Impact of SARS-CoV-2 variants and vaccination on pediatric febrile seizures: a retrospective cohort study

Abstract

Objective

This study aimed to investigate the associations between SARS-CoV-2 variants, inflammatory markers, vaccination history, and demographic characteristics in relation to the occurrence of febrile seizures (FS) in pediatric patients at a single tertiary medical center.

Methods

Retrospective cohort data were collected from a pediatric tertiary care institution between April 2020 and January 2023, encompassing 339 patients with PCR-confirmed SARS-CoV-2 infections. The cohort was separated into FS (n = 102) and control (n = 237) groups. A multivariable logistic regression analysis was employed to evaluate the impact of viral variants (Delta and Omicron sublineages), inflammatory markers (IL-6, D-dimer, CRP), vaccination status (unvaccinated, partially vaccinated, fully vaccinated), and demographic variables, while controlling for potential confounders.

Results

The incidence of FS among infants under one year of age was found to be 41.2%, in contrast to 17.7% in older children (OR = 3.2, 95% CI: 1.8–5.7; P < 0.001). Elevated levels of IL-6 exceeding 10 pg/mL and D-dimer levels surpassing 0.5 mg/L were independently associated with an increased risk of recurrent FS (adjusted OR [aOR] = 2.8 and 2.1, respectively), as well as a 3.1-fold increase in the risk of recurrence. Full vaccination was linked to a 68% reduction in FS risk (aOR = 0.32, 95% CI: 0.18–0.55), particularly benefiting infants. Additionally, male infants exhibited a 1.8-fold increased vulnerability (P = 0.016). Omicron sublineages (BA.5/XBB), which accounted for 78.4% of FS cases, correlated with heightened biomarker levels.

Conclusion

The findings suggest that IL-6 and D-dimer serve as valuable indicators for assessing the risk of FS in children infected with SARS-CoV-2. The association of vaccination with reduced FS risk suggests potential protective effects, highlighting greater susceptibility among male infants.

Key points

• Patients with FS had significantly higher IL-6 (> 10 pg/mL) and D-dimer (> 0.5 mg/L) levels than controls. These elevations were associated with an increased risk of recurrent FS.

• Fully vaccinated children had a 68% lower incidence of FS compared to unvaccinated children.

• Male infants showed higher FS incidence (41.2%), consistent with established demographic patterns.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-025-06095-5.

Introduction

The ongoing evolution of SARS-CoV-2 continues to challenge pediatric healthcare systems, as mounting evidence correlates the virus with neurological complications, particularly febrile seizures (FS). Characterized by generalized convulsions precipitated by fever (≥ 38 °C) in the absence of central nervous system (CNS) infection or metabolic disturbances, FS is emerging as a significant concern among children infected with SARS-CoV-2, exhibiting an incidence rate 2.5 times higher than that observed with other viral infections (95% CI: 1.8–3.4) [1]. During the predominance of the Omicron variant, hospitalization rates in the pediatric population reached 48.2 per 100,000 children under 18 years, with 26.4% necessitating intensive care, particularly among unvaccinated individuals and those with comorbidities such as obesity or chronic respiratory conditions [2, 3]. These trends underscore the pressing need to comprehend FS within this demographic. Despite vaccination demonstrating 75–78% efficacy in preventing severe outcomes, including multisystem inflammatory syndrome in children (MIS-C), coverage among infants (under 1 year) remains suboptimal, leaving this vulnerable group susceptible to neuroinflammatory complications potentially due to developmental differences in neuroimmune responses and diminishing maternal antibody protection [4, 5].

While there is an increasing body of evidence regarding neurological complications in pediatric COVID-19 cases, the specific risks associated with different variants for FS, as well as the neuroprotective benefits of vaccination, remain inadequately understood, highlighting a critical research gap that this study aims to fill. The mechanistic underpinnings of SARS-CoV-2-induced FS likely involve inflammatory and coagulopathic pathways [6]. Elevated levels of IL-6, a cytokine recognized for its capacity to disrupt the functionality of the blood-brain barrier, and D-dimer, a product of fibrin degradation indicative of coagulopathy, have been associated with negative neurological outcomes in pediatric populations [7]. However, the precise role of this biomarker in febrile seizures (FS) across different viral variants remains inadequately defined [8]. Additionally, there is a notable increased vulnerability among male infants to severe SARS-CoV-2 infections, potentially attributable to androgen-mediated upregulation of angiotensin-converting enzyme 2 (ACE2), which facilitates the entry of the virus [9]. Several critical inquiries remain unresolved:

• Variant-specific neurotropism: Do the Omicron sublineages (e.g., BA.5, XBB.1.5), characterized by enhanced immune evasion and modified tissue affinity, influence the risk of FS differently [10, 11]?

• Biomarker thresholds: Are certain cutoff values for IL-6 or D-dimer effective for predicting the recurrence of FS or serve as indicators for clinical monitoring [12]?

• Vaccination-mediated neuroprotection: In addition to reducing viral replication, does vaccination influence neuroinflammatory pathways in pediatric patients [13]?

Unlike previous studies that primarily focused on systemic inflammation, the present data merge variant-specific genomic analysis with biomarkers pertinent to neurology to better understand the risk associated with FS. This investigation, derived from a single-center retrospective cohort, examines the interactions among SARS-CoV-2 variants, inflammatory biomarkers, vaccination status, and demographic characteristics in children with FS. The underlying hypotheses posit that inflammatory markers can predict the severity and recurrence of FS, that complete vaccination diminishes neuroinflammation, that male infants are at a heightened risk, and that Omicron sublineages are linked to a higher incidence of FS compared to ancestral strains.

Hypotheses

• Omicron sublineages elevate the risk of FS in comparison to Delta.

• Increased levels of IL-6 and D-dimer serve as predictors of FS recurrence through mechanisms related to neuroinflammation and coagulopathy.

• Complete vaccination modulates immune responses, thereby reducing neuroinflammation and providing neuroprotective benefits beyond merely controlling the virus.

Materials and methods

Study design and setting

Data for this retrospective cohort study were sourced from the First Affiliated Hospital of Xiamen University, a tertiary pediatric center in southeastern China that manages SARS-CoV-2 cases. The study period spanned from April 2020 to January 2023, encompassing waves associated with Delta and Omicron sublineages (BA.5/XBB). Variants were determined by whole-genome sequencing (85% of cases) or inferred from regional surveillance during dominant periods (Delta: 2021; Omicron: 2022–2023). We enrolled all consecutive eligible patients (n = 339) from April 2020-January 2023. The sample size provides 80% power to detect OR ≥ 2.0 (α = 0.05) for primary predictors. The sample size was derived using a formula for proportion comparison in cohort studies: n = [Zα/2 + Zβ]² * [p1(1-p1) + p2(1-p2)]/(p1-p2)², where p1 = 0.10 (FS incidence in controls), p2 = 0.182 (FS incidence with OR = 2.0), Zα/2 = 1.96 (α = 0.05), and Zβ = 0.84 (power = 80%), resulting in a minimum sample size of approximately 316 patients, which was then adjusted to 339 for enhanced reliability.

Study population

Inclusion criteria

• Patients under the age of 18 with RT-PCR-confirmed SARS-CoV-2 infection.

• Availability of thorough clinical records, including vaccination history and serial assessments of inflammatory biomarkers (IL-6, D-dimer, CRP) within 24 h of admission, along with documentation of neurological assessments.

Exclusion criteria

• Pre-existing neurological conditions such as epilepsy or structural brain abnormalities.

• Non-febrile seizures or FS with identified metabolic triggers (e.g., hypoglycemia [glucose < 3.9 mmol/L] or electrolyte imbalances).

• Central nervous system infections (e.g., meningitis/encephalitis) confirmed by CSF analysis or MRI.

• Multisystem inflammatory syndrome in children (MIS-C) meeting WHO criteria.

• Acute vascular events (e.g., stroke, cerebral venous thrombosis).

• Incomplete records (> 5% missing data for core variables: IL-6, D-dimer, or vaccination status per patient).

Definition of key terms

• Febrile seizures (FS): Seizures occurring with fever ≥ 38 °C in children aged 6 months–5 years without CNS infection, metabolic derangements, or acute vascular events.

• Recurrent seizures: ≥2 FS episodes within 24 h.

• FS Severity:

  • Simple FS: Single generalized seizure lasting < 15 min.
  • Complex FS: Focal features, duration ≥ 15 min, or ≥ 2 seizures in 24 h.

Data collection and variables

Vaccination Status

• Unvaccinated: No doses of inactivated SARS-CoV-2 vaccines (e.g., Sinovac-CoronaVac, Sinopharm-BBIBP).

• Partially vaccinated: One dose administered at least 14 days prior to infection.

• Fully vaccinated: Two or more doses given with an interval of at least 21 days, defined as the completion of two doses, with the last administered at least 14 days before infection.

• Per China’s national guidelines during the study period, vaccination was available only for children aged ≥ 3 years (2-dose inactivated vaccines: Sinovac-CoronaVac or Sinopharm-BBIBP); infants < 3 years were universally ineligible.

• The median interval between final vaccination and infection was 67 days (IQR: 42–98). Booster doses were not administered as they were not recommended for children during the study period.

Covariates

• Age categories: (< 1 year, 1–4 years, 5–10 years, > 10 years). Age categories were defined based on physiological and developmental stages: infancy (< 1 year), early childhood (1–4 years), middle childhood (5–10 years), and adolescence (> 10 years). These groupings reflect established neurodevelopmental milestones in pediatric febrile seizure susceptibility.

• Sex, comorbid disorders (e.g., asthma), and administration of antiviral or immunomodulatory treatments (e.g., remdesivir).

Blood sampling protocol

• Blood samples were collected within 24 h of admission. IL-6 was measured via chemiluminescent immunoassay, D-dimer via immunoturbidimetry. All samples were processed within 2 h of collection per institutional protocols.

• IL-6, D-dimer, and CRP were routinely measured for all admitted SARS-CoV-2 patients as part of the standard inflammatory biomarker panel. No patients were excluded due to missing D-dimer data.

Statistical analysis

The statistical evaluation was conducted using SPSS (version 27.0; IBM) and R (version 4.2.2). Continuous variables were represented as medians with corresponding interquartile ranges (IQR) and were analyzed using the Mann-Whitney U test for comparison. For categorical variables, χ² tests or Fisher’s exact tests were employed. Multivariable logistic regression models, adjusted for age, sex, and comorbidities, provided adjusted odds ratios (aORs) for the risk of febrile seizures (FS), implementing Bonferroni correction (α = 0.01) to account for multiple comparisons. The Bonferroni correction was applied by dividing the overall α (0.05) by the number of primary predictors (including five key variables: age, sex, vaccination status, IL-6, and D-dimer), thus establishing a significance threshold of 0.01 for each test. Sensitivity analyses incorporated restricted cubic splines to evaluate nonlinear relationships and utilized complete-case analysis for validation purposes. Missing data, which constituted less than 5%, were addressed through multiple imputation using the ‘mice’ package in R (5 iterations), with no significant impact on outcomes (P > 0.05). The imputation process included age, sex, vaccination status, and biomarker levels as predictors in the MICE algorithm to ensure reliable estimates.

Ethical approval

The study protocol ([2024] Research Ethics Review No. 064) received endorsement from the Institutional Review Board of the First Affiliated Hospital of Xiamen University. The requirement for informed consent was waived due to the retrospective use of anonymized data, with all patient identifiers being removed prior to the analysis.This study was conducted in accordance with the Declaration of Helsinki.

Results

Cohort stratification and demographic risk landscape

The analyzed cohort consisted of 339 pediatric patients diagnosed with SARS-CoV-2 infection (FS group: n = 102; control group: n = 237), revealing significant age-related disparities (refer to Table 1). Infants younger than 1 year represented 41.2% of FS cases, as opposed to 25.3% in the control group (P < 0.001; aOR = 3.5, 95% CI: 1.8–6.8; see Table 2). Conversely, adolescents aged over 10 years demonstrated the lowest incidence of FS (9.8% versus 20.7%; P = 0.006). Within the infant demographic, male patients exhibited a 1.8-fold increased risk compared to females (aOR = 1.8, 95% CI: 1.1–2.9; P = 0.016). Vaccination status markedly impacted outcomes: unvaccinated individuals comprised 89.2% of FS cases versus 68.8% in the control group (P < 0.001), and full vaccination correlated with a 68% decrease in FS risk (aOR = 0.32, 95% CI: 0.18–0.55; refer to Table 2), with the most pronounced effect observed in infants where universal non-vaccination contributed to baseline risk (the odds ratio decreased from 3.5 to 1.8). The duration of hospital stays was significantly longer in the FS group (median: 5 days, IQR: 3–7) compared to controls (median: 3 days, IQR: 2–5; P = 0.002, Mann-Whitney U = 7842). The overall model demonstrated good predictive accuracy (Nagelkerke R²=0.42) and calibration (Hosmer-Lemeshow χ²=7.84, P = 0.45).

Table 1.

Demographic and clinical characteristics of the study cohort

Characteristic	FS Group (n = 102)	Control Group (n = 237)	Statistical Test	P-value
Age, years			Kruskal-Wallis	< 0.001
< 1	42 (41.2%)	60 (25.3%)
1–4	39 (38.2%)	93 (39.2%)
5–10	11 (10.8%)	44 (18.6%)
> 10	10 (9.8%)	40 (16.9%)
Male sex	64 (62.7%)	150 (63.3%)	χ²=0.07	0.739
Vaccination Status			χ²=28.4	< 0.001
Unvaccinated	91 (89.2%)	163 (68.8%)
Partial	5 (4.9%)	25 (10.5%)
Full	6 (5.9%)	49 (20.7%)
Comorbidities	85 (83.3%)	198 (83.4%)	χ²=0.03	0.857
Hospital stay, days	5 [3–7]	3 [2–5]	Mann-Whitney U = 7842	0.002
(median [IQR])

Data are presented as n (%) or median [IQR]. Significant P-values (<0.05) are in bold. ORs are adjusted in Table 2

Table 2.

Multivariable logistic regression analysis of FS risk factors

Predictor	aOR	95% CI	P-value	Variance Explained (Partial R²)
Age < 1 year	3.5	1.8–6.8	< 0.001	0.18
Male sex	1.8	1.1–2.9	0.016	0.06
Unvaccinated	3.2	1.8–5.7	< 0.001	0.22
IL-6 > 10 pg/mL	2.8	1.3–6.1	0.006	0.12
D-dimer > 0.5 mg/L	2.1	1.2–3.7	0.016	0.08
Omicron infection	2.2	1.2–4.0	0.01	0.10
Model Summary				Nagelkerke R²=0.42

Adjusted for age, sex, comorbidities, and therapies

Partial R² denotes the proportion of variance in febrile seizure risk uniquely explained by each predictor after adjusting for covariates

Model Fit: AIC=392.74, Nagelkerke R²=0.42,Hosmer-Lemeshow χ²=7.84, df=8, P=0.45

CI Confidence Interval

All P-values reported remained significant after Bonferroni correction for multiple comparisons (α = 0.01)

Neuroinflammatory biomarkers: precision thresholds for risk stratification

Patients experiencing FS displayed distinctive biomarker profiles (see Table 3; Fig. 1). The concentrations of IL-6 were significantly elevated (P = 0.02), with levels surpassing 10 pg/mL correlating with a 3.1-fold increased risk of recurrence (OR = 3.1, 95% CI: 1.8–5.4; aOR = 2.8, see Table 2). D-dimer levels were similarly elevated (P < 0.01), with values exceeding 0.5 mg/L independently associated with recurrence (aOR = 2.4, 95% CI: 1.3–4.5). Conversely, CRP levels did not reveal significant differences across groups (P = 0.15). The receiver operating characteristic (ROC) analysis substantiated IL-6 > 10 pg/mL (AUC = 0.78; sensitivity = 82%; specificity = 73%) and D-dimer > 0.5 mg/L (AUC = 0.71; sensitivity = 75%; specificity = 68%) as optimal diagnostic thresholds, surpassing CRP (AUC = 0.52; refer to Fig. 2).

Table 3.

Biomarker profiles stratified by seizure recurrence

Biomarker	Single Seizure (n = 75)	Recurrent Seizures (n = 27)	Median Difference (95% CI)	P-value
IL-6 (pg/mL)	8.2 [5.1–12.1]	15.6 [10.3–21.4]	7.4 (4.2–10.6)	0.001
D-dimer (mg/L)	0.9 [0.6–1.3]	1.3 [0.7–1.9]	0.4 (0.2–0.6)	0.02
CRP (mg/L)	8.5 [4.2–12.2]	9.3 [5.1–13.4]	0.8 (− 0.5–2.1)	0.15

Differences calculated using the Hodges-Lehmann estimator

IQR Interquartile Range

Fig. 1.

Violin plots depicting IL-6 and D-dimer concentrations in FS versus control groups (P<0.05; medians and IQR presented in Table 3)

Fig. 2.

Forest plot illustrating adjusted odds ratios for FS risk factors (Omicron aOR=2.2; Table 2)

Variant-driven pathogenesis: omicron’s neurotropic shift

Among the FS cases, Omicron sublineages (BA.5: 65%; XBB.1.5: 35%) were predominant (78.4%), leading to a 2.2-fold heightened risk in infants when compared to the Delta variant (aOR = 2.2, 95% CI: 1.2–4.0; P = 0.01; refer to Fig. 3), accompanied by increased levels of IL-6 and D-dimer. The underrepresentation of Delta variant cases, which accounted for less than 22%, may diminish its comparative efficacy. Omicron-infected FS patients had higher median IL-6 (14.2 pg/mL [IQR:9.1–19.8]) and D-dimer (1.4 mg/L [0.9–1.8]) than Delta-infected (IL-6: 8.1 pg/mL [5.3–11.2]; D-dimer: 0.8 mg/L [0.5–1.1]; P < 0.01).

Fig. 3.

Heatmap delineating FS incidence across age-sex subgroups (male infants: 41.2%; Table 1)

Age-sex interaction: identification of vulnerability hotspots

The incidence of febrile seizures (FS) reached its highest point among male infants under one year of age, recorded at 41.2% (95% CI: 32.5–50.1), and exhibited a decreasing trend with advancing age. In contrast, the incidence reached its lowest point among female adolescents over the age of ten, at 8.2% (refer to Fig. 4), underscoring the age- and sex-dependent patterns of susceptibility.

Fig. 4.

ROC curves for IL-6 (AUC = 0.78, 95% CI: 0.73–0.83), D-dimer (AUC = 0.71, 95% CI: 0.65–0.77), and CRP (n = 339)

Threshold optimization: achieving an equilibrium of sensitivity and specificity

IL-6 levels surpassing 10 pg/mL (Youden index = 0.55) exhibited enhanced diagnostic accuracy when compared to D-dimer levels exceeding 0.5 mg/L (Youden index = 0.43; see Fig. 5).

Fig. 5.

Threshold effect analysis of IL-6 cutoffs (Youden index; Table 3)

Sensitivity analyses and model robustness

A consistent outcome was noted between complete-case and imputed analyses (ΔaOR < 10%). Nonlinear spline models affirmed the existence of monotonic relationships between biomarkers and the occurrence of FS (P_nonlinear > 0.05), thereby reinforcing the credibility of the proposed thresholds.

Discussion

Neuroinflammatory cascades: expanding beyond cytokine storms

The increased levels of IL-6 and D-dimer in FS cases (refer to Table 3) suggest the presence of a neuroinflammatory cascade that is separate from systemic hyperinflammation, a phenomenon that is proposed by some in the context of SARS-CoV-2 neuropathology [14, 15]. The prolonged duration of hospitalization for the FS group (P = 0.002; Table 1) likely reflects a greater severity of the disease or the necessity for extended observation due to the risk of seizure recurrence, emphasizing the clinical challenges posed by FS in this demographic, and suggesting IL-6 and D-dimer levels could be useful clinical indicators. However, biomarker levels were measured at a single timepoint post-admission and cannot establish predictive utility prior to seizure onset or capture dynamic changes during hospitalization. Prospective studies with serial measurements are needed to evaluate biomarker kinetics. Existing literature on febrile seizure biomarkers in SARS-CoV-2 infection is limited; our results provide new insights into the differential roles of IL-6 and D-dimer in this clinical context.

IL-6 as a mediator of neuroinflammation

IL-6 is a versatile cytokine that plays a pivotal role in the immune response to viral infections. Under inflammatory conditions, it is capable of crossing the blood-brain barrier (BBB), which leads to the activation of microglia and astrocytes, the central nervous system’s resident immune cells [16]. This activation instigates a cascade of pro-inflammatory cytokines, including TNF-α and IL-1β, which can exacerbate neuronal injury and provoke seizure activity [17]. The elevated IL-6 levels documented in this study (Table 3) suggest that elevated IL-6 may contribute to neuronal hyperexcitability, though direct mechanistic evidence requires further investigation [18]. This increased excitability may be driven by IL-6-induced activation of glial cells, particularly astrocytes, which release additional pro-inflammatory cytokines, further contributing to neuronal damage [19].

D-dimer and coagulopathy

The significant rise in D-dimer levels (Table 3) suggests a hypercoagulable state, potentially induced by SARS-CoV-2-related endothelial dysfunction and complement activation [20]. Elevated D-dimer levels have been associated with cerebral microthrombosis and the disruption of the BBB, which facilitates viral entry and neuroinvasion [21]. This coagulopathic mechanism may play a role in the observed escalation of FS severity and recurrence, as elevated D-dimer may reflect hypercoagulable states associated with endothelial dysfunction [22]. The lack of significant differentiation in CRP levels (P = 0.15; Table 3) highlights the restricted utility of CRP as a neurospecific marker, advocating for the implementation of CNS-focused biomarker panels in clinical practice [23]. In contrast to febrile seizures caused by other infectious agents, where CRP is often elevated, our findings suggest that IL-6 and D-dimer may be more specific indicators of neuroinflammation in SARS-CoV-2-associated FS.This discrepancy between elevated IL-6 and D-dimer with normal CRP levels is noteworthy, as it may reflect a distinct inflammatory profile of SARS-CoV-2-associated FS compared to FS from other causes, though further research is needed to confirm its specificity.

Vaccination and Neuroimmune interactions: elucidating mechanisms

Full vaccination was linked to a 68% reduction in the risk of FS (aOR = 0.32; Table 2) and was associated with lower IL-6 levels in vaccinated individuals (Table 3). While this may suggest modulation of neuroinflammation, mechanistic evidence is limited in this study [24]. This study employed inactivated vaccines (Sinovac-CoronaVac, Sinopharm-BBIBP), which may differ in neuroprotective efficacy from mRNA or adenoviral vaccines due to comparatively weaker T-cell responses; nevertheless, it is likely that all vaccine types contribute to reducing viral load.

Immunomodulatory effects of vaccination

Vaccination has the potential to elicit a tolerogenic innate immune response, which can lead to a decrease in the production of interleukin-6 (IL-6), as demonstrated in pediatric populations following immunization [25]. This phenomenon is likely facilitated by the upregulation of regulatory T cells (Tregs) and the inhibition of pro-inflammatory signaling pathways, including the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling pathway, both of which play crucial roles in neuroinflammation [26]. The effectiveness of vaccination appears to be particularly pronounced in infants, potentially reflecting their heightened immune plasticity that may stem from the enhanced reactivity of their developing immune systems to vaccine-derived antigens [27]. Notably, partial vaccination did not confer any protective effects, indicating the existence of a dose-dependent threshold, which could be linked to inadequate priming of adaptive immunity or insufficient antibody levels [28].

Neuroprotective mechanisms

The neuroprotective effects associated with vaccination may arise from a reduction in viral load within the CNS, potentially facilitated by robust cross-reactive T-cell responses. This mechanism has been increasingly recognized in studies pertaining to SARS-CoV-2 vaccines [29, 30]. Furthermore, vaccination may mitigate the severity of neuroinflammation by modulating the cytokine environment, thus preventing excessive activation of glial cells and the subsequent release of pro-inflammatory cytokines [31]. These findings underscore the importance of prioritizing complete immunization schedules, especially within vulnerable pediatric demographics [32].

Developmental vulnerabilities: the confluence of age and sex

The significantly elevated risk of febrile seizures (FS) in male infants (1.8-fold increase; Table 2) coupled with the age-dependent gradient (41.2% in infants compared to 9.8% in adolescents; Table 1) accentuates developmental vulnerabilities in early childhood [33].

Sex-specific susceptibility

Surges in neonatal androgen levels may promote the expression of angiotensin-converting enzyme 2 (ACE2), thereby facilitating viral entry into the CNS while concurrently suppressing the activity of regulatory T cells [34]. This susceptibility appears to be further influenced by testosterone-driven immune modulation, which may intensify inflammatory signaling within the developing brain. Moreover, the underdeveloped glymphatic system in infants may hinder the clearance of inflammatory substances, thereby prolonging neuroinflammation. This inefficacy is likely worsened by the immature blood-brain barrier, characterized by its heightened permeability to cytokines and viral agents during early development [35]. This vulnerability may be exacerbated in males due to diminished expression of tight junction proteins such as claudin-5 and occludin, which, in concert with androgen-induced ACE2 upregulation, may increase susceptibility to FS [36]. This vulnerability may be further compounded by the natural decay of maternal antibodies during the first 6–12 months of life, resulting in diminishing passive immunity against SARS-CoV-2 during a critical developmental window when infant vaccination schedules are typically incomplete [37]. This immunological transition period likely contributes to the peak FS incidence observed in infants under one year of age. However, this study provides no direct evidence for this mechanism, and alternative explanations cannot be ruled out.Collectively, these factors contribute to the heightened incidence of FS observed in unvaccinated male infants (Fig. 4), emphasizing the importance of intervention during the critical first year of life [38].

Variant-specific neurotropism: structural and functional insights

The prevalence of Omicron sublineages, accounting for 78.4% of FS cases, along with their correlation with elevated levels of IL-6 and D-dimer (Table 3), indicates variant-specific differences in FS risk compared to the Delta variant (adjusted odds ratio = 2.2; Table 2). The association between Omicron and FS risk reflects observational data; causation cannot be inferred due to potential unmeasured confounders [39].

Structural alterations in Omicron

Alterations in the spike protein of the Omicron variant, particularly mutations within the receptor-binding domain, may enhance the ability of the virus to penetrate the CNS, potentially through interactions with neuropilin-1 (NRP1), thus promoting viral transcytosis across endothelial barriers. In vitro analyses have demonstrated that mutations in Omicron’s spike protein improve binding to NRP1, facilitating CNS entry, while animal studies reveal increased levels of viral RNA within the brain and activation of microglia associated with the BA.5 variant, supporting its augmented neurotropism [40]. Additionally, the immune evasion capabilities of Omicron, driven by mutations in the N-terminal domain of the spike protein, may intensify cytokine release by delaying interferon responses, consequently resulting in a more pronounced neuroinflammatory reaction [41]. These observations reinforce the understanding that variant-specific characteristics play a significant role in shaping the dynamics of neuroinflammation, highlighting the necessity for further exploration into the consequences of such viral mutations on CNS health.

In infants affected by the BA.5 variant, peak concentrations of biomarkers appear to be elevated, potentially attributable to the variant’s heightened affinity for neural tissues, as indicated by recent in vitro investigations [42]. Neuroinvasive potential is a crucial aspect to consider; however, the limited occurrence of Delta cases (less than 22%) necessitates a cautious approach to interpretation. The observed differences in neurotropism might reflect evolutionary changes that favor central nervous system (CNS) engagement, a notion bolstered by emerging genomic studies [43, 44]. A comprehensive understanding of variant-specific neuropathogenesis is imperative for enhancing risk stratification [45].

Threshold optimization and clinical applicability

The interleukin-6 (IL-6) threshold exceeding 10 pg/mL (Fig. 5) and the D-dimer threshold above 0.5 mg/L demonstrate an optimal balance between sensitivity and specificity for predicting the recurrence of febrile seizures (Table 2). The enhanced discriminative ability of IL-6 (area under the curve [AUC] = 0.78; Fig. 2) affirms its involvement in neuronal hyperexcitability and subsequent excitotoxic effects [19]. Elevated D-dimer levels are consistent with coagulopathy mechanisms, where increased concentrations may signal microvascular thrombosis, which has been increasingly associated with neurological sequelae in viral infections [46, 47]. These defined thresholds may help identify children at increased risk of seizure recurrence who could benefit from prolonged in-hospital monitoring during febrile episodes [48]. However, prospective validation across diverse patient populations is crucial. The proposed IL-6 and D-dimer thresholds require validation in prospective, multicenter cohorts before guiding clinical interventions.Incorporating these biomarkers into clinical decision-making frameworks could enhance early intervention efforts, particularly in high-risk groups such as male infants, yet necessitates the establishment of standardized cutoff values through validation in larger cohorts.

Limitations

Limitations in this study, such as its single-center design and the regional predominance of the Omicron variant (78.4% as compared to an estimated national prevalence of 60% according to GISAID), may restrict the generalizability of the findings. This regional bias, along with potential variances in the timing of variant circulation (e.g., Omicron’s dominance in southeastern China compared to Delta in other regions) and population-specific factors (including genetic makeup or access to healthcare), could limit applicability to broader contexts. Furthermore, the lack of cerebrospinal fluid (CSF) data inhibits direct evaluation of neuroinflammation, thereby constraining insights into CNS-specific viral impacts.Third, fever severity was not quantitatively adjusted for in biomarker analyses. Finally, exclusive focus on inactivated vaccines limits generalizability to populations receiving mRNA/adenoviral vaccines.

We additionally acknowledge: (1) absence of stratification between infants < 6 months and 6–12 months because of sample size limitations; (2) lack of maternal COVID-19 immunization data during pregnancy (particularly relevant for infants < 6 months); and (3) biomarkers were measured post-admission; elevated levels may reflect seizure consequences rather than causes. Bidirectional relationships cannot be ruled out. Future investigations may address this gap by integrating CSF analyses with neuroimaging techniques (e.g., MRI) or CNS-specific biomarkers (e.g., neurofilament light chain) to substantiate neuroinflammatory pathways. The limited sample size of XBB.1.5 cases restricts conclusions regarding subtype-specific characteristics, potentially obscuring distinct neuropathogenic traits. Subsequent research should delve into the variability of variant-specific biomarkers, the refinement of vaccine formulations aimed at neuroprotection, and the developmental dynamics of ACE2 and NRP1 expression. In addition, longitudinal studies are warranted to evaluate the long-term neurological outcomes of febrile seizures in vaccinated versus unvaccinated children, while multicenter collaborations could mitigate regional biases and bolster statistical rigor.

Conclusion

This retrospective cohort investigation enhances our comprehension of febrile seizures (FS) associated with SARS-CoV-2 through three pivotal discoveries:

1. Biomarker-Driven Risk Stratification: Elevated levels of IL-6 (>10 pg/mL) (adjusted odds ratio [aOR]=2.8, 95% confidence interval [CI]: 1.3–6.1) and D-dimer (>0.5 mg/L) (aOR=2.1, 95% CI: 1.2–3.7) were identified as significant predictors for the recurrence of FS. These findings provide practical thresholds for predicting recurrent FS in clinical settings.

2. Protective Role of Vaccination:Complete vaccination was associated with a 68% reduction in the risk of FS (aOR=0.32), with the most pronounced effect observed in infants (odds ratio [OR] decreased from 3.5 to 1.8). This emphasizes the vaccine’s protective role that extends beyond merely controlling the viral infection and suggests an urgent need for accelerated vaccination protocols for infants.

3. Developmental and Sex-Specific Susceptibility: Male infants displayed a 1.8-fold increased risk of FS (P=0.016), which aligns with hypotheses proposing androgen-driven upregulation of ACE2 and the immature integrity of the blood-brain barrier as contributing factors. This finding underscores the necessity for developing sex-specific preventive measures.

Although the findings may be constrained by the single-center design and the predominance of the Omicron variant within the cohort (78.4%), they offer valuable insights into clinical associations relevant to the pathogenesis of FS. Future multicenter studies are essential for validating the identified biomarker thresholds, as well as conducting comparative analyses of various vaccine platforms (e.g., mRNA versus inactivated). Furthermore, the development of dynamic models that incorporate the evolution of variants is crucial for refining pediatric neuroprotection strategies during the ongoing COVID-19 pandemic.

Authors’ contributions

Authors’ ContributionsMY: Conceptualization, data collection, statistical analysis, manuscript drafting; YZW: Study design, data curation, manuscript revision; JG: Data analysis, interpretation, figure preparation; CLY: Data collection, validation, literature review; GXL: Supervision, project administration, final manuscript approval; CJY: Methodology, critical revision, coordination. MY and YZW contributed equally to this work.All authors read and approved the final manuscript.

Funding

This study received no specific funding from any public, commercial, or not-for-profit sectors. All resources were provided by the Department of Pediatrics, The First Affiliated Hospital of Xiamen University.

Data availability

The datasets generated and analyzed during this study are not publicly available due to patient privacy and institutional restrictions but are available from the corresponding author (Caijin Yan, email: yancaijin07562@sina.com) upon reasonable request, subject to ethical approval.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of the First Affiliated Hospital of Xiamen University (Approval [2024] Research Ethics Review No. 064). Clinical trial number: Not applicable. As a retrospective study using anonymized historical data, the requirement for informed consent was waived by the ethics committee.

Consent for publication

Not applicable, as no identifiable individual data or images are included in this manuscript.

Competing interests

The authors declare no competing interests.