Utility of ECGs in the pediatric emergency department for patients presenting with a seizure

Anthony G. Pompa, MD ^a, Peter LaRossa, MD ^b, Lee B. Beerman, MD ^a, Yoshimi Sogawa, MD ^a,

Johanna Rosen, MD ^a, Gaurav Arora, MD ^a,?

^a UPMC Children‘s Hospital of Pittsburgh, University of Pittsburgh School of Medicine, Pittsburgh, PA, United States

^b C.S. Mott Children‘s Hospital, University of Michigan, Ann Arbor, MI, United States

Keywords:

Electrocardiogram Seizure

long QT syndrome Cardiology Neurology

Introduction

Seizure is a common presenting complaint in the pediatric emer- gency department (ED). Initial diagnostic investigation beyond a thor- ough history and physical examination typically focuses on searching for reversible causes of a seizure, Predisposing factors which may have precipitated a seizure, and diagnoses which could mimic seizures. Initial evaluation is variable based on the clinical scenario and available resourced in the ED, but is generally minimal, sometimes involving neu- roimaging or biochemical evaluation [1,2]. An electrocardiogram (ECG) can be part of this initial evaluation looking for primary cardiac arrhyth- mias such as Long QT syndrome, which traditional teaching suggests may lead to syncope that can mimic seizures, mostly identified in case-reports [3,4]. Long QT syndrome is rare with a prevalence between 1 in 2000 and 1 in 3000 in the general population [5].

There are no guidelines to direct ECG use after a seizure for pediatric patients. The institutional practice in our ED is to obtain an ECG in a ma- jority of these patients. Data about the yield of ECG testing in pediatric patients with seizure is lacking. Importantly, there are known nonspe- cific electrocardiographic changes associated with seizures which do not correlate with underlying cardiac disease. These include ST depres- sion, ST elevation, T wave inversions, premature atrial and ventricular contractions, sinus pauses, nodal escape rhythms, and non-specific con- duction delays [6-10].

This study sought to evaluate the utility of ECGs in determining pri- mary cardiac disease in post-seizure patients in the pediatric ED and in- vestigate the frequency and characteristics of known nonspecific post- seizure ECG changes. We also sought to characterize any potential

* Corresponding author at: Heart Institute, UPMC Children’s Hospital of Pittsburgh, Fifth Floor Faculty Pavilion, 4401 Penn Ave., Pittsburgh, PA 15224, United States.

E-mail address: [email protected] (G. Arora).

correlation between time from seizure to time of ECG with the presence of known post-seizure ECG changes.

Methods

Study design

We performed an institutional review board approved retrospective review of all patient encounters in our ED over a 5-year period that were associated with an International Statistical Classification of Diseases and Related Health Problems (ICD) code for seizure. Both ICD-9 and ICD-10 codes were needed for the initial database inquiry since our period of in- terest spanned 2015, the year United States hospitals widely adopted ICD-10. From ICD to 9, codes 345.x (Epilepsy and recurrent seizure),

780.3 (convulsions), and 779.0 (convulsions in newborn) were in- cluded. Additionally, from ICD to 10, code G40.x (Epilepsy and recurrent seizures) was included.

Patient selection and data collection

Patient charts were excluded if an ECG was not performed during the encounter, seizure was an incidental chronic diagnosis in a pa- tient presenting for a different chief complaint, or if the ED physician documented that the presentation was not consistent with a seizure after they had obtained their initial history and physical examina- tion. Patients taking a QT prolonging medication, those with known structural heart disease, or those with a prior Abnormal ECG in the electronic health record were also excluded. The list of QT prolonging medications was taken from the “Combined List of Drugs that Prolong QT and/or Cause Torsades de pointes (TDP)” from www.crediblemeds.org [11].

The final cohort was narrowed to patients who presented to the pe- diatric ED for evaluation of seizure and had an ECG available for review. Patient demographic information was collected from this cohort includ- ing age, sex, height, and weight. Past medical history was collected as well as personal seizure history, history of prior ECGs, and current med- ications or known ingestions. The time of the seizure event was ob- tained from Emergency Medical Services records where available and time of ECG acquisition was also acquired.

https://doi.org/10.1016/j.ajem.2019.11.016

0735-6757/(C) 2019

Definition of patient groupings

Final interpretations of ECGs by an attending pediatric cardiologist were available in the electronic health record. ECGs were considered normal with final reads of normal sinus rhythm, sinus tachycardia, Sinus bradycardia, or sinus arrhythmia. All other ECGs were labelled as “flagged for review.” This definition was chosen as it was felt that most ED providers would be comfortable classifying the first category as normal, while other reads such as “ST changes” would result in a more variable reaction from the ED providers. ECGs flagged for review (n = 157) were then reread by an attending pediatric electrophysiolo- gist blinded to the details of the patient history. They were grouped into “normal variants” and “Abnormal.” The abnormal ECGs were further categorized as “Abnormal requiring further evaluation” and “Abnormal not requiring further evaluation.” The rereads grouped into each of these categories are listed in Table 1. In addition, a sample of forty nor- mal ECGs (or about 10% of that original grouping) were randomly se- lected for reread by a pediatric electrophysiologist for confirmation.

Separately, the interpretations of the ECGs read as abnormal by an

attending pediatric cardiologist were evaluated for known nonspecific post-seizure abnormalities, other than rate. Based on findings from prior studies these abnormalities included ST depression, ST elevation, T wave inversion, premature atrial contractions, premature ventricular contractions, and ventricular conduction delays. We reviewed patient records to determine if any subsequent cardiac evaluation was per- formed, what that evaluation entailed, and if there was any eventual cardiac diagnosis.

Statistical analysis

Continuous variables were reported as means +- standard deviation. Discrete variables were detailed as whole numbers and/or percentages as indicated. We analyzed continuous variables without assumed nor- mal distribution by using Mann-Whitney U tests and unpaired categor- ical variables with Fisher?s exact tests. All statistical analyses were performed using GraphPad Prism version 7.0d for Mac OS X (GraphPad

Table 1

ECG read categorization.

Software, La Jolla, California) and p-values of b0.05 were considered to be significant.

Results

Characteristics of study subjects

A total of 2767 patient encounters between January 1 2011 and De- cember 31 2015 were initially considered for inclusion and 192 (7%) were excluded because an ECG was not associated with the encounter or available for review. Another 1996 (72%) were excluded because ei- ther seizure was a historical diagnosis but not the chief complaint, the patient had a history of known heart disease, the patient was taking a QT prolonging medication, or the ED provider documented that they were not evaluating for a seizure based on their history and physical ex- amination (Fig. 1).

This left 579 patient encounters as the study population. Among the 579 patients, 316 (55%) were male and the mean age was 6.6 +-

5.2 years. Furthermore, 369 (64%) patients were presenting with a first-time aFebrile seizure, 95 (16%) presented with a febrile seizure (be- tween 6 months and 6 years of age), and the remaining 115 (20%) were patients who had previously been diagnosed with a seizure disorder or who had had at least one unprovoked seizure in the past. Of those pa- tients, 69 were currently prescribed an antiepileptic medication. A total of 68 (12%) patients had a prior normal ECG in the medical record (Table 2).

Review of ECGs

As demonstrated in Fig. 2, a total of 579 ECGs were included, of which 422 (73%) were normal. The remaining 157 (27%) were flagged for review and reread by a pediatric electrophysiologist. Of these, 94 (16%) were normal variants and 63 (11%) were considered “Abnormal ECGs”. These 63 abnormal EKGs were further classified as “Abnormal not requiring further evaluation” (4, 1%, which included limb lead rever- sal and mild left axis deviation), or “Abnormal requiring further evalua- tion” (59, 10%). The specific diagnoses in each category are listed in Table 1. The specific rereads of the originally abnormal ECGs are outlined in Tables 3 and 4. Additionally, all of the forty normal ECGs re- read by an electrophysiologist were confirmed to be normal.

Original ECG read categorization

Normal Flagged for further review by electrophysiologist

Outcomes of further cardiac evaluation

Normal sinus rhythm

Sinus tachycardia Sinus bradycardia Sinus arrhythmia

All other reads

In total, 31 patients had additional cardiac evaluation. This included 8 repeat ECGs, 6 Holter monitors, 18 echocardiograms, and 24 outpa- tient cardiology clinic evaluations. Further cardiac work-up was then separated into patients with a known nonspecific post-seizure ECG

change and those without one of those changes. Those with known

Electrophysiologist ECG Reread Categorization

“Normal variants” “Abnormal not requiring “Abnormal requiring further further evaluation” evaluation“

Total # of Patients Presenting

Reread as normal Limb lead reversal QRS abnormality: complete or

incomplete Right bundle branch block, intraventricular conduction delay

with ICD Code of Seizure: n = 2767

Total # of Patients Presenting

No ECG performed n = 192

Nonspecific ST/T wave changes

ST elevation consistent with early repolarization

Wandering atrial pacemaker

Mild left axis deviation (>=-20? if N1 year or N0? if b1 year)

QTc N450 msec

Chamber hypertrophy: atrial or ventricular

AV block

Rhythm abnormality: premature atrial or ventricular contraction Left axis deviation b-20?

Wolff-Parkinson-White pattern T wave inversion in whole lead

with ICD Code of Seizure and ECG Obtained:

n = 2575

Presented with diagnosis of seizure and ECG was obtained:

n = 579

Other chief complaint n = 897

Evaluated as a non-epileptiform event n = 755

Taking QT prolonging medication n = 99

Prior cardiac disease or abnormal ECG n = 242

Duplicate encounters n = 3

groups

Fig. 1. Cohort selection and demographics.

Table 2

Characteristics and likelihood of an abnormal ECG.

Table 3

Electrophysiologist “Normal variant” and “Abnormal not requiring further review”

Characteristics	Total, N (%)	Abnormal ECG, N (% of	Normal ECG, N (% of	Odds ratio (95% confidence interval)		rereads. Diagnoses on ECGs reread by electrophysiologist as “normal variant” and Count “abnormal not requiring further review”
		Total)	total)			No abnormality 48
Sex						Nonspecific ST/T wave changes 19
Male	316 (55)	92 (29)	224 (71)	0.80 (0.55-1.16)		Early repolarization pattern 18
Female Prior Normal ECG Yes	263 (45) 68 (12)	65 (25) 17 (25)	198 (75) 51 (75)	p = 0.26 1.13 (0.65-2.00)		Mild left axis deviation 9 Limb lead reversal 3 Wandering atrial pacemaker 1
No	511 (88)	140 (27)	371 (73)	p = 0.77
Antiepileptic Medication

Yes	69 (12)	23 (33)	46 (67)	0.71 (0.42-1.22)
No	510 (88)	134 (26)	376 (74)	p = 0.24

First afebrile seizure

Yes 369 (64) 102 (28) 267 (72) 0.93 (0.64-1.35)

No 210 (36) 55 (26) 155 (74) p = 0.77

b2 years old

Yes 138 (24) 39 (28) 99 (72) 0.93 (0.60-1.42)

No 441 (76) 118 (27) 323 (73) p = 0.74

was no significant statistical difference between these 2 groups when it came to the likelihood that they would have an abnormal ECG in the ED (odds ratio: 1.13; confidence interval: 0.65 to 2.00; p = 0.77, Fisher?s exact test). There was no statistically significant difference in the likelihood of having a normal or abnormal ECG in any of the charac- teristics analyzed (Table 2).

Total Abnormal ECG

Normal ECG

Median difference (95% confidence interval)

3.5. Analysis of patients with febrile seizures

There were 95 patients who presented with a febrile seizure be- tween 6 months and 6 years of age. Of those, 25 had an abnormal ECG, 13 of these abnormalities were associated with a known post-

Age, Mean years +- SD,	6.6	6.4 +- 5.1,	7.0	1 (-25 to 20)
Median	+- 5.2, 5	6	+- 5.6, 5	p = 0.55

post-seizure changes accounted for 50% of the Holter Monitors, 39% of echocardiograms, and 46% of the outpatient cardiology clinic evalua- tions. The outcome of testing these 31 patients was that 2 patients were incidentally diagnosed with vagally mediated first-degree heart block (a benign phenomenon) [12], one had myocarditis, and one pa- tient (0.2% of the entire initial cohort) was diagnosed with long QT syndrome.

Of the patients with ECGs reread as “Abnormal requiring further evaluation” only 15 of 59 actually underwent cardiac evaluation. Fur- thermore, the likelihood of follow-up was not significantly different be- tween patient’s whose ECGs were placed in the “Abnormal requiring further evaluation” group versus the “Normal variant” plus “Abnormal not requiring further evaluation” groups. (odds ratio: 1.75; confidence interval: 0.82 to 3.69; p = 0.21, Fisher?s exact test).

Analysis of patients with prior normal ECGs

The likelihood of patients having an abnormal ECG was compared between the group of patients with a normal prior ECG in the medical record and those without a prior ECG. Of the 68 patients with a prior normal ECG, 17 had an abnormal ECG in the ED and from the remaining 511 patients without a prior normal ECG, 140 were abnormal. There

seizure change, and 4 had additional cardiac evaluation. All of these pa- tients had normal additional cardiac work-up, including 4 repeat ECGs, 3 outpatient cardiology clinic evaluations, and an echocardiogram.

Frequency of nonspecific post-seizure ECG changes

A separate analysis of the study population of 579 patients was done to determine frequency of known associated post-seizure changes. From the 157 patients with an ECG flagged for review, 94 (60%) had an abnormality known, or based on prior research likely, to be associ- ated with a post-seizure change. This means only 10% of patients in the study population had an abnormal ECG that was not confounded by a post-seizure change.

As mentioned earlier, previously described post-seizure changes were defined as ST depression, ST elevation, T wave inversion, prema- ture atrial or ventricular contractions, and ventricular conduction de- lays. For the purposes of our analysis, we considered non-specific ST depression and ST elevation consistent with early repolarization to be “Normal variants.” We labeled T wave inversions in whole lead groups, intraventricular conduction delay, right bundle branch block, incom- plete right bundle branch block, premature ventricular contractions, and premature atrial contractions as “Abnormal requiring further

All patient encounters included in the study group n = 579

Normal ECG n = 422

Flagged for Further Review by an Electrophysiologist

n = 157

27 % of included encounters

n = 94

n = 63

11% of included encounters

n = 4

n = 59

Fig. 2. Electrophysiologist rereads of initially abnormal ECGs.

Table 4

Electrophysiologist “abnormal requiring further evaluation” rereads.

Diagnoses on ECGs reread by electrophysiologist as “abnormal requiring further evaluation”

Count

Prior literature on the utility of ECGs in the pediatric ED in patients presenting after a seizure is lacking. A large study did examine the QT interval in the pediatric ED for patients presenting with all complaints and found that one-third had QT interval greater than 440 ms and no

Chamber hypertrophy 24

QTc N450 msec 10

AV Block 8

Left axis deviation b-20? 8

QRS abnormality 6

Premature atrial or ventricular contractions 3

Wolff-Parkinson-white pattern 1

T wave inversions 1

evaluation.” Though these findings have been previously published as potential post-seizure changes, we did not feel these could be ignored in pediatric patients. The total of the post-seizure change group was 94 patients, this is made up of some of the patients from the “Normal variants” group as well as the “Abnormal requiring further evaluation” group.

Correlation of time from seizure with presence of ECG changes

The mean time from seizure event to ECG acquisition for the total co- hort was 208 +- 165 min. There was no statistical difference between patients with a normal ECG, mean of 207 +- 160 min and a median of 150 min, and patients with an abnormal ECG with known post-seizure changes, mean of 198 +- 154 min and a median of 160 min (Mann- Whitney U = 11881, 95% confidence interval, -25 to 20; p = 0.63)

(Fig. 3).

Discussion

This retrospective review is the first study of its kind to analyze the utility of ECGs in post-seizure pediatric patients presenting to the ED and characterize the frequency of known nonspecific post-seizure ECG changes in this population.

The initial database inquiry was broad to ensure that all patients who may have presented to the ED with a seizure within the search pe- riod was included. The preliminary search was then carefully manually reviewed to remove patients for the stated exclusion criteria including those who were not presenting with a seizure for the encounter in ques- tion but had an ICD code of seizure attached to their chart for other reasons.

1500 p = 0.63

Minutes from Seizure to ECG

1000

500

0

patients were diagnosed with a long QT syndrome [13]. Another study showed that of 8 patients with long QT syndrome who had a delayed di- agnosis none of them were given this diagnosis based on ECGs in the ED, even though they all had one obtained [3]. There has never been a study of this size to evaluate the frequency of post-seizure ECG changes in the ED or elsewhere. Additionally, the overwhelming majority of initial en- counters presenting for evaluation of a seizure event had an ECG ob- tained (93%); therefore, this cohort can be considered to represent almost all patients presenting to the pediatric ED in the early post- ictal period.

The frequency of nonspecific post-seizure ECG changes has not been characterized in a study of this scale in the past. We have shown that these confounding ECG abnormalities are present in the majority, or 60%, of patients presenting to the ED after a seizure. While we cannot determine a causative relationship because most of these patients did not have a baseline or follow-up ECG, these specific changes have been previously reported to be associated with seizures. Their presence calls into question the utility of obtaining a routine ECG in these patients in the immediate post-seizure period since these changes confound that interpretation. False positive ECG reads led to unnecessary testing in- cluding time and resource intensive tasks for patients, families, and pro- viders. While these interventions pose minimal harm to the patients, in a healthcare environment increasingly focused on time and cost effi- ciency, these low-yield interventions draw more scrutiny.

The frequency of these nonspecific post-seizure ECG changes is not completely understood and the changes themselves have not fully been characterized in a large study. Recent research has even shown that QT prolongation itself may represent a nonspecific post-seizure ECG change [10].

Prior guidelines have recommended that patients presenting with a classic febrile seizure do not need further evaluation such as neuroimag- ing or blood studies [14]. However, in our series it is apparent that pro- viders continue to obtain ECGs in this setting so these were included. Of note, the rate of abnormal ECGs in these patients was the same as the rate for the entire cohort. Thus, we do not believe the data is inherently skewed by including this population. Moving forward, this data will hopefully contribute to the evidence that patients with first time febrile seizures do not require an ECG.

Overall, the frequency of diagnosing long QT syndrome in patients in our population presenting to the pediatric ED with a seizure is low, 1 out of 579, or 0.2%. This proportion could potentially be even lower since the one patient who was found to have a long QT syndrome presented in a unique way as this was her first encounter since immigrating to the United States and this patient carried a previous diagnosis of epilepsy from an unknown provider in a developing country. However, this pa- tient was initially evaluated for a breakthrough seizure in the ED based on her reported past medical history and was therefore included in the cohort. She had Recurrent episodes of loss of consciousness with Seizure activity in the ED and was noted to have intermittent torsades de pointes on the bedside monitor.

The one patient found to have myocarditis was discovered in a sim- ilarly incidental way with many abnormalities noted on continuous car- diorespiratory monitor which was in place due to his presentation with

altered mental status and presumed seizure.

Normal ECG

ECG with known post- seizure change

There were 10 additional patients (1.7% of the initial cohort) who were found to have a Corrected QT interval (QTc) greater than 450 ms by the electrophysiologist’s interpretation. Using the 95% confidence in- terval of 450 ms as the Upper limit of normal means that 2.5% of the pop- ulation is above this cut-off. Therefore, in a group of 579 patients one

Fig. 3. Time from seizure to ECG. There was no significant difference in the time from

seizure to ECG between patients with a normal ECG (207 +- 160 min) and those with an abnormal ECG with known post-seizure changes (198 +- 154 min), (Mann-Whitney U = 11881, 95% confidence interval, -25 to 20; p = 0.63).

could expect up to 14 patients to have a QTc greater than 450 ms by ran- dom sampling alone. Additionally, only 3 of these patients had a QTc greater than 480 ms which is the 99% confidence interval mark. Only

one of these patients had further evaluation, with a follow-up ECG and Holter monitor, which were normal. Since the other patients did not have further cardiac evaluation it is impossible to definitively say that their event was not attributed to an arrhythmia. However, it has been 2-6 years since these patients initially presented and they have not returned with another event, making this less likely.

We were also able to demonstrate that the likelihood of an ECG hav- ing one of these post-seizure changes in the ED is not temporally depen- dent as there was no significant difference in the time from seizure to ECG acquisition between patients with a normal ECGs and those with abnormal ECGs with a known post-seizure change. This suggests that the time since the seizure event cannot be used to determine if an ECG is less likely to have nonspecific post-seizure change and therefore more likely to detect true cardiac disease.

As for the limitations of this study, first it was performed at a single institution. There were a large number of patients with abnormal ECGs who did not have any further cardiac evaluation. While up to the time this data was collected (2-6 years after their presentation) they did not present to the same institution with cardiac disease it is unknown if they will present at a later date or if they have already presented to an- other facility with primary cardiac disease. This is less likely since our in- stitution is the only pediatric hospital in the region but is still a possibility.

Conclusions

Currently, there are no guidelines for obtaining ECGs in pediatric pa- tients after a seizure and the data about post-seizure ECGs in these pa- tients is lacking. The majority of ECGs obtained in the post-seizure period are either normal or confounded by prevalent nonspecific post- seizure changes which do not correlate with primary cardiac disease. The likelihood that an ECG will be abnormal cannot be predicted by the amount of time that has passed since the seizure episode, the pa- tient?s personal history of a normal ECG, or evaluated patient or seizure characteristics. In this retrospective review, the yield of ECGs to detect

cardiac disease in post-seizure patients was low. Further study is needed to determine a more judicious use of ECGs in post-seizure pa- tients presenting to the pediatric ED, if there is a role for them at all.

Disclosures

No external funding for this manuscript. The authors have no finan- cial relationships relative to this article to disclose.

References

1. Gaillard WD, Chiron C, Harvey AS, et al. Guidelines for imaging infants and children with recent-onset epilepsy. Epilepsia 2009;50(9):2147-53.
2. Morton L. Evaluation of seizure disorders in the pediatric patient. Adv Studies Med 2001;1(4):129-34.
3. MacCormick JM, McAlister H, Crawford J, et al. Misdiagnosis of long QT syndrome as epilepsy at first presentation. Ann Emerg Med 2009;54(1):26.
4. Hazle MA, Shellhaas RA, Bradley DJ, et al. Arrhythmogenic channelopathy syn-

   dromes presenting as refractory epilepsy. Pediatr Neurol 2013;49:134-7.

   Stramba-Badiale M, Crotti L, Goulene K, et al. Electrocardiographic and genetic screening for long QT syndrome: results from a prospective study on 44,596 neo- nates. Circulation 2007;116:II_377.

5. Tigaran S, Pasmussen V, Dam M, et al. ECG changes in Epilepsy patients. Acta Neurol Scand 1997;96:72-5.
6. Zijlmans M, Flanagan D, Gotman J. Heart rate changes and ECG abnormalities during Epileptic seizures: prevalence and definition of an objective clinical sign. Epilepsia 2002;43:847-54.
7. Opherk C, Coromilas J, Hirsch LJ. Heart rate and EKG changes in 102 seizures: anal- ysis of influencing factors. Epilepsy Res 2002;52:117-27.
8. Nei M, Ho RT, Sperling MR. EKG abnormalities during Partial seizures in refractory

   epilepsy. Epilepsia 2000;41(5):542-8.

   Brotherstone R, Blackhall B, McLellan A. Lengthening of corrected QT during epilep- tic seizures. Epilepsia 2010;51(2):221-32.

9. Woosley RL, Heise CW, Romero KA. QTdrugs List. Innovation Park Dr., Oro Valley: AZCERT, Inc.; 2017 AZ 85755.
10. Alboni P, Holz A, Brignole M. Vagally mediated atrioventricular block: pathophysiol- ogy and diagnosis. Heart 2013;99(13):904-8.
11. Van Dorn CS, Johnson JN, Taggart NW, et al. QTc values amoung children and adoles- cents presenting to the emergency department. Pediatrics 2011;128(6):1395-401.
12. Seizures F. Guideline for the neurodiagnostic evaluation of the child with a simple febrile seizure. Pediatrics 2011;127(2):389-94.