Concordance between prehospital Glasgow Coma Scale–based classification and diagnosis-based injury severity in destination decisions for suspected severe trauma: a linked EMS–hospital cohort study

The Glasgow Coma Scale (GCS) measures level of consciousness. Its agreement with hospital diagnosis-based injury severity among patients identified by emergency medical services (EMS) as having suspected severe trauma remains uncertain. This population-based cohort study analyzed 6,179 patients aged ≥ 15 years with suspected severe trauma in South Korea (2022–2023), linking 119 EMS run-sheet records with National Emergency Department Information System data. Destination patterns were evaluated using two complementary severity definitions: prehospital GCS 3–8 as the primary classification and a provider-supplied binary indicator of International Classification of Diseases-based Injury Severity Score (ICISS)-2020 < 0.9 as a secondary diagnosis-based reference. Destination undertriage was quantified for each severity definition. Level 1–2 transport proportions were reported descriptively and separately for patients with GCS 13–15 and for patients with ICISS ≥ 0.9. Associations between each prehospital time interval and initial-hospital mortality were assessed in separate regression models adjusted for age, sex, and weighted Revised Trauma Score (RTS). Among 5,841 patients with complete data on both severity criteria, prehospital GCS-based classification and post-hospital ICISS-based severity showed weak positive concordance (Cohen’s κ = +0.192, 95% CI + 0.167 to + 0.217; sensitivity of GCS 3–8 for detecting ICISS < 0.9, 54.2%). One in ten patients (10.0%) met the anatomical criterion despite GCS 9–15 — a subgroup not captured by GCS-based classification alone. Destination undertriage was substantial under both definitions (GCS-based 33.6%; ICISS-based 47.3%), with a marked rural disparity in each (adjusted odds ratios 2.65 and 2.04, respectively). Crude inverse associations between transport time and mortality were attenuated after adjustment for age, sex, and weighted RTS. Among patients identified by EMS as suspected severe trauma, prehospital GCS-based classification and post-hospital ICISS-based severity showed only weak positive concordance, with approximately one in ten patients meeting the anatomical criterion despite GCS 9–15. Destination undertriage was substantial under both definitions, particularly in rural areas. These findings concern GCS-based classification within an EMS-selected cohort and do not establish the performance of the complete 119 field triage protocol. They support further evaluation of destination decisions and rural access to higher-capability care.