www.elsevier.com/locate/jns

Journal of the Neurological Sci

Abnormalities on ECG and telemetry predict stroke outcome at 3 months

Hanne Christensena,*, Anders Fogh Christensenb, Gudrun Boysena

aDept. of Neurology, Bispebjerg Hospital, University of Copenhagen 2400 Copenhagen NV, DenmarkbDept. of Radiology, Rigshospitalet, University of Copenhagen, Denmark

Received 8 February 2005; accepted 16 March 2005

Available online 1 June 2005

Abstract

Background: ECG is a useful tool in monitoring vital functions in patients with acute stroke; however, fairly little evidence is available

concerning the prevalence and the prognostic impact of ECG findings in patients with acute cerebral infarction and acute intracerebral

haemorrhage (ICH).

Methods: This analysis was based on data from 692 patients with acute cerebral infarction, 155 patients with intracerebral haemorrhage

(ICH), and 223 patients with transient ischaemic attack (TIA), who were admitted to hospital within 6 h of symptom onset. A 12 lead ECG

was obtained on admission, and the patient was on telemetry for the first 12–24 h of hospitalisation.

Results: ECG abnormalities were observed in 60% of patients with cerebral infarction, 50% of patients with ICH, and 44% of patients with

TIA. In multivariate analyses 3-month mortality in patients with ischaemic stroke was predicted by atrial fibrillation OR 2.0 (95% CI 1.3–

3.1), atrio-ventricular block OR 1.9 (95% CI 1.2–3.9), ST-elevation OR (2.8, 95% CI 1.3–6.3), ST-depression OR 2.5 (95% CI 1.5–4.3),

and inverted T-waves OR 2.7 (95% CI 1.6–4.6). This was independent of stroke severity, pre-stroke disability and age. In patients with ICH,

sinus tachycardia OR 4.8 (95% CI 1.7–14.0), ST-depression OR 5.2 (95% CI 1.1–24.9), and inverted T-wave 5.2 (95% CI 1.2–22.5)

predicted poor outcome. None of the changes reached significance in patients with TIA. In patients with severe cerebral infarction or ICH,

heart rate did not decrease within the first 12 h after admission, which was the case in patients with mild to moderate stroke. Rapid heart rate

predicted 3-month mortality in multivariate testing OR 1.7 (95% CI 1.02–2.7).

Conclusions: ECG abnormalities are frequent in acute stroke and may predict 3-month mortality.

D 2005 Elsevier B.V. All rights reserved.

Keywords: Stroke; ECG

1. Introduction

Abnormal ECGs are frequently recorded in patients with

acute stroke. Little evidence, however, exists concerning the

prevalence of ECG changes and their prognostic impact for

patients with cerebral infarction and intracerebral haemor-

rhage [1].

The aim of this retrospective analysis was to describe the

prevalence of common ECG abnormalities on admission 12-

lead ECG and on telemetry in a large patient population

admitted to the hospital within 6 h of the onset of a focal

neurological deficit, and to evaluate the impact of these

findings on functional outcome and mortality 3 months

0022-510X/$ - see front matter D 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.jns.2005.03.039

* Corresponding author. Tel.: +45 35 31 61 14; fax: 45 35 31 25 95.

E-mail address: hc04@bbh.hosp.dk (H. Christensen).

later. Patients with and without history of cardiac disease

were included.

2. Patients and methods

The study population includes 1192 consecutive patients

with cerebrovascular disease who were admitted to an acute

stroke unit within 6 h of symptom onset. Data was

registered as previously described [2]. In short, diagnosis

was based on clinical and CT-scan findings on admission.

Clinical data were collected prospectively.

Stroke severity was assessed on admission by the

Scandinavian Stroke Scale (SSS) [3] in which no neuro-

logical deficit equals 58 points, and 0 points represent a

comatose paralysed state. Outcome was evaluated by the

ences 234 (2005) 99 – 103

Table 1

Clinical data in 1070 patients with acute cerebrovascular disease

Diagnose ACI

(N =692)

ICH

(N =155)

TIA

(N =223)

Age (years) 76 (67–82) 74 (62–81) 70 (58–79)

Delay from symptom

onset to admission

and 12 lead ECG (h)

2 (1.1–3.5) 2 (1–3.1) 2.2 (1.1–3.5)

Scandinavian Stroke

Scale on admission

39 (22–49) 25 (10–35) 55 (50–58)

Modified Rankin Scale

(3 months after stroke)

3 (2–4) 4 (2–6) 1 (0–2)

3 months fatality rate 16.5% (114) 38.1% (59) 4% (9)

Recurrent stroke

<=12 months

9% (60) 3.9% (6) 8.9% (19)

History of MI 13.3% (78) 7.1% (11) 8.1% (18)

History of angina 14.6% (101) 5.2% (8) 13% (29)

History of congestive

heart failure

14.2% (98) 5.8% (9) 7.6% (17)

History of hypertension 35.4% (245) 38.7% (60) 35% (78)

Median and interquartile range or percent and number are presented.

ACI: Acute cerebral infarction.

ICH: Intracerebral haemorrhage.

TIA: Transient ischaemic attack.

H. Christensen et al. / Journal of the Neurological Sciences 234 (2005) 99–103100

modified Rankin Scale (mRS) [4] including death within 3

months after stroke onset. Severe stroke was defined as

SSS�25 and mild to moderate stroke as SSS>25.

Routine hospital procedure was followed for obtaining

ECGs. This includes the nurses obtaining a 12 lead ECG on

admission, followed by 12–24 h of telemetry (Teleguard

3200, Danica Biomedical A/S, Denmark). Abnormal

rhythms on ECG-telemetry were automatically printed and

placed in the patients’ chart. Normal telemetry tracings were

not stored in the chart.

12 lead ECGs were available for analysis in 1070 of the

1192 patients. 12 lead ECG data was either missing or

unreadable (due to resolution of the hospital scanner) in the

other 122 patients. The 122 patients with no admission ECG

were excluded from the analysis. This may introduce some

bias, as these excluded patients had more severe neuro-

logical deficits (SSS admission 31 vs. 41 in 1070 included

patients). There were no differences in age and pre-stroke

mRS. 675 of the 1070 included patients had abnormalities

on telemetry, which were included in the analysis.

When ECG abnormalities occurred in the acute stroke

unit, they were assessed by the neurologist in charge or by a

consultant cardiologist as appropriate, and treated according

to usual practice.

ECG’s were analysed retrospectively by one observer,

AFC, who was blinded to all clinical data. The following

changes were recorded when abnormal, according to general

guidelines [5]: atrial fibrillation, atrial flutter, sinus tachy-

cardia: HR>120, sinus bradycardia: HR<45, atrio-ventric-

ular block (all degrees), ventricular tachycardia>5 s, ectopic

beats, ST-elevation, ST-depression, isoform T-wave,

inverted T-wave, U-wave, and QTc>0.44 s. QTc was

calculated as QT/�R�R.

Cerebral infarction was diagnosed in 692 patients, ICH in

155 patients, and TIA in 223 patients. Statistical analysis

was performed using SPSS for Windows 9.0 (SPSS Inc.

Chicago, USA) and included descriptive statistics and

multiple logistic regression analysis, which were performed

as enter analyses. Outcome was death at 3 months. Beside

the various investigated ECG-abnormalities, we also

included SSS on admission, pre-stroke mRS, and age.

These possibly confounding variables were selected as

possible confounders based on literature and our previous

findings. The significance level was set at 0.05. Student’s t-

test was used in comparing continuous variable, and chi-

square-test was used in assessing the independence of row

and column variables in a cross tab.

The Scientific–Ethics committee of the Copenhagen and

Frederiksberg area reviewed the study protocol and had no

objections to its conduct.

3. Results

ECG-abnormalities of some kind were found in 55.3%

(592) of this acute stroke unit population. Patients’

characteristics are summarised in Table 1; 36.7% of all

patients had a history of hypertension, 10.1% a history of

acute myocardial infarction, 12.7% a history of angina,

and 12.0% had history of congestive heart failure.

Increasing stroke severity augmented the frequency of

ECG-abnormalities; median SSS on admission was 37 in

patients with ECG-abnormalities in comparison to 44 in

patients without ECG-abnormalities, Mann–Whitney’s test

p <0.001.

4. Patients with cerebral infarction (N =692)

An abnormal ECG was observed in 416 patients

(60.1%). The most frequent findings were sinus tachy-

cardia (24.3%), atrio-ventricular block (21.4%), and

ectopic beats (30.9%), Table 2. Repolarization changes

were observed in 32.5% of patients. Some ECG

abnormalities were related to severity of neurological

deficits; atrial fibrillation median SSS 28 vs. 39, p =0.12,

prolonged QTc median SSS 26 vs. 39, p =0.007, atrio-

ventricular block median SSS 34.5 vs. 39 p =0.038, ST-

depression median SSS 33.5 vs. 39 p =0.016, and ST-

elevation median SSS 30 vs. 39 p=0.046. In multivariate

logistic regression analysis, atrial fibrillation, OR 2.0

(95% CI 1.3–3.1), A-V block OR 1.9 (95% CI 1.2–

3.9), ST-elevation OR 2.8 (95% CI 1.3–6.3), ST-

depression OR 2.5 (95% CI 1.5–4.3), and inverted T-

wave OR 2.7 (95% CI 1.6–4.6) predicted 3-month

mortality independent of pre-stroke handicap, stroke

severity, and age. Ectopic beats >10%, atrial flutter,

sinus bradycardia, isoform T-wave, and ventricular and

sinus tachycardia did not affect prognosis in multivariate

Table 2

The frequencies of ECG abnormalities in 1070 patients with acute

cerebrovascular disease

Diagnosis ACI %

(N =692)

ICH %

(N =155)

TIA %

(N =223)

v2 test

ECG-changes, any. 60.1 (416) 49.7 (77) 44.4 (99) <0.001

Atrial fibrillation 11.3 (78) 1.9 (3) 7.2 (16) 0.003

Atrial flutter 2 (14) 0.6 (1) 1.3 (3) 0.508

Sinus tachycardia 24.3 (168) 19.4 (30) 14.3 (32) 0.001

Sinus bradycardia 13.7 (95) 9.7 (15) 12.2 (27) 0.310

A-V block, any 21.4 (148) 9.7 (15) 14.8 (33) 0.011

A-V block, 3rd degree 11.7 (81) 8.4 (13) 9 (20) 0.005

VT>5 s 1.2 (8) 0 0.4 (1) 0.209

Ectopic beats>10% 30.9 (214) 25.8 (40) 18.4 (41) 0.005

ST-elevation 5.3 (37) 5.2 (8) 2.2 (5) 0.067

QTc>0.44 s 6.9 (48) 5.8 (9) 4 (9) 0.094

ST-depression 16 (111) 7.1 (11) 9.0 (20) 0.001

Inverted T-wave 18.2 (126) 8.4 (13) 11.2 (25) 0.002

Isoform T-wave 5.9 (41) 6.5 (10) 6.3 (14) 0.909

U-wave 0.4 (3) 0 1.3 (3) 0.702

Any repolarisation

change

32.5 (225) 23.2 (36) 23.7 (53) 0.003

Percent and numbers are presented together.

ACI: Acute cerebral infarction.

ICH: Intracerebral haemorrhage.

TIA: Transient ischaemic attack.

A-V: Atrioventricular.

H. Christensen et al. / Journal of the Neurological Sciences 234 (2005) 99–103 101

testing. Sudden death did not occur during the observa-

tion period.

5. Patients with intracerebral haemorrhage (N =155)

Abnormal ECG was observed in 77 patients (49.7%).

The most frequent abnormal ECG findings were sinus

tachycardia (19.4%), and ectopic beats>10% (25.8%),

Table 2. Repolarization changes were observed in 23.2%

of patients. No ECG-abnormalities were significantly

related to neurological deficit. In multivariate logistic

regression analysis sinus tachycardia OR 4.8 (95% CI

1.7–14.0), ST-depression OR 5.2 (CI 95% 1.1–24.9), and

inverted T-wave OR 5.2 (95% CI 1.2–22.5) predicted

mortality at 3 months, independent of pre-stroke handicap,

stroke severity, or age. The presence of atrial fibrillation,

atrial flutter, sinus bradycardia, atrio-ventrivcular block, ST

elevation or depression, inverted T-wave, isoform T-wave,

or more than 10% ectopic beats did not affect prognosis in

multivariate testing. No patients suffered ventricular tachy-

cardia or sudden death during the observation period.

6. Patients with TIA (N =223)

ECG changes were found in 44.4% of patients with TIA,

ectopic beats >10% and atrio-ventricular block being the

most frequent findings, Table 2. Repolarization abnormal-

ities were observed in 23.8% of patients. None of the

analysed ECG-findings predicted mortality at 3 months in

multivariate analysis. We also analysed the influence of the

combined presence of ectopic beats, ST-depression and

inverted/isoform T-wave, and atrial fibrillation and block on

mortality at 3 months after the vascular insult but found no

significant results.

7. Comparison between diagnoses

The frequency of ECG-abnormalities differed signifi-

cantly between diagnoses, the frequency being highest in

patients with cerebral infarction. This was primarily due

to higher frequencies of atrial fibrillation, atrio-ventricular

block, ST-depression, and T-wave inversion. However,

rates of sinus tachycardia, ectopic beats, and ST-elevation

were higher in both patients with cerebral infarction and

intracerebral haemorrhage than in patients with TIA,

Table 2.

8. Heart rate in patients with stroke (N =847)

Heart rate was significantly higher in severe stroke than

in mild to moderate stroke and followed a different time

course: in mild to moderate stroke, heart rate rapidly

declined after admission, whereas a slow decline at a higher

heart rate was seen in severe stroke, Fig. 1. From 6 h to 32 h

after admission, the pulse rate was significantly higher in

severe stroke patients compared to mild to moderate stroke

patients at all times tested by t-tests. We tested the relation

between heart rate 6–14 h after stroke onset and outcome as

to death or dependency 3 months after stroke in a multi-

variate logistic regression model also including pre-stroke

mRS, age, stroke severity, and body temperature measured

at the same time points as heart rate. We found a consistent

relation within this time span, and a risk increase of 1.2–1.3

for death 3 months after stroke with each increase in heart

rate of 10/min, Table 3. At pulse rates higher than the

median=>76 bpm 12 h after admission the risk of death

within 3 months after stroke onset increased, OR 1.7 (95%

CI 1.02–2.7) in a multivariate logistic regression analysis

also including pre-stroke mRS, age, stroke severity, and

body temperature measured at the same time points as heart

rate.

9. Discussion

In this population from an acute stroke unit that included

patients with and without previously known cardiac disease

ECG-abnormalities were observed in 55.3% of patients. In

patients with ischaemic stroke, atrial fibrillation, atrio-

ventricular block, ST-elevation and depression, and inverted

T-waves predicted mortality at 3 months in multivariate

testing. ECG abnormalities may relate to the aetiology of the

patient’s vascular event [6]. Some [7] but not all inves-

Hours after hospital admission

48 hours24 hours12 hoursAdmission

95%

CI

Hea

rt r

ate

100

90

80

70

Hours after hospital admission

48 hours24 hours12 hoursAdmission

95%

CI

Hea

rt r

ate

100

90

80

70

1a 1b

Fig. 1. a) Represents the time course of heart rate in patients with severe stroke (N =279) in the first 48 h after admission (within 6 h of stroke onset). b)

Represents the time course of heart rate in patients with mild to moderate stroke (N =568) in the first 48 h after hospital admission (within 6 h of stroke onset).

Heart rates are significantly different in multiple t-tests from 6 h after admission to 32 h after admission.

H. Christensen et al. / Journal of the Neurological Sciences 234 (2005) 99–103102

tigators [8,9] have suggested that cardiac arrhythmias have

little influence on subsequent recovery. In this large-scale

study we confirmed that specific ECG-changes predict poor

outcome.

In patients with haemorrhagic stroke, sinus tachycardia,

ST-depression, and inverted T-waves predicted increased

mortality at 3 months in multivariate testing. Atrial

fibrillation and atrio-ventricular block may reflect co-

existing cardiac conditions, while ST-segment changes most

likely are cerebrogenic cardiac effects. The time course of

heart rate changes was different in severe stroke from that in

mild to moderate stroke; this may be associated with the

onset of the adrenocorticoid stress response to stroke, as this

response is the determining factor of heart rate in the resting,

non febrile person [10] and this would fit with the timing of

the tachycardia.

It appears from this study that changes resulting from

manifest cardiac disease (e.g. atrial fibrillation, heart block)

were mainly found in patients with ischaemic lesions, while

ST-segment changes resulting from the stroke were frequent

in both patients with ischaemic and haemorrhagic stroke.

ECG-abnormalities related to stroke severity in patients with

Table 3

Multivariate logistic regression analysis of the effect of an increase in heart

rate 6–14 h after hospital admission of 10 beats/min on 3 months mortality

independent of age, Scandinavian Stroke Scale, pre-stroke modified Rankin

Scale, and body temperature

Time after admission (h) OR 95% CI Significance

6 1.2 1.05–1.3 0.0091

8 1.2 1.1–1.3 0.0039

10 1.3 1.1–1.4 0.0003

12 1.2 1.1–1.4 0.0071

14 1.3 1.1–1.4 0.0010

Body temperature and heart rate were measured simultaneously (N =847).

OR: odds ratio.

95% CI: 95% confidence intervals.

ischaemic but not haemorrhagic stroke. The small sample

size of the ICH patient compared to large numbers of

patients with ischemic stroke cannot be excluded as an

explanation for the differences.

The strength of the present study is that it was based on a

large patient population that is representative of the stroke

population in the country of the authors, and that the

frequencies of a large number of significant ECG changes

were investigated.

A limitation of the study is that the results may be

affected by selection bias as patients without retrievable

ECG’s tended to have worse outcome in comparison to

patients with ECG’s. However, this would most likely result

in an underestimation of both the frequencies and the impact

of ECG abnormalities on outcome.

Another weakness of the study is that follow up ECGs

were not obtained for comparison. Analysis of cerebral

lesion location has not been done in this stroke population

as CT-scan was performed only once and before the full

infarct size was demarcated. It is therefore not possible to

relate ECG-findings to specific structures such as the insula.

Cardiac monitoring in acute cerebral infarction is useful

in detecting paroxysmal atrial fibrillation and acute MI

without chest pain, and the occurrence of ventricular

tachycardia. One pilot study in an acute stroke unit setting

has indicated that intensive monitoring, including ECG

surveillance, improves outcome as to death or dependency

[11]. However, whether in fact ECG-monitoring contributes

to improve outcome in acute stroke patients is less obvious,

and our study was not designed to evaluate this question.

In our patients with TIA, none of the ECG changes had

prognostic significance, whereas atrial fibrillation was

predictive of new cerebro-vascular events [12,13]. In

patients with stroke the reported frequencies of ECG-

abnormalities vary from 50% to 92% and in controls from

22% to 65% [7,14–17]. Ectopic beats, atrial fibrillation,

prolonged QTc, atrio-ventricular block, ST-depression, and

H. Christensen et al. / Journal of the Neurological Sciences 234 (2005) 99–103 103

T-wave changes are reported more frequently in patients

with acute stroke. A causal relationship between the acute

brain lesion and the ECG abnormalities is supported by the

decreasing frequency of abnormalities in the weeks and

months following stroke [18–20]. The rates of ECG-

abnormalities that we found were low in comparison to

some studies [14,16,21–23] and comparable to others

[7,8,24]. The relatively lower frequency of ECG abnormal-

ities (55%) could be due to our patients having their 12 lead

ECG’s recorded and ECG monitoring started within 6 h of

stroke onset, before effects from brain swelling are fully

developed. The frequencies might have been higher, if the

observation period had started later. Another possible

explanation remains that the frequency of ECG findings

was higher in the patients that were excluded from analysis

due to lack of retrievable ECG recordings. Our reported

frequencies are therefore to be regarded as minimum

figures, however, if we assume that all the excluded 122

patients had abnormal ECGs the total frequency of

abnormal ECG’s would increase from 55% to 60%, only.

In conclusion, we found that ECG abnormalities are

frequent in acute stroke and often reflect cardiac morbidity.

Some ECG abnormalities and increasing heart rate predict

poor recovery.

Funding

Augustinusfonden

Dagmar Marshalls Fond

Else og Mogens Wedel-Wedelsborgs Fond

Foundatation for Research in Neurology

Ludvig; Sara Elsass’ Fond

Aase og Ejnar Danielsens Fond

Acknowledgements

We are grateful to Dr. Bjarne Sigurd, MD, DMSc,

Department of Cardiology, Bispebjerg Hospital, and Dr.

Nancy Futrell, Intermountain Stroke Center, Salt Lake City,

USA, for kindly reviewing the manuscript and to secretary

Inger Hedegaard for retrieval of ECG’s.

No conflicts of interest.

References

[1] Khechinashvili G, Asplund K. Electrocardiographic changes in

patients with acute stroke: a systematic review. Cerebrovasc Dis

2002;14:67–76.

[2] Christensen H, Meden P, Overgaard K, Boysen G. The course of blood

pressure in acute stroke is related to the severity of the neurological

deficits. Acta Neurol Scand 2002;106:142–7.

[3] Lindenstrøm E, Boysen G, Christiansen LW, a Rogvi Hansen B,

Nielsen PW. Reliability of Scandinavian neurological stroke scale.

Cerebrovasc Dis 1991;1:103–7.

[4] van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJA, van Gijn J.

Interobserver agreement for the assessment of handicap in stroke

patients. Stroke 1988;19:604–7.

[5] Sandøe E, Sigurd B. Arrhythmia: a clinical electrocardiographic guide.

Verlag’ Bingen Publishing Partners; 1991.

[6] McHenry Jr LC, Toole JF, Miller HS. Long-term EKG monitoring in

patients with cerebrovascular insufficiency. Stroke 1976;7:264–9.

[7] Norris JW, Froggatt GM, Hachinski VC. Cardiac arrhythmias in acute

stroke. Stroke 1978;9:392–6.

[8] Lavy S, Yaar I, Melamed E, Stern S. The effect of acute stroke on

cardiac functions as observed in an intensive stroke care unit. Stroke

1974;5:775–80.

[9] Miah K, von Arbin M, Britton M, de Faire U, Helmers C, Maasing R.

Prognosis in acute stroke with special reference to some cardiac

factors. J Chronic Dis 1983;36:279–88.

[10] Desborough JP. The stress response to trauma and surgery. Br J

Anaesth 2000;85(1):109–17.

[11] Sulter G, Elting JW, Langedijk M, Maurits NM, De Keyser J.

Admitting acute ischemic stroke patients to a stroke care monitoring

unit versus conventional stroke unit. A randomized pilot study. Stroke

2003;34:101–4.

[12] Elkins JS, Sidney S, Gress DR, Go AS, Bernstein AL, Johnston SC.

Electrocardiographic findings predict short-term cardiac morbidity

after transient ischaemic attack. Arch Neurol 2002;59:1437–41.

[13] Fujishima M, Tanaka K, Omau T. Electrocardiographic changes in

cerebral transient ischemic attack. Angiology 1973;24:310–5.

[14] Goldstein DS. The electrocardiogram in stroke: relationship to

pathophysiological type and comparison with prior tracings. Stroke

1979;10:253–9.

[15] Myers MG, Norris JW, Hachinski VC, Weingert ME, Sole MJ.

Cardiac sequelae of acute stroke. Stroke 1982;13:838–42.

[16] Dimant J, Grob D. Electrocardiographic changes and myocardial

damage in patients with acute cerebrovascular accidents. Stroke 1977;

8:448–55.

[17] Lavy S, Stern S, Herishianu Y, Carmon A. Electrocardiographic

changes in ischaemic stroke. J Neurol Sci 1968;7:409–15.

[18] Orlandi G, Fanucchi S, Strata G, Pataleo L, Landucci Pellegrini L,

Prontera C, et al. Transient autonomic nervous system dysfunction

during hyperacute stroke. Acta Scand Neurol 2000;102:317–21.

[19] Tonet JL, Frank R, Ducardonnet A, Fillette F, Fontaine G, Komadja

M, et al. L’enregistrement de Holter dans les accidents ischemiques

cerebraux. Nouvelle Presse Med 1981;10:2491–4.

[20] Babalis D, Maisonblanche P, Leclercq JF, Coumel Ph. Interet d’un

electrocardiogramme de longue duree chez les patients ayant eu un

accident cerebral ischemique. Arch Mal Cæur 1984;77:100–5.

[21] Britton M, de Faire U, Helmers C, Miah K, Ryding C, Wester PO.

Arrhythmias in patients with acute cerebrovascular disease. Acta Med

Scand 1979;205:425–8.

[22] Lindgren A, Wohlfart B, Pahlm O, Johansson BB. Electrocardio-

graphic changes in stroke patients without primary heart disease. Clin

Physiol 1994;14:223–31.

[23] McDermott MM, Lefevre F, Arron M, Martin GJ, Biller J. ST segment

depression detected by continuous electrocardiography in patients

with acute ischemic stroke or transient ischemic attack. Stroke 1994;

25:1820–4.

[24] Kreus KE, Kemila SJ, Takala JK. Electrocardiographic changes in

cerebrovascular accidents. Acta Med Scand 1969;185:327–34.