ESJ Comment: Mortality After Young Stroke: Can we predict it from the simple ECG?

Comment Author: Alastair Webb, Associate Clinical Fellow, Centre for Prevention of Stroke and Dementia, Oxford

Original article: Pirinen et al Twelve-lead electrocardiogram and mortality in young adults after ischaemic stroke European Stroke Journal Vol 2, Issue 1, pp. 77 – 86 First published date: December-14-2016

Original article available here.

Stroke incidence in younger patients is increasing, despite an overall reduction in stroke incidence.1 This is associated with significant working years lost to disability and death. However, our understanding of the reason for this increase is unclear and prediction of the outcome of stroke in this important group is poor, with common prognostic scales being dominated by outcomes in older patients.2 In a recent study from the European Stroke Journal, Pirinen and colleagues use the large Helsinki Young Stroke Registry to determine whether the simple ECG can help us to predict the risk of death in these patients.3

A number of ECG parameters have been shown to predict the risk of death in the general population, including heart rate, AF, interventricular conduction delay and t-wave inversion,4 but these have not been systematically applied to young stroke patients. As the large majority of patients with stroke undergo a simple ECG, this is an attractive tool for risk prediction. If abnormalities on the ECG identify high risk patients, and particularly if these abnormalities identified treatable pathophysiological processes, then this could represent both a predictive tool and a method to reduce fatalities in this important group.

In 690/1008 patients in the Helsinki Young Stroke Registry under 50 years old, 16.1% had died after nearly 9 years of follow-up with 9.1% dying from cardiovascular causes. The authors used sequentially more adjusted cox regression models to determine the relationship between multiple potentially predictive ECG parameters and death, independent of stroke aetiology, demographic factors and stroke severity. As expected, age, diabetes, stroke severity, heavy drinking and stroke aetiology were predictive of death, whilst heart rate, a short p-wave duration and longer corrected QT interval also predicted all cause death, although the authors concluded that with a relatively small number of events, short p-wave duration was likely a chance finding. Ultimately, only heart rate independently predicted the risk of cardiovascular death (HR 1.35 per 10 beats / minute, 95% CI 1.21-1.49).

This study benefits from a large population of young patients with stroke, and a high mortality for the assessed age group. The authors have used reasonable statistical methods to limit confounding by demographic and stroke-related indices through sequentially adjusted Cox models and used a relatively sophisticated method to correct for multiple co-varying ECG parameters by applying a simple Bonferroni correction for underlying number of independent components in a principal components analysis. Although further steps to model significant ECG parameters together to exclude overestimation of the significance of individual parameters due to residual co-variation would be ideal, we can nonetheless be reasonably confident that heart rate at least is a reliable predictor of cardiovascular mortality in young patients with ischaemic stroke.

Perhaps the most surprising finding of this study is the limited number of factors that predicted cardiovascular mortality. Even atrial fibrillation was not predictive, despite a cardioembolic aetiology predicting the risk of death. This may reflect over-correction of ECG parameters by both the Bonferroni correction and by adjustment for demographic and stroke-related indices. For example, by adjusting ECG evidence of AF for ‘known AF’ we may fail to recognise the importance of AF or related characteristics on the single ECG because a single ECG will be less sensitive than the dichotomous classification of ‘any AF’ for a specific patient. Similarly, some ECG characteristics may partially mediate the relationship between heavy drinking and death through predicting alcohol related arrhythmia or cardiomyopathy, but may be lost through the lower sensitivity of the ECG after adjustment. There is therefore a risk of not appreciating potentially important pathophysiological mechanisms through this form of analysis. Furthermore, by definition, this is a heterogeneous population of patients with a complex combination of strokes due either to classic aetiological factors or to rare aetiologies. It is likely that ECG parameters would have differential importance dependent upon the aetiology of the stroke and due to the stroke location (e.g. insular cortex) or stroke severity. Such relationships would be best identified by stratification of the population. Finally, the study is unable to address whether an increased heart rate, or prolonged QTc interval, is specific to a population with stroke, or whether we are simply identifying indices that are relevant to the entire population regardless of whether they have had a stroke or not. If so, heart rate would not represent a stroke-related modifiable risk factor.

Nonetheless, this important study has many strengths due to the large number of young stroke patients included, and has identified a novel predictive measure in this population that could represent potentially modifiable pathophysiological processes to improve outcome. The pathophysiological mechanism underlying the predictive value of an increased heart rate in the general population is debatable, including suggestions that it is a marker of physical fitness, cardiovascular reserve, frailty, sympathetic over activity, end-organ damage or related to an increased risk of cardiac arrhythmia.5 However, the mechanism may yet be modifiable and elements are likely to be stroke dependent, as indicated by the predictive value of baroreceptor sensitivity for mortality in stroke patients.6 This study should therefore prompt further research to determine why patients with an increased heart rate after stroke are at an increased risk of death, whether heart rate is increased in stroke patients compared to similar patients with similar risk profiles without stroke and whether steps to ameliorate the factors underlying an increased heart rate also reduce the risk of future mortality, whether by pharmacological modification, a focus on cardiovascular fitness or by increased monitoring for associated complications such as the risk of AF. Therefore, although further work is required, this is an important step to improve our care of younger patients with stroke.


  1. Krishnamurthi RV, Moran AE, Feigin VL, Barker-Collo S, Norrving B, Mensah GA, Taylor S, Naghavi M, Forouzanfar MH, Nguyen G, Johnson CO, Vos T, Murray CJ, Roth GA and Group GBDSPE. Stroke Prevalence, Mortality and Disability-Adjusted Life Years in Adults Aged 20-64 Years in 1990-2013: Data from the Global Burden of Disease 2013 Study. Neuroepidemiology. 2015;45:190-202.
  2. Ntaios G, Gioulekas F, Papavasileiou V, Strbian D and Michel P. ASTRAL, DRAGON and SEDAN scores predict stroke outcome more accurately than physicians. Eur J Neurol. 2016;23:1651-1657.
  3. Pirinen J PJ, Aarnio K, Aro AL, Mustanoja S, Sinisalo J, Kaste M, Haapaniemi E, Tatlisumak T, Lehto M. Twleve-lead electrocardiogram and mortality in young adults after ischaemic stroke. European Stroke Journal. 2016;2:9.
  4. Aro AL, Anttonen O, Tikkanen JT, Junttila MJ, Kerola T, Rissanen HA, Reunanen A and Huikuri HV. Intraventricular conduction delay in a standard 12-lead electrocardiogram as a predictor of mortality in the general population. Circ Arrhythm Electrophysiol. 2011;4:704-10.
  5. Custodis F, Schirmer SH, Baumhakel M, Heusch G, Bohm M and Laufs U. Vascular pathophysiology in response to increased heart rate. J Am Coll Cardiol. 2010;56:1973-83.
  6. Robinson TG, Dawson SL, Eames PJ, Panerai RB and Potter JF. Cardiac baroreceptor sensitivity predicts long-term outcome after acute ischemic stroke. Stroke. 2003;34:705-12.