DTaP-induced infantile epileptic spasms syndrome high-risk window: Insights from school-age children

ABSTRACT

To determine the specific high-risk characteristics and temporal window associated with Diphtheria, Tetanus, and acellular Pertussis (DTaP) vaccine-precipitated infantile epileptic spasms syndrome (IESS), providing evidence for optimizing vaccination safety strategies, we conducted a multicenter retrospective cohort study of 564 patients with long-term follow-up (6.5–10.5 y). Patients were stratified by seizure onset interval into Vaccination-proximate (≤72 hours) and distant (>72 hours) groups. The proximate group (n = 53, 9.4%) exhibited a significantly younger age distribution (median [IQR]: 5.0 [3.0–6.0] vs. 5.0 [4.0–7.0] months, p = .04) and higher DTaP exposure (71.7% vs. 51.3%). Multivariable logistic regression in an etiologically analytic cohort (n = 541) identified the DTaP primary series as a significant independent precipitant for rapid onset (Dose 1: OR = 4.00; Doses 2 & 3: OR = 4.67) compared to other inactivated vaccines. Crucially, no increased risk was observed for the DTaP booster (Dose 4) or non-DTaP inactivated vaccines. Kaplan-Meier analysis confirmed a steep rise in seizure incidence within 72 hours post-primary DTaP. Long-term follow-up revealed no significant differences in all-cause mortality, seizure remission, or educational placement (p > .05) between groups, indicating that vaccination merely precipitated disease onset without altering the etiology-determined prognosis. This study, based on the full-course observation of school-aged children, is the first to validate the specificity of the DTaP-IESS association. The DTaP primary series is a critical precipitating factor for rapid onset (≤72 hours) in infants with underlying epileptogenic susceptibility. We recommend a stratified immunization strategy for high-risk infants: adhering to schedules for non-DTaP vaccines while adopting a prudent “delay-but-not-refuse” approach for DTaP to balance neurological safety with herd immunity.

Introduction

The global resurgence of pertussis presents a severe public health challenge and remains a significant cause of childhood mortality.¹ The diphtheria, tetanus, and acellular pertussis (DTaP) vaccine serves as the cornerstone of prevention strategies.² However, the peak age of onset for infantile epileptic spasms syndrome (IESS), typically between 3 and 12 months (range 1–24 months),³ coincides precisely with the schedule for the primary DTaP immunization series (administered at 3, 4, and 5 months of age). Consequently, since the initial report in 1983,⁴ the relationship between vaccination and IESS has remained a focal point of academic research and a subject of widespread public concern.⁵

As a catastrophic age-dependent epileptic syndrome, IESS is frequently accompanied by developmental delay and drug-resistant epilepsy.⁶ With an incidence of approximately 30 per 100,000 live births, IESS often evolves into other refractory epilepsy types or syndromes and carries an average mortality rate of 12%, imposing a substantial burden on families and society.⁷ While previous studies have suggested a potential association between vaccination and IESS, the lack of long-term follow-up validation has led to equivocal conclusions. This uncertainty exacerbates public vaccine hesitancy,⁸ paradoxically leaving these vulnerable infants unprotected against pertussis.

Pertussis toxin (PTx), a potent adjuvant component of DTaP vaccines, serves not only as the core immunogen^9,10 but is also widely utilized to induce experimental autoimmune models of the nervous system (e.g., EAE/EAN).^11,12 The intrinsic neuro-immune modulating potential of PTx provides biological plausibility for its role in precipitating seizures. However, with advancements in diagnostic technologies, specific underlying etiologies have been identified in patients previously considered to have “vaccine encephalopathy”.^13,14 Notably, SCN1A mutations have been detected in the majority of vaccine-associated Dravet syndrome cases.^15–17 This mechanistic insight compels us to reexamine the role of DTaP within the etiologically more heterogeneous IESS population, necessitating a risk assessment that transcends mere temporal coincidence.

The rapid advancement of neuroimaging and next-generation sequencing (NGS) technologies¹⁸ has enabled the identification of etiologies in IESS patients previously classified as cryptogenic. Therefore, utilizing multimodal diagnostic techniques involving multi-strategy tiered genetic testing and (PET-)MRI/CT, we established a rigorously screened cohort of school-aged IESS patients who had completed their childhood immunization. This study aimed to reexamine the true association between DTaP-containing vaccines and seizure onset, specifically investigating their potential risk as an independent precipitating factor and their dose-specific effects. Our findings provide high-quality evidence to dispel public misconceptions and optimize immunization strategies for infants susceptible to epilepsy.

Methods

Study design and patient population

This retrospective cohort study collected clinical data from patients diagnosed with IESS across 31 administrative regions of China between January 1, 2015, and December 31, 2018, with follow-up extending up to school age (6–8 y). This study was conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the Ethics Committee of the Chinese PLA General Hospital (Approval No. 2025-400), which waived the need for individual patient consent due to the retrospective nature of the analysis.

Inclusion criteria were patients meeting the IESS diagnostic criteria, having completed routine childhood vaccination, and possessing complete vaccination records (verified by medical records or immunization certificates). IESS was diagnosed based on typical epileptic spasms occurring in clusters or individually (flexor, extensor, or mixed type), accompanied by typical or atypical hypsarrhythmia on electroencephalography (EEG).³ Exclusion criteria included duplicate records, incomplete clinical data, a time interval exceeding 40 d between the last vaccination and seizure onset, or the coexistence of other types of epilepsy or neurological overlapping disorders. The 40-d threshold aligns with the WHO Causality Assessment framework and Canadian IMPACT surveillance guidelines.^19,20

Data collection and definitions

Clinical data extracted from medical records included sex, age at seizure onset, seizure frequency (standardized to daily episodes), number of anti-seizure medications (ASMs), and family history. According to the 2017 International League Against Epilepsy (ILAE) classification,²¹ etiologies were categorized into six groups: genetic, structural, metabolic, immune, infectious, and unknown. Structural abnormalities were primarily evaluated using a standardized 3T MRI protocol (Siemens Prisma), which included 3D-T1WI (TR/TE = 2300/2.98 ms), T2-FLAIR, and DWI sequences. A tiered genetic testing strategy was implemented for patients without perinatal structural abnormalities, using methods such as high-resolution G-banded karyotyping, targeted panel sequencing, whole-exome sequencing (WES), and SNP array analysis. The pathogenicity of variants was assessed according to the guidelines of the American College of Medical Genetics and Genomics (ACMG).²²

Detailed vaccination histories were verified using medical records or vaccination booklets to collect data on vaccine type, dose sequence, and the interval between the last vaccination and the first seizure (with the day of vaccination defined as Day 0). To minimize recall bias, caregiver reports regarding the exact date of seizure onset were, whenever possible, cross-validated against objective medical records such as multiple previous admission records, accessible outpatient notes, or previous EEG reports, prioritizing these objective anchors in cases of discrepancy. To eliminate exposure trajectory bias, the primary exposure variable for analysis was strictly defined as the specific vaccine dose administered most recently prior to the onset of the first epileptic spasm. Based on the temporal relationship between vaccination and seizure onset, patients were divided into two groups: the Vaccination-proximate Group (onset ≤72 hours post-vaccination) and the Vaccination-distant Group (onset >72 hours to 40 d post-vaccination). This 72-hour threshold was established a priori based on the biological timeline of immediate innate immune reactogenicity, as acute systemic inflammation following adjuvanted vaccinations typically peaks and resolves within this 3-d window.²³ This ensures high specificity for isolating immediate vaccine-triggered physiological stress from delayed reactions.²⁴ Notably, this 0–3 d window is aligned with established clinical precedents.²⁵

To investigate dose-specific risks, vaccines were further categorized into seven distinct subgroups: (1) DTaP Dose 1; (2) DTaP Dose 2 or 3; (3) DTaP Booster (Dose 4); (4) Live vaccine Dose 1; (5) Live vaccine Dose 2 and subsequent doses; (6) Inactivated vaccine Dose 1; and (7) Inactivated vaccine Dose 2 and subsequent doses. Additionally, to account for multiple vaccines administered on the same day, exposures were explicitly categorized into standalone and concurrently administered groups, as detailed in Supplementary Table 4.

During the follow-up period, data on mortality, seizure freedom, and educational status were collected. Survival outcomes were evaluated based on all-cause mortality. Neurological outcomes were defined as the seizure-free status of surviving patients at the last follow-up. Functional outcomes were assessed based on educational placement (attending regular school, special education, or home care) to serve as a proxy measure for long-term cognitive function and social adaptability.

Statistical analysis

Descriptive analyses were conducted for all baseline variables. Continuous data were presented as medians with interquartile ranges (IQR) due to non-normal distribution and compared using the Mann-Whitney U test. Categorical variables were expressed as frequencies and percentages, with comparisons performed using the Chi-square test or Fisher’s exact test as appropriate. To identify independent predictors of rapid seizure onset, a multivariable logistic regression analysis was performed. Patients with metabolic, immune, or infectious etiologies (n = 23) were excluded from the multivariable logistic regression and time-to-event analyses. Biologically, these etiologies involve active endogenous immune or metabolic crises that would profoundly confound the exogenous immune stress induced by vaccination, precluding the separation of vaccine-specific reactogenicity from the acute natural course of the primary disease. Furthermore, this exclusion was a statistical necessity to prevent the “complete separation” phenomenon in the modeling, as there were zero instances of vaccination-proximate events (≤72 hours) within these specific subgroups. The model was adjusted for sex and etiology. Age at onset was excluded from the final multivariable model to avoid multicollinearity, as vaccination schedules are inherently age-dependent. Time-to-event analysis was conducted using the Kaplan-Meier method to visualize the cumulative incidence of seizures over the first 10 d post-vaccination, stratified by vaccine subgroups. All statistical analyses were performed using R software (version 4.4.2, with gtsummary, survival, and survminer packages) and GraphPad Prism (version 9.0). A two-sided p value < 0.05 was considered statistically significant.

Results

Study population and baseline characteristics

A total of 769 patients with IESS were initially enrolled in this study. Following the exclusion of 205 ineligible patients, 564 patients constituted the total cohort for the initial descriptive analysis (Figure 1). Based on the temporal relationship between vaccination and seizure onset, patients were categorized into the Vaccination-proximate Group (n = 53) and the Vaccination-distant Group (n = 511). All included children had reached school age and completed their routine childhood immunization schedules.

Figure 1.

Flowchart of study participants.

A total of 769 patients were enrolled, of whom 564 were included in the study (total cohort). Long-term prognosis was evaluated in this total cohort. For further statistical modeling, 23 patients with metabolic, immune, or infectious etiologies were excluded due to the absence of vaccination-proximate events. The remaining 541 patients constituted the Analytic Cohort for multivariable logistic regression and time-to-event analyses. Patients in this cohort were categorized into the Vaccination-proximate Group (seizure onset ≤72 hours post-vaccination) and the Vaccination-distant Group (onset >72 hours to 40 d).

The baseline demographic and clinical characteristics of the total cohort are summarized in Table 1. There were no statistically significant differences between the two groups regarding sex distribution (p = .77), daily seizure frequency (p = .43), or the number of ASMs used (p = .88). However, a significant difference was observed in the age at onset (p = .040). Although the median age at onset was 5.0 months for both groups, statistical analysis revealed that the age distribution in the Vaccination-proximate Group (IQR = 3.0–6.0 months) was significantly shifted toward a younger age compared with the Vaccination-distant Group (IQR = 4.0–7.0 months).

Table 1.

Demographic and clinical characteristics of all included patients.

Characteristic	Total (N = 564)a	Vaccination-proximate Group (N = 53)a	Vaccination-distant Group (N = 511)a	p valueb
Age at Onset (months)	5 (4, 7)	5 (3, 6)	5 (4, 7)	.04
Sex				.77
Male	332 (58.9%)	30 (56.6%)	302 (59.1%)
Female	232 (41.1%)	23 (43.4%)	209 (40.9%)
Etiology (All Categories)				.69
Genetic	96 (17.0%)	11 (20.8%)	85 (16.6%)
Structural	271 (48.0%)	24 (45.3%)	247 (48.3%)
Infectious	4 (0.7%)	0 (0.0%)	4 (0.8%)
Unknown	174 (30.9%)	18 (34.0%)	156 (30.5%)
Metabolic	18 (3.2%)	0 (0.0%)	18 (3.5%)
Immune	1 (0.2%)	0 (0.0%)	1 (0.2%)
Seizure Frequency (per day)	40 (15, 90)	40 (25, 81)	40 (15, 93)	.43
Unknown	10	1	9
Number of ASMs	2 (1, 3)	2 (1, 3)	2 (1, 3)	.88

^aData are presented as n (%) for categorical variables and median (IQR) for continuous/ordinal variables. Categorical variables were compared using Fisher’s exact test; continuous/ordinal variables (including the number of ASMs) were compared using the Wilcoxon rank-sum test.

^bWilcoxon rank sum test; Fisher’s exact test.

The etiological distribution is presented in Table 1, with specific subtypes detailed in Supplementary Tables 1 and 2. While genetic, structural, and unknown etiologies were observed in both groups, patients with metabolic (n = 18), immune (n = 1), or infectious (n = 4) etiologies were identified exclusively within the Vaccination-distant Group. Within the Vaccination-proximate Group (n = 53), multimodal diagnostic evaluation revealed that 66.0% (35/53) of patients had a definitive genetic or structural pathogenic background. Given the absence of rapid-onset events in the metabolic, immune, and infectious etiology subgroups, these 23 patients were excluded from subsequent multivariable logistic regression and time-to-event analyses to prevent complete separation and bias. This selection process established an Analytic Cohort of 541 patients, comprising only those with genetic, structural, or unknown etiologies (Figure 1).

Notably, nine patients (17.0%, 9/53) in the Vaccination-proximate Group developed a fever prior to the onset of spasms, and in six of these cases, both the fever and the seizure occurred on the day of vaccination. More importantly, seven of these nine febrile patients (including 3 cases of congenital cerebral dysplasia, 3 cases of hypoxic-ischemic encephalopathy, and 1 case with a TSC gene variant) were confirmed to have a clear underlying organic or genetic etiology.

Vaccine exposure profiles and spatiotemporal distribution of seizure onset

Vaccine exposure profiles differed significantly between the two groups (Supplementary Table 3). DTaP-containing vaccines were the predominant exposure in the Vaccination-proximate Group (71.7%, 38/53), a proportion significantly higher than in the Vaccination-distant Group (51.3%, 262/511). In contrast, other inactivated vaccines, such as the inactivated poliovirus (IPV) and hepatitis B (HepB) vaccines, were more prevalent in the Vaccination-distant Group (IPV: 14.1% vs. 7.5%; HepB: 12.3% vs. 1.9%). Notably, live attenuated vaccines, including the measles-mumps-rubella (MMR) vaccine, were identified exclusively in the Vaccination-distant Group (7.2%).

Figure 2(A) illustrates the temporal and age distribution of seizure onset following vaccination. To highlight early characteristic patterns, the analysis focused on data from the first 21 d post-vaccination. A significant clustering of cases was observed within the Vaccination-proximate Group (≤72 h). This high-density cluster was concentrated among infants aged 3 to 6 months, coinciding with the schedule for the primary immunization series. Notably, this temporal feature was primarily driven by DTaP-containing vaccines. In contrast, cases associated with live attenuated vaccines or other inactivated vaccines exhibited a sparse and scattered distribution along the timeline, lacking distinct focal clustering.

Figure 2.

Temporal and age distribution of seizure onset post-vaccination.

(A) Scatter plot illustrating the relationship between the vaccination-to-seizure interval (X-axis, days) and age at onset (Y-axis, months). The X-axis is truncated at Day 21 to optimize visual resolution. Bubble size corresponds to the patient count at each data point, and colors represent the specific vaccine type administered (see legend). The red dashed line demarcates the vaccination-proximate period (≤72 hours). Within this period, a dense cluster of cases is observed among infants aged 3–6 months, predominantly associated with DTaP-containing vaccines (orange bubbles). In contrast, cases associated with other vaccines are more dispersedly distributed.

(B) Detailed temporal distribution of seizure onset following DTaP-containing vaccinations. Upper panel: scatter plot showing age at onset (months) versus the vaccination-to-onset interval (days) for patients receiving any DTaP-containing vaccine. Histogram with a Loess trend line illustrating the daily fluctuations in case numbers. A significant clustering of cases is observed within the vaccination-proximate period (≤72 hours), followed by a secondary peak on Day 6 and a subsequent decline, highlighting the specific temporal pattern associated with DTaP exposure.

Abbreviations: DTaP, diphtheria-tetanus-acellular pertussis vaccine; MR, measles-rubella combined vaccine; IPV, inactivated poliovirus vaccine; JEV, Japanese encephalitis vaccine; MPV-A, group A meningococcal polysaccharide vaccine; Hib, Haemophilus influenzae type b vaccine; LAVV, live attenuated varicella vaccine; HepB, hepatitis B vaccine; EV71, enterovirus 71 vaccine.

A specific analysis of DTaP-containing vaccines was performed (Figure 2(B)). The histogram with a trend line revealed a distinctive bimodal-like pattern: while the primary surge of cases occurred steeply within the first 72 hours, a secondary clustering of reported cases was observed around Day 6, followed by a gradual decline in the daily number of cases. Despite the presence of this delayed peak, the steep concentration of cases within 72 hours post-vaccination remains the core epidemiological feature distinguishing DTaP-associated seizures from the background incidence.

Association between vaccine type/dose and vaccination-proximate seizure onset

To investigate the specific impact of vaccine types and doses, we analyzed the Analytic Cohort (n = 541), which included patients with genetic, structural, or unknown etiologies (Table 2). As shown in Supplementary Table 4 and Supplementary Figure 1, the vaccine exposure profile in the Vaccination-proximate Group exhibited a highly specific concentration of DTaP-containing vaccines. Specifically, standalone DTaP vaccine accounted for 71.7% (38/53) of vaccinations in the Vaccination-proximate Group, significantly higher than the 50.8% (248/488) observed in the Vaccination-distant Group. Additionally, the Vaccination-proximate Group included other DTaP-containing combination vaccines, such as DTaP + IPV (3.8%) and pentavalent vaccines (1.9%).

Table 2.

Characteristics of the Analytic Cohort.

Characteristic	Total (N = 541)a	Vaccination-proximate Group (N = 53)a	Vaccination-distant Group (N = 488)a	p valueb
Age at Onset (months)	5.0 (4.0, 7.0)	5.0 (3.0, 6.0)	5.0 (4.0, 7.0)	.03
Sex				.66
Male	322 (59.5%)	30 (56.6%)	292 (59.8%)
Female	219 (40.5%)	23 (43.4%)	196 (40.2%)
Etiology				.72
Genetic	96 (17.7%)	11 (20.8%)	85 (17.4%)
Structural	271 (50.1%)	24 (45.3%)	247 (50.6%)
Unknown	174 (32.2%)	18 (34.0%)	156 (32.0%)
Primary Vaccine Type (Broad Category)				<.001
DTaP-containing Vaccines	308 (56.9%)	42 (79.2%)	266 (54.5%)
Live Attenuated Vaccines	55 (10.2%)	5 (9.4%)	50 (10.2%)
Inactivated Vaccines (Non-DTaP)	178 (32.9%)	6 (11.3%)	172 (35.2%)
Detailed Vaccine Status (7 Groups)				<.01
DTaP Dose 1	87 (16.1%)	12 (22.6%)	75 (15.4%)
Live Vaccine Dose 1	51 (9.4%)	5 (9.4%)	46 (9.4%)
Inactivated Dose 1	76 (14.0%)	3 (5.7%)	73 (15.0%)
Inactivated Dose 2+	102 (18.9%)	3 (5.7%)	99 (20.3%)
Live Vaccine Dose 2+	4 (0.7%)	0 (0.0%)	4 (0.8%)
DTaP Dose 2&3	200 (37.0%)	30 (56.6%)	170 (34.8%)
DTaP Booster (Dose 4)	21 (3.9%)	0 (0.0%)	21 (4.3%)
Seizure Frequency (per day)	40.0 (15.0, 94.5)	40.0 (25.0, 80.5)	40.0 (15.0, 99.0)	.49
Unknown	9	1	8
Number of ASMs	2 (1, 3)	2 (1, 3)	2 (1, 3)	.86

^aData are presented as n (%) for categorical variables and median (IQR) for continuous/ordinal variables. Categorical variables were compared using Fisher’s exact test; continuous/ordinal variables (including the number of ASMs) were compared using the Wilcoxon rank-sum test.

^bWilcoxon rank sum test; Fisher’s exact test.

Regarding dose sequence, 22.6% of patients in the Vaccination-proximate Group received the first dose of DTaP, and 56.6% received the cumulative second or third doses. Notably, no patients in the Vaccination-proximate Group received the DTaP booster (dose 4) or subsequent doses of live attenuated vaccines, whereas these accounted for 4.3% and 0.8% in the Vaccination-distant Group, respectively (p = .003).

To quantify these differences, a multivariable logistic regression model was constructed, adjusting for sex and etiology (Table 3 and Figure 3). Using the first dose of other inactivated vaccines as the reference, DTaP Dose 1 demonstrated a significant association with an increased risk of vaccination-proximate onset (OR = 4.00, 95% CI = 1.20–18.20, p = .04). This risk remained significantly elevated for the cumulative DTaP Doses 2 & 3 (OR = 4.67, 95% CI = 1.58–20.06, p = .01), whereas subsequent doses of other inactivated vaccines (Inactivated Dose 2+) did not confer increased risk (OR = 0.78, p = .80). Consistent with Figure 3 and Supplementary Table 4, odds ratios for the DTaP Booster and subsequent live attenuated vaccine doses were not estimable (NE) because of the complete absence of vaccination-proximate events in these categories. Clinically, this “not estimable” status directly translates to an extremely low triggering risk. This negative finding statistically corroborates the relative safety of vaccinations administered to older infants.

Table 3.

Multivariable logistic regression analysis.

Characteristic	OR	95% CI	p value
Vaccine Type & Dose
Inactivated Dose 1 (Ref)	—	—
DTaP Dose 1	4.00	1.20, 18.20	.04
Live Vaccine Dose 1	2.91	0.67, 14.87	.16
Inactivated Dose 2+	0.78	0.14, 4.36	.77
Live Vaccine Dose 2+	NE	—	>.99
DTaP Dose 2&3	4.67	1.58, 20.06	.01
DTaP Booster	NE	—	.99
Etiology
Unknown	—	—
Genetic	1.24	0.53, 2.79	.61
Structural	0.78	0.41, 1.52	.46
Gender
Male	—	—
Female	1.20	0.66, 2.16	.54

Abbreviations: CI = Confidence Interval, OR = Odds Ratio, NE: Not Estimable.

No patients in the DTaP Booster and Live Vaccine Dose 2+ categories were classified into the Vaccination-proximate Group, precluding the calculation of odds ratios.

Figure 3.

Forest plot of multivariable logistic regression analysis for vaccination-proximate seizure onset.

The plot displays adjusted odds ratios (ORs) with 95% confidence intervals (CIs) for factors associated with vaccination-proximate onset (≤72 hours post-vaccination) in the Analytic Cohort (n = 541). Reference groups (male, unknown etiology, inactivated Dose 1) are indicated by open circles at OR = 1.00. DTaP Dose 1 and DTaP Dose 2 & 3 (solid orange circles) show statistically significant associations with vaccination-proximate onset (p < .05). The DTaP Booster is not shown due to zero events (OR not estimable).

Abbreviations: DTaP, diphtheria-tetanus-acellular pertussis vaccine; Ref, reference group; OR, odds ratio; CI, confidence interval.

Time-to-event analysis and risk window

Kaplan-Meier cumulative incidence curves were utilized to dynamically characterize the trajectory of seizure risk during the first 10 d post-vaccination (Figure 4). The analysis revealed significant temporal heterogeneity stratified by vaccine type and dose sequence.

Figure 4.

Kaplan-Meier cumulative incidence of seizure onset stratified by vaccine type and dose.

The curves illustrate the cumulative probability of seizure onset over the first 10 d post-vaccination. The vaccination-proximate period (≤72 hours) is highlighted in red. DTaP Dose 1 and DTaP Dose 2 & 3 exhibit a steep vertical rise within this period, indicating a high concentration of immediate risk. In contrast, DTaP Booster and Live Vaccine Dose 2+ show zero events (maintaining a flat line at 0%) within the vaccination-proximate period, with the DTaP Booster showing a delayed rise only after Day 3. Other vaccine types display a gradual, low-rate accumulation of cases.

Abbreviation: DTaP, diphtheria-tetanus-acellular pertussis vaccine.

The curves for DTaP Dose 1 and DTaP Doses 2 and 3 exhibited a steep rise within the initial 72 hours. Following this period, a distinct inflection point was observed, after which the slopes of both curves plateaued. In contrast, the curves for other inactivated vaccines and live attenuated vaccines displayed a slow, low-level linear accumulation pattern. Notably, the curve for the DTaP Booster (Dose 4) showed no event within the vaccination-proximate window.

Long-term follow-up and prognostic outcomes

The follow-up duration for the 564 patients ranged from 6.5 to 10.5 y (Table 4). Regarding survival outcomes, the all-cause mortality proportion was 13.2% (7/53) in the Vaccination-proximate Group and 14.7% (75/511) in the Vaccination-distant Group, with no statistically significant difference observed between the groups (p > .99).

Table 4.

Long-term clinical and developmental outcomes of the study cohort stratified by vaccination-onset interval.

Outcome Variables	Vaccination-proximate Group N = 53a	Vaccination-distant Group N = 511a	p valueb
Survival Status (Total Cohort)			>.99
Alive	46/53 (86.8%)	436/511 (85.3%)
Dead	7/53 (13.2%)	75/511 (14.7%)
Seizure Outcomes (Among Survivors)			.74
Seizure Free	16/46 (34.8%)	141/436 (32.3%)
Persistent Seizures	30/46 (65.2%)	295/436 (67.7%)
Educational Status (Among Survivors)			.88
No Schooling/Home Care	24/46 (52.2%)	210/436 (48.2%)
Regular Schooling	9/46 (19.6%)	87/436 (20.0%)
Special Education	13/46 (28.3%)	139/436 (31.9%)

^aData are n/N (%). Functional outcomes exclude deceased patients.

^bFisher’s exact test.

Analysis of neurological outcomes among surviving patients indicated that the seizure-free proportion at the last follow-up was 34.8% (16/46) in the Vaccination-proximate Group and 32.3% (141/436) in the Vaccination-distant Group (p = .74). Concerning educational status, 19.6% (9/46) of school-aged survivors in the Vaccination-proximate Group attended regular schools, compared with 20.0% (87/436) in the Vaccination-distant Group; this difference was not statistically significant (p = .88). Similarly, there were no significant differences in the proportions of patients receiving special education or home care between the two groups.

Discussion

This study provides a comprehensive long-term evaluation of the association between IESS and vaccination based on full-course observation up to school age, bridging a critical gap in long-term follow-up data. Through detailed temporal analysis of the vaccine spectrum, we identified a significant clustering of cases within 72 hours post-vaccination, a phenomenon specifically concentrated following the primary DTaP immunization series. Although the median age of onset (5 months) highly overlaps with the standard DTaP schedule (3, 4, and 5 months), creating a statistical coincidence, epidemiological patterns in our cohort suggest that the underlying mechanism may extend beyond mere temporal chance. Observations supporting this view include the significant shift toward a younger age of onset in the Vaccination-proximate Group (IQR = 3.0–6.0 months) compared with the Vaccination-distant Group (IQR = 4.0–7.0 months), suggesting that vaccination may have precipitated the clinical manifestation of the disease earlier than its natural course. Furthermore, the overwhelming dominance of DTaP vaccines in the proximate group stands in sharp contrast to the scarcity of other inactivated vaccines (e.g., Hepatitis B, IPV), pointing toward a potential vaccine-specific reactogenicity.

Leveraging the application of multimodal etiology tracing technology,²⁶ we found that in the Vaccination-proximate Group, where the temporal association was most pronounced, up to 66.0% (35/53) of infants had definitive preexisting epileptogenic etiologies. This indicates that the neurological vulnerability of these children far predates the vaccination window. Crucially, based on this, we propose that the robust immune response induced by DTaP may serve as a critical stressor or “second hit,” tipping a developmentally vulnerable brain into the chaotic network state of hypsarrhythmia.

Unlike other epilepsies, IESS exhibits a unique responsiveness to ACTH therapy, indicating that immune factors play a pivotal role in its pathogenesis.²⁷ A phenomenon warranting deeper investigation is the observation of a significant secondary peak on Day 6 (one week post-vaccination), in addition to the immediate onset peak within 72 hours post-vaccination. This “dual-peak pattern” suggests the possibility of two distinct immune triggering mechanisms: The immediate peak within 72 hours may be primarily associated with vaccine-induced innate immune responses, acute nonspecific inflammation, or fever (9 patients in the proximate group experienced fever prior to seizure onset), which aligns with the rapid stress component of the “second hit” model.¹⁶ In contrast, the secondary peak on Day 6 is temporally closer to the initiation of adaptive immune responses and the peak of T cell activation.²⁸ Notably, we observed that cases with onset on Day 6 frequently involved exposure to multivalent DTaP combinations (e.g., pentavalent vaccines) and measles-containing vaccines (MMR/MR). This suggests that the secondary peak may be driven by the complex immunogenicity of combined formulations or the characteristic delayed reactogenicity of live vaccines.^4,29 The interpretation of this delayed clustering, however, warrants epidemiological caution. The apparent narrowness of this peak and the corresponding dearth of observations on adjacent days strongly suggest the compounding influence of digit preference. In retrospective settings, caregivers frequently round the onset interval to approximately one week, resulting in artificial data heaping around days 6 and 7. Consequently, this secondary peak likely reflects a genuine, broader biological window of delayed immune stress that has been observationally compressed into a focal point by parental recall bias.

The multivariable regression model identified DTaP Dose 1 and DTaP Doses 2 and 3 as independent risk factors for rapid onset. Importantly, this model inherently utilized concurrent vaccinations as an internal control to disentangle age confounding. As demonstrated in Table 3, subsequent doses of other inactivated vaccines (Inactivated Dose 2+) administered during the same 3-to-4-month chronological window did not trigger an acute onset peak. This contrasting outcome suggests that the precipitating effect is specific to the DTaP formulation rather than a passive artifact of age. While the absence of booster doses in the Vaccination-proximate Group is partly attributable to the peak incidence of IESS coinciding with the primary immunization schedule, a biologically significant control finding emerged: among the 21 older toddlers who received the DTaP booster, none exhibited rapid seizure onset post-vaccination. This different clinical outcome under the same vaccine exposure delineates a specific, age-dependent window of biological vulnerability, suggesting that the triggering effect may be contingent upon brain maturity— specifically, that immature neural networks are more susceptible to depolarization by immune stimulation. Consequently, the DTaP-IESS association appears to involve the convergence of three critical dimensions: chronological age, vaccine status, and time to onset. The 3-to-6-month neurodevelopmental window provides the essential baseline neural vulnerability. The primary DTaP formulation acts as the specific immunological stressor, an effect not observed with concurrent inactivated vaccines. The strict 72-hour onset interval reflects the immediate innate immune response bridging this underlying age-related susceptibility to the acute clinical manifestation.

Studies have reported that PTx, an indispensable immune adjuvant in DTaP vaccines, may increase the risk of seizures in individuals with specific susceptibility backgrounds,³⁰ particularly in children with underlying encephalopathy.⁴ Similarly, Goodman et al.³¹ observed an increased incidence of infantile spasms following vaccination using a temporal shift model, suggesting a potential non-coincidental association. Furthermore, the acute reactogenicity of DTaP likely involves the synergistic effect of aluminum adjuvants. These adjuvants are known to rapidly activate the NLRP3 inflammasome, triggering the release of pro-inflammatory cytokines such as IL-1β within hours of administration.³² Mechanistically, PTx can induce polyclonal activation and effector function of CD8+ T cells, upregulate CD28 and CD69, and promote the production of IFNγ and IL-17.³³ Acting as a Th1/Th17 pathway activator, PTx may transiently upregulate pro-inflammatory cytokines via the IL-6/STAT3 axis. Although this effect typically resolves within 24–72 hours and is insufficient to induce sustained epileptogenic network remodeling,^34,35 for susceptible infants whose brains already harbor a “first hit” (e.g., TSC mutations, cortical dysplasia),¹⁶ the combined immune-enhancing effects of PTx and aluminum-induced innate signaling, along with vaccine-induced fever and nonspecific systemic inflammation, may be sufficient to destabilize the neuro-immune network and precipitate spasms. Since post-vaccination fever could not be adjusted in our regression model, we cautiously propose that this “second hit” functions as a composite physiological stressor, rather than exclusively through the direct neurotoxicity of specific vaccine components. Against the severe backdrop of global pertussis resurgence, this study provides evidence-based rationale for achieving safer vaccination practices in vulnerable populations. More importantly, we found that non-DTaP inactivated vaccines (e.g., Hepatitis B, IPV) did not increase seizure risk during the primary immunization period. This suggests that clinicians need not cease all vaccinations for high-risk infants, clinically defined as those harboring known preexisting neurological vulnerabilities or definitive epileptogenic etiologies (e.g., structural brain abnormalities, specific genetic defects, or previous severe encephalopathy). To ensure proactive identification before the first seizure occurs, this high-risk group should also encompass infants presenting with subtle “pre-symptomatic” indicators. Beyond a positive family history of early-onset epilepsy, such indicators include mild neurodevelopmental delays (e.g., motor or social milestones), persistent muscle tone abnormalities, or atypical neuro-behavioral patterns (such as exaggerated startle responses) identified during routine pediatric screenings. Instead, they should prioritize the completion of non-DTaP vaccines, exercising prudent management solely for DTaP until the child’s nervous system crosses the 3-to-6-month window of vulnerability. The long-term follow-up of 6.5 to 10.5 y in this study also revealed that although the Vaccination-proximate Group exhibited an earlier age of onset, it did not worsen the disease’s natural trajectory by causing additional damage. The long-term cognitive prognosis of IESS patients is fundamentally predetermined by their underlying etiology.³⁶ DTaP vaccination may have accelerated the clinical manifestation of the disease, but it did not fundamentally alter the developmental potential of the neural network. This study offers insights into a long-standing clinical dilemma, allowing clinicians to maintain long-term herd immunity defenses without sacrificing the neurological stability of vulnerable groups such as IESS patients.

Our study has several limitations. First, although vaccination dates were rigorously verified using medical records to minimize recall bias, the assessment of seizure onset timing relied partially on caregiver reports, which may introduce potential inaccuracies. Second, while the total sample size (n = 564) is significantly larger than in previous cohort studies, sample sizes in certain subgroups remained relatively small, and larger prospective studies are needed to definitively rule out rare risks. Third, constrained by the incompleteness of data records, we were unable to include post-vaccination fever as a covariate in the regression models for adjustment. Consequently, we cannot fully dissociate whether the triggering mechanism stems from pyrogenicity or direct neurotoxicity; however, this lack of distinction does not affect the clinical conclusion identifying DTaP as a risk factor.

Conclusion

Based on the full-course observation of school-aged children with IESS, this study identifies the DTaP vaccine as a specific precipitating factor for seizure onset within 72 hours post-vaccination in infants with underlying epileptogenic susceptibility. Crucially, this risk is strictly confined to the primary immunization series (Doses 1-3). We recommend implementing a risk-stratified immunization strategy for high-risk infants, adopting an approach of “timely administration of non-DTaP vaccines and deferred administration of DTaP” until the neurodevelopmental window of vulnerability closes. By identifying high-risk individuals through precision diagnosis and engaging in scientific risk communication, clinicians can effectively mitigate vaccine hesitancy, thereby maintaining herd immunity against pertussis while ensuring the neurological safety of vulnerable individuals.

Biographies

Jing Wang, M.D., Ph.D., is an Associate Chief Physician at the Department of Pediatrics, Chinese PLA General Hospital. She received her Ph.D. degree from the PLA Medical School. Dr. Wang specializes in pediatric neurology, and her research interests primarily focus on neuro-immunology and vaccination safety. She has authored over 30 papers in the field. She currently serves as a Council Member of the Beijing Association Against Epilepsy.

Li-Ping Zou, M.D., Ph.D., is a Chief Physician, Professor, and Doctoral Supervisor. She currently serves as the Director of the Academic Committee of the Pediatric Medical Department at the Chinese PLA General Hospital. She obtained her Ph.D. from the Karolinska Institute, Sweden, in 1999. With extensive experience in pediatric neurology and immunology, Professor Zou has actively presided over and participated in the handling and appraisal of numerous national public health emergencies related to infectious diseases and vaccination. Her research interests center on the intersection of neuro-immunology and vaccine safety. She has authored over 500 papers. She also serves as a Standing Committee Member of the Pediatrics Branch of the Chinese Medical Association and was honored with the Pediatric Outstanding Contribution Award in 2019.

Funding Statement

This work was supported by the National Key Research and Development Program of China [2023YFC2706304] and Innovation Talent Fund of Senior Department of Pediatrics, The Seventh Medical Center of PLA General Hospital [QZX-04-EKLHJH-2].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

De-identified data that support the findings and a data dictionary will be made available upon reasonable request via e-mail to the corresponding author.

Ethics statement

This retrospective cohort study was conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the Ethics Committee of the Chinese PLA General Hospital (Approval No. 2025–400), which waived the need for individual patient consent due to the retrospective nature of the analysis.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2026.2651584.