Abstract
Abstract
Background
Severe traumatic brain injury (TBI) is a leading cause of trauma death, yet reported mortality varies widely. We estimated the pooled mortality of severe TBI and explored sources of variation.
Methods
We searched PubMed/MEDLINE, Embase, Scopus, Web of Science, the Cochrane Library and Google Scholar from inception to January 2026 (PROSPERO CRD420261417287). Observational and interventional studies reporting the number of deaths among patients with severe TBI (Glasgow Coma Scale [GCS] ≤8) were eligible. Two reviewers independently selected studies, extracted data and assessed risk of bias with the Newcastle-Ottawa Scale. Logit-transformed proportions were pooled with a DerSimonian-Laird random-effects model; we report 95% confidence intervals (CI) and 95% prediction intervals (PI). Heterogeneity was explored by subgroup analysis, meta-regression and pre-specified sensitivity analyses, and certainty was rated with GRADE for prognosis.
Results
Twenty-four studies (11,544 patients; 5,234 deaths) from 15 countries, published 2001–2024, were included. Study-level mortality ranged from 18.2% to 71.1%. Pooled mortality was 40.9% (95% CI 35.7–46.3%; I² = 96.3%), with a 95% PI of 18.9–67.2%. Mortality differed by region (p = 0.002): 47.1% (37.6–56.8%) in Africa, 44.5% (40.2–48.8%) in Europe, 38.6% (25.4–53.7%) in Asia and 31.6% (26.5–37.2%) in the Americas. Mortality did not differ significantly by country income group, outcome time point, age group or publication year. Estimates were robust to alternative random-effects estimators, leave-one-out analysis (range 39.6–41.7%) and restriction to studies with independently verified counts (41.7%) or low risk of bias (38.4%). Certainty of evidence was low.
Conclusions
Approximately two in five patients with severe TBI die, but the wide prediction interval shows that the expected mortality in any single setting may lie anywhere between roughly one in five and two in three. Mortality was highest in African cohorts. Standardised definitions, outcome time points and reporting are needed to allow meaningful comparison across trauma systems.