Question: Does a prognostic model accurately estimate risk for early death in patients with traumatic bleeding?
Design: 2 cohort studies: Clinical Randomisation of an Antifibrinolytic in Significant Haemorrhage (CRASH-2) trial (derivation) and Trauma Audit and Research Network (TARN) dataset (validation).
Setting: 274 hospitals in 40 countries (derivation cohort) and 60% of hospitals that received trauma patients in England and Wales and some hospitals in Europe (validation cohort).
Patients: Derivation cohort: 20 127 trauma patients with, or at risk for, significant bleeding within 8 hours of injury (median age 30 y, median 2 h from injury). Validation cohort: 14 220 patients > 15 years of age (median age 39 y, median 1 h from injury) who arrived at hospital alive and had ≥ 1 of death from injury during admission, hospitalization > 3 days, need for intensive or high-dependency care, or need for interhospital transfer for specialist care. Patients who had isolated closed limb injuries or were > 65 years of age with isolated fractured neck of femur or pubic ramus fracture were excluded. DESCRIPTION OF PREDICTION GUIDE: The prognostic model included country (low, middle, or high income), age, time since injury, Glasgow coma score, systolic blood pressure, respiratory rate, heart rate, and type of injury (blunt or penetrating) (available at www.crash2.lshtm.ac.uk).
Outcome: Early death (in-hospital death within 4 wk of injury).
Main results: 15% and 12% of patients from the derivation and validation cohorts died. The operating characteristics of the model in the derivation and validation cohorts are in the Table.
Conclusion: An 8-factor prognostic model predicted early death in patients with traumatic bleeding.