Journal of Clinical Neuroscience 20 (2013) 554–559

Contents lists available at SciVerse ScienceDirect

Journal of Clinical Neuroscience

journal homepage: www.elsevier .com/ locate/ jocn

Clinical Study

Should transesophageal echocardiography be performed in acute strokepatients with atrial fibrillation?

Juliane Herm a, Maria Konieczny b, Gerhard Jan Jungehulsing a,b, Matthias Endres a,b, Arno Villringer c,d,Uwe Malzahn e, Peter U. Heuschmann e, Karl Georg Haeusler a,b,⇑a Center for Stroke Research Berlin, Charité – Universitätsmedizin Berlin, Berlin, Germanyb Department of Neurology, Charité – Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, Berlin D-12200, Germanyc Max-Planck Institute and University Hospital, Leipzig, Germanyd Berlin School of Mind and Brain, Mind and Brain Institute, Humboldt-University, Berlin, Germanye Institute of Clinical Epidemiology and Biometry, Universität Würzburg, Würzburg, Germany

a r t i c l e i n f o

Article history:Received 2 November 2011Accepted 10 March 2012

Keywords:AnticoagulationAtrial fibrillationCardiac sources of embolismEchocardiographyIschemic stroke

0967-5868/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.jocn.2012.03.049

⇑ Corresponding author. Tel.: +49 30 84454244; faxE-mail address: georg.haeusler@charite.de (K.G. H

a b s t r a c t

The diagnostic need for echocardiography in acute stroke patients with documented atrial fibrillation(AF) is controversial because the index stroke per se is an indication for therapeutic anticoagulationaccording to guidelines. We retrospectively analyzed medical records of 2390 stroke patients consecu-tively admitted over a 2-year period to three different stroke units at university hospitals in Berlin, Ger-many. AF was diagnosed in 21.2% (n = 506) of 2185 patients with acute ischemic stroke or transientischemic attack. Overall, 36.6% (n = 185) of all AF patients underwent transesophageal echocardiography(TEE) or transthoracic echocardiography within days of hospital admission. According to multivariateanalysis, age and in-hospital conventions determined the diagnostic use of TEE in stroke patients withknown AF, while the existing cardiovascular risk profile had no impact. Major cardiac sources of embo-lism were identified by echocardiography in 9.7% (n = 18) of all AF patients with acute stroke, includingnon-AF-related sources of embolism in 3.8% (n = 7). However, echocardiographic findings did not result inany therapeutic intervention other than immediate anticoagulation. Furthermore, echocardiographicfindings had no impact on the prescription of anticoagulants at hospital discharge or long-term survival.Taken together, our data indicate that diagnostic echocardiography offers only a little additional informa-tion and does not impact clinical management and outcome in acute stroke patients with known AF.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Post-stroke diagnostic work-up is of particular importance foreffective secondary stroke prevention, but diagnostic standardsvary considerably between stroke centers.1 Guidelines recommendechocardiography for stroke patients with suspected cardiacsources of embolism,2 and transesophageal echocardiography(TEE) reveals relevant pathologic findings in up to one-third ofthese patients.3–7 While TEE is superior to transthoracic echocardi-ography (TTE) for the detection of left atrial (LA) thrombi, endocar-ditis and aortic plaque,6,8,9 it is a time-consuming, semi-invasiveprocedure and serious complications may rarely occur.10 In addi-tion, echocardiographic findings seem to have less therapeutic im-pact in elderly stroke patients.11,12

One might argue that echocardiographic findings do not haveany therapeutic impact in patients with known AF because the in-dex stroke per se is an indication to administer anticoagulants to

ll rights reserved.

: +49 30 84454264.aeusler).

these patients according to current guidelines.13,14 However, echo-cardiography might reveal AF-independent sources of embolism,such as endocarditis, left ventricular thrombi, aortic dissection orsevere atherosclerosis in the aortic arc, affording appropriate ther-apy to prevent further strokes. Moreover, a proven cardiac throm-bus might need immediate therapeutic anticoagulation, which isoften delayed in daily clinical practice due to the risk of intracra-nial hemorrhage in the acute phase of stroke.15 While the guide-lines of the European Stroke Organisation state that TEE ‘‘allowsrisk stratification for further thromboembolic events’’2 in strokepatients with AF, current US guidelines do not recommend TEE instroke patients with ‘‘known cardiac source’’ of embolism ‘‘inwhich a TEE would not change management’’.16 Other guidelinesdo not give specific recommendations for diagnostic echocardiog-raphy in stroke patients with AF.14,17

The aims of this study were: (i) to identify factors influencingthe physicians’ choice to perform diagnostic TEE in stroke patientswith documented AF prior to echocardiography; (ii) to evaluate thefrequency of pathologic echocardiographic findings in acute strokepatients with documented AF before hospital discharge; and (iii) to

J. Herm et al. / Journal of Clinical Neuroscience 20 (2013) 554–559 555

determine the impact of echocardiographic findings on early sec-ondary stroke prevention and long-term survival.

2. Methods

2.1. Study design and study population

The study was approved by the Ethics Committee of the CharitéUniversity of Medicine, Berlin, Germany (EA1/186/07). We retro-spectively analyzed medical records of all stroke patients treatedat three different university stroke units in Berlin, Germany. Thehospitals are located in the south (Benjamin Franklin Hospital),the north (Rudolf Virchow Hospital) and the centre of Berlin (Cha-rité Mitte Hospital), providing stroke unit services to the local pop-ulation without any selection criteria. In order to identify strokepatients, we used relevant World Health Organization (WHO)International Classification of Diseases (ICD-10) discharge diagno-ses (I61.x; I63.x; G45.x) as described in a recent publication.18

Overall, 2390 stroke patients were consecutively admitted be-tween 1 January 2003 and 31 December 2004. This analysis was re-stricted to all patients with AF (including paroxysmal, persisting orpermanent AF) and ischemic stroke or transient ischemic attack(TIA; according to the WHO criteria19). We assessed whether AFwas diagnosed pre-admission or prior to in-hospital echocardiog-raphy. The following information was taken from medical records:age, gender, antithrombotic medication on admission and dis-charge, comorbid congestive heart failure, hypertension, diabetes,previous stroke or TIA, intracerebral hemorrhage or non-strokevascular events and diagnostic results during hospital stay. Pa-tients on full-dose heparin (n = 2) as well as patients on a combina-tion of oral anticoagulation and antiplatelets (n = 8) were assignedto the anticoagulation group for data analysis. The pre-admissionCHADS2 score20 was determined according to medical records.Functional outcome was assessed using the modified Rankin scale(mRS)21 and the National Institute of Health Stroke Scale (NIHSS).22

Stroke severity was defined according to NIHSS score on admissionas ‘‘mild’’ (0–5 points), ‘‘moderate’’ (6–14 points) or ‘‘severe’’ (P 16points).23

Echocardiographic reports were reviewed and pathologic find-ings were defined as previously described.8,24 Major sources ofembolism (SoE) included: endocarditis, left atrial (LA) or left ven-tricular (LV) tumor or thrombus, LV dyskinesia and ejection frac-tion (EF) < 30%. Minor SoE included: spontaneous echo contrast,mitral valve prolapse, patent foramen ovale (and coexisting atrialseptum aneurysm), ventricular aneurysm, aortic plaques (definedas irregular intimal thickening of P 2 mm) as well as ascendingaortic aneurysm. There were no joint standard operating proce-dures for performing echocardiography in acute stroke patientsat the different university stroke units at that time. The use ofechocardiography in stroke patients with AF was based on clinicaljudgment. Survival of all patients was assessed in August 2011,based on data on vital status from the local Residents’ RegistrationOffice.

2.2. Statistical analysis

The results were reported as absolute and relative frequenciesfor categorical variables, mean and standard deviation (SD) forage as well as median and interquartile range (IQR; first to thirdquartile) for variables of time duration. Fisher’s exact test was usedto test differences in proportions for dichotomous characteristicsbetween independent groups. The exact Mann–Whitney U-testwas applied to detect differences in the distribution of metrical,not normally distributed variables between independent groups.Based on the results taken from univariate analysis we used mul-

tivariate logistic regression to detect factors influencing the prob-ability of TEE performance. Because of small sample sizes weapplied the method of forward stepwise regression to identifythe subset of influencing factors. We performed a multivariateanalysis of covariance (including age, categorized NIHSS on admis-sion and place of hospitalization) to examine whether TEE perfor-mance significantly prolonged mean in-hospital stay. To agree withthe assumption of approximately normal distribution for thedependent variable, we applied the square root transformationfor duration of in-hospital stay. Levene’s test of equality of errorvariance did not reveal a violation of the homogeneity of variancesassumption. To calculate the odds of undergoing TEE according tothe admitting stroke unit, we used a multivariable logistic regres-sion adjusted for age group and prior anticoagulation.

A p-value of < 0.05 was considered to be significant. Data anal-ysis was exploratory and we did not use methods for multiple test-ing and consequently did not control ‘‘familywise error rates’’when applying Fisher’s exact test and the Mann–Whitney U-test.Data were analyzed using Predictive Analytics Software (PASW)Statistics 19 (IBM, Armonk, NY, USA).

3. Results

Out of 2185 patients with ischemic stroke, 506 (21.2%) hadeither documented AF during hospital stay or a past medical his-tory of AF. In 336 (66.4%) of these 506 patients AF was alreadyknown prior to admission. Newly detected AF was diagnosed in110 (21.7%) stroke patients by 12-lead resting electrocardiography(ECG) on admission, in 40 (7.9%) patients by an additional 24-hourHolter ECG and in 20 (4.0%) patients by bedside ECG monitoring.The mean age of the AF patients was 75.9 years (SD, 9.5; range,39–97) and 57% were female. During a median in-hospital stay of10 (IQR, 7–14) days, 35 (6.9%) stroke patients with AF died. Sev-enty-five AF patients (14.8%) died within 1 month, and 90(17.8%) patients died within 3 months after stroke onset.

In total, 185 (36.6%) of 506 AF patients underwent echocardiog-raphy during hospital stay (Fig. 1). In detail, 46 (9.2%) patients re-ceived TEE and additional TTE, 29 (5.7%) TEE only and 110 (21.7%)received TTE only. No major complication of TEE or TTE was re-ported. Delay from stroke onset to TEE and TTE varied significantly(p = 0.006). While TTE was done after a median of 3.0 (IQR, 1.0–5.0)days, patients underwent TEE after 4.0 (IQR, 2.0–7.0) days. Multi-variate analysis showed borderline significance (p = 0.049) for aprolonged hospital stay of AF patients who underwent TEE. Table 1summarizes the characteristics of all patients with AF, and thosepatients undergoing TEE, TTE or no echocardiography.

3.1. Use of transesophageal echocardiography in stroke patients withknown atrial fibrillation before echocardiography

Of 185 AF patients undergoing echocardiography, 159 (85.9%)patients had documented AF prior to echocardiography, while AFwas documented during or after the procedure in 26 (14.1%). Ofthose 159 patients with known AF, 37 (23.2%) received TEE andTTE, 25 (15.7%) TEE and 97 (61.0%) TTE. We compared stroke pa-tients with known AF prior to TEE (n = 62) with those who didnot receive TEE (i.e. TTE only or no echocardiography at all;n = 431). According to multivariate analysis (Table 2), age < 65years and in-hospital conventions were independently associatedwith TEE use. Compared with the other stroke units, the odds ofundergoing TEE was increased more than three-fold (OR, 3.3;95% confidence interval [CI], 1.9–5.9), if a stroke patient wasadmitted to Stroke Unit 2. There was a trend for AF patients with-out anticoagulation on admission to undergo TEE (p = 0.057), whilesex, CHADS2 score, mRS and NIHSS score had no impact.

Fig. 1. Flow diagram showing study profile. AF = atrial fibrillation, TEE = trans-esophageal echocardiography, TIA = transient ischemic attack, TTE = transthoracicechocardiography.

556 J. Herm et al. / Journal of Clinical Neuroscience 20 (2013) 554–559

3.2. Pathologic echocardiographic findings in stroke patients withatrial fibrillation

Echocardiography revealed pathologic findings in 67 (36.2%) of185 stroke patients with AF who underwent echocardiography.These findings were considered as major cardiac SoE in 18 (9.7%)patients, minor cardiac SoE in 44 (23.8%) patients and minornon-cardiac SoE in 26 (14.1%) patients (Table 3). Patients undergo-

Table 1Characteristics of patients with ischemic stroke with atrial fibrillation

P

(n = 506)

Age in years; mean (SD) 75.9 (9.5)Female sex; n (%) 289 (57.1)Transient ischemic attack; n (%) 40 (7.9)Comorbidities; n (%)

Previous stroke or TIA 126 (24.9)Diabetes 172 (34.0)Hypertension 386 (76.3)Heart failure 165 (32.6)Coronary artery disease 149 (29.4)Peripheral artery disease 29 (5.7)Artificial heart valve 8 (1.6)

CHADS2 before admission; mean (SD) 2.6 (1.4)mRS on admission; mean (SD) 3.1 (1.7)NIHSS on admission; mean points (SD) 8.4 (6.8)Anticoagulation on admission; n (%) 73 (14.4)Anticoagulation at discharge; n (%) 271 (57.5)Hospital admission; n (%)

Stroke Unit 1 104 (20.6)Stroke Unit 2 133 (26.3)Stroke Unit 3 269 (53.2)

Hospital stay in days; median (IQR) 10 (7–14)

CHADS2 score was calculated according to Gage et al.20

IQR = interquartile range, mRS = modified Rankin scale, NIHSS = National Institute of HeaTIA = transient ischemic attack, TTE = transthoracic echocardiography.

ing TEE had major SoE in 12.0% and minor SoE in 50.7%, while TTErevealed major SoE in 8.2% and minor SoE in 5.5%. The in-hospitaldelay to echocardiography did not differ for those patients with orwithout proven major SoE (p = 0.74).

Cardiac thrombi (n = 8; 4.3%) provided almost half of the majorSoE, while there were no patients with cardiac tumor or endocar-ditis. AF-related cardiac SoE (i.e. mitral valve stenosis [n = 4] orLA thrombi [n = 8]) were present in 11 (5.9%) of all patients under-going echocardiography. Non-AF-related major SoE were detectedin 7 (3.8%) AF patients (LV dyskinesia with EF < 30% [n = 6]; LVthrombus (n = 1]). Moreover, about one-third of TEE-patients hadatherosclerotic plaques in the aortic arch (Table 3).

3.3. Impact of echocardiographic findings in stroke patients with atrialfibrillation

According to univariate analysis (Table 4), there was a trend(p = 0.035) for comorbid heart failure in patients with major SoE.None of age, sex, diabetes, arterial hypertension, coronary arterydisease, peripheral artery disease, CHADS2 score or anticoagulationon admission was associated with major cardiac SoE. A proven ma-jor SoE had no statistically significant effect on the prescription ofanticoagulants at hospital discharge (p > 0.995) because 11 out of18 (61.1%) surviving patients with major cardiac SoE and 98 outof 162 (60.5%) surviving patients without proven cardiac SoE re-ceived anticoagulation at hospital discharge. Anticoagulation wasstarted immediately after hospital admission in the majority ofstroke patients available for anticoagulation. Subsequently, we ob-served no impact of diagnostic echocardiography on the timing ofanticoagulation. The presence of major cardiac SoE did not affectlong-term mortality. During the first year post-stroke, 16.7% of pa-tients with major cardiac SoE and 17.4% without proven major car-diac SoE died (p > 0.995). Five years post-stroke, 41.2% of patientswith major cardiac SoE and 44.8% of patients without were re-ported dead (p > 0.995). Comparing patients with proven majorcardiac SoE with all other stroke patients with AF, there were sim-ilar rates of anticoagulation at hospital discharge (p = 0.813) andcomparable 5-year mortality rates (p = 0.219).

TEE(n = 75)

TTE(n = 110)

No echocardiography(n = 321)

70.1 (8.7) 75.3 (9.5) 77.5 (9.2)34 (45.3) 57 (51.8) 198 (61.7)5 (6.7) 4 (3.6) 31 (9.7)

15 (20.3) 23 (20.9) 88 (27.3)20 (26.7) 50 (45.5) 102 (31.8)50 (68.0) 84 (76.4) 252 (78.2)23 (30.7) 43 (39.1) 99 (30.8)21 (28.0) 41 (37.3) 87 (27.1)6 (8.0) 8 (7.3) 15 (4.7)1 (1.3) 5 (4.5) 2 (0.6)2.0 (1.4) 2.6 (1.4) 2.7 (1.4)2.9 (1.7) 3.2 (1.6) 3.1 (1.7)6.9 (6.0) 8.2 (6.6) 8.8 (7.0)5 (6.7) 20 (18.2) 48 (15.0)52 (70.3) 57 (54.3) 162 (55.7)

10 (13.3) 62 (56.4) 32 (10.0)36 (48.0) 37 (33.6) 60 (18.7)29 (38.7) 11 (10.0) 229 (71.3)11 (8–15) 10 (7–14) 10 (7–14)

lth Stroke Scale, SD = standard deviation, TEE = transesophageal echocardiography,

Table 2Factors associated with diagnostic use of transesophageal echocardiography (TEE) in stroke patients with documented AF prior to TEE

Univariate analysis Multivariate analysis

TEE(n = 62)

No TEE(n = 431)

pa OR (95% CI) p

Age group <0.0005* <0.0005*

< 65 years 17 (27.4) 42 (9.7) 165–84 43 (69.4) 299 (69.4) 0.33 (0.17–0.65)P 85 years 2 (3.2) 90 (20.9) 0.05 (0.01–0.21)

Female sex 30 (48.4) 255 (59.2) 0.130 NDCHADS2 score before admission 0.151 ND

CHADS2 = 0–1 21 (33.9) 99 (23.0)CHADS2 = 2–3 28 (45.2) 209 (48.5)CHADS2 P 4 13 (21.0) 123 (28.5)

NIHSS on admission; points 0.384 NDNIHSS = 0–5 30 (48.4) 191 (44.3)NIHSS = 6–15 24 (38.7) 152 (35.4)NIHSS P 16 8 (13.9) 88 (20.4)

modified Rankin Scale (mRS) score on admission 0.719 NDmRS = 0–1 13 (21.0) 85 (19.7)mRS = 2–3 23 (37.1) 142 (32.9)mRS P 4 26 (41.9) 204 (47.3)

Anticoagulation on admission 5 (8.1) 68 (15.8) 0.127 0.39 (0.15–1.03) 0.057Hospital admission <0.0005* <0.0005*

Stroke Unit 1 10 (9.6) 94 (90.4) 1Stroke Unit 2 29 (23.0) 97 (77.0) 5.62 (1.48–7.41)Stroke Unit 3 23 (8.7) 240 (91.3) 0.99 (0.45–2.21)

Values are expressed as n (%). CHADS2 score calculated according to Gage et al.20

AF = atrial fibrillation, CI = confidence interval, mRS = modified Rankin scale, ND = variable not included within the selected model, NIHSS = National Institute of Health StrokeScale, OR = odds ratio, TEE = transesophageal echocardiography.* Statistical significance. ap values are based on Fisher’s exact test or exact Mann–Whitney U-test, as appropriate.

Table 3Pathologic echocardiographic findings in acute ischemic stroke patients with atrialfibrillation

TEE ± TTE(n = 75)

TTE(n = 110)

Patients with major cardiac sources of embolism* 9 (12.0) 9 (8.2)Left ventricular thrombus 0 1 (0.9)Left atrial thrombus 2 (2.7) 0Left atrial appendage thrombus 6 (8.0) 0Mitral valve stenosis 1 (1.3) 3 (2.7)Left ventricle hypokinesia and EF < 30% 1 (1.3) 5 (4.5)

Patients with minor cardiac sources of embolism* 38 (50.7) 6 (5.5)Spontaneous echo contrast 26 (34.7) 1 (0.9)Isolated PFO 8 (10.6) 0PFO/ASD & atrial septum aneurysm 2 (2.7) 3 (2.7)Mitral valve prolapsed 5 (6.7) 1 (0.9)Ventricular aneurysm 0 1 (0.9)

Patients with non-cardiac sources of embolism* 24 (32.0) 2 (1.8)Aortic aneurysm 1 (1.3) 0Aortic plaques 24 (32.0) 2(1.8)

CHADS2 score calculated according to Gage et al.20

ASD = atrial septal defect; EF = ejection fraction; PFO = patent foramen ovale;TEE = transesophageal echocardiography; TTE = transthoracic echocardiography.* Some patients had more than one source of embolism. Values are expressed as n (%).

J. Herm et al. / Journal of Clinical Neuroscience 20 (2013) 554–559 557

4. Discussion

To investigate whether semi-invasive TEE is a useful diagnostictool in AF patients with acute ischemic stroke, we retrospectivelyanalyzed the frequency and clinical relevance of pathologic echo-cardiographic findings in 506 stroke patients, consecutively admit-ted to different stroke units in Berlin, Germany.18

We observed considerable variation regarding the diagnosticuse of echocardiography within the analyzed stroke units, reflect-ing the variance of guideline recommendations2,14,16,17 and under-lining the need for joint standard operating procedures. Overall,echocardiography was performed in about one-third of all AF pa-tients with acute stroke (Table 1) and significantly prolonged the

in-hospital stay. If AF was known before echocardiography, the pa-tient’s age and in-hospital conventions, but not the existing cardio-vascular risk profile, had an impact on whether TEE wasperformed. Furthermore, there was an inverse trend for TEE usein patients receiving anticoagulation on admission, while the car-diovascular risk profile and stroke-related disability on admissionhad no impact (Table 2).

Echocardiography revealed a major cardiac SoE in about one outof ten AF patients who underwent TEE and/or TTE. While TTE re-vealed a major cardiac SoE in 8% of patients investigated, therewere major cardiac SoE in 12% of all AF patients undergoing TEE(Table 3). Comparable TEE studies reported major cardiac SoE in30–33% of AF patients3,25 and 20–28% in patients with ‘‘crypto-genic’’ stroke.4–6 The TEE-detected rate of LA (appendage) thrombi(11%) and the detected rate of LV thrombi (0.5%) were consistentwith previous studies in AF patients, reporting LA (appendage)thrombi in 10–16% and LV thrombi in 1.1–1.5%, respectively.25,26

None of age, NIHSS score on admission or pre-admission CHADS2

score predicted the occurrence of major cardiac SoE. Pre-definedminor cardiac SoE were detected in about 50% of all patientsundergoing TEE, which is in line with previous studies of TEE instroke patients with AF reporting minor cardiac SoE in 58–62%.3,25 However, definitions of minor cardiac SoE varied to a cer-tain extent, limiting the comparability of these studies.

Interestingly, TEE revealed non-AF-related SoE (mainly aorticplaques) in about one-third of all AF patients (Table 3), while anon-AF-related major SoE was present in only 4% of all stroke pa-tients with AF undergoing echocardiography. However, such find-ings do not prove that AF was not the cause of stroke but furtheradded to the individual stroke risk. Notably, echocardiographydid not reveal a cardiac or non-cardiac SoE that required immedi-ate therapeutic intervention other than anticoagulation. Furtherindicating the low impact of echocardiography on the clinical man-agement in this subset of stroke patients, the prescription rate ofanticoagulants was similar at hospital discharge in stroke patientswith or without proven major cardiac SoE. This was primarily

Table 4Characteristics of stroke patients with atrial fibrillation with a proven major cardiac source of embolism

Major cardiac SoE(n = 18)

No major cardiac SoE(n = 167)

pa

Age group 0.159< 65 years 7 (38.9) 32 (19.2)65–84 years 10 (55.6) 118 (70.7)P 85 years 1 (5.6) 17 (10.2)

Female sex 6 (33.3) 85 (50.9) 0.215Anticoagulation on admission 1 (5.6) 24 (14.4) 0.475Comorbidities

Diabetes 7 (38.9) 63 (37.7) > 0.995Hypertension 14 (77.8) 121 (72.5) 0.784Heart failure 11 (61.1) 55 (32.9) 0.035Coronary artery disease 6 (33.3) 56 (33.5) > 0.995Peripheral artery disease 2 (11.1) 12 (7.2) 0.631

CHADS2 score before admission 0.465CHADS2 = 0–1 7 (38.9) 44 (26.3)CHADS2 = 2–3 7 (38.9) 87 (52.1)CHADS2 P 4 4 (22.2) 36 (21.6)

NIHSS on admission; points 0.102NIHSS = 0–5 11 (61.1) 78 (46.7)NIHSS = 6–15 3 (16.7) 67 (40.1)NIHSS P 16 4 (22.2) 22 (13.2)

Hospital stay in days; median (IQR) 10 (7–14) 12 (8–18) 0.399

CHADS2 score calculated according to Gage et al.20 Values are expressed as n (%) if not otherwise stated.a p-values are based on Fisher’s exact test or exact Mann–Whitney U-test as appropriate. IQR = interquartile range, NIHSS = National Institute of Health Stroke Scale,

SoE = source of embolism.

558 J. Herm et al. / Journal of Clinical Neuroscience 20 (2013) 554–559

caused by contraindications for anticoagulation (e.g. risk of fallingor recent intracranial hemorrhage). Furthermore, mitral valve ste-nosis was defined as a major cardiac SoE in patients with AF8,24 butthese patients did not receive immediate anticoagulation in mostcases. In addition, the estimated risk of bleeding complications inthe acute phase of stroke might have lowered the rate of anticoag-ulation at hospital discharge in our cohort of patients. Whethernew thrombin or factor Xa inhibitors will increase prescriptionrates of anticoagulants on hospital admission is an important ques-tion, because recent trials did not focus on acute stroke patientswith AF.

Finally, echocardiographic findings had no effect on long-termsurvival, although residual confounding has to be taken into ac-count because information on recurrent stroke and cause of deathwas missing. Further limitations of our study are the retrospectivestudy design and the limited number of TEE patients. The rate ofcardiac SoE in the TTE group may be artificially low because pa-tients with suspicious findings might have preferentially under-gone TEE afterwards. Notably, echocardiography was notroutinely performed, limiting the ability to draw final conclusions,but reflecting daily clinical practice in stroke patients with knownAF. The strengths of our study include the ability to discriminatebetween known and unknown AF prior to diagnostic echocardiog-raphy, the long-term follow up and the pooled analysis of datafrom three university stroke units. However, large prospectivemulticenter studies are needed to replicate our findings.

Conflicts of interest/disclosures

GJJ reports lecture fees by Bayer Healthcare, Boehringer Ingel-heim, Genzyme, Novartis, Pfizer, Sanofi-Aventis and Takeda. MEhas received grant support from AstraZeneca and Sanofi-Aventis,has participated in advisory board meetings of Boehringer Ingel-heim and Sanofi-Aventis, and has received honoraria from Novar-tis, Pfizer, EVER, Bayer Healthcare, AstraZeneca, BoehringerIngelheim, Sanofi-Aventis, Trommsdorff, Berlin-Chemie, Glaxo-SmithKline, and Bristol-Myers-Squibbs. AV reports lecture fees byBoehringer Ingelheim, Sanofi-Aventis, Ipsen, Bayer (Schering) andBerlin Chemie, and has participated in advisory board meetings

of Boehringer Ingelheim and Bristol-Myers-Squibbs. KGH receivedlecture fees by Sanofi-Aventis and Bayer Healthcare.

Acknowledgements

The project has received funding from the German Federal Min-istry of Education and Research via the Competence Net Stroke anda grant from the Center for Stroke Research Berlin. ME receives fur-ther support from the Volkswagen-Stiftung, Deutsche Forschungs-gemeinschaft and the European Union.

References

1. Heidrich J, Heuschmann PU, Kolominsky-Rabas P, et al. Variations in the use ofdiagnostic procedures after acute stroke in Europe: results from the BIOMED IIstudy of stroke care. Eur J Neurol 2007;14:255–61.

2. Ringleb P, Bousser M, Ford G, et al. Guidelines for management of ischaemicstroke and transient ischaemic attack. The European Stroke Organisation (ESO).Cerebrovasc Dis 2008;25:457–507.

3. Strandberg M, Marttila RJ, Helenius H, et al. Transoesophagealechocardiography should be considered in patients with ischaemic stroke ortransient ischaemic attack. Clin Physiol Funct Imaging 2008;28:156–60.

4. De Castro S, Papetti F, Di Angelantonio E, et al. Feasibility and clinical utility oftransesophageal echocardiography in the acute phase of cerebral ischemia. AM JCardiol 2010;106:1339–44.

5. Harloff A, Handke M, Reinhard M, et al. Therapeutic strategies afterexamination by transesophageal echocardiography in 503 patients withischemic stroke. Stroke 2006;37:859–64.

6. de Bruijn SF, Agema WR, Lammers GJ, et al. Transesophageal echocardiographyis superior to transthoracic echocardiography in management of patients of anyage with transient ischemic attack or stroke. Stroke 2006;37:2531–4.

7. Yaghoubi E, Nemati R, Aghasadeghi K, et al. The diagnostic efficiency oftransesophageal compared to transthoracic echocardiographic findings from405 patients with ischemic stroke. J Clin Neurosci 2011;18:1486–9.

8. Kapral MK, Silver FL. Preventive health care, 1999 update: 2. Echocardiographyfor the detection of a cardiac source of embolus in patients with stroke. CMAJ1999;161:989–96.

9. Strandberg M, Marttila RJ, Helenius H, et al. Transoesophagealechocardiography in selecting patients for anticoagulation after ischaemicstroke or transient ischaemic attack. J Neurol Neurosurg Psychiatry2002;73:29–33.

10. Hilberath JN, Oakes DA, Shernan SK, et al. Safety of transesophagealechocardiography. J Am Soc Echocardiogr 2010;23:1115–27. quiz 1220–1.

11. Wolber T, Maeder M, Atefy R, et al. Should routine echocardiography beperformed in all patients with stroke? J Stroke Cerebrovasc Dis 2007;16:1–7.

J. Herm et al. / Journal of Clinical Neuroscience 20 (2013) 554–559 559

12. Vitebskiy S, Fox K, Hoit BD. Routine transesophageal echocardiography for theevaluation of cerebral emboli in elderly patients. Echocardiography2005;22:770–4.

13. Camm AJ, Kirchhof P, Lip GYH, et al. Guidelines for the management of atrialfibrillation: the Task Force for the Management of Atrial Fibrillation of theEuropean Society of Cardiology (ESC). Eur Heart J 2010;31:2369–429.

14. Deutsche Gesellschaft für Neurologie (DGN). Primär- und Sekundärpräventionder zerebralen Ischämie. Die wichtigsten Empfehlungen auf einen Blick. In:Diener HC, NP, editors. Leitlinien für die Diagnostik und Therapie in derNeurologie. Georg Thieme Verlag; 2008.

15. Micheli S, Agnelli G, Caso V, et al. Clinical benefit of early anticoagulation incardioembolic stroke. Cerebrovasc Dis 2008;25:289–96.

16. Douglas PS, Garcia MJ, Haines DE, et al. ACCF/ASE/AHA/ASNC/HFSA/HRS/SCAI/SCCM/SCCT/SCMR 2011 appropriate use criteria for echocardiography. A reportof the American College of Cardiology Foundation Appropriate Use Criteria TaskForce, American Society of Echocardiography, American Heart Association,American Society of Nuclear Cardiology, Heart Failure Society of America, HeartRhythm Society, Society for Cardiovascular Angiography and Interventions,Society of Critical Care Medicine, Society of Cardiovascular ComputedTomography, Society for Cardiovascular Magnetic Resonance, AmericanCollege of Chest Physicians. J Am Soc Echocardiogr 2011;24:229–67.

17. Adams RJ, Albers G, Alberts MJ, et al. Update to the AHA/ASA recommendationsfor the prevention of stroke in patients with stroke and transient ischemicattack. Stroke 2008;39:1647–52.

18. Haeusler KG, Konieczny M, Endres M. Villringer A HP. Impact of anticoagulationbefore stroke on stroke severity and long term survival. Int J Stroke2012;7:544–50.

19. Hatano S. Experience from a multicentre stroke register: a preliminary report.Bull World Health Org 1976;54:541–53.

20. Gage BF, Waterman AD, Shannon W, et al. Validation of clinical classificationschemes for predicting stroke: results from the national registry of atrialfibrillation. JAMA 2001;285:2864–70.

21. van Swieten JC, Koudstaal PJ, Visser MC, et al. Interobserver agreement for theassessment of handicap in stroke patients. Stroke 1988;19:604–7.

22. Brott T, Adams Jr HP, Olinger CP, et al. Measurements of acute cerebralinfarction: a clinical examination scale. Stroke 1989;20:864.

23. Savitz SI, Lew R, Bluhmki E, et al. Shift analysis versus dichotomization of themodified Rankin scale outcome scores in the NINDS and ECASS-II trials. Stroke2007;38:3205–12.

24. Pepi M, Evangelista A, Nihoyannopoulos P, et al. Recommendations forechocardiography use in the diagnosis and management of cardiac sources ofembolism: European Association of Echocardiography (EAE) (a registeredbranch of the ESC). Eur J Echocardiogr 2010;11:461–76.

25. Kim YD, Park B, Cha MJ, et al. Stroke severity in concomitant cardiac sources ofembolism in patients with atrial fibrillation. J Neurol Sci 2010;298:23–7.

26. Han SW, Nam HS, Kim SH, et al. Frequency and significance of cardiac sourcesof embolism in the TOAST classification. Cerebrovasc Dis 2007;24:463–8.