American Journal of Emergency Medicine (2012) 30, 1080-1088

Original Contribution

Clinical assessment of children with first-attack seizures admitted to the ED

Chun-Yu Chen MD^a, Wen-Chieh Yang MD^a, Kang-Hsi Wu MD^b, Han-Ping Wu MD^c,d,e,?

^aDepartment of Pediatrics, Changhua Christian Hospital, Changhua 50006, Taiwan, ROC

^bDepartment of Pediatrics, Children’s Hospital and School of Chinese Medicine, China Medical university hospitals,

Taichung 40447, Taiwan, ROC

^cInstitute of Clinical Medicine, National Yang-Ming University, Taipei 11221, Taiwan, ROC

^dDepartment of Pediatrics, Buddhist Tzu-Chi General Hospital, Taichung Branch, Taichung 42743, Taiwan, ROC

^eDepartment of Medicine, Tzu Chi University, Hualien 97004, Taiwan, ROC

Received 3 April 2011; revised 5 July 2011; accepted 13 July 2011

Abstract

Purpose: This study aims to evaluate clinical values and determine the function of a pediatric observation unit (POU) as an alternative to Inpatient unit admission for children with newly onset seizures.

Basic Procedures: Pediatric patients who were sent to the emergency department (ED) with new-onset seizure were retrospectively analyzed in a 6-year study period. All patients were divided into 3 groups: POU-discharged, unplanned inpatient admission, and required admission. Basic demographics, clinical course, biologic data, and radiologic findings were analyzed among the 3 groups.

Main Findings: From the 910 children admitted to the ED with first attack of seizure, 405 (44.5%; mean age, 2.86 +- 2.64 years) were admitted to the POU. Of them, 184 (45.4%) were later discharged. Using multivariate logistic regression analysis, patients with febrile seizure, those without elevated serum C- reactive protein level, and those who did not require first-line anticonvulsants in the ED were associated with an increased trend of POU discharge.

Principal Conclusions: The POU may be an alternative to immediate admission in selected cases of first seizures. Related information such as age, use of anticonvulsants in the ED, serum C-reactive protein value, and clinical diagnosis of febrile seizure are important factors for determining whether pediatric patients with first seizure attack should be admitted or discharged.

(C) 2012

Introduction

* Corresponding author. Department of Pediatrics, Buddhist Tzu-Chi General Hospital, Taichung Branch #66, Taichung County 42743, Taiwan,

R.O.C. Tel.: +886 4 606 0666×2798; fax: +886 4 3602 1123.

E-mail address: [email protected] (H.-P. Wu).

A pediatric observation unit (POU) provides an alterna- tive treatment site for patients with selected diagnoses [1]. It may be necessary for some patients who require observation for a specified time to make a definitive diagnosis, thereby avoiding unnecessary hospital admission and misdiagnosis [2,3]. Children are commonly admitted to the POU because of respiratory illnesses (ie, asthma, croup, bronchiolitis, and

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2011.07.008

pneumonia), gastrointestinal disorders (eg, gastroenteritis, abdominal pain), dehydration, infections, overdose or poisoning, and seizures [2].

Seizures are the most common and frightening neurologic disorder in childhood and account for approximately 1% of all emergency department (ED) visits [4]. There are many possible etiologies of a first seizure attack in children, including infection, neurologic/developmental causes, trau- matic head injury, toxins, and metabolic disturbances [5]. Hospital admission may provide some diagnostic and therapeutic benefits, and most patients are admitted because of concerns of potentially life-threatening sequelae. Howev- er, the high admission rate may lead to high costs and inefficient use of resources.

This study analyzed the clinical spectrum of first seizures in children who presented at the ED and determined the role of the POU as an alternative to inpatient admission for such cases.

Materials and methods

This study was a single-center medical chart review of patients 18 years or younger who presented with a first attack of seizure disorder at the pediatric ED of Changhua Christian Hospital, with approximately 30 000 pediatric visits per year and a 12.5% general admission rate. The hospital POU was a 19-bed unit adjacent to the pediatric ED. Patients were admitted to the POU after evaluation by pediatric attending physicians. The charts of all eligible patients during the 6- year period from January 2003 to December 2008 were reviewed.

Potentially eligible patient visits were identified by searching the Changhua Christian Hospital health records database. Charts with any febrile convulsions or other convulsions (International Classification of Diseases, Ninth Revision, 780.31, 780.39) and epilepsy-associated diagnoses (International Classification of Diseases, Ninth Revision, 345.00 to 345.91) in the primary or secondary discharge diagnosis fields were selected. Patients with recurrent seizure disorders or who discharged themselves against medical advice were excluded.

The primary outcome analyses were patients who were admitted to and then discharged from the POU. The secondary outcome analyses were patients with unplanned inpatient admission (UIA) or those who required admission (RA) and were then discharged from the wards. In the study hospital, patients presenting with a first seizure had to be admitted to a ward or POU for further evaluation. Patients presenting with a first seizure were initially admitted to a POU if they fulfilled the following criteria:

(1) mechanical ventilation was not required, (2) the duration of seizure was less than 30 minutes, (3) no focal neurologic deficits, (4) prolonged postictal period 30 minutes or less, (5) no toxic appearance, (6) no signs of

increased intracranial pressure, (7) no meningeal signs, (8) no persistent Altered consciousness level, (9) no intractable vomiting, and (10) no Severe headache.

All of the included patients were divided into 3 groups based on the unit of admission and discharge: (1) POU- discharged (POUD) group, or patients successfully dis- charged from the POU and without readmission to the hospital within 48 hours of discharge; (2) UIA group, or patients who required transfer from the POU to a general ward or intensive care unit (ICU) or who were readmitted to the hospital within 48 hours of POU discharge; and (3) RA group, or patients who were admitted to a general ward or ICU without POU stay.

For each patient, the following information was obtained: age, sex, type of seizure, associated symptoms (ie, fever, cough, rhinorrhea, vomiting, diarrhea, and headache), family history of seizure or epilepsy, developmental history, Laboratory test results (white blood count, C-reactive protein [CRP], stool rotavirus antigen test, Serum electrolytes, blood glucose, and cerebrospinal fluid [CSF] analysis), neuroim- aging findings (cranial magnetic resonance imaging [MRI] or computed tomography [CT]), electroencephalography (EEG) findings, duration of hospital stay, final diagnosis, anticonvulsants given in the ED, and admission to the ICU, general ward, or POU.

Seizure type, including generalized tonic-clonic and generalized tonic and partial, was classified based on the Commission on Epidemiology and Prognosis, 1993 Interna- tional League Against Epilepsy [6]. status epilepticus was defined as “a single Epileptic seizure of more than 30 min duration or a series of epileptic seizures during which function was not regained between ictal events in a period of more than 30 min [6].” Febrile seizure was defined as “an epileptic seizure occurring in childhood after 1 month of age, associated with febrile illness not caused by a central nervous system (CNS) infection, without previous neonatal seizure or previous unprovoked seizure, and not meeting the criteria for other acute symptomatic seizures [6].”

In addition, febrile seizures were classified as either simple or complex. A simple febrile seizure lasted less than 15 minutes, was initially generalized in nature, and occurred once during a 24-hour period. In contrast, a Complex febrile seizure lasted more than 15 minutes, had focal features at any time, or recurred within a 24-hour period [7].

Statistical analysis

Data of categorical variables were analyzed by the ?2 or Fisher exact test, where appropriate. Data of continuous variables were analyzed by the Student t test. P b .05 was regarded as statistically significant. Multivariate logistic regression analysis was then developed using variables that were significantly associated with the decision to which unit the patient was admitted. Distributions of variables were reported as percentages and mean +- SD.

Results

Demographics

During the 6-year study period, 910 patients (mean age, 3.45 +- 3.71 years; 513 boys and 397 girls) who presented at the ED with a first attack of seizure were enrolled. Of the 910 patients, 786 (86.4%) were younger than 6 years, and 578 (63.5%) children presented with fever. Generalized tonic- clonic seizure was the most common seizure type (n = 630, or 69.2%) and occurred more commonly in the POUD and UIA groups than in the RA group (P b .001) (Table 1). There was a family history of seizure-related disorders in 74 (8.1%) patients, whereas 34 (3.7%) had a history of developmental delay, with language development delay being the most common type. The associated clinical symptoms, including fever, cough, rhinorrhea, diarrhea, vomiting, and headache, were statistically different among the 3 groups.

Comparison between patients with hospital stay and discharged patients

The mean hospital stay of patients in the UIA and RA groups was 5.44 +- 5.10 days (range, 1-66 days). Febrile seizure was the most common etiology in the 3 groups. The underlying diagnoses in children with febrile seizure included upper respiratory tract infections (URI) (45%), lower

Table 1 Demographic data and clinical presentations of the patients with a first seizure attack

Unit of hospital stay and discharge

P

GTC indicates generalized tonic-clonic seizure; GT, generalized tonic seizure.

* Nonconvulsive type or types of seizure not well recorded in the patient’s chart.

respiratory tract infections (16%), acute gastroenteritis (19%), urinary tract infections (8%), and enterovirus in- fections (12%). In addition, anticonvulsants were given to 199 (21.9%) of 910 patients in the ED, and seizures subsided in 86.6% after receiving benzodiazepines (diazepam, mid- azolam, and lorazepam). However, seizures persisted in 21 patients even after 3 doses of benzodiazepines.

Stool rotavirus antigen tests were performed in 131 patients with watery diarrhea, and positive findings were recorded in 34 (26%). Of the 184 patients in the POUD group, there were no positive stool rotavirus antigen test results. Cerebrospinal fluid analysis was performed in 148 patients, and 18 had abnormal results. All of the 148 children who underwent lumbar puncture were admitted to a ward or the ICU. Imaging studies of the head (CT or MRI) were arranged for 216 children. The indications were history of trauma, prolonged postictal period, partial Seizure activity, signs of increased intracranial pressure, focal neurologic deficits, persistent headache, and persistent altered consciousness level. Among them, 65 (30%) had abnormal findings.

An EEG was performed in patients suspected with underlying epileptogenic brain pathology within 24 hours of POU stay or inpatient admission. An EEG was performed 10 days after a febrile seizure. Electroencephalography was performed in 519 of the 910 patients, and there were abnormal EEG results in 219 (37.8%), 204 (93.2%) of whom were in the UIA and RA groups (P = .003) (Table 2).

POUD (n = 184)				UIA (n = 221)			RA (n = 505)
		n	%		n	%		n	%
Sex	Female	67	36.4	95		43.0	235		46.5	.059
	Male	117	63.6	126		57.0	270		53.5
Age (y)	b1	7	3.8	20		9.0	92		18.2	b.001
	1-6	164	89.1	183		82.8	320		63.4
	7-12	10	5.4	15		6.8	52		10.3
	13-18	3	1.6	3		1.4	41		8.1
Type of seizure	GTC	155	84.2	179		81.0	296		58.6	b.001
	GT	3	1.6	5		2.3	10		2.0
	Partial	10	5.4	9		4.1	54		10.7
	Status epilepticus	1	0.5	0		0.0	20		4.0
	Others ?	15	8.3	28		12.6	125		24.7
Associated symptoms	Fever	152	82.6	181		81.9	245		48.5	b.001
	Cough	84	45.7	118		53.4	188		37.2	b.001
	Rhinorrhea	81	44.0	108		48.9	162		32.1	b.001
	Headache	2	1.1	5		2.3	29		5.7	.007
	Vomiting	27	14.7	47		21.3	120		23.8	.036
	Diarrhea	20	10.9	49		22.2	86		17.0	.011
Family history of seizure		17	9.2	25		11.3	32		6.3	.065
or epilepsy
Developmental delay	4		2.2	3		1.4	27		5.3	.015

		n	%		n	%
Sex	Female	67	36.4		330	45.5	.027 a
	Male	117	63.6		396	54.5
Age (y)	b1	7	3.8		112	15.4	b.001 a
	1-6	164	89.1		503	69.3
	7-12	10	5.4		67	9.2
	13-18	3	1.6		44	6.1
Type of seizure	GTC	155	84.2		475	65.4	b.001 a
	GT	3	1.6		15	2.1
	Partial	10	5.4		63	8.7
	Absence	1	0.5		3	0.4
	Status epilepticus	1	0.5		20	2.8
	Others b	14	7.6		150	20.7
Associated symptoms	Fever	152	82.6		426	58.7	b.001 a
	Cough	84	45.7		306	42.1	NS
	Rhinorrhea	81	44.0		270	37.2	NS
	Headache	2	1.1		34	4.7	.020 a
	Vomiting	27	14.7		167	23.0	.014 a
	Diarrhea	20	10.9		135	18.6	.013 a
Stool rotavirus antigen	Negative	4	100.0		93	73.2	NS
	Positive	0	0.0		34	26.8
Brain image	Normal	5	83.3		146	69.5	NS
	Abnormal	1	16.7		64	30.5
EEG	Normal	55	78.6		305	59.9	.003 a
	Abnormal	15	21.4		204	40.1
Diagnosis	Febrile seizure	152	82.6		352	48.5	b.001 a
	Epilepsy	12	6.5		134	18.5
	CNS infection	0	0.0		28	3.9
	congenital anomaly	0	0.0		3	0.4
	metabolic disease	0	0.0		15	2.1
	Trauma	3	1.6		22	3.0
	Rotavirus related	0	0.0		33	4.5
	Unprovoked seizure	17	9.2		139	19.1
Anticonvulsant in ED		17	9.2		182	25.1	b.001 a

Clinical analysis between patients in the POUD and UIA groups

Table 2 Clinical analysis of the patients with first seizure attack between hospital stay and discharge

Admission

P

POUD (n = 184) (UIA + RD) (n = 726)

^a Statistically significant by the ?² test or Fisher exact test, where appropriate; NS, Not significant.

^b Nonconvulsive type or type of seizure not well recorded in the patient’s chart.

During the study period, 405 (44.5%) of the 910 patients were admitted to the POU. Of these 405 patients (mean age, 2.86 +- 2.64 years), 184 (45.4%) were in the POUD group. The length of POU stay of the 405 patients was 18.86 +-

10.33 hours, and the length of stay of the POUD group was longer than that of the UIA group (21.38 +- 8.22 vs 16.76 +- 11.40 hours, respectively; P b .001). Generalized tonic- clonic seizure was the most common seizure type (334/405, or 82.5%). Diarrhea was more common in patients in the UIA group than in patients in the POUD group (P = .003) (Table 3).

A CSF analysis was performed in 21 of the 221 patients in the UIA group, and 4 patients had abnormal results. Of the 4 patients, 3 were younger than 2 years. Of the 38 children who underwent head imaging studies (CT or MRI), 9 (23.7%) had abnormal findings (Table 4). Of these 9 children, 7 were given anticonvulsants in the ED. Febrile seizure was the most common diagnosis (319/405, or 78%) and was more frequent in patients in the POUD group than in the UIA group (P = .042). In addition, complex febrile seizures were found in 64 (20.3%) of the 316 patients and were more frequent in the UIA group (P = .008). Details of diagnosis were shown in Fig. 1.

Anticonvulsants were given to control seizures in 71 (17.5%) of the 405 patients in the ED. Anticonvulsants were

Table 3

Clinical analysis of patients between POUD and UIA groups

Admitted to POU (n = 405) POUD (n = 184)

P

UIA (n = 221)

^a Statistically significant by the ?² test or Fisher exact test, where appropriate.

Table 4 Details of abnormal brain imaging studies performed in the POU (n = 9)

n			%		n	%
Associated symptoms	Fever	152	82.6	181		81.9	NS
	Cough	84	45.7	118		53.4	NS
	Rhinorrhea	81	44.0	108		48.9	NS
	Headache	2	1.1	5		2.3	NS
	Vomiting	27	14.7	47		21.3	NS
	Diarrhea	20	10.9	49		22.2	.003 a
Stool rotavirus antigen	Negative	4	100.0	39		73.6	NS
	Positive	0	0.0	14		26.4
Brain image	Normal	5	83.3	24		75.0	NS
	Abnormal	1	16.7	8		25.0
EEG	Normal	55	78.6	100		78.7	NS
	Abnormal	15	21.4	27		21.3
Diagnosis	Febrile seizure	152	82.6	164		74.2	.001 a
	Epilepsy	12	6.5	18		8.1
	CNS infection	0	0.0	4		1.8
	Congenital anomaly	0	0.0	1		0.5
	Trauma	3	1.6	2		0.9
	Rotavirus related	0	0.0	14		6.3
	Unprovoked seizure	17	9.2	18		8.1
Febrile seizure	Yes	152	82.6	164		74.2	.042 a
	No	32	17.4	57		25.8
Type of febrile seizure	Simple	131	86.2	121		73.8	.008 a
	Complex	21	13.8	43		26.2
Anticonvulsant in ED	Yes	17	9.2	54		24.4	b.001 a

No.	Age	Sex	Physical and neurologic examination	Cause	Brain CT or MRI results
1 a	4 mo	Male	Normal	Asceptic meningitis	Mild ventricular enlargement with proportional widening of
2 a 3 a	4 y 8 mo 3 y 10 mo	Female Female	Normal Hyperreflexia	Epilepsy Epilepsy (remote	the cortical sulci Mild communicating hydrocephalus Focal brain tissue loss, left frontoparietal lobes
4 a	2 y 1 mo	Male	Normal	brain insult) Dandy-Walker	A large retrovermian CSF collection on the fourth ventricle
5 a 6 a	3 y 1 mo 1 y 4 mo	Female Female	Normal Normal	complex Epilepsy Complex febrile	Focal low attenuation in the right centrum semiovale A prominent enhancing vascular structure in the left frontal
7 b	1 y 1 mo	Male	Consciousness	seizure Meningoencephalitis	region Increased signal intensity on T2W1 and flair of MRI, left
8 b	5 mo	Male	disturbance Consciousness	Shaken baby	temporal lobes Subacute subdural hematomas overlying the bilateral
			disturbance retinal	syndrome	frontotemporoparietal regions
9 b	8 mo	Male	hemorrhage Consciousness	Encephalitis caused	Faint patchy areas of increased signal intensity on T2-weighted
			disturbance	by EV71 infection	image and flair predominantly involving the dorsal pons
EV indicates enterovirus. a Brain CT. b brain MRI.

Unprovoked seizure Rotavirus related

Trauma Congenital anomaly CNS infection

Epilepsy Febrile convulsion

8.1

9.2

6.3

0.0

0.9

1.6

0.5

0.0

1.8

0.0

8.1

6.5

74.2

82.6

POUD UIA

(n = 184) (n = 221)

0 20 40 60 80 100

%

Multivariate logistic regression analysis was developed using variables that revealed differences among the POUD, UIA, and RA groups in univariate analyses (P b .05). Two factors were associated with increased inpatient manage- ment: elevated CRP (adjusted OR, 1.464; 95% CI, 1.209- 1.773; P b .001) and use of first-line anticonvulsants at the ED (adjusted OR, 2.725; 95% CI, 1.462-5.079; P = .002).

Two factors were also associated with decreased inpatient management: age of 1 to 6 years (adjusted OR, 0.235; 95% CI, 0.085-0.649; P = .005) and a clinical diagnosis of febrile seizure (adjusted OR, 0.042; 95% CI, 0.006-0.308; P = .002) (Table 6).

Fig. 1 Details of diagnosis of patients in the UIA and POUD groups.

also administered to patients in the UIA group more frequently than in the POUD group (P b .001).

Laboratory data and treatments in patients with first attack of seizure

Laboratory evaluations showed a significant difference in potassium and CRP levels between the UIA and POUD groups. Multivariate logistic regression analysis was devel- oped using variables that showed significant differences between the POUD and UIA group in univariate analyses (P b .05). Two factors were associated with increased rate of UIA: elevated CRP (adjusted odds ratio [OR], 1.432; 95% confidence interval [CI], 1.187-1.728; P b .001) and first line of anticonvulsants administered in the ED (adjusted OR, 2.913; 95% CI, 1.478-5.743; P = .002). The

diagnosis of febrile seizure was associated with decreased rate of UIA (adjusted OR, 0.448; 95% CI, 0.230-0.872;

P = .018) (Table 5).

In the UIA group, the positive predictive value and negative predictive value of elevated serum CRP levels were 65.0% and 74.8%, respectively; 76.1% and 90.1%, respec- tively, for use of first-line anticonvulsants at the ED; and 52.1% and 17.4%, respectively, for a clinical diagnosis of febrile seizure. There were significant differences in sodium, potassium, and CRP levels between patients in the POUD group and those in the UIA and RA groups.

In the UIA and RA group, the positive predictive value and negative predictive value of elevated serum CRP levels were 85.1% and 62.0%, respectively; 81.0% and 74.8%, respectively, for use of first-line anticonvulsants at the ED; 75.4% and 10.9%, respectively, for age of 1 to 6 years; and 69.8% and 17.4%, respectively, for a clinical diagnosis of febrile seizure.

Discussion

Observation Units can provide an alternative to inpatient admission, continued patient management, reduction in costs, and decreased missed diagnoses and medicolegal risks [2,8]. Retrospective studies on the use of POUs report that patients have been treated successfully for common or selected clinical conditions, including asthma, acute gastro- enteritis with dehydration, and croup [9-12]. However, the role and the clinical value of a POU as an alternative to inpatient admission for patients with first seizure remain unclear. The previous literature have suggested that the POU inpatient admission rate is between 10.4% and 20.3% [9,13,14]. Scribano et al [9], in 2001, showed an overall POU inpatient 10.4% admission rate, and patients with seizures had the highest rate (18.8%).

In the present study, the UIA rate of patients with first seizures is 54.6% from the total 405 patients. Each UIA requires time to transfer patients’ care information to other health care providers, which inconveniences patients and families. However, there has been no study on the UIA rate

Table 5 Multivariate logistic regression analysis of factors associated with UIA
	(n = 405)	Total	UIA	%	OR	95% CI	P
Potassium	Mean +- SD	4.08 +- 0.52	4.15 +- 0.61		1.377	0.823-2.305	.223
CRP	Mean +- SD	1.27 +- 2.58	1.79 +- 3.33		1.432	1.187-1.728	b.001
Diarrhea	No	336	172	51.2	1.000
	Yes	69	49	71.0	1.728	0.893-3.344	.104
Febrile seizure	No	89	57	64.0	1.000
	Yes	316	164	51.9	0.448	0.230-0.872	.018
Anticonvulsant in ED	No	334	167	50.0	1.000
	Yes	71	54	76.1	2.913	1.478-5.743	.002

Table 6 Multivariate logistic regression analysis of factors associated with UIA and RA
	n = 910	Total	UIA + RA	%	OR	95% C.I.	P
Age (y)	b1	119	112	94.1	1.000
	1-6	667	503	75.4	0.235	0.085-0.649	.0050
	7-12	77	67	87.0	0.316	0.079-1.265	.104
	13-18	47	44	93.6	0.542	0.054-5.477	.604
Fever	No	332	300	90.4	1.000
	Yes	578	426	73.7	1.310E+08	0.000	.997
Headache	No	874	692	79.2	1.000
	Yes	36	34	94.4	3.121	0.348-28.001	.309
Vomiting	No	716	559	78.1	1.000
	Yes	194	167	86.1	1.250	0.693-2.254	.459
Diarrhea	No	755	591	78.3	1.000
	Yes	155	135	87.1	1.587	0.841-2.996	.154
Sodium	Mean +- SD	135.98 +- 4.21	136.10 +- 4.58		0.997	0.925-1.076	.947
Potassium	Mean +- SD	4.10 +- 0.55	4.12 +- 0.59		1.249	0.774-2.014	.362
CRP	Mean +- SD	1.25 +- 2.54	1.42 +- 2.82		1.464	1.209-1.773	b.001
Febrile seizure	No	406	374	92.1	1.000
	Yes	504	352	69.8	0.042	0.006-0.308	.002
Anticonvulsant in ED	No	718	551	76.7	1.000
	Yes	192	175	91.1	2.725	1.462-5.079	.002

of patients with first seizure or the definite risk factors associated with increased UIA. multiple logistic regression analysis in this study indicates that elevated CRP and first- line anticonvulsant use in the ED are significantly associated with increased UIA rate. These 2 factors are also significantly associated with an increase in the need for ward or ICU admission in the total 910 patients.

Serum CRP, an acute-phase protein, rises in an acute inflammation, such as bacterial or some viral infections. Moreover, CRP may show an increase in serial serum levels after 12 to 24 hours of severe inflammation. In the present study, decision as admitting or discharging a patient is a multifactorial process. Elevated serum CRP levels may indicate severe inflammation or infection. The underlying causes of febrile seizure also play an important role in such decision making for patients with seizures. Based on the results here, febrile seizure is the major etiology of children with first attack of seizure in the ED, whereas the most common underlying disease is URI (45%), followed by acute gastroenteritis (19%). Serum CRP levels in patients with URI or acute gastroenteritis appear normal or slightly elevated. In the POUD group, 152 (82.6%) of 184 patients with febrile seizure were discharged from the hospital uneventfully, and most of them had normal serum CRP levels. Therefore, analyzing CRP in patients with first febrile seizure will be very helpful.

Anticonvulsant use in the ED may indicate that patients have prolonged or recurrent seizure attacks, which do not stop spontaneously. With such clinical condition, potential risks of life-threatening sequelae such as hypoxia may increase. However, age between 1 and 6 years and a clinical diagnosis of febrile seizure significantly reduce the need for inpatient admission. In addition, patients diagnosed as

having febrile seizure and admitted to the POU have a lower rate of UIA. The duration of POU stay in first seizures has not been well defined in literature. In the present study, the duration of POU stay is 21.38 +- 8.22 hours in the POUD group with no severe complication noted after discharge. A 24-hour observation in a POU may be an alternative recommendation for patients who present with first seizure but without any indication of immediate admission. These findings also suggest that every child with first seizure attack should have specific Laboratory workup.

Before the age of 5 years, 2% to 5% of children will experience at least 1 febrile seizure, which is the most common type of first seizure in young children [7,15]. In a previous study, febrile seizure (62.1%) is the main etiology of a first attack of seizure in the ED, and URI and systemic viral infections (45%) are the main underlying diagnoses in this group of children [16]. In the current study, the male-to-female ratio in first seizure is 1.29:1 with statistical significance in male sex among the POUD, UIA, and RA groups. This may be because the most common diagnosis in this study is febrile seizure, and previous Case-control studies have found a 1.4:1 male-to-female ratio in febrile seizure [17].

A retrospective cohort study has reported that diarrhea is a highly associated symptom in patients with seizures and that rotavirus is an identified etiology [18-22]. In the present study, rotavirus is confirmed by stool antigen analysis in 26% of patients with seizure and diarrhea, which is more frequent in patients in the UIA group than in the POUD group (P = .003). Rotavirus-related seizures may be associated with Severe dehydration or recurrent seizure attacks and an increase in the need for inpatient services. However, by multivariate logistic regression analysis, there is no significant increase in the risk of UIA.

In young children, Bacterial meningitis can be the cause of seizure and is often associated with poor outcomes [23,24]. Once CNS infection is suspected, children should be admitted to the hospital for further evaluation and treatment. The American Academy of Pediatrics recommends lumbar puncture as a consideration in patients younger than 18 months who present with febrile seizure [25]. In the present study, lumbar puncture has been performed in 21 of the 405 patients admitted to the POU with progressive abnormal findings in Neurologic examinations, recurrent seizures combined with fever, and persistent headache and vomiting, but without a definite organism cultured or isolated. Of these 21 patients, meningoencephalitis is diagnosed in 2 patients, encephalitis in 1, and Aseptic meningitis in another. All 4 patients have abnormal findings in neurologic examinations (Consciousness disturbance or drowsiness) during their POU stay, and 3 are younger than 18 months. Therefore, this study suggests that Lumbar punctures be performed based on detailed examination rather than routinely in children with febrile seizure.

Brain imaging studies may help emergency physicians identify possible causes of first-attack seizures [26]. In the current study, 38 patients admitted to the POU underwent head CT or brain MRI examinations, and 9 with abnormal imaging findings were in the infant or preschool age groups. There were no CT abnormalities in 6 patients who presented with complex febrile seizures. It has been reported that patients presenting with complex febrile seizures may not be at increased risk for abnormalities in brain imaging studies [16,26]. Therefore, emergency brain imaging should be arranged for children who present with first seizure and with known head injury, abnormal physical and neurologic examinations, and acute symptomatic causes other than fever. Because of the potential risks of recurrent seizure and life- threatening events such as bacterial meningitis or encepha- litis, young children who present with new-onset seizure should be admitted to the POU or be an inpatient for further observation or survey even if they initially show normal physical and neurologic examinations or diagnosed as

febrile seizure.

The present study has some limitations. First, physicians made the diagnosis of seizure based on the presenting seizure activity at the ED or on descriptions of witnesses. Second, although all of the patients in this study were managed and evaluated by pediatric emergency attending physicians who decided where to admit the patients based on major indications, there might be some bias among physicians in making clinical judgment. Third, in a retrospective single- center review of medical records, some details of history and physical examination may not be rigorously documented. Such limitations may cause some bias in analyzing the first seizure attack in the ED.

In conclusion, age, use of anticonvulsants in the ED, serum CRP, and a clinical diagnosis of febrile seizure are important factors for determining whether children with a first attack of seizure should be admitted or discharged. A

POU stay may provide an alternative to immediate admission in selected patients, especially in those who present with first febrile seizure without evidence of severe bacterial infection.

Acknowledgment

The study was partly funded by grants from the China Medical university hospital (DMR-100-167).

References

1. Wiley 2nd JF, Friday JH, Nowakowski T, et al. Observation units: the role of an out-patient extended treatment site in pediatric care. Pediatr Emerg Care 1998;14:444-7.
2. Mace SE. Pediatric observation medicine. Emerg Med Clin North Am 2001;19:239-54.
3. American College of Emergency Physicians. Management of observation units. Ann Emerg Med 1988;17:1348-52.
4. Roth HL, Drislane FW. Seizures. Neurol Clin 1998;16:257-84.
5. Friedman MJ, Sharieff GQ. Seizures in children. Pediatr Clin North Am 2006;53:257-77.
6. Commission on Epidemiology and Prognosis, International League against Epilepsy. Guideline for epidemiologic studies on epilepsy. Epilepsia 1993;34:592-6.
7. Academy of Pediatrics. Provisional Committee on Quality Improve- ment, Subcommittee on Febrile Seizures. Practice parameter: the neuro-diagnostic evaluation of the child with a first simple febrile seizure. Pediatrics 1996;97:769-72.
8. Kemper KJ. Medically inappropriate hospital use in a pediatric population. N Engl J Med 1988;318:1033-7.
9. Scribano PV, Wiley II JF, Platt K. Use of an observation unit by a pediatric emergency department for common pediatric illnesses. Pediatr Emerg Care 2001;17:321-3.
10. Miescier MJ, Nelson DS, Firth SD, et al. Children with asthma admitted to a pediatric observation unit. Pediatr Emerg Care 2005;21: 645-9.
11. Mallory MD, Kadish H, Zebrack M, et al. Use of a pediatric observation unit for treatment of children with dehydration caused by gastroenteritis. Pediatr Emerg Care 2006;22:1-6.
12. Greenberg RA, Dudley NC, Rittichier KK. A reduction in hospital- ization, length of stay, and hospital charges for croup with the institution of a pediatric observation unit. Am J Emerg Med 2006;24: 818-21.
13. Alpern ER, Calello DP, Windreich R, et al. Utilization and unexpected hospitalization rates of a pediatric emergency department 23-hour observation unit. Pediatr Emerg Care 2008;24:589-94.
14. Zebrack M, Kadish H, Nelson D. The pediatric hybrid observation unit: an analysis of 6477 consecutive patient encounters. Pediatrics 2005;115:e535-42.
15. McAbee GN, Wark JE. A practical approach to uncomplicated seizures in children. Am Fam Physician 2000;62:1109-16.
16. Chen CY, Chang YJ, Wu HP. New-onset seizures in pediatric emergency. Pediatr Neonatol 2010;51:103-11.
17. Pal DK, Kugler SL, Mandelbaum DE, et al. Phenotypic features of familial febrile seizures: case-control study. Neurology 2003;60: 410-4.
18. Zerr DM, Blume HK, Berg AT, et al. Non-febrile illness seizures: a unique seizure category? Epilepsia 2005;46:952-5.
19. Lin SC, Hsu HY, Wang PJ, et al. Rotavirus gastroenteritis associated with afebrile seizure in childhood. Acta Paediatr Sin 1996;37:204-7.
20. Contino MF, Lebby T, Arcinue EL. Rotaviral gastrointestinal infection causing afebrile seizures in infancy and childhood. Am J Emerg Med 1994;12:94-5.
21. Wang YC, Hung KL. Benign seizures associated with mild diarrhea: clinical analysis of 20 cases. Acta Paediatr Sin 1993;34:451-7.
22. Narchi H. Benign afebrile cluster convulsions with gastroenteritis: an observational study. BMC Pediatr 2004;4:2.
23. Goetghebuer T, West TE, Wermenbol V, et al. Outcome of meningitis caused by Streptococcus pneumoniae and Haemophilus

influenzae type b in children in The Gambia. Trop Med Int Health 2000;5:207-13.

1. Inoue S, Nakazawa T, Takahashi H, et al. Seizures in the acute phase of aseptic and bacterial meningitis. No To Hattatsu 1998;30:494-9.
2. Leung AKC, Robson WLM. Febrile seizures. J Pediatr Health Care 2007;21:250-5.
3. Maytal J, Krauss JM, Novak G, et al. The role of brain computed tomography in evaluating children with new onset of seizures in the emergency department. Eplilepsia 2000;41:950-4.