial fibrillation [20].

The main limitations of our study are the low prevalences of spontaneous echocardiographic contrast, thrombi of the left atrium or left atrial appendage, and stroke or embolism. The transesophageal echocardiographic technology used in our study is now out of date, a limitation inherent in every long-term follow-up study. Measurement of the size of the left atrial appendage was impeded by this limitation. We also used monoplane and biplane probes, which are inferior to the currently available multiplane probes in visualizing the complex anatomy of the left atrial appendage [25, 27]. These problems are also reflected by the considerable interobserver and intraobserver deviations. It is uncertain, however, whether the use of monoplane probes has influenced the number of overlooked thrombi. It probably did not have an effect because the rate of stroke or embolism was similar in patients investigated with monoplane and biplane probes. The currently used multiplane probes are reported to improve the accuracy in diagnosing thrombi of the atria or atrial appendages [28]. No studies have examined whether the sensitivity and specificity of multiplane probes in detecting such thrombi are higher than those of biplane probes. To our knowledge, multiplane probes do not improve the diagnosis of spontaneous echocardiographic contrast. Neither assessment of the thoracic aorta nor Doppler studies of left atrial appendage flow were part of the protocol because neither had been standardized sufficiently for use in a multicenter trial at the beginning of our study. Finally, we did not perform laboratory tests to assess the efficacy of platelet inhibition, and we did not monitor the intensity of anticoagulation.

In outpatients without recent stroke and thrombi of the left atrium or left atrial appendage, length and width of the left atrial appendage were associated with stroke or embolism in univariate analysis. According to multivariate analysis, age, hypertension, and previous stroke were risk factors for stroke or embolism, and thrombi of the left atrium or left atrial appendage may be a risk factor. In these patients, the history may be more useful than transesophageal echocardiography for the assessment of risk for embolism.

Appendix 1: Study Group Participants

Principal investigator: Claudia Stollberger, MD.

Krankenanstalt Rudolfstiftung, Vienna, Austria: Christina Abzieher; Friedrich Abzieher. MD; Gunther Ernst; Gabriele Krutisch, MD; Thomas Langer; Waltraud Lutz; Elisabeth Punz, MD; Stanislaus Rakusan, MD; Raimund Reisinger; Jorg Slany, MD; and Walther B. Winkler, MD.

Krankenhaus Florisdorf, Vienna, Austria: Christian Bachl, MD; Michael Mischuretz, MD; and Leopold Peschl, MD.

Krankenhaus Krems, Krems. Austria: Gerhard Kronik, MD; Maria Schuh, MD (deceased); and Petra Steininger, MD.

Ustav kardiovaskularnych chorob, Bratislava, Slovakia: Pavel Chnupa, MD; Robert Hatala, MD; Kristina Havlinova, MD; Igor Riecansky, MD; and Michal Simo, MD (deceased).

Additional participants: Michael Brainin, MD; Volker Enzenhofer, MD; Josef Finsterer, MD; Zofia Reinbacher; Victor Scheiber, PhD; and Barbara Schneider, PhD.

Appendix 2: Definitions of Clinical Characteristics

Constant atrial fibrillation was diagnosed if it persisted for at least 3 weeks on repeated electrocardiography before study entry and if the patient did not report symptoms suggestive of intermittent atrial fibrillation.

Coronary heart disease was categorized as "definite" or "possible." Definite disease referred to resting or exercise angina pectoris; myocardial ischemia had to be documented by exercise electrocardiography or thallium imaging. Either the patient had to have a history of definite myocardial infarction or coronary angiography must have shown greater than 50% stenosis of at least one coronary artery. "Possible" coronary heart disease was diagnosed if clinical symptoms were suggestive of coronary heart disease but no objective testing was done.

Diabetes was considered to be present if the patient was receiving oral antidiabetic agents or insulin.

Duration of atrial fibrillation was estimated by the history given by the patient and the referring physician.

Heart failure was diagnosed if the patient showed clinical or radiologic evidence of impaired cardiac function. Failure was graded according to the New York Heart Association classification.

Hypertension was defined as a systolic blood pressure exceeding 160 mm Hg or a diastolic blood pressure exceeding 90 mm Hg on repeated observations over 3 months; cut-off blood pressure was lower in patients receiving long-term antihypertensive therapy.

Hyperthyroidism was defined as current or previous treatment of thyrotoxicosis.

Intermittent atrial fibrillation was diagnosed if there were at least two episodes of atrial fibrillation alternating with sinus rhythm documented by electrocardiography within 1 month before study entry.

Myocardial infarction was diagnosed when one of the following criteria was documented: 1) serial electrocardiographic or enzyme changes compatible with myocardial infarction or 2) a history of compatible clinical syndromes plus abnormal Q waves or diminishing R-wave amplitude in two or more adjacent precordial leads.

Previous stroke was defined as acute onset of a focal ischemic neurologic deficit lasting more than 24 hours without evidence of other causes. This event must have occurred more than 1 year before study entry.

Type of atrial fibrillation was labeled "hypertensive" if hypertension was the most probable cause, "ischemic" if coronary heart disease was the most probable cause, and "dilative" in patients with dilative cardiomyopathy. It was considered "postinflammatory" in patients with a history of endocarditis or myocarditis. If none of these cardiovascular diseases were present and no other cause of atrial fibrillation could be found, "lone atrial fibrillation" was diagnosed.

From Ustav kardiovaskularnych chorob, Bratislava, Slovakia; Krankenhaus Krems, Krems, Austria; Neurologisches Krankenhaus, Maria Gugging, Austria; and Neurologisches Krankenhaus Rosenhugel, Institut fur medizinische Statistik, and Krankenanstalt Rudolfstiftung, Vienna, Austria.

Dr. Chnupa: Ustav kardiovaskularnych chorob, Partizanska 2, 81105 Bratislava, Slovakia.

Dr. Kronik: Krankenhaus Krems, Mitterweg 10, A-3500 Krems, Austria.

Dr. Brainin: Neurologisches Krankenhaus, Hauptstrasse 2, A-3400 Maria Gugging, Austria.

Dr. Finsterer: Neurologisches Krankenhaus Rosenhugel, Riedelgasse 5, A-1130, Vienna, Austria.

Dr. Schneider: Institut fur medizinische Statistik, Schwarzspanierstrasse 17, A-1090 Vienna, Austria.

Dr. Slany: Krankenanstalt Rudolfstiftung, Juchgasse 25, A-1030 Vienna, Austria.