Predictors for intracranial hemorrhage in mild traumatic brain injury

<p dir="ltr">Traumatic brain injury (TBI) is a significant contributor to morbidity and mortality globally, especially in its most severe forms. However, the most prevalent type is mild traumatic brain injury (mTBI), which constitutes a significant proportion of patients presenting to emergency departments (EDs) worldwide for assessment. There are existing guidelines to support management of these patients. However, there have been changes in both epidemiology and risk factors in recent years, particularly in the western world where an aging population with medically managed thrombotic disease is becoming an increased proportion of TBI patients. This thesis aims to explore the role of predictors available from the outset at the ED towards prediction of clinically significant outcomes such as traumatic intracranial hemorrhage (tICH), neurosurgical TBI cases and death due to TBI.</p><p dir="ltr">In Study I, a protocol was designed for the Stockholm scorE of LEsion detection on Computed Tomography following mild Traumatic Brain Injury (SELECT-TBI) study. The aim of the study is to generate an overarching retrospective observational population-based mTBI patient cohort using medical records from the Greater Stockholm Metropolitan Area. The cohort enables a detailed and comprehensive study of how clinically available variables perform as predictors against both the radiological outcome of intracranial lesions and the clinically definite outcome of neurosurgical TBI cases. In total, 73 key variables were determined to be relevant. The study was registered in Clinicaltrials.gov (NCT04995068) and ethical approval was obtained from the Swedish Ethical Review Authority (DNR 2020-05728).</p><p dir="ltr">In Study II, a systematic review and meta-analysis was performed to identify, quantify, and critically appraise risk factors associated with CT-confirmed tICH in mTBI patients, with the goal of informing future ED management guidelines. Seventeen studies, encompassing 26 040 patients with 2 054 cases of CT- verified tICH (7.9%), were included. Skull base fracture (odds ratio, OR 11.71, 95% confidence interval, CI 5.51-24.86) and Glasgow Coma Scale upon ED presentation <15 (OR 4.69, 95% CI 2.76-7.98) emerged as the strongest predictors of tICH. Additional significant predictors included loss of consciousness, post-traumatic amnesia, vomiting, antiplatelet therapy, and male sex. By contrast, commonly assumed risk factors such as anticoagulant use, headache, and intoxication were not predictive. Emerging evidence suggested possible roles for scalp lesions and older age, though inconsistencies precluded pooled synthesis. Risk of bias was moderate to high in most included studies due to inadequate control of confounding.</p><p dir="ltr">Study III is an interim analysis and statistical analysis plan based on the initial 5 000 patients included in the main SELECT-TBI cohort. Three modeling approaches were evaluated: generalized linear models (GLMs), random forest (RF) algorithms, and Lasso-regularized logistic regression (LR). Model performance was assessed using area under the receiver operating characteristic curve (AUC), calibration curves and Brier scores. Across models, the most consistent predictors of ICL were Glasgow Coma Scale (GCS) deterioration and score, signs of basilar skull fracture, high energy trauma mechanisms, and vomiting. S100B emerged as a strong predictor in biomarker-inclusive models. The Lasso regression model incorporating hemoglobin, platelet count, and S100B demonstrated the best performance (AUC 0.807 for any ICL, 0.903 for clinically significant ICLs, defined as ICLs resulting in intubation, neurosurgical transfer, or death).</p><p dir="ltr">In Study IV, a post-hoc analysis was conducted using data from the SELECT-TBI cohort from six emergency departments in the Stockholm region between 2015- 2020 focusing on the risk of tICH with antithrombotic medication. Logistic regression analyses, both univariable and multivariable, were applied to estimate ORs and 95% CIs, adjusting for confounding variables. Among 28 973 included mTBI patients, 9.6% had CT-verified tICH. Apixaban was associated with a reduced risk of tICH (adjusted OR 0.74, 95% CI 0.62-0.87), while acetylsalicylic acid was associated with increased risk (adjusted OR 1.20, 95% CI 1.07-1.35). In the CT-positive sub-cohort (n = 2 948), no antithrombotic agent predicted transfer to neurosurgical department. However, both warfarin (adjusted OR 3.62, 95% CI 1.76-7.15) and apixaban (adjusted OR 2.36, 95% CI 1.00-5.06) were independent predictors of TBI-related mortality. Over the study period, apixaban use increased markedly in the mTBI population, surpassing warfarin after 2018.</p><p dir="ltr">In summary, the studies demonstrate findings that reaffirm the validity of some established markers while challenging others, such as the complex risk profiles across antithrombotic subtypes. While aspirin use increased the likelihood of tICH, apixaban appeared safer with respect to initial hemorrhage but was associated with elevated mortality once a traumatic hemorrhage is present.</p><p dir="ltr">These findings provide some support for potential updates in ED management guidelines for mTBI patients to improve specificity. However, further prospective trials are necessary to validate our results before clinical implementation.</p><h3>List of scientific papers</h3><p dir="ltr">The present thesis is based on the following studies, which will be referred to by their Roman numerals (*shared authorship):</p><p dir="ltr">I. Fletcher-Sandersjöö, A., Tatter, C., <b>Yang, L.</b>, Pontén, E., Boman, M., Lassarén, P., Forsberg, S., Grönlund, I., Tidehag, V., Rubenson-Wahlin, R., Strömmer, L., Westberg, K., Ängeby, K., Djärv, T., Lundblad, O., Bartek, J., Jr, Thelin, E. P. Stockholm score of lesion detection on computed tomography following mild traumatic brain injury (SELECT-TBI): study protocol for a multicentre, retrospective, observational cohort study. BMJ Open, 2022;12(9):e060679. Published 2022 Sep 1. <a href="https://doi.org/10.1136/bmjopen-2021-060679" rel="noreferrer" target="_blank">https://doi.org/10.1136/bmjopen-2021-060679</a></p><p dir="ltr">II. <b>Yang, L.</b> J., Lassarén, P., Londi, F., Palazzo, L., Fletcher-Sandersjöö, A., Ängeby, K., Thelin, E. P .* , Rubenson Wahlin, R .* Risk factors for traumatic intracranial hemorrhage in mild traumatic brain injury patients at the emergency department: a systematic review and meta-analysis. Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine, 2024;32(1):91. Published 2024 Sep 17. <a href="https://doi.org/10.1186/s13049-024-01262-6" rel="noreferrer" target="_blank">https://doi.org/10.1186/s13049-024-01262-6</a></p><p dir="ltr">III. <b>Yang, L.</b> J .* , Tatter, C .* , Fletcher-Sandersjöö, A., Froese, L., Lassarén, P., Tjerkaski, J., Bergman, E. E., Björkman, F. E., Bronge, J., Antonsson, J., Teromaa, K., Nylander, M., Örtqvist, S., Kylander, W., Lindqvist, W., Ängeby, K., Rubenson Wahlin, R., Thelin, E. P. Stockholm Score of Lesion Detection on Computed Tomography following Mild Traumatic Brain Injury (SELECT-TBI) Study: Pilot Analysis and Statistical Analysis Plan Acta Neurochirurgica, 2025;167(1). Published 2025 Jul 1. <a href="https://doi.org/10.1007/s00701-025-06598-1" rel="noreferrer" target="_blank">https://doi.org/10.1007/s00701-025-06598-1</a></p><p dir="ltr">IV. <b>Yang, L.</b> J., Fletcher-Sandersjöö, A., Tatter, C., Froese, L., Lassarén, P., Patel, A., Ibrahim, A., Nykvist, A., Bivner Johansson, A., Brandt, E., Wager, E., Norman, E., Bergman, E. E., Björkman, F. E., Kratz, G., Lidström, H., Bergqvist, H., Hallden Kullander, I., Bronge, J., Antonsson, J., Wallin, K., Teromaa, K., Olofsson, L., Sandgren, L., Nylander, M., Ghorab, M., Mai, M., Hallongren, M., Wellén, R., Vesterlund, R., Clementsson, S., Faisal, S., Hassan, S., Örtqvist, S., Wocalewski, V., Hasselberg, V., Kylander, W., Lindqvist, W., Hallstedt, Z., Ängeby, K., Rubenson Wahlin, R., Thelin, E. P. The Effect of Antithrombotic Therapy on Traumatic Intracranial Hemorrhage and Additional Adverse Outcomes in Mild Traumatic Brain Injury Patients: A Post-hoc Analysis of the SELECT-TBI Cohort [Manuscript]</p>