Lee SH, Lim D, Kim DH, Kim SC, Kim TY, Kang C, Jeong JH, Park YJ, Lee SB, Kim RB. J. Emerg. Med. 2018; 54(4): 427-434.
Affiliation
Department of Emergency Medicine, Gyeongsang National University School of Medicine, Gyeongsang National University Hospital, Jinju, Gyeongsangnam-Do, Republic of Korea; Center for Regional Cardiocerebrovascular Disease, Gyeongsang National University Hospital, Jinju, Gyeongsangnam-Do, Republic of Korea.
BACKGROUND: Mortality prediction in patients with brain trauma during initial management in the emergency department (ED) is essential for creating the foundation for a better prognosis.
OBJECTIVE: This study aimed to create a simple and useful survival predictive model for patients with isolated blunt traumatic brain injury that is easily available in the ED.
METHODS: This is a retrospective study based on the trauma registry data of an academic teaching hospital. The inclusion criteria were age ≥ 15 years, blunt and not penetrating mechanism of injury, and Abbreviated Injury Scale (AIS) scores between 1 and 6 for head and 0 for all other body parts. The primary outcome was 30-day survival probability. Internal and external validation was performed.
RESULTS: After univariate logistic regression analysis based on the derivation cohort, the final Predictor of Isolated Trauma in Head (PITH) model for survival prediction of isolated traumatic brain injury included Glasgow Coma Scale (GCS), age, and coded AIS of the head. In the validation cohort, the area under the curve of the PITH score was 0.970 (p < 0.0001; 95% confidence interval 0.960-0.978). Sensitivity and specificity were 95% and 81.7% at the cutoff value of 0.9 (probability of survival 90%), respectively.
CONCLUSIONS: The PITH model performed better than the GCS; Revised Trauma Score; and mechanism of injury, GCS, age, and arterial pressure. It will be a useful triage method for isolated traumatic brain injury in the early phase of management.