Background: Evaluation of injury severity and prognosis in TBI must be fast, yet scoring systems have different purposes, predictors, outcomes, validation populations, and clinical value.
Methods: This structured narrative review compiled knowledge about well-established and novel scoring systems assessing TBI severity and prognosis. PubMed/MEDLINE, Scopus, Web of Science, and Embase databases were systematically searched from the publication of major scoring systems up until December 2025, with further citation chasing. The comparison of the systems was performed by population, clinical setting, outcome, predictive ability, reliability, validation, and limitations of each.
Results: GCS, Marshall, Rotterdam, RTS, TRISS, FOUR, and APACHE II are still clinically useful; the Helsinki and Stockholm scoring systems add to CT-based characterization of TBI. IMPACT and CRASH utilize both clinical and radiological variables for formal prognostication, GOS-E is a functional outcome measure and BIG is useful for risk stratification of mild TBI pathways. Modern AI/ML models exhibit promising discriminatory ability, yet calibration, validation, reproducibility and generalizability are their shortcomings.
Conclusion: No single tool applies to all scenarios; GCS and RTS facilitate initial assessment. While, CT-based scores add radiologic information after neurologic examination. IMPACT/CRASH are more suited to prognostic evaluation.