Central nervous system infections and status epilepticus association, clinical profile and outcomes in children hospitalized with new-onset acute symptomatic seizures

Abstract

This study investigated the clinical characteristics and risk factors for poor clinical outcomes among children presenting with a first seizure episode to the Emergency Department (ED). This study further assessed the association between convulsive status epilepticus (CSE) and central nervous system (CNS) infections. This was a single-center, cross-sectional study of children admitted to a tertiary pediatric hospital in southern Vietnam between June and October 2024. Propensity score matching (PSM) was applied to balance the baseline characteristics, and logistic regression models were used to identify the determinants of poor clinical outcomes. Of the 2728 ED admissions, 425 children with new-onset seizures met the inclusion criteria. CSE occurred in 52 patients (12.2%). Abnormal cerebrospinal fluid findings suggestive of CNS infections were identified in 49/131 (37.4%) patients who underwent lumbar puncture. Poor outcomes, defined as death or neurological sequelae at discharge, were observed in 11.8% of hospitalized children. Multivariable logistic regression with PSM adjustment demonstrated a significant association between CSE and CNS infections (adjusted odds ratio 30.9 [95% confidence interval 12.4–77.4]; P < .001). The determinants of poor outcomes included focal seizure onset, lower Glasgow Coma Scale scores at admission, delayed presentation from seizure onset to hospitalization, and confirmed CNS infections. These findings highlight the substantial burden of severe neurological complications in children presenting with first-time seizures and underscore the critical role of timely recognition of CSE and CNS infections. This study, based on data from resource-limited hospitals, strengthens diagnostic pathways and informs early management strategies for acute symptomatic seizures in resource-limited settings.

Key points.

• Among children admitted with the first seizure attack, convulsive status epilepticus accounted for 12.2%, and 37.4% of patients had abnormal cerebrospinal fluid results, consistent with central nervous system (CNS) infections.

• At discharge, 11.8% of the patients experienced poor clinical outcomes, including death, survival with neurological sequelae and disability.

• Using propensity score matching for baseline confounders, we identified a significant association between CNS infections and convulsive status epilepticus.

• Determinants for poor clinical outcomes in children hospitalized for new-onset seizures included focal seizures, lower Glasgow Coma Scale scores at admission, prolonged intervals from seizure onset to hospitalization, and CNS infections.

1. Introduction

Seizures are characterized by transient manifestations due to abnormal, synchronous or excessive activity of nerve cells in the brain.^[1] Seizures are frequently observed neurological symptoms among hospitalized children, with a reported prevalence of approximately 18.3% of all admissions.^[2] Seizures can be classified as acute symptomatic seizures, which occur in temporal association with an acute insult to the central nervous system (CNS), or as unprovoked seizures associated with epilepsy. Acute symptomatic seizures account for approximately 40% of afebrile seizures and have an annual incidence of 29 to 39 per 100,000 individuals.^[3,4] Acute symptomatic seizures are associated with higher odds of mortality than unprovoked seizures.^[5] Determining the underlying etiology of seizures is essential, as appropriate treatment of the precipitating cause is critical for seizure control and prevention of recurrence. The etiology of acute symptomatic seizures varies with age. In neonates, common causes include hypoxic-ischemic encephalopathy, cerebral infarction, and intracranial hemorrhage.^[6] In infants and young children, CNS infections, head trauma, and electrolyte disturbances predominate, with infections accounting for up to 82% of cases.^[7] Brain imaging abnormalities are identified in approximately 62.8% of patients with acute symptomatic seizures,^[7] whereas cerebrovascular disease is the leading cause in older adults.

Approximately 12.5% of acute symptomatic seizures can progress to convulsive status epilepticus (CSE).^[7] CSE may result in brain damage, long-term sequelae, and death.^[8] A high prevalence of CNS infections among pediatric patients with acute seizures has been reported, ranging from 80% to 82% in Kenyan and Indian cohorts, respectively.^[2,7] Importantly, CNS infections have greatly contributed to an increase in the mortality rate by approximately 21 to 25% among children admitted to the pediatric intensive care unit.^[9–11] Common short-term complications among survivors include cerebral edema and hemorrhage, while long-term neurological outcomes are remarkable, including epilepsy, cranial nerve palsies, hydrocephalus, learning disabilities, and behavioral disorders.^[9–11] Despite the high prevalence of CNS infections among children with CSE, robust evidence of their association remains limited, particularly in resource-constrained settings.^[12–15] Previous studies have been constrained by their retrospective designs, limited statistical power, and inadequate adjustment for potential confounding factors.^[12–15] Therefore, this study aimed to determine the prevalence and associations of CSE and CNS infections among hospitalized children with new-onset acute symptomatic seizures. We also characterized the clinical features and outcomes and identified factors associated with poor clinical outcomes. Using data from resource-limited hospitals, this study provides evidence to strengthen diagnostic pathways and facilitate the timely management of acute symptomatic seizures, with the potential to improve outcomes in resource-limited settings.

2. Methods

2.1. Study design, setting and participants

This single-center, cross-sectional study was conducted at Children’s Hospital 2, a large tertiary referral pediatric institution in southern Vietnam. We enrolled all children admitted to the Emergency Department (ED) with an acute seizure episode between June and October 2024. The eligibility criteria were age <16 years and new-onset acute symptomatic seizures.^[16] Patients were excluded from the study if they met any of the following criteria: a previous diagnosis of epilepsy, or a documented history of neurological disorders, including developmental delay, cerebral palsy, or permanent neurological sequelae resulting from prior brain injury.

2.2. Study outcomes

The primary study outcomes were the prevalence of status epilepticus, CNS infections, and poor clinical outcomes in children admitted to the ED. The clinical outcomes were evaluated at the time of discharge. Poor clinical outcomes were defined as a composite of in-hospital death, anticipated death following discharge against medical advice due to a terminal illness, and survival with neurological sequelae or disability. Neurological sequelae/disability were defined as the presence of new persistent neurological deficits at discharge, including cognitive impairment, focal limb weakness or paralysis, spastic quadriparesis, or other clinically significant motor or neurological impairments attributable to acute symptomatic seizures or underlying diseases.

2.3. Study definitions

Acute symptomatic seizure was defined as an episode of clinical seizure arising in close temporality with an acute CNS insult, which may be metabolic, toxic, structural, infectious, or inflammatory in origin.^[16,17] CSE was defined as continuous generalized seizures lasting more than 30 minutes or shorter duration with recurrent seizures in the absence of consciousness recovery between seizures.^[16]

CNS infections include meningitis, meningoencephalitis, encephalitis, and brain abscesses. Cases were classified according to the level of diagnostic certainty as confirmed, probable, or possible CNS infection.^[18,19]

• Confirmed CNS infection was defined as the identification of a causative pathogen in the cerebrospinal fluid (CSF) using Gram staining, culture, or polymerase chain reaction (PCR).

• Probable CNS infection was defined by a compatible clinical presentation (e.g., meningeal signs, seizures, and/or altered consciousness) together with CSF findings consistent with CNS infection, including pleocytosis (white blood cell count >30 cells/mm³ in neonates or >10 cells/mm³ in children beyond the neonatal period), elevated protein concentration (>1.0 g/L in neonates or >0.45 g/L in older children), decreased CSF glucose (<2.2 mmol/L), or a CSF-to-serum glucose ratio <0.5 in the absence of microbiological confirmation.

• Possible CNS infection was defined as a compatible clinical syndrome suggestive of CNS infection without microbiological confirmation or characteristic CSF abnormalities, but supported by neuroimaging, electroencephalography (EEG), or the treating physician’s clinical assessment after the exclusion of alternative diagnoses.

Meningitis was diagnosed according to the above criteria,^[18] while meningoencephalitis and encephalitis were diagnosed when meningitis was accompanied by clinical evidence of brain parenchymal involvement, such as altered mental status, focal neurological deficits, or compatible neuroimaging or EEG findings.^[19]

In addition, encephalitis diagnosis was made based on a combination of clinical, laboratory, neuroimaging, and EEG findings, and established with 1 main criterion required plus at least 2 minor criteria^[19,20]:

• Major criterion: Patients with mild to severe altered consciousness lasting ≥24 hours, including drowsiness, lethargy, agitation, confusion, coma, or changes in behavior and personality, without alternative causative explanations.

• Minor criteria: Fever or history of fever ≥38^oC during the illness; seizures and/or new-onset focal neurological signs; CSF leukocyte count ≥5 cells/mm³; EEG abnormalities consistent with encephalitis; and radiological findings (computed tomography scan and/or magnetic resonance imaging [MRI]) of the brain parenchyma indicative of encephalitis.

2.4. Indications for neurological investigations:

Lumbar puncture, neuroimaging (MRI), and electroencephalography were performed only when clinically indicated, rather than routinely. Lumbar puncture was considered in children with suspected CNS infection after excluding any clinical contraindications. Neuroimaging was performed for focal neurological features, prolonged or recurrent seizures, CSE, persistently impaired consciousness, suspected increased intracranial pressure, recent head trauma, or suspected structural brain lesions. EEG was performed at the treating clinician’s discretion, primarily for focal seizures, recurrent seizures, suspected epilepsy, persistent altered consciousness, and possible nonconvulsive seizures.

2.5. Data measurements and collection

Case report forms were manipulated to prospectively gather clinical and laboratory data from medical records upon admission to the ED and post-admission. The collected data were entered into an electronic database for further analysis. The Glasgow Coma Scale (GCS) was used to assess the consciousness of patients upon admission.^[21] Additionally, the clinical outcomes of the patients were scrutinized during their stays in the Emergency and Treatment Departments and at hospital discharge.

2.6. Statistical analysis

Continuous variables were described using the median and interquartile range (IQRs), while categorical variables were presented as numerical counts (n) and corresponding percentages (%). Missing data on the main known risk factors for acute symptomatic seizures were considered the primary source of bias (Figure S1, Supplemental Digital Content 1). In this study, the high proportion of missing data (>50%) in metabolic evaluations and neuroimaging may have introduced selection bias, as patients receiving complete workups may systematically differ from those without workups in terms of disease severity and access to diagnostic resources. Such bias could affect both the distribution of the identified etiologies and the clinical outcomes. In this respect, complete case analyses were performed. PSM was performed to reduce confounding effects by balancing the baseline characteristics of children with and without CNS infections. Covariates were selected a priori based on clinical relevance and potential association with both CNS infection and seizure severity. These included age, white blood cell count, GCS score at admission, and interval between seizure onset and hospital presentation. Age reflects differences in the epidemiology of pediatric seizures and CNS infections, whereas white blood cell counts and GCS score serve as markers of disease severity. The time from seizure onset to hospital presentation was included because delayed evaluation may influence both the detection of CNS infection and the risk of CSE. PSM was implemented using the nearest-neighbor method (Figure S2, Supplemental Digital Content 2). Propensity scores were estimated using logistic regression analysis. Participants were matched 1:1 by nearest-neighbor matching without replacement using a caliper of 0.2 SD of the logit of the propensity score. Of the 52 treated participants and 371 controls before matching, 24 pairs were successfully matched, leaving 28 treated participants and 347 unmatched controls. Covariate balance was evaluated using standardized mean difference (Supplementary Data, Supplemental Digital Content 3). Based on the PSM cohort, bivariate and multivariable logistic regressions were performed to identify the determinants of poor clinical outcomes. Complete case backward stepwise model selection based on the Akaike information criterion (AIC), Least Absolute Shrinkage and Selection Operator, and ridge regression were employed to determine the most significant risk factors. Statistical significance was set at P < .05 for all comparisons. R statistical software (version 4.3.3, Boston) was used for all the analyses.

2.7. Ethics statement

This study was approved by the Institutional Review Board of the Children’s Hospital 2, Ho Chi Minh City, Vietnam, with the approval number 517/GCN-BVND2, signed on 17 June, 2024. Informed consent was obtained from the patients’ parents. This study strictly adhered to the principles of Good Clinical Practice and the ethical guidelines of the Declaration of Helsinki.

3. Results

3.1. Baseline characteristics of patients’ cohort on hospital admission

We identified 425 eligible children with acute symptomatic seizures among 2728 children admitted to the ED during the study period (Fig. 1). Table 1 shows the clinical characteristics of patients with new-onset seizures upon admission to the ED and their outcomes at discharge. The median patient age was 1.8 years (interquartile range, IQR: 1.1–3.8), and a large proportion (61.7%) of the participants were aged 1 to 5 years. Males accounted for 56% of participants. The median duration from disease onset to hospital admission was 2 days (IQR, 1–2). Of the 425 patients, 407 (95.8%, 95% CI: 93.5–97.3%) experienced generalized seizures, while 18 (4.2%, 95% CI: 2.7–6.5%) experienced focal motor seizures. Febrile seizures were observed in 83.3% of participants. CSE accounted for 12.2% of all cases. Notably, 75 patients (17.7%) presented with altered mental status (GCS score < 14) upon hospital admission. Only 1 patient had hypoglycemia (< 50 mg/dL), and the remaining patients had normal or mildly elevated blood glucose levels. A total of 69 patients experienced electrolyte disturbances, including hyponatremia (5/375, 1.3%), hypocalcemia (14/371, 3.8%), and hypomagnesemia (50/370, 13.5%). Metabolic biomarkers (blood lactate and ammonia) were measured in 44 patients, and the results were within the normal reference ranges. Of the 125 patients with new-onset seizures who underwent electroencephalogram (EEG) examination, 59/125 (47.2%) had abnormal EEG results. Nevertheless, there were no statistical differences in abnormal EEG findings between patients with and without CSE (63% vs 43%, P = .09) and those with 1 or multiple episodes of seizures (27% vs 50%, P = .10). The median length of hospital stay was 4 days (IQR, 3–7).

Figure 1.

Flow chart of study participants.

Table 1.

Clinical characteristics and outcomes at discharge of patients with new-onset seizures upon admission to the ED.

Characteristics	No. participants	Statistics
Median patient age, (yr)	425	1.8 (1.1–3.8)
Patient age group	425
<1 yr		98 (23.1)
1–5 yr		262 (61.7)
>5 yr		65 (15.3)
Male sex, n (%)	425	236 (56)
Duration of illness, (d)	425	2 (1–2)
Type of seizures, n (%)	425
Generalized		407 (95.8)
Focal motor		18 (4.2)
Febrile seizure, n (%)	425	354 (83.3)
Convulsive status epilepticus, n (%)	425	52 (12.2)
Median GCS score (lowest-highest levels)	425	15 (8–15)
Altered mental status (GCS score <14), n (%)	425	75 (17.7)
White blood cell count, (x 109/L)	423	10.8 (8–14)
Hemoglobin, (g/dL)	423	11.8 (10.9–12.6)
Platelet count, (x 109/L)	423	285 (229–355)
Aspartate aminotransferase, (IU/L)	370	43 (35–53)
Alanine aminotransferase, (IU/L)	370	19 (14–26)
Hypoglycemia (< 50 mg/dL), n (%)	425	1 (0.2)
Electrolyte disturbances, n (%)
Hyponatremia	375	5 (1.3)
Hypocalcemia	371	14 (3.8)
Hypomagnesemia	370	50 (13.5)
Serum ammonia level, (µg/dL)	44	44 (36–51)
Blood lactate, (mmol/L)	44	1.6 (1.2–2.3)
Serum creatinine, (µmol/L)	370	37 (32–45)
Abnormal EEG, n (%)	125	59 (47.2)
Abnormal neuroimaging findings, n (%)	138	60 (43.5)
Clinical presentations of CNS infections, n (%)	425	30 (7.1)
Abnormal findings on CSF analysis, n (%)	131	49 (37.4)
Length of hospital stay, (d)	425	4 (3–7)
Outcomes of patients at discharge, n (%)	425
Good outcome-alive with no sequelae		375 (88.2)
Poor outcomes		50 (11.8)
Died in-hospital		10 (2.4)
Discharged and died at home		2 (0.5)
Alive with sequelae		38 (8.9)

Summary statistics are presented as median (interquartile range, IQR) for continuous variables and frequency (%) for categorical variables.

CNS = central nervous system, CSF = cerebrospinal fluid, ED = emergency department, EEG = electroencephalogram, GCS = Glasgow Coma Scale.

3.2. Patients’ outcomes at hospital discharge

Overall, 50 (11.8%) patients had poor clinical outcomes at discharge, including in-hospital death (2.4%), anticipated death following discharge at home (0.5%), and survivors with neurological sequelae (8.9%). Subgroup analysis showed that poor outcomes occurred in 6 of 18 children with focal seizures (33.3%, 95% CI 13.3–59.0) compared with 44 of 407 children with generalized seizures (10.8%, 95% CI 8.0–14.4). Focal seizures were associated with poor clinical outcomes (odds ratio [OR] 4.13, 95% CI 1.38–11.2; P = .012).

3.3. Brain imaging findings

Among the 139 patients who underwent brain radiology, including transfontanelle sonography, CT scan, and MRI, 60/139 (43%) participants among the patients who underwent MRI showed abnormal brain imaging findings (Table 2). Of the 9 infants who underwent transfontanelle sonography, 3 had aberrant findings, including cerebral venous sinus thrombosis, cerebral hemorrhage, and brain abscesses. Common aberrant findings on brain CT scans included cerebral hemorrhage, hypodensity of the cerebral parenchyma, ventricular enlargement, brain tumors, and edema. Notably, encephalitis was the most common brain abnormality on MRI, as observed in 13/139 (9.3%) patients, followed by enlargement and bleeding of the ventricular and subarachnoid spaces, microcephaly, hypoxic-ischemic encephalopathy, and other congenital brain diseases.

Table 2.

Brain radiological findings of patients admitted with new-onset seizure (n = 139).

Brain imaging findings	Statistics
Transfontanelle ultrasonography (n = 9)
Normal findings	6 (66.7)
Brain abscess	1 (11.1)
Cerebral hemorrhage	1 (11.1)
Cerebral venous sinus thrombus	1 (11.1)
CT-Scan (n = 87)
Normal findings	55 (63.2)
Cerebral hemorrhage	11 (12.6)
Hypodensity of cerebral parenchyma	6 (6.9)
Enlargement of ventricles	3 (3.4)
Brain edema	3 (3.4)
Brain tumor	2 (2.3)
Bilateral injury of lentiform nucleus	1 (1.1)
Enlargement of subarachnoid space	1 (1.1)
Subarachnoid cyst	1 (1.1)
Tuberculous meningitis	1 (1.1)
Hypoxic-ischemic encephalopathy	1 (1.1)
MRI (n = 52)
Normal findings	25 (48.1)
Encephalitis	13 (25)
Enlargement of ventricles	5 (9.6)
Bleeding of subarachnoid space	2 (3.8)
Subarachnoid cyst	2 (3.8)
Enlargement of subarachnoid space	1 (1.9)
Cerebral malacia	1 (1.9)
Bilateral hippocampal sclerosis	1 (1.9)
Microcephalus	1 (1.9)
Hypoxic-ischemic encephalopathy	1 (1.9)

Summary statistics are frequency and percentage (%).

CT-Scan = computed tomography scan, MRI = magnetic resonance imaging.

3.4. CNS infections and cerebrospinal fluid analysis

Thirty patients presented with clinical manifestations of CNS infection, and 92 patients had clinical signs of infection and dramatically elevated white blood cell counts upon ED admission. Overall, 131 patients underwent lumbar puncture and CSF analysis, of whom 49/131 (37.4%) showed abnormal CSF findings consistent with encephalitis and meningitis (Table 3). The median CSF leukocyte count was 06 (IQR 3–30) cells/mm³. Leukocytosis in the CSF (>10 cells/mm³) was present in 39/131 (29.8%) patients. The median CSF protein level was 0.45 g/L (IQR 0.27–0.69). The median CSF glucose concentration was 3.7 mmol/L (IQR 3.3–3.8). Five patients had CSF samples which were positive for N-methyl-D-aspartate receptor (NMDAR) antibodies. PCR of the CSF revealed pathogens of acute viral encephalitis in 4 patients (one patient with herpes simplex virus and 3 patients with Japanese encephalitis virus).

Table 3.

Cerebrospinal fluid analysis of patients undergoing lumbar puncture (n = 131).

CSF characteristics	Statistics
No. patients with abnormal CSF analysis	49/131 (0.37)
White cell counts (cells/mm3)	6 (3–30)
White cell count >10 cells/mm3, n (%)	39 (29.8)
CSF protein, (g/L)	0.45 (0.27–0.69)
CSF glucose, (mmol/L)	3.7 (3.3–3.8)
Positive autoimmune brain NMDAR antibody, n (%)	5/131 (3.8)
Polymerase chain reaction test, n (%)	4/131 (3.1)
Herpes simplex virus	1 (0.8)
Japanese encephalitis virus	3 (2.2)

Summary statistics are median (IQR) for continuous data, and frequency (%) for categorical data.

CSF = cerebrospinal fluid, NMDAR = N-methyl-D-aspartate receptor.

3.5. Associations between CNS infections and CSE

As indicated in Table 4, univariate and adjusted multivariable logistic analyses showed strong associations between CNS infections and CSE in children with new-onset seizures admitted to the ED. CNS infections had a greater odds of status epilepticus occurrence (OR, 31.6 [95% CI, 12.6–79.2], P < .001). After adjustment by propensity score matching, CNS infections continued to be strongly associated with CSE, with 30.9 times greater odds (OR 30.9 [95% CI, 12.4–77.4], P < .001).

Table 4.

Associations of CNS infections and convulsive status epilepticus in hospitalized children with the new-onset seizures.

Statistical methods	Bivariate analysis*		Multivariable analysis†
	OR (95% CI)	P	OR (95% CI)	P
Without PSM	33.27 (13.7–81.1)	<0.001	31.6 (12.6–79.2)	<0.001
With PSM	32.6 (13.4–79.2)	<0.001	30.9 (12.4–77.4)	<0.001

OR = odds ratio, 95% CI = confidence interval, CNS = central nervous system, PSM = Propensity score matching method.

PSM was performed to balance the baseline confounders including patient age, white blood cell counts, Glasgow Coma Score at admission, and the interval (days) between seizure onset and hospital presentation.

Before PSM, 52 participants with CSE and 371 controls were included. Logistic regression used the full cohort; 1:1 PSM matched 24 CSE participants to 24 controls.

^*Bivariate logistic analyses.

^†Multivariable logistic analyses adjusted by electrolyte disturbances and newly diagnosed epilepsy.

3.6. Risk factors for poor clinical outcomes in children hospitalized with the first attack of acute symptomatic seizures

As shown in Table 5, the main risk factors associated with poor clinical and neurological outcomes included focal seizures (compared with generalized seizures), CNS infections, prolonged interval (days) from seizure onset to hospitalization, and lower Glasgow Coma Scale scores upon admission to the ED. Notably, the AIC-based model failed to reveal that CSE was associated with poor outcomes in patients with new-onset seizures. However, CSE was identified as a significant risk factor for CNS infections after PSM adjustment for confounding factors.

Table 5.

Factors associated with poor clinical outcomes in children hospitalized with new-onset seizures.

Variables	Bivariate analyses*		Multivariable model 1†		Multivariable model 2‡
	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Age (yr)	1.12 (1.04–1.21)	<.01	–	–	–	–
Female patients (compared to males)	1.12 (0.61–2.04)	.72	–	–	–	–
Presence of underlying diseases (yes)	2.69 (0.92–7.92)	.07	–	–	–	–
Time between seizure onset and hospital presentation, (d)	1.94 (1.34–2.82)	<.001	1.34 (1.07–1.68)	.01	1.35 (1.07–1.71)	.01
GCS score at admission, (+1 point)	0.27 (0.2–0.36)	<.001	0.3 (0.2–0.47)	<.001	0.28 (0.2–0.4)	<.001
Focal seizure (compared to generalized)	3.93 (1.38–11.2)	.01	11.4 (2.83–46.2)	<.001	11.5 (2.85–46.1)	<.001
Convulsive status epilepticus (yes)	24.6 (11.9–51.1)	<.001	1.36 (0.45–4.11)	.58	–	–
Electrolyte disturbances (yes)	2.9 (1.48–5.69)	<.01	–	–	–	–
Hyponatremia	0.1 (0.02–0.59)	.01	–	–	–	–
Hypocalcemia	0.25 (0.08–0.79)	.02	–	–	–	–
Hypomagnesemia	0.48 (0.23–1.01)	.05	–	–	–	–
Abnormal neuroimaging findings§, (yes)	9.35 (3.84–22.7)	<.001	–	–	–	–
Newly diagnosed epilepsy upon this admission (yes)	2.45 (1.22–4.92)	.01	–	–	–	–
CNS infections, (yes)	26.0 (10.9–62.1)	<.001	2.99 (1.05–8.53)	.04	3.05 (1.08–8.63)	.03

Complete case analysis was performed for bivariate and multivariable logistic analyses.

OR = odds ratio, 95% CI = confidence interval, CNS = central nervous system, GCS = Glasgow Coma Scale.

^*Bivariate logistic regression analyses (n = 425).

^†Model 1: Multivariable logistic analysis with variable selection by Lasso regression (n = 370).

^‡Model 2: Multivariable logistic analysis with variable selection by Akaike Information Criterion (n = 370).

^§Abnormal neuroimaging data were available on 138/425 patients; hence, they were not analyzed in the multivariable analyses.

4. Discussion

CSE and CNS infections are major contributors to mortality and neurological disability in children, accounting for approximately 15% and 25%, respectively, among pediatric patients hospitalized in the ED and pediatric intensive care units.^[8–11] In the present study, 11.8% of children with new-onset seizures experienced poor clinical outcomes. We identified key factors independently associated with poor clinical outcomes in this population. This study provides novel evidence from a real-world cohort of children hospitalized with new-onset acute symptomatic seizures at a tertiary referral hospital in a resource-limited setting. Unlike previous studies that have primarily focused on specific seizure syndromes or high-income settings, our study comprehensively characterizes the epidemiology, clinical features, burden of CSE and CNS infections, and predictors of poor clinical outcomes within the context of routine clinical practice, where access to advanced diagnostic investigations is often limited. These findings address an important evidence gap by providing locally relevant data that can inform risk stratification, strengthen diagnostic pathways, and guide early management strategies for children with acute symptomatic seizures in resource-limited healthcare settings.

To date, data on the epidemiology, clinical characteristics, outcomes, and risk factors of acute symptomatic seizures, CSE, and CNS infections among children hospitalized with a first episode of acute symptomatic seizures remain limited, particularly in resource-constrained settings. Therefore, this study aimed to characterize the burden of acute symptomatic seizures, CSE, and CNS infections at a tertiary pediatric referral hospital in southern Vietnam. In this study, the prevalence of acute symptomatic seizures was 15.6% of pediatric ED admissions. This prevalence was higher than that reported in studies from a general pediatric hospital in Mali and an urban tertiary hospital ED in Tanzania.^[22,23] Despite the higher prevalence, the mortality rate in our cohort (2.9%) was substantially lower than that reported in the African cohorts.^[22,23] The differences observed between our cohort and those from Mali and Tanzania may reflect variations in disease epidemiology and patient demographics. Mali and Tanzania are malaria-endemic countries in Africa, where cerebral malaria remains a major cause of acute neurological complications and acute symptomatic seizures. In contrast, malaria transmission in Vietnam has declined substantially, with sporadic cases occurring mainly in highland and forested areas bordering Cambodia and Laos.^[24,25] Significantly, severe dengue complications continue to impose a substantial burden on public health in Vietnam.^[26–29] In addition, our cohort included a broader age range, from neonates to adolescents (≤16 years), than that of the African cohorts.^[22,23] Collectively, these differences in the epidemiological context and age distribution are likely to influence the underlying etiologies of acute symptomatic seizures.

In this study, febrile seizures were most frequently observed in children aged 1 to 6 years, and generalized tonic-clonic seizures and recurrent seizures upon admission were prominent. Notably, a high proportion of children with new-onset acute symptomatic seizures had abnormal neuroimaging findings, observed in 60 of 139 (43%) patients who underwent brain imaging. This prevalence is broadly consistent with that reported by Soni et al (62.8%).^[7] Cerebrospinal fluid abnormalities compatible with meningitis or encephalitis were identified in 37% of patients. Among those investigated, autoimmune neuronal antibodies were detected in the CSF of 5 patients (3.8%), while 4 patients (3.0%) were diagnosed with viral encephalitis caused by the Japanese encephalitis virus or herpes simplex virus. The prevalence of autoimmune and viral encephalitis in our cohort was lower than that reported by Huong et al,^[30] likely reflecting the differences in the study populations. While our study included children presenting with new-onset acute symptomatic seizures, Huong et al enrolled patients who met the clinical diagnostic criteria for encephalitis, in whom a higher prevalence of autoimmune and viral etiologies was expected.^[30] We also observed a considerable proportion of patients with abnormal CSF findings in whom no specific etiology was identified. This likely reflects the limited availability of advanced diagnostic techniques for CSF analysis, including multiplex PCR, GeneXpert Ultra, and metagenomic sequencing, which can improve pathogen detection in CNS infections.^[31] Collectively, these findings underscore the importance of early diagnostic evaluation and timely intervention in children with new-onset seizures and suspected meningoencephalitis. Previous research has shown that the predominant causes of CNS infections in Vietnam include the Japanese encephalitis virus, dengue virus, herpes simplex virus, and enteroviruses.^[32] At our tertiary pediatric referral hospital, these pathogens are frequently identified in children with CNS infections and have been associated with substantial mortality and a high burden of neurological sequelae among survivors.^[33]

Additionally, CSE is a neurological emergency that may result in multiorgan dysfunction, cardiorespiratory arrest, and death if not recognized and treated promptly. In our cohort, CSE occurred in 12.2% of children with new-onset acute symptomatic seizures, consistent with previous reports.^[2,7] Among children presenting with new-onset seizures, structural brain abnormalities and electroencephalographic (EEG) abnormalities were the most common findings, identified in 43.5% and 47.2% of the patients, respectively. These were followed by electrolyte disturbances, CNS infections, and metabolic disorders. The contribution of metabolic disorders could not be adequately evaluated because only a small proportion of patients underwent metabolic testing. A large screening study showed that 13.8% of pregnant Vietnamese women harbor genes associated with metabolic diseases, including glucose-6-phosphate dehydrogenase deficiency, phenylketonuria, and galactosemia.^[34] In this study, a substantial proportion of missing biochemical data precluded the assessment of their association with acute symptomatic seizures. Nevertheless, hypoglycemia was uncommon in our cohort, with only 1 patient presenting with a blood glucose level <50 mg/dL, who recovered promptly following intravenous glucose administration. Notably, electrolyte disturbances, including hyponatremia, hypocalcemia, and hypomagnesemia, were frequently identified. Overall, 16.4% of patients had at least 1 electrolyte abnormality, including severe hyponatremia (1.3%), hypocalcemia (3.3%), and hypomagnesemia (11.8%). Most cases of hypomagnesemia were mild to moderate and responded well to magnesium replacement therapy. These electrolyte abnormalities were largely attributed to gastroenteritis and were promptly corrected after hospitalization. Evidence regarding the burden of CNS infections in children with CSE remains limited, particularly in resource-constrained settings.^[12,13] In our study, CNS infections were identified in 22 of 52 (42.3%) children with CSE, consistent with previous reports from low-resource African countries.^[13,35,36] Importantly, we used propensity score matching to account for baseline confounding factors, thereby strengthening the evidence for an association between CNS infections and CSE in children with new-onset acute symptomatic seizures.^{[12,13,35,36]} These findings provide important epidemiological data on the burden of CNS infections and CSE among hospitalized children in resource-limited settings and support the need for early recognition and prompt management of CNS infections to reduce morbidity and mortality.

In this study, 11.8% of children presenting with their first episode of acute symptomatic seizures experienced poor clinical outcomes at the time of hospital discharge. This proportion was higher than that reported by Richard et al (4.4%)^[2] but lower than that reported by Soni et al (27.6%).^[7] Notably, our study was conducted in a resource-constrained setting in Southeast Asia, where demographic characteristics, socioeconomic conditions, healthcare resources, and disease epidemiology may differ substantially from those in other regions. Such variations are likely to influence both the etiological spectrum and prognosis of acute symptomatic seizures and CSE. Therefore, we aimed to identify the key determinants of poor outcomes, with the premise that targeted interventions addressing modifiable risk factors may contribute to improved patient outcomes. We found that a longer interval from seizure onset to ED admission, lower GCS scores on presentation, focal seizures, and CNS infections were independently associated with poor clinical outcomes. These findings are consistent with previous reports and underscore the importance of early recognition and timely management in children presenting with acute seizures or CSE.^[7,23]

Our findings have important clinical implications. Children with suspected CNS infections should undergo expedited diagnostic evaluations and close clinical monitoring to enable the timely initiation of targeted therapy. Initial management should prioritize rapid stabilization, including airway and respiratory support, optimization of oxygenation and ventilation, control of hyperthermia, correction of acid–base and electrolyte disturbances, and early diagnosis and treatment of meningoencephalitis. Collectively, these measures may improve clinical outcomes in children with new-onset acute symptomatic seizures and CSE, particularly in resource-limited settings.

5. Limitations

This study had several limitations. First, it was conducted at a single tertiary institution, which may limit the generalizability of the findings to other healthcare settings. Second, the cross-sectional design precluded the inference of temporal or causal relationships between the determinants and outcomes. Third, CSE was defined using the historical criterion of seizure duration >30 minutes rather than the current International League Against Epilepsy operational definition (≥5 minutes or recurrent seizures without recovery of consciousness), which may have led to an underestimation of the CSE prevalence. Fourth, clinical, laboratory, and neurodiagnostic investigations were performed as part of routine clinical care, resulting in measurement variability and missing data, particularly for brain magnetic resonance imaging and metabolic investigations. Because MRI data were missing in 67.5% of patients and were unlikely to be missing at random, multiple imputation was considered inappropriate; therefore, complete case analyses were performed, reducing the effective sample size.^[37] In addition, neurological investigations (lumbar puncture, neuroimaging, and EEG) were performed according to clinical indications rather than systematically. Children with CSE were more likely to undergo neurological investigations, and the differential ascertainment of CNS infection (verification bias) may have overestimated its association with CSE. This limitation cannot be fully addressed using propensity score matching methods. Although propensity score matching was used to reduce measured confounding, residual confounding and selection bias could not be excluded. An important methodological observation was that CSE was not independently associated with poor outcomes in the AIC-based multivariable logistic regression models; however, it emerged as a significant risk factor for CNS infections after propensity score matching was applied to reduce confounding factors. This may reflect the limited number of outcome events relative to the number of predictors. Previous studies have shown that when the events-per-variable ratio is low, AIC-based variable selection can yield unstable coefficient estimates, reduced precision, and suboptimal model performance.^[38] Finally, pathogen identification was limited by restricted access to advanced microbiological and molecular diagnostic tests. Consequently, infectious etiologies, particularly viral pathogens, were likely underdiagnosed, and some cases classified as noninfectious or of unknown etiology may have represented unrecognized CNS infections.^[39] Despite these limitations, our study provides important real-world evidence from a resource-limited setting and highlights the key diagnostic and management challenges in children presenting with new-onset acute symptomatic seizures and CSE. Prospective multicenter studies incorporating standardized diagnostic protocols and comprehensive pathogen testing are warranted to validate and extend these results.

6. Conclusions

This study provides comprehensive data on the burden of CNS infections in children hospitalized with new-onset acute symptomatic seizures in a resource-limited setting. CNS infections were associated with CSE, and delayed presentation, lower GCS scores, focal seizures, and CNS infections were identified as the major risk factors for poor clinical outcomes. These findings underscore the importance of early recognition, timely diagnostic evaluation, and prompt management of children with suspected CNS infections and acute symptomatic seizures and may inform clinical practice and resource allocation in similar resource-constrained settings.

Acknowledgments

We are grateful to the patients, Tuong Trong-Hanh Tran (MD), and administrative staff at the Children’s Hospital No.2 for their support in this study.

Author contributions

Conceptualization: Thanh Tat Nguyen, Kieu Thi-Thuy Huynh, Van Thi-Khanh Le.

Formal analysis: Thanh Tat Nguyen.

Investigation: Thanh Tat Nguyen, Kieu Thi-Thuy Huynh.

Writing – original draft: Thanh Tat Nguyen, Kieu Thi-Thuy Huynh.

Writing – review & editing: Thanh Tat Nguyen, Kieu Thi-Thuy Huynh, Van Thi-Khanh Le.

Data curation: Kieu Thi-Thuy Huynh.

Funding acquisition: Kieu Thi-Thuy Huynh, Van Thi-Khanh Le.

Methodology: Kieu Thi-Thuy Huynh, Van Thi-Khanh Le.