QT dispersion (QTD), the difference between the maximum and minimum QT interval on the 12-lead electrocardiogram (ECG), is a marker of heterogeneity of ventricular repolarization.¹ Previous studies have shown increased QTD to be a predictor of adverse outcomes in various cardiac disease states. Increased QTD has been found to be associated with cardiac arrhythmias and sudden cardiac death in patients with myocardial infarction, left ventricular hypertrophy, congestive heart failure, coronary artery disease, diabetes and end-stage renal disease.^2–10

Acute brain injury has been shown to result in altered autonomic regulation of the cardiovascular system.^1–13 Whether QTD measurement is influenced by acute brain injury remains unknown. Accordingly, the objective of this study was to determine the clinical significance of QTD measurement in patients hospitalized with acute cerebrovascular accidents (CVA) and transient ischemic attacks (TIA).

Methods
We retrospectively studied 140 consecutive patients (age, 72 ± 13 years) who were admitted to Winthrop University Hospital with acute neurological events between January 1, 1998 and April 30, 1998. Neurological events were defined as neurological symptoms, such as visual or speech disturbances, paresis or paralysis, loss of sensation, gait or equilibrium disturbance. The patients were excluded from the study in the case of altered mental status caused by metabolic disturbances, seizures or presence of brain masses.

Neurological evaluation
A complete neurological evaluation was performed on all patients by an experienced neurologist. Three separate functional scales were used to re-evaluate patients on discharge: the National Institute of Health Stroke Scale (NIHSS), the Barthel Index and the Modified Rankin Scale.^14–16 Clinical data were collected continuously during and after hospitalization and entered into the stroke database using standard definitions for each variable. The following clinical and demographic variables were analyzed: age, gender, diabetes, hypertension, hypercholesterolemia, current smoking, prior CVA or TIA, history of congestive heart failure, history of coronary disease, recent cardiac surgery (within 1 month), and carotid disease (defined as >= 40% stenosis in one or both carotid arteries by carotid Doppler imaging). The primary clinical endpoints analyzed in the study were mortality during hospitalization and the degree of neurological impairment determined by the NIHSS, Barthel Index and Modified Rankin scales upon discharge.

Electrocardiographic data
The ECG data were retrospectively collected. The QTD measurements were calculated from the admission ECG, which was routinely obtained with a three-channel Burdick E310 electrocardiograph (Burdick Corporation, Milton, Wisconsin). The ECG tracings were recorded during quiet respiration at a paper speed of 25 mm/second and a manual calibration of 1 mV equal to 10 mm. The first ECGs available on presentation were analyzed manually by observers blinded to clinical data. The QT interval was measured to the nearest 10 msec in each of the presenting ECGs from the onset of the QRS complex to the end of the T-wave. In the presence of a U-wave, the end of the T-wave was determined using the tangent method on the descending limb of the T-wave and determined at which point it intersected the baseline.¹ QTD was calculated as the difference between the maximum and minimal QT intervals. Heart rate and maximum QRS duration were also recorded. Interobserver and intraobserver variability of QTD measurements were evaluated in 10 randomly selected patients. For interobserver comparisons, a second investigator was blinded to the results of the first investigator. For intraobserver variability, an investigator was blinded to the measurements of his first analysis. Interobserver and intraobserver variability were expressed as the mean absolute difference between observations.

Statistics. Data are expressed as means ± standard deviations. Univariate analyses were performed in categorical data using the Spearman correlation and the rank-sum and Kruskal-Wallis tests. Multivariate analyses were performed with the use of stepwise and forced multiple logistic regressions in order to determine independent correlates of hospital mortality. The presence of carotid disease was considered a categorical variable for multivariate analysis. Statistical significance was defined as a p-value < 0.05.

Results
Patient characteristics are shown in Table 1. The mean age was 72 ± 13 years and 48% were male. The neurological events were distributed as follows: 49% CVA, 43% TIA and 8% intercerebral hemorrhage (ICH). Serum electrolytes, including potassium, were normal in all patients. Only one patient had been taking an anti-arrhythmic medication. The ECG of one patient with acute myocardial infarction was uninterpretable. A total of 120 of the 140 patients presenting with acute neurological events had interpretable ECGs with measurable QT intervals in at least 11 of 12 leads. There were no significant differences in clinical characteristics between patients with interpretable and uninterpretable ECGs.

Table 1

Patient characteristics

QTD values ranged from 0–160 msec (median, 45 msec). Mean QTD values were higher in patients with ICH as compared to CVA and TIA (70 ± 15 msec versus 53 ± 27 msec versus 48 ± 31 msec, respectively; p = 0.03). There was a trend toward QTD correlating with age (r = 0.18; p = 0.056). Mean QTD values were higher in patients with congestive heart failure (80 ± 43 msec versus 47 ± 24 msec; p = 0.006) and in patients with carotid artery disease (59 ± 32 msec versus 46 ± 27 msec; p = 0.045) than in patients without these conditions. Mean QTD values were not related to gender, hypertension, diabetes, smoking, alcohol and hypercholesterolemia. QTD values were not associated with atrial fibrillation, prior myocardial infarction, coronary disease or recent cardiac surgery.

Hospital mortality was 9.2% for the entire group. Mean QTD values were higher in non-survivors compared to survivors (83 ± 20 msec versus 50 ± 15 msec; p = 0.004). With regard to functional outcomes, increasing QTD values weakly correlated with all three functional scales upon discharge: NIHSS (r = 0.20; p = 0.042), Barthel Index (r = -0.20; p = 0.032) and Modified Rankin scale (r = 0.22; p = 0.017). All patients with TIA survived. On multivariate analysis, other independent predictors of hospital mortality included QTD (odds ratio = 1.35; 95% confidence interval = 1.08–1.68) and a trend toward age (odds ratio = 1.07; 95% confidence interval = 0.99–1.16). On age-adjusted logistic regression analysis, mortality increased by an odds ratio of 1.28 and 95% confidence interval was 1.02–1.61 for every 10 msec increase in QTD. The intraobserver variability was 5 ± 4 msec and interobserver variability was 12 ± 16 msec.

Discussion
This study found considerable variability in the measurement of QT dispersion on admission ECG in patients presenting with acute neurological events. Increasing QTD values were associated with greater hospital mortality and weakly correlated with lower functional outcomes on three standard scales of functional disability.

Over the past decade, QTD has been used to prognosticate patients with cardiovascular disease at risk for ventricular tachyarrhythmias and sudden death. Few studies have included patients with primary disorders that are non-cardiac, such as end-stage renal disease and diabetes.^8,9,17 The rationale for this study is based on several lines of evidence suggesting that autonomic regulation of the cardiovascular system is altered by acute brain injury. First, QT prolongation, T-wave changes and supraventricular and ventricular tachyarrhythmias are common ECG manifestations of stroke, which are unrelated to ischemic heart disease.¹⁸ Moreover, stroke significantly decreases heart rate variability for up to 6 months after brain injury.^19,20 In addition, greater variability exists in systolic blood pressure measured by power spectral analytic techniques in patients with acute stroke as compared to controls.²¹ The present study extends these findings and suggests increased QTD to represent another manifestation of cardiovascular autonomic dysregulation. Only one prior study specifically examined the significance of QTD measurement in subarachnoid hemorrhage.²² Although the cardiovascular system plays a major role in the regulation of cardiovascular function, the effects of non-hemorrhagic stroke on QTD had not been previously considered.

Table 2

Univariate analysis of clinical characteristics and QT dispersion

Higher QTD was associated with intercerebral hemorrhage, greater hospital mortality and worse functional outcomes on discharge. QTD values directly correlated with functional scores assessed by the NIHSS and Modified Rankin scores and inversely correlated with Barthel scores. These findings are concordant since higher functional status is associated with lower scores on the former two scales and with higher scores on the latter instrument. QTD was unrelated to coronary disease, history of myocardial infarction and prior bypass surgery. Although there was a trend toward QTD values correlating with age, multivariate analyses revealed QTD to be the only independent predictor of hospital mortality.

Several explanations, such as centrally mediated sympathetic hyperactivity, reduced cardiac parasympathetic innervation and abnormal baroreceptor function, might account for these findings. QTD has been proposed to reflect heterogeneity of ventricular repolarization, which not only reflects underlying structural heart disease but also cardiac regulation by the sympathetic nervous system. Several studies have evaluated the relationship between the sympathetic activity and QTD. Yoshida et al. studied 29 patients with prior myocardial infarction and 17 controls and found QTD directly related to the degree of cardiac (123) I-metaiodobenzyl guanidine (MIBG) uptake, a marker of cardiac sympathetic nervous system activity.23 However, a similar relationship between QTD and I-123 MIBG uptake has not been confirmed in studies of patients with diabetes and hypertrophic cardiomyopathy.^9,17,24 With regard to cerebrovascular disease, Randall found greater QTD in 26 patients with subarachnoid hemorrhage as compared to controls with nonruptured cerebral aneurysms.²² In this study, QTD measurements directly correlated with plasma concentrations of DHPG, a metabolite of norepinephrine. A direct correlation between QTD and plasma norepinephrine levels was found in healthy subjects undergoing tilt-table testing.²⁵ Direct evidence linking reduced parasympathetic control and increased QTD is lacking despite reduced parasympathetic controlled heart rate variability in patients with ischemic stroke.¹³ Although carotid baroreceptor dysfunction has been demonstrated after acute stroke,²¹ baroreceptor function is also impaired in patients with congestive heart failure and carotid artery disease.^26,27 In the present study, higher QTD values were associated with both congestive heart failure and carotid disease. Therefore, a number of reasons exist for baroreceptor dysfunction and the specific mechanism by which acute neurological events increase QTD remains unknown. Nevertheless, QTD measurements in patients presenting with CVA and TIA likely reflect neurological injury as well as underlying heart disease.

Mean QTD values obtained in the subset of patients with ICH in the present study were similar to those obtained in patients with subarachnoid hemorrhage [70 versus 78 msec (median)].²² Other studies that have evaluated the ECG manifestations of acute neurological events preceded QTD measurement, whereas studies assessing the significance of QTD in cardiovascular disease had not considered the central nervous system. The fact that QTD was influenced by the presence of congestive heart failure, but not by coronary disease, recent coronary artery bypass graft surgery or atrial fibrillation, probably relates to the patient population studied. Although cardiac disease is the leading cause of long-term mortality in stroke patients, the long-term prognostic value of QTD remains unknown.

Finally, the prognostic value of QTD may relate to the magnitude of brain injury, which was not considered in this study. Similar results were obtained in a study of healthy, elderly subjects, in whom baseline QTD measurements predicted stroke mortality (relative risk = 3.21; 95% confidence interval = 1.09–9.47) during 10-year follow-up.²⁸ This suggests that QTD may reflect occult as well as obvious central nervous system pathologies. Although the presence of LVH on ECG increased the predictive value of QTD for stroke mortality, QTD was unrelated to cardiac mortality.

Study limitations. This study is subject to the inherent limitations of a retrospective study. Six deaths occurred in patients with uninterpretable ECGs, leaving relatively few mortality events in patients with interpretable ECGs. All patients with TIA survived. Although QTD was not corrected for heart rate and QRS duration,²⁹ there were no differences in these parameters between survivors and non-survivors. Although laterality was found to be predictive of stroke-related arrhythmias,^13,30 presenting symptoms and infarct location on computed tomographic and magnetic resonance imaging studies were not considered. Medication effects cannot be completely excluded, as angiotensin-converting inhibitors, angiotensin receptor blockers and beta-blockers have been shown to reduce QTD in certain patient subgroups.^31–34 QTD was not compared with other measures of autonomic regulation, but no gold standard remains for the interpretation of cardiovascular autonomic dysfunction in patients with cardiac disease or stroke.³⁵ Moreover, the weak association between functional outcomes and QTD values preclude its routine use in predicting functional outcomes. Whether correction for heart rate would improve the predictive value of QTD remains unknown.

Despite these limitations, we conclude that in patients presenting with acute neurological events, increased QTD on admission ECG was significantly related to hospital mortality and to functional outcomes on hospital discharge. Furthermore, increased QTD was related to the type of neurological event as QTD values were greater in patients with subarachnoid hemorrhages as compared to those with CVA and TIA. In this patient population, increasing QTS was also associated with advancing age, congestive heart failure and carotid disease, but not with atrial fibrillation, coronary disease or recent bypass surgery. Therefore, QTD would appear to reflect not only underlying heart disease but also acute neurological injury. The study population provides an accurate demographic representation of patients with acute neurological events encountered in clinical practice. It appears that both structural heart disease as well as altered cardiac neuroregulation can increase heterogeneity of ventricular repolarization. Further studies are needed for prospective validation and to determine the specific mechanisms by which QTD is influenced by acute neurologic events. Whether QTD is influenced by the presence of carotid disease independent of CVA or TIA remains unknown. In addition, serial changes in QTD and the long-term prognostic value of QTD in this setting merit further study.