Spontaneous intracranial hemorrhage (sICH) is a rare but critical neurological emergency in children (1, 2). Despite its low overall incidence, the risk of death and long-term neurological complications remains high (3, 4). Survivors frequently suffer from permanent neurological deficits, which limit their daily functional independence and place a significant economic, emotional, and social burden on their families (5–7). With the widespread use and advancement of imaging technology, the survival rate and overall prognosis of children with sICH have improved (8, 9). However, due to the varied etiology, the cause of hemorrhage in some children remains unknown even after extensive imaging and laboratory examinations, highlighting the limitations of current diagnostic techniques and the incomplete understanding of the disease pathophysiology (10–12).

Pediatric sICH is highly variable due to the interplay of clinical and imaging factors. Previous research has shown that hematoma volume and location, and altered mental status are important independent predictors of neurological outcome (7, 13). Clinical scoring systems, such as the Pediatric Intracerebral Hemorrhage (PICH) score and the modified PICH (mPICH) score, have been developed for risk classification in sICH among children (13, 14). These models have substantial therapeutic utility, however, they are primarily concerned with anatomical and physiological variables, leaving out etiological and clinical elements that may have a significant impact on prognosis. Combining etiological classification with early clinical features can improve the accuracy and clinical utility of prognostic predictions, allowing for the timely identification of high-risk children (7, 14, 15). This is critical for making clinical judgments about neurosurgical procedures and intensive care monitoring (4). Furthermore, survivors may require long-term multidisciplinary care, which includes neurosurgery follow-up, rehabilitation, and cognitive support, emphasizing the importance of treatment planning and customized prognostic assessment (6). Understanding how etiology and early clinical factors influence prognosis is critical for informing treatment decisions and developing public health policies (2, 3, 16).

Our study sought to define the etiological spectrum of pediatric sICH and investigate the association between early clinically related features and neurological outcomes. We used an eight-year cohort to examine how underlying causes and initial presentation affect prognosis. These findings aim to support risk stratification and guide evidence-based clinical management (3).

In all, 148 children with sICH were enrolled. Patients were divided into four etiological groups: unknown causes (n = 50), vascular causes (n = 52), other defined causes (n = 18), and blood-related causes (n = 28) (Figure 2). Vascular causes accounted for the majority of sICH cases (n = 52, 35.1%), including arteriovenous malformations (AVMs, n = 44), cavernous hemangiomas (n = 6), Moyamoya syndrome, and one AVM with thrombosis. Hematologic disorders were identified in 28 patients (18.9%), primarily due to vitamin K deficiency (n = 16), hemophilia (n = 7), fibrinogen deficiency (n = 3), and thrombocytopenia (n = 2). Other defined causes (n = 18, 12.2%) comprised central nervous system infections (n = 11), intracranial tumors (n = 6) and one case secondary to hypertension. The remaining 50 cases (33.8%) had no identifiable etiology despite comprehensive evaluation and were classified as idiopathic or cryptogenic.

We described the etiological distribution of pediatric sICH and identified early clinical predictors of neurological sequelae, emphasizing the importance of combining clinical insights with etiological information (3, 7, 10). Within this approach, we first looked at the distribution of underlying causes and their clinical significance. Vascular abnormalities emerged as the major cause in our sample, mostly affecting school-aged children and adolescents (10). Age-related differences in cerebrovascular development and hemodynamic stress may contribute to the higher hemorrhage tendency observed in pediatric AVMs compared with adults (17, 18). This trend has important implications for therapy. Most children with vascular abnormalities require surgery or endovascular treatment to avoid rebleeding, and early, focused intervention is typically associated with better results than those with secondary causes (19). Children with coagulopathies or blood abnormalities, on the other hand, have a greater risk of complications and death. Impaired coagulation and widespread bleeding can expand the hematoma, raise intracranial pressure, and cause subsequent brain damage. Prompt identification and correction of coagulation issues is consequently critical for stabilizing patients and enhancing recovery (11, 20, 21).

Secondary intracerebral hemorrhage caused by factors other than coagulopathies, such as infections, tumors, and hypertension, accounted for 12.2% of cases and was a strong predictor of poor prognosis in multivariate models, demonstrating the clinical importance of recognizing systemic and secondary etiologies in pediatric sICH (10, 11). Although the hemorrhage volume in these children is often small, the primary disease carries a significant systemic burden, and their poor prognosis may reflect the combined effects of the underlying pathology and the hemorrhage itself. In many such patients, neurological deterioration may be primarily caused by the primary disease itself, rather than the size or location of the hemorrhage. Furthermore, symptoms of the primary disease often mask early neurological manifestations, leading to delays in diagnosis and treatment. Given the relatively small number of cases in each subtype, further stratified analysis was not possible; therefore, the observed increased risk should be interpreted as reflecting the combined severity of these systemic etiologies rather than a single hemorrhage mechanism (11, 12). In addition, we found a high proportion of secondary or systemic causes in this study, such as hematologic disorders, infections, and tumors, primarily affecting infants and young children. These causes themselves generally have poor prognoses. Therefore, when interpreting the age-prognosis relationship in spontaneous intracranial hemorrhage in children, it is crucial to consider the heterogeneity of underlying causes, rather than solely relying on age itself (22).

Children with unknown etiology accounted for approximately one-third of the cohort, representing a substantial proportion of pediatric sICH cases. Their outcomes fell between those of the vascular and secondary etiology groups. These children may have hidden vascular malformations or temporary coagulation problems that are difficult to detect with standard imaging (10, 11, 23). In our series, 29 patients (58%) in the unknown group showed cerebrovascular developmental anomalies on MRI. During the follow-up period, two individuals (4%) had clear vascular abnormalities identified. This suggests that many cryptogenic cases involve mild or progressive vascular abnormalities that are missed at first diagnosis. These cases highlight both the limitations of existing imaging techniques and our incomplete understanding of pediatric cerebrovascular development. Future advances in imaging resolution and analytical techniques may help improve diagnostic accuracy and etiologic classification (1, 9, 12).

Further investigation indicated disparities in the location of bleeding between etiologies, and the position of the lesion was strongly associated with prognosis. Deep intracerebral hemorrhages, particularly those involving the basal ganglia and thalamus, were associated with poorer neurological outcomes. This finding is supported by Park and Jang (24). In contrast, lobar and subdural hemorrhage generally indicated a better prognosis (6, 7, 16). Vascular malformations most commonly cause lobar hemorrhage, which has good surgical accessibility and a more favorable prognosis (10, 19). Hematologic disorders cause multi-site bleeding, with lobar hemorrhage combined with subdural hemorrhage being especially common in this group. Patients with multifocal lesions, especially those with hematologic disorders, have a higher incidence of adverse outcomes, with a mortality rate of 17.9%, indicating the impact of diffuse lesions and coagulation disorders (6, 25, 26). We found no association between poor functional outcomes and intraventricular extension or infratentorial, which is consistent with previous findings (27). These imaging patterns yield significant prognostic clues. The anatomical location of the lesion can direct acute-phase treatment, such as decompression for lobar hemorrhage, as well as long-term risk prediction (8, 16).

Vomiting at onset was associated with favorable outcomes in our cohort, a finding not previously reported in pediatric sICH. The underlying mechanism remains unknown, however, several explanations are possible. Vomiting implies that brainstem reflexes are intact, as is some autonomic regulation. Furthermore, vomiting may prompt patients to seek medical attention immediately, allowing for timely treatment before neurological deterioration begins. These characteristics could assist in explaining the relationship between vomiting and a better prognosis. However, these interpretations remain hypothetical, as no prior evidence has demonstrated a protective effect of vomiting in pediatric sICH. Additional multicenter studies are needed to corroborate this discovery and explain its physiological basis. In contrast, early seizures occurred in 30.4% of patients and were independently associated with poor prognosis. This differs from the findings of Williams et al. (28), who reported seizures as protective against in-hospital mortality. The disparity could be attributed to differing study endpoints: we evaluated long-term neurological outcomes, whereas their research focused on short-term survival. Seizures significantly increase brain metabolic demand and oxygen consumption, resulting in secondary hypoxia and worsened neuronal damage. Early seizure management and ongoing EEG monitoring are critical (29–31).

A low GCS score upon admission was consistently related to a poor prognosis, supporting the GCS score’s significant predictive value (2, 3, 22). Combining GCS with clinical and etiologic factors can further improve predictive accuracy, as demonstrated in models such as mPICH (13, 32). Complications and overall systemic status also play a major role in outcomes. Complications occurred in 66.2% of patients and included both neurological and systemic problems. Hydrocephalus occurred in 9.5% of cases, mostly associated with intraventricular extension. Early drainage or shunt procedures were beneficial for recovery (33, 34). Other reversible problems, including anemia, electrolyte imbalance, and pneumonia, typically lengthen hospital stays but are easily managed (22, 26, 29). These complications likely reflect the overall severity of the disease, highlighting the importance of close monitoring. Laboratory test findings can also indicate the body’s systemic stress level and the degree of bleeding. In univariate analysis, lower hemoglobin and fibrinogen levels were related to a worse prognosis; however, in multivariate models, they lost significance, suggesting that they primarily reflect disease severity (35–37). Maintaining adequate hemoglobin and coagulation function remains essential to prevent hematoma expansion.

Mortality in this cohort was 8.8%, which differs from the mortality rate previously reported in prior pediatric ICH studies (6, 7). The difference likely stems from variations in patient mix, inclusion criteria, and follow-up duration. Deaths mainly occurred in children with secondary etiologies, emphasizing that systemic or primary disease rather than hemorrhage volume is often the critical factor (10, 15). No deaths were observed among unknown cases, though this may reflect small sample size or selection bias rather than a truly benign nature. Rebleeding occurred in 6.1% of patients, most within the vascular or hematologic groups. Reports on pediatric sICH recurrence are limited and inconsistent, and no standardized follow-up imaging protocol currently exists (8, 38). Targeted vascular imaging and hematologic monitoring may help reduce recurrence risk. Continuous post-discharge surveillance is essential for these high-risk patients (16).

In our study, symptomatic epilepsy occurred in 6.8% of patients, primarily among those with secondary etiologies. Early cortical irritation may contribute to the development of persistent epilepsy. Persistent epilepsy greatly affects quality of life and long-term rehabilitation, emphasizing the need for early seizure prevention and continuing neurological monitoring. Epilepsy may represent a continuum from acute brain injury to sustained functional impairment. Structured follow-up for high-risk children, including EEG monitoring and neuropsychological examination, may help detect subclinical epileptiform activity before significant functional decline occurs, but direct evidence is lacking (29, 39). Data on post-ICH epilepsy and cognitive outcomes in children remain scarce, and unified models for long-term prognostic prediction are lacking. Large multicenter prospective studies with standardized follow-up are needed to develop robust prognostic models incorporating etiologic, imaging, and neurophysiological measures.

Treatment should be individualized according to etiology. Surgical or endovascular repair of vascular malformations, early correction of coagulopathy, and infection control were all associated with improved outcomes, underscoring the value of etiology-based management (40, 41). Despite these findings, our single-center retrospective design has inherent limitations, including a relatively small sample size and evolving diagnostic resources over the study period. The absence of dynamic laboratory and imaging data limits the assessment of disease progression; ongoing monitoring of hematologic and inflammatory markers may provide better insight into evolving injury (16, 42). Future studies should incorporate enhanced neuroimaging, radiomics, and molecular biomarkers to improve diagnostic accuracy. Machine learning models can help with personalized prognosis and decision-making, resulting in more precise early intervention (43, 44). Establishing a multi-center collaborative database will help overcome sample size constraints and support the development and validation of standardized risk assessment tools for pediatric spontaneous cerebral hemorrhage (45). These efforts will lay a solid platform for more precise and personalized management of children’s sICH.

This study focuses on the role of etiology and early neurological status in defining the outcomes of children with spontaneous intracranial hemorrhage. Onset seizures, lower GCS scores, and secondary causes such as tumor- or infection-related bleeding were all associated with a poor prognosis, but vomiting appeared to be protective. These findings demonstrate the importance of early etiologic identification, rapid neuroimaging, and timely multidisciplinary intervention. Optimizing early management and referral pathways may improve neurological recovery and survival in affected children. Future multicenter, prospective studies with larger cohorts are warranted to validate these results and explore underlying pathophysiological mechanisms.