Predictors of Intracranial Hemorrhage on Initial Head Computed Tomography After Mild Traumatic Brain Injury

Abstract

Background/Aim

Traumatic intracranial hemorrhage after mild traumatic brain injury (mTBI) remains a major clinical challenge, particularly in aging societies. This study characterized the incidence, clinical relevance, and associated clinical factors of computed tomography (CT)-positive traumatic intracranial lesions in adults with mTBI.

Patients and Methods

Relevant data were retrospectively collected from adults with mTBI who underwent head CT at our Emergency Department between April 2023 and April 2024. Clinical characteristics, medication exposure, CT findings, and the need for neurosurgical intervention were analyzed. CT images were interpreted in routine practice by Board-certified radiologists, and CT-positive cases were classified according to the Marshall CT criteria.

Results

Of 471 patients (mean age: 70 years), 66 (14.0%) had CT-confirmed traumatic intracranial hemorrhage. Male sex and antiplatelet use were more frequent in the CT-positive group than in the CT-negative group (59.1% vs. 39.8%, p=0.003; and 9.1% vs. 2.7%, p=0.021, respectively). Falling was the predominant mechanism of injury (86.6%). Traumatic subarachnoid hemorrhage (50.0%) and intracerebral hemorrhage or contusion (42.4%) were the most common CT findings. Most CT-positive cases were Marshall grade II (93.9%), and eight patients (12.1% of CT-positive cases; 1.7% of the total cohort) required neurosurgical intervention. In a prespecified descriptive subgroup analysis of patients with numerically documented triage Glasgow Coma Scale scores of 13-15, clinically relevant CT-positive lesions and neurosurgical interventions were still observed in the GCS 15 subgroup.

Conclusion

In this CT-selected older cohort with mTBI, CT-positive traumatic intracranial lesions were not rare. Although most lesions were radiologically mild, a clinically important minority required neurosurgical intervention. These findings support emphasizing radiological severity and downstream management, not CT positivity alone, when evaluating older adults with mTBI.

Introduction

Mild traumatic brain injury (mTBI) is a common event at Emergency Departments (EDs). It is defined by a Glasgow Coma Scale (GCS) score of 13 to 15, loss of consciousness for ≤30 min, and post-traumatic amnesia for ≤24 h (1). The incidence of traumatic intracranial hemorrhage (tICH) after mTBI ranges from 3.7% to 33.9% (2). Key risk factors for mTBI-associated tICH include a low GCS score, signs suggesting skull fracture, vomiting, advanced age, higher-risk mechanisms, and medication-related bleeding risk (1-5). Even with oral anticoagulant use, the overall incidence of tICH after mTBI remains 9.4% (3). Although most patients with mTBI-associated tICH can be managed conservatively, only a minority ultimately require neurosurgical intervention (1, 6).

Several clinical decision rules, such as the Canadian CT Head Rule (CCHR) and the New Orleans Criteria (NOC), have been proposed to achieve a balance between radiation exposure from computed tomography (CT) and mTBI-associated tICH diagnosis (4, 5). However, reported rates of CT-positive tICH vary considerably across studies due to differences in patient selection, definitions of mTBI, and institutional CT imaging protocols (1, 7). These decision tools are designed to prioritize sensitivity for clinically important outcomes, which can yield low specificity and variable CT utilization patterns across different settings (4, 5, 7). Moreover, CT-positive findings in older adults are not necessarily homogeneous in severity or downstream management implications (1). Therefore, in this study, we aimed to characterize the observed rate of CT positivity, radiological severity, and clinical relevance of traumatic intracranial lesions in local ED practice and to determine how commonly reported risk signals align with CT-positive findings in an older, fall-prone cohort.

Patients and Methods

Design and participants. Ethical approval for this study was obtained from the Institutional Review Board (approval number: CGH-P114070). This single-center, retrospective cohort study was conducted between April 2023 and April 2024. Adults who presented with blunt head trauma and an initial GCS score of 13-15 were screened for eligibility. For this study, mTBI was operationally defined as a triage GCS score of 13-15 in the absence of penetrating injury or polytrauma requiring immediate resuscitation.

A brain CT scan was performed at the attending emergency physician’s discretion, primarily guided by the CCHR and NOC and by other institutional protocols for evaluating head injuries. Patients were included in the analysis when their index emergency visit occurred during the study period and sufficient clinical and radiological data were available. Patients with penetrating head injury, those who were transferred from outside hospitals without accessible CT images or reports, those with a history of neurosurgical procedures or known intracranial lesions that might interfere with radiological interpretation, those with duplicate encounters, and those with missing key variables required for the primary analysis were excluded.

Outcome measure. Clinical data, including demographic characteristics (age and sex), injury mechanisms, presenting symptoms, external signs of head injury, comorbidities, and medication histories, were reviewed and recorded. Radiological findings were obtained from official reports interpreted by Board-certified radiologists. Patients were categorized into two groups based on their initial brain CT findings: CT-positive and CT-negative. CT positivity was defined by the presence of any acute traumatic intracranial lesion or condition, such as subarachnoid hemorrhage, intracerebral hemorrhage, subdural hematoma, epidural hematoma, intraventricular hemorrhage, or skull fracture. Hemorrhage type and severity were further classified according to the Marshall CT criteria (8).

Statistical analysis. All statistical analyses were performed using SPSS (version 29; IBM, Armonk, NY, USA). Continuous variables are presented as the mean±standard deviation, whereas categorical variables are presented as counts with percentages. Between-group comparisons were performed using the independent-samples t-test or Mann–Whitney U-test for continuous variables, and the chi-square test or Fisher’s exact test for categorical variables, as appropriate. Given the limited number of events in several exposure subgroups, the analyses were primarily descriptive and univariate, intended to identify factors associated with CT positivity. A two-tailed value of p<0.05 indicated statistical significance.

Results

Patient characteristics and clinical presentation. A total of 613 patients with mTBI were initially identified. After applying the selection criteria, 471 patients who underwent initial head CT were included in the final analysis (Table I). The patients’ mean age was 70.0±19.4 years. Of them, 42.5% were men. Falling emerged as the most common mechanism of injury (86.6%), followed by motor vehicle accidents (10.0%). Scalp laceration and subgaleal hematoma were reported in 12.1% and 41.4% of all patients, respectively. Hypertension (45.4%) and diabetes mellitus (28.0%) were identified as the most prevalent comorbidities, whereas coronary artery disease (12.3%), hyperlipidemia (6.2%), Parkinson’s disease (6.6%), and dementia (6.8%) were less frequent. Use of antiplatelet and anticoagulant medications was reported by 3.6% and 1.9% of patients, respectively.

Table I. Characteristics of the computed tomography (CT)-negative and CT-positive groups (N=471).

CAD: Coronary artery disease; CKD: chronic kidney disease; CLD: chronic lung disease. GCS: Glasgow Coma Scale; DM: diabetes mellitus; ESRD: end-stage renal disease; MVA: motor vehicle accident. Statistically significant p-values are shown in bold.

Clinical factors associated with CT positivity. Of all 471 patients, 66 (14.0%) had CT-positive traumatic intracranial findings. No significant difference in mean age was observed between the CT-positive and CT-negative groups (70.1±18.6 vs. 69.9±19.6 years, p=0.940). However, male sex was significantly associated with CT positivity (59.1% vs. 39.8%, p=0.003). The distribution of triage GCS categories 13-15 differed significantly between the CT-negative and CT-positive groups (p=0.024), but this difference did not indicate a clinically meaningful trend toward lower GCS in the CT-positive group, as most patients in both groups had a GCS score of 15. By contrast, no significant between-group difference was observed in the overall distribution of injury mechanisms (p=0.228) or in external scalp lesions, such as scalp lacerations or subgaleal hematomas. Similarly, no correlations were observed between common comorbidities and a positive CT result. Notably, antiplatelet use was significantly more common in the CT-positive group than in the CT-negative group (9.1% vs. 2.7%, p=0.021), whereas anticoagulant use exhibited no significant effect.

Patterns of tICH in CT-positive patients. Among the 66 patients with mTBI-associated tICH, most CT-positive cases were Marshall grade II (93.9%), indicating diffuse lesions without mass effect, followed by grade I (3.0%), grade III (1.5%), and grade VI (1.5%). No patients were categorized as having Marshall grade IV or V lesions. For the analysis of tICH type, subarachnoid hemorrhage emerged as the most common lesion (50.0%), followed by intracerebral hemorrhage (42.4%) and subdural hematoma (24.2%) (Table II). Epidural hematoma was observed in 9.1% of the patients. However, intraventricular hemorrhage was observed in only 1.5% of the patients. Skull fractures were identified in 4.5% of all patients in the CT-positive group. Mixed hemorrhagic patterns (two or more lesion types) were identified in 10.6% of the patients. Overall, these findings indicated that traumatic subarachnoid hemorrhage and intracerebral hemorrhage were the predominant radiological patterns observed in patients with mTBI-associated tICH.

Table II. Hemorrhage subtypes in the computed tomography (CT)-positive group.

Neurosurgical intervention. Overall, eight patients (12.1% of the CT-positive cases and 1.7% of the total cohort) required neurosurgical intervention. The comparison of conservative treatment and neurosurgical intervention in 66 CT-positive patients is summarized in Table III. Significant differences were observed in the distributions of injury mechanisms and triage GCS categories between the two groups (p=0.025 and p<0.001, respectively). Falls remained the predominant mechanism of injury in both groups, accounting for 84.5% of conservatively managed patients and 75.0% of surgically treated patients. Although a triage GCS score of 15 was the most common presentation overall, patients requiring neurosurgical intervention were more likely to present with a GCS score of 14 than those managed conservatively (37.5% vs. 1.7%). Subdural hematoma was significantly more frequent among patients who underwent neurosurgical intervention than among those managed conservatively (62.5% vs. 19.0%, p=0.017). These findings suggest that a triage GCS score of 14 or the presence of subdural hematoma may be associated with an increased likelihood of requiring neurosurgical intervention in patients with CT-positive mTBI.

Table III. Characteristics of the computed tomography-positive group by treatment (N=66).

CAD: Coronary artery disease; CKD: chronic kidney disease; CLD: chronic lung disease. GCS: Glasgow Coma Scale; DM: diabetes mellitus; EDH: epidural hematoma; ESRD: end-stage renal disease; ICH: intracerebral hemorrhage; IVH: intraventricular hemorrhage; MVA: motor vehicle accident; N/A: not calculated because these variables were constant (no events); SAH: subarachnoid hemorrhage; SDH: subdural hematoma. Statistically significant p-values are shown in bold.

Discussion

Clinical decision rules and contemporary challenges in mTBI evaluation. Major clinical decision rules for adult mTBI consistently prioritize changes in neurological status and a limited set of high-yield symptoms or signs to identify patients at risk for tICH (4, 5, 7). In external validation work, the CCHR is typically operationalized using high- or medium-risk features such as failing to reach a GCS score of 15 within 2 h, having suspected open or depressed skull fracture, having any signs of basilar skull fracture, vomiting, being aged ≥65 years, having retrograde amnesia, and being exposed to dangerous mechanisms of injury (5). By contrast, the NOC emphasizes having headaches, vomiting, being aged >60 years, having drug or alcohol intoxication, having persistent anterograde amnesia (short-term memory deficit), experiencing a seizure, and sustaining a visible trauma above the clavicles, reflecting a strong focus on clinical instability (4). Current decision rules may lead to the overuse of CT, resulting in unnecessary radiation exposure, prolonged ED stays, increased healthcare costs, and unnecessary use of healthcare imaging resources (2, 9). In addition, evolving risk profiles, such as the widespread use of direct oral anticoagulants and the increasing prevalence of an aging mTBI population, may limit the applicability of these guidelines (3).

Incidence and clinical predictors of CT-positive tICH. In a systematic review and meta-analysis of 26,040 patients with mTBI, Yang et al. reported that 2,054 patients had tICH (total prevalence: 7.9%) and the significant risk factors were signs of skull base fracture [odds ratio (OR)=11.71], GCS scores of <15 (OR=4.69), consciousness loss (OR=2.57), post-traumatic amnesia (OR=2.13), vomiting (OR=2.04), antiplatelet treatment (OR=1.54), and male sex (OR=1.28) (2). In our cohort, the incidence of mTBI-associated tICH was 14%, within the reported range but higher than the overall prevalence. This discrepancy may be attributable to the overall population being older (mean age: approximately 70 years), although no significant age difference was observed between the CT-negative and CT-positive groups (69.9±19.6 years vs. 70.1±18.6 years, p=0.94). Overall, our findings indicate that age functioned more as a background risk marker than as a discriminator once the patients met the mTBI criteria and underwent CT evaluation. The findings further confirmed that males were significantly over-represented in the CT-positive group (59.1% vs. 39.8% in the CT-negative group, p=0.003).

In another systematic review and meta-analysis of 11,172 patients with mTBI treated with anticoagulants, Karamian et al. reported immediate ICH rates of 6.4% and 10.5% among patients taking oral anticoagulants and vitamin K antagonists, respectively, with an overall immediate ICH incidence of 8.5% (3). The reviewers further indicated that delayed ICH occurred in 1.6% and 1.9% of all patients, respectively, yielding an overall rate of 1.7%. In our present study, a significantly higher prevalence of antiplatelet use was observed in the CT-positive group than in the CT-negative group (9.1% vs. 2.7%, p=0.021). However, we found no significant association between anticoagulant use and CT-confirmed ICH. This discrepancy suggests differences in patient selection and contemporary prescription patterns, whereby antiplatelet therapy remains prevalent in older patients with mTBI and vascular comorbidities, and may contribute to a distinct hemorrhage risk profile not fully captured by anticoagulation-based risk stratification.

Injury mechanisms and radiological patterns of tICH. In their ED cohort study conducted in Finland, Isokuortti et al. reported that ground-level falls accounted for more than half of all confirmed mTBI cases, whereas motor vehicle accidents and other traffic-related collisions accounted for smaller fractions (1). In our study, the mechanism of injury was primarily falls (86.6% falls, 10.0% traffic-related collisions, and 3.4% other causes), aligning with our diverse cohort being composed of older patients and reinforcing the notion that mTBI positivity benchmarks cannot be interpreted without accounting for who is being scanned and why (1, 10). Notably, tICH can be categorized into extra-axial lesions (epidural hematoma, subdural hematoma, and traumatic subarachnoid hemorrhage), intra-axial lesions (contusion and intraparenchymal hematoma), and intraventricular hemorrhage, each reflecting distinct vulnerability profiles of the meningeal layers, bridging veins, cortical surfaces, and parenchyma (10). In the study by Isokuortti et al.,subdural hematoma (11.7%), traumatic subarachnoid hemorrhage (7.0%), and contusion (5.5%) were among the most common abnormalities in patients with confirmed mTBI, with epidural hematoma (0.2%) being comparatively uncommon (1). By contrast, Riemann et al. reported that traumatic subarachnoid hemorrhage (48%), contusion (40%), and epidural hematoma (37%) were their most common abnormalities (9). In our CT-positive group, subarachnoid hemorrhage (50.0%) and intraparenchymal hemorrhage or contusion (42.4%) were the most common findings, followed by subdural hematoma (24.2%) and epidural hematoma (9.1%). Of our patients, 10.6% had combined hemorrhage patterns, and 4.5% had skull fractures identified on CT. Overall, this pattern of mTBI-associated tICH highlights the heterogeneity of CT findings in mTBI across injury mechanisms and hospital levels.

Neurosurgical intervention and clinical implications. Progressive tICH may be a life-threatening condition in patients with mTBI, who particularly require surgical intervention (3). A large-scale external validation study revealed that both the CCHR and NOC achieved 100% sensitivity for neurosurgical intervention, but specificity was markedly lower for the NOC than for the CCHR (5). Another study reported a neurosurgical intervention rate of approximately 0.4% to 1.3% among patients with mTBI-associated tICH (7, 11). In our cohort, most CT-positive lesions were grade II (93.9%), indicating diffuse lesions without mass effect. Nevertheless, eight patients required neurosurgical intervention, corresponding to 1.7% of the total cohort and 12.1% of CT-positive cases. Our results also suggest that although most CT-positive lesions following mTBI are radiologically mild and can be managed conservatively, patients undergoing neurosurgical intervention were more likely to present with a triage GCS score of 14 or subdural hematoma (62.5% vs. 19.0%, p=0.017). Taken together, these findings indicate that even among patients classified as having mTBI, subtle reductions in neurological status and specific hemorrhagic patterns may identify individuals at increased risk for clinical deterioration and the need for subsequent surgical intervention. Repeat CT scans, rather than purely routine surveillance tests, are advised in cases of acute neurological deterioration among patients with mTBI with CT positivity (6, 12).

Study limitations. This study has several limitations that should be acknowledged, particularly in regard to aging societies. Firstly, this was a single-center, retrospective study conducted in a population with a markedly advanced mean age. This design may limit the generalizability of our findings to younger or more diverse mTBI cohorts. Additionally, the high prevalence of older patients inherently increases exposure to antiplatelet and anticoagulant therapies, which potentially alter the risk profile and clinical presentation of tICH. Secondly, medication use was captured from medical records at presentation. This approach may not fully capture adherence, dosage, or recent changes, limiting granular assessments of hemorrhage risk across different agents. Thirdly, because event counts were limited in several exposure subgroups, particularly anticoagulant users, our analyses were primarily univariate and descriptive, which precludes firm inferences about independent predictors after adjustment for confounders. Fourthly, this study focused on immediate CT findings and did not systematically evaluate delayed tICH, which is particularly relevant in older patients and those receiving antithrombotic therapy. Finally, functional outcomes and long-term neurological sequelae were not assessed, which restricts the interpretation of the broader clinical implications of CT-positive mTBI in super-aging populations. Collectively, these limitations underscore the need for prospective, multicenter studies tailored to older populations with contemporary antithrombotic exposure to better define risk stratification and management strategies for mTBI-associated tICH.

Conclusion

In this CT-selected cohort of patients with mTBI, tICH on initial head CT was not uncommon. Male sex and antiplatelet exposure were more frequently observed among patients with CT-positive findings. Although most CT-positive lesions, particularly traumatic subarachnoid hemorrhage and intracerebral hemorrhage/contusion, were radiologically mild and predominantly Marshall grade II, a clinically meaningful minority still required neurosurgical intervention. These findings underscore that CT-positive mTBI in older, fall-prone patients is clinically heterogeneous rather than uniformly benign and support a context-specific approach to risk assessment that considers radiologic severity and downstream management, not CT positivity alone.

Conflicts of Interest

No potential conflicts of interest relevant to this article were reported.

Authors’ Contributions

Conceptualization: CYL, CCL, and CTH. Data curation: CYL and MHH. Formal analysis: CYL, MHH, and TLY. Methodology: CYL, CCL, and CTH. Visualization: CYL. Writing - original draft: CYL. Writing - review and edit: CCL and CTH. All Authors have read and agreed to the published version of the manuscript.

Artificial Intelligence (AI) Disclosure

No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.