Professional Documents
Culture Documents
Twelve-Lead Electrocardiogram and Mortality in Young Adults After Ischaemic Stroke
Twelve-Lead Electrocardiogram and Mortality in Young Adults After Ischaemic Stroke
Twelve-Lead Electrocardiogram and Mortality in Young Adults After Ischaemic Stroke
Abstract
Introduction: Ischaemic stroke at young age carries an increased risk for mortality in comparison to the general
population, but factors associated with mortality have been poorly studied. We studied the role of electrocardiogram
in mortality risk stratification in young stroke patients.
Patients and methods: The Helsinki Young Stroke Registry encompasses 1008 patients aged <50 years with ischaemic
stroke. We included 690 patients for this electrocardiogram substudy. Our endpoints were all-cause and cardiovascular
mortality. Cox regression models – adjusted for clinical and demographic characteristics – were used to identify the
electrocardiogram parameters associated with these endpoints.
Results: At a mean follow-up of 8.8 years, cumulative all-cause and cardiovascular mortality were 16.1 and 9.1%,
respectively. Factors associated with both endpoints included diabetes (type 1 for all-cause, type 2 for cardiovascular
mortality), heavy drinking, malignancy, as well as stroke severity and aetiology. Of the electrocardiogram parameters,
higher heart rate (hazard ratio 1.35 per 10/min, 95% confidence interval 1.21–1.49), a shorter P-wave (hazard ratio
0.78 per 10 ms decrement, 0.64–0.92) and longer QTc interval (1.09 per 10 ms, 1.03–1.16) were associated with
increased all-cause mortality. Only a higher heart rate (1.42 per 10/min, 1.24–1.60) was associated with death from
cardiovascular causes.
Conclusions: A higher heart rate during the subacute phase after stroke is associated with an elevated risk of all-cause
and cardiovascular mortality in young adults. A longer QTc interval is associated only with higher all-cause mortality.
P-wave characteristics and their possible association with mortality need further studies.
Keywords
Electrocardiogram, stroke, young stroke, prognosis, mortality
Date received: 22 September 2016; accepted: 27 November 2016
Introduction
1
Department of Cardiology, Heart and Lung Center, Helsinki University
Ischaemic stroke (IS) is known to decrease survival in Hospital, Helsinki, Finland
young adults.1,2 In this patient population, vascular dis- 2
Clinical Neurosciences, Neurology, University of Helsinki and
eases are the most common cause of excess mortality, Department of Neurology, Helsinki University Hospital, Helsinki,
while known factors increasing the risk for death Finland
3
Department of Clinical Physiology and Nuclear Medicine, HUS Medical
include more severe stroke, increasing age, higher
Imaging Center, Helsinki University Hospital and University of Helsinki,
CRP levels, malignancy, heart failure, myocardial Helsinki, Finland
infarction, heavy drinking, type 1 diabetes (T1D), dif- 4
Department of Neurology, Institute of Neuroscience and Physiology,
ferent stroke aetiologies, recurrent stroke, leukoaraiosis Sahlgrenska Academy, University of Gothenburg and Sahlgrenska
and accumulation of multiple risk factors.1,3–8 University Hospital, Gothenburg, Sweden
Since many electrocardiogram (ECG) findings are
Corresponding author:
related to increased mortality in the general population, Jani Pirinen, Department of Cardiology, Heart and Lung Center, Helsinki
we hypothesised that ECG could be useful in assessing University Hospital, Haartmaninkatu 4, FIN-00029 Helsinki, Finland.
the risk of mortality in young patients with IS. Email: jani.pirinen@helsinki.fi
78 European Stroke Journal 2(1)
Such findings in the general population have included size lesion was defined as a lesion of a cortical superficial
intraventricular conduction delay (IVCD), left ven- branch of the anterior, middle or posterior cerebral
tricular hypertrophy (LVH), T-wave inversions, artery. A large anterior lesion involved the complete ter-
P-terminal force (PTF), atrial fibrillation (AF), a wide ritory of the anterior or middle cerebral artery and a
QRS-T angle and bundle branch blocks (BBBs).9–16 In large posterior lesion involved the complete territory of
older stroke populations, pathological Q-waves, AF the posterior cerebral artery with border zone territories.
and a higher heart rate have been found as markers
of higher mortality.17–20 In contrast, a large study on
young stroke patients observed no association with AF
Follow-up data
and increased mortality.21 We obtained data on deaths and causes of death from
To our knowledge, risk factors for cardiovascular Statistics Finland. The reliability of this register is
death after stroke in young adults have not been sys- described elsewhere.25 Follow-up data were obtained
tematically studied and the value of ECG findings asso- until 31 December 2011.
ciated with increased mortality has neither been
studied. In this study, we assessed the risk factors for
cardiovascular death in particular, and whether ECG
Definition of endpoints
abnormalities are associated with long-term risk of The primary endpoint was death from any cause. Our
death after stroke in young adults. secondary endpoint was death from cardiovascular dis-
ease, defined as any of the three primary reasons
(underlying cause of death, intermediate cause of
Patients and methods death or immediate cause of death) being a diagnosis
from 390 to 459 in ICD-9 or from category I in ICD-10.
Study population Death due to index stroke was counted as both primary
The relevant authorities approved the study, and it was and secondary endpoint regardless of time from stroke
carried out at the Departments of Neurology and onset.
Cardiology, Helsinki University Hospital. This retro-
spective study investigated data from the Helsinki
ECG analysis
Young Stroke Registry, with 1008 consecutive patients
with first-ever IS aged 15–49 years and admitted to our JPi, blinded to clinical data at time of analysis, analysed
hospital between 1994 and 2007.22 the ECGs. In uncertain cases he consulted two senior
Patients were examined according to a standard cardiologists (ML and AA) for consensus.
protocol, involving a 12-lead ECG on admission and, We considered the continuous ECG parameters
in most cases, also during the stay at the hospital.22 To heart rate, P-wave duration, P-wave axis, PR interval,
exclude changes related to the acute phase of stroke and QRS complex duration, QRS frontal axis, T-wave fron-
obtaining an ECG close to baseline, we analysed ECGs tal axis, corrected QT interval and QRS-T angle. The
taken 1–14 days after the onset of stroke symptoms and dichotomised ECG parameters we considered were
at least one day after admission.23 IVCD, first-degree AV block, P-wave duration
<80 ms, PTF, LVH, pathological Q-waves, J-waves
(early repolarisation pattern), T-wave inversions,
Risk factors BBBs, prolonged QTc, rhythm, QRS duration
We included in our analyses the following demographic >110 ms and frontal QRS-T angle >110 . Detailed def-
factors: age and sex, and the following comorbidities: initions of ECG characteristics used in this registry are
obesity, hypertension, smoking, dyslipidaemia, known previously published.23
AF, cardiovascular disease, T1D, type 2 diabetes (T2D)
mellitus, malignancy and heavy drinking. Definitions of
Statistical analysis
the comorbidities are described in detail in the online-
only supplement (Table e1). Stroke characteristics We used the Cox proportional hazards model to inves-
analysed were NIH Stroke Scale (NIHSS) score on tigate the factors associated with the endpoints. First,
admission, stroke aetiology according to Trial of Org we performed a univariate analysis and determined
10172 in Acute Stroke Treatment classification (with which of the demographic factors, predefined risk fac-
modifications to better adapt the classification for tors and index stroke characteristics fell below signifi-
young patients; Supplementary Table e3), lesion size cance level of P < 0.10. For both endpoints, a unique
(ordinal scale ranging from 1 to 4) and lesion multipli- set of covariates was thus created for the multivariate
city (one or more lesions).24 A lesion was defined as analysis. Second, we tested all of the significant param-
small when it was measured as <1.5 cm. A medium eters in multivariate models, one model for each
Pirinen et al. 79
endpoint. The ECG parameters were then added – one Table 1. Causes of 111 deaths during the follow-up.
at a time, in order to avoid overcorrection – into these
N (%)
models, creating an adjusted model for each ECG par-
a
ameter. To correct for the performed multiple partly Underlying cause of death cardiovascular 55 (50)
correlated ECG parameters, we applied a principal Coronary artery disease or myocardial infarction 17 (15)
component analysis of all ECG parameters to define Index stroke 13 (12)
the amount of underlying uncorrelated components. New ischaemic or haemorrhagic stroke 15 (14)
The resulted number of components needed to explain Cardiomyopathy 7 (6)
95% of the overall variability in ECG parameters was
Thrombophlebitis of lower extremities 1 (1)
8, which we used as a correction factor in Bonferroni
Endocarditis 1 (1)
correction. P < 0.05 was considered statistically signifi-
cant. Kaplan–Meier survival plots were finally created Aortic valve stenosis 1 (1)
to graphically demonstrate the major findings. We per- Underlying cause of death other 56 (50)
formed all analyses on SPSS 22.0 for Windows (SPSS than cardiovascular
Inc., IBM, Armonk, USA). Malignancy 24 (22)
Alcoholic liver disease 3 (3)
Alcoholic pancreatitis 3 (3)
Results Infection 3 (3)
A total of 690 patients fulfilled our inclusion criteria Accident, suicide or poisoning 8 (7)
and were included in the present study. Of the original Complications of type 1 diabetes 8 (7)
1008 patients, four patients were excluded due to false Systemic lupus erythematosus 2 (2)
primary diagnosis (stroke mimic), 48 due to unknown Muscle diseases 2 (2)
exact stroke date, 25 were treated only as outpatients Chronic obstructive pulmonary disease 2 (2)
and 241 had no ECG of good quality obtained during
Duodenal ulcer 1 (1)
the required time window. Thirty-three patients (4.6%)
a
received thrombolysis. Brain imaging was performed by In addition to these, eight other patients had a cardiovascular intermedi-
either CT or MRI on all patients: CT was performed on ate or immediate of death, hence 63 patients in the cardiovascular cause
of death group.
662 patients (95.9%) and MRI on 452 patients (65.5%).
On 424 patients (61.4%), both imaging modalities were
used. Compared with the excluded patients, the
included patients had higher NIHSS score, longer cor- drinking, malignancy, stroke aetiology, NIHSS score
rected QT interval, wider QRS-T angle, more cardio- and lesion size. Factors associated with cardiovascular
vascular disease and a different distribution of stroke death were age, hypertension, cardiovascular disease at
aetiology (Supplementary Table e3). Mean follow-up baseline, T2D, AF, heavy drinking, malignancy, stroke
of the study cohort was 8.8 years (interquartile range aetiology, NIHSS score and lesion size (data not
6.3–12.1). Supplementary Table e2 shows baseline shown). The multivariate analyses on demographic
comorbidities and medication data. and clinical factors were constructed including these
parameters (Tables 2 and 3). Factors associated with
all-cause mortality in the multivariate analysis were
Mortality age, T1D, heavy drinking, malignancy, NIHSS score,
Of our patients 111 (16.1%) died during follow-up of lesion size and stroke aetiology. Factors associated with
whom 63 (9.1%) died due to cardiovascular causes. cardiovascular mortality in the multivariate analysis
Fifteen patients died during the first 30 days after the were cardiovascular disease at baseline, T2D, heavy
index stroke. The most frequent primary causes of drinking, malignancy, NIHSS score, lesion size and
death were recurrent stroke, malignancy and coronary stroke aetiology.
artery disease. The most common cardiovascular causes
of death were coronary artery disease, index stroke and
recurrent stroke (Table 1).
ECG findings associated with long-term mortality
In the Bonferroni-corrected multivariate analysis, heart
Association between demographic and clinical factors rate, QTc interval and P-wave duration were associated
with death from any cause (Table 4). Only heart rate
and long-term mortality
was statistically significantly related to cardiovascular
In the univariate analyses, factors associated with all- death (Table 5). Kaplan–Meier survival plots of heart
cause mortality were age, hypertension, smoking, car- rate tertiles are presented for all-cause mortality and
diovascular disease at baseline, AF, T1D, heavy cardiovascular mortality (Figures 1 and 4).
80 European Stroke Journal 2(1)
Table 2. Risk factors for death from any cause. Multivariate analysis with Cox regression of different
parameters and their hazard ratio: demographics, comorbidities and index stroke characteristics, only those
significant in univariable analysis included.
Table 3. Risk factors for death from cardiovascular cause. Multivariate analysis with Cox regression of different
parameters and their hazard ratio: demographics, comorbidities and index stroke characteristics, only those
significant in univariate analysis included.
Table 3. Continued.
Table 4. ECG parameters’ relation to death from any cause. Multivariate analysis with Cox regression, adjusted for age,
hypertension, smoking, type 1 diabetes, cardiovascular disease, known AF (except for rhythm), malignancy, heavy
drinking, stroke aetiology, NIHSS and lesion size.
Table 5. ECG parameters’ relation to death from cardiovascular disease. Multivariate analysis with Cox regression,
adjusted for age, hypertension, type 2 diabetes, cardiovascular disease, known AF (except for rhythm), malignancy, heavy
drinking, stroke aetiology, NIHSS and lesion size.
Figure 1. Kaplan–Meier plot of heart rate tertiles and all-cause mortality. Log rank P < 0.001.
Figure 2. Kaplan–Meier plot of QTc interval tertiles and all-cause mortality. Log rank P < 0.001.
A longer QTc interval is also associated with higher between longer QTc interval and cardiovascular mor-
mortality and cardiovascular mortality in the general tality, yet one would specifically think the causes of
population and even in older stroke patients,27–29 but death would be cardiovascular with this ECG
quite surprisingly we did not find an association abnormality.
84 European Stroke Journal 2(1)
Figure 3. Kaplan–Meier plot of P-wave duration tertiles and all-cause mortality. Log rank P ¼ 0.129.
Figure 4. Kaplan–Meier plot of heart rate tertiles and cardiovascular mortality. Log rank P < 0.001.
It is rather surprising that we did not find an asso- and cardiovascular) in the general population, and AF
ciation of any other ECG parameters with mortality, and pathological Q-waves have been associated with
since many of the parameters we analysed have earlier higher mortality also in older stroke populations.9–19
been associated with increased mortality (both all-cause As our patients’ causes of death were heterogeneous
Pirinen et al. 85
6. Naess H, Waje-Andreassen U and Nyland H. Risk factor 20. Erdur H, Scheitz JF, Grittner U, et al. Heart rate on
burden predicts long-term mortality in young patients admission independently predicts in-hospital mortality
with arterial cerebral infarction. Acta Neurol Scand in acute ischemic stroke patients. Int J Cardiol 2014;
2013; 127: 92–96. 176: 206–210.
7. Putaala J, Haapaniemi E, Kurkinen M, et al. Silent 21. Pezzini A, Grassi M, Lodigiani C, et al. Predictors of
brain infarcts, leukoaraiosis, and long-term prognosis in long-term recurrent vascular events after ischemic
young ischemic stroke patients. Neurology 2011; 76: stroke at young age: the Italian project on stroke in
1742–1749. young adults. Circulation 2014; 129: 1668–1676.
8. Putaala J, Haapaniemi E, Kaste M, et al. How does 22. Putaala J, Metso AJ, Metso TM, et al. Analysis of 1008
number of risk factors affect prognosis in young patients Consecutive patients aged 15 to 49 with first-ever ische-
with ischemic stroke? Stroke 2012; 43: 356–361. mic stroke: the Helsinki young stroke registry. Stroke
9. Aro AL, Anttonen O, Tikkanen JT, et al. Intraventricular 2009; 40: 1195–1203.
conduction delay in a standard 12-lead electrocardiogram 23. Pirinen J, Putaala J, Aro AL, et al. 12-lead resting ECG
as a predictor of mortality in the general population. Circ reveals high-risk sources of cardioembolism in young
Arrhythm Electrophysiol 2011; 4: 704–710. adult ischemic stroke. Int J Cardiol 2015; 198: 196–200.
10. Ishikawa J, Ishikawa S and Kario K. Levels of Cornell 24. Adams HP Jr, Bendixen BH, Kappelle LJ, et al.
voltage and Cornell product for predicting cardiovascular Classification of subtype of acute ischemic stroke.
and stroke mortality and morbidity in the general Definitions for use in a multicenter clinical trial.
Japanese population. Circ J 2014; 78: 465–475. TOAST. Trial of Org 10172 in Acute Stroke Treatment.
11. Aro AL, Anttonen O, Tikkanen JT, et al. Prevalence and Stroke 1993; 24: 35–41.
prognostic significance of t-wave inversions in right pre- 25. Official Statistics of Finland (OSF): Causes of death
cordial leads of a 12-lead electrocardiogram in the [e-publication], http://www.stat.fi/til/ksyyt/2013/ksyyt_
middle-aged subjects. Circulation 2012; 125: 2572–2577. 2013_2014-12-30_laa_001_en.html (2013, accessed 30
12. Eranti A, Aro AL, Kerola T, et al. Prevalence and prog- April 2016).
nostic significance of abnormal P terminal force in lead 26. Custodis F, Schirmer SH, Baumhäkel M, et al. Vascular
V1 of the ECG in general population. Circ Arrhythm pathophysiology in response to heart rate. J Am Coll
Electrophysiol 2014; 7: 1116–1121. Cardiol 2010; 56: 1973–1983.
13. Ohsawa M, Okayama A, Okamura T, et al. Mortality 27. Noseworthy PA, Peloso GM, Hwang S-J, et al. QT inter-
risk attributable to atrial fibrillation in middle-aged and val and long-term mortality risk in the Framingham
elderly people in the Japanese general population. Heart Study. Ann Noninvasive Electrocardiol 2012; 17:
Nineteen-Year Follow-up in NIPPON DATA80. Circ J 340–348.
2007; 71: 814–819. 28. Ishikawa J, Ishikawa S and Kario K. Relationships
14. Aro AL, Huikuri HV, Tikkanen JT, et al. QRS-T angle between the QTc interval and cardiovascular, stroke, or
as a predictor of sudden cardiac death in a middle-aged sudden cardiac mortality in the general Japanese popula-
general population. Europace 2012; 14: 872–876. tion. J Cardiol 2015; 65: 237–242.
15. Eriksson P, Wilhelmsen L and Rosengren A. 29. Wong KYK, Mac Walter RS, Fraser HW, et al. Long
Bundle-branch block in middle-aged men: risk of compli- QTc predicts future cardiac death in stroke survivors.
cations and death over 28 years. The primary prevention Heart 2003; 89: 377–381.
study in Göteborg, Sweden. Eur Heart J 2005; 26: 30. Soedamah-Muthu SS, Fuller JH, Mulnier HE, et al. All-
2300–2306. cause mortality rates in patients with type 1 diabetes mel-
16. Bussink BE, Holst AG, Jespersen L, et al. Right bundle litus compared with a non-diabetic population from the
branch block: prevalence, risk factors, and outcome in UK general practice research database, 1992–1999.
the general population: results from the Copenhagen Diabetologia 2006; 49: 660–666.
City Heart Study. Eur Heart J 2013; 34: 138–146. 31. Tancredi M, Rosengren A, Svensson AM, et al. Excess
17. Tanaka M, Nakayama Y, Maeda Y, et al. mortality among persons with type 2 diabetes. N Engl J
Electrocardiographic Q-waves as a predictor of mortality Med 2015; 373: 1720–1732.
in patients with cerebral infarction. Neurology 2004; 62: 32. Allemann S, Saner C, Zwahlen M, et al. Long-term car-
1818–1821. diovascular and non-cardiovascular mortality in women
18. Candelise L, Pinardi G and Morabito A. Mortality in and men with type 1 and type 2 diabetes mellitus: a 30-
acute stroke with atrial fibrillation. The Italian acute year follow-up in Switzerland. Swiss Med Wkly 2009;
stroke study group. Stroke 1991; 22: 169–174. 139: 576–583.
19. Kaarisalo MM, Immonen-Räihä P, Marttila RJ, et al. 33. Hillier TA and Pedula KL. Complications in young
Atrial fibrillation and stroke. Mortality and causes of adults with early-onset type 2 diabetes: losing the relative
death after the first acute ischemic stroke. Stroke 1997; protection of youth. Diabetes Care 2003; 26: 2999–3005.
28: 311–315.