Secondary Logo

Journal Logo

Institutional members access full text with Ovid®

Applicability of the predictors of the historical trauma score in the present Dutch trauma population: Modelling the TRISS predictors

Gunning, Amy C. MD, MSc; Leenen, Luke P.H. MD, PhD

Journal of Trauma and Acute Care Surgery: October 2014 - Volume 77 - Issue 4 - p 614–619
doi: 10.1097/TA.0000000000000415
Original Articles

BACKGROUND The extensively used trauma scores to evaluate trauma center performances have been derived decades ago. This leaves its applicability in the current trauma population a subject of discussion.

In this study, we evaluate the applicability of the current trauma scores in today’s trauma population.

METHODS This study was performed in the central trauma region (Utrecht) of the Netherlands. Data from all admitted trauma patients were collected and split according to trauma mechanism, that is, blunt or penetrating trauma. The number of events in the penetrating trauma population was too small to derive or validate a model. We have validated the original predictors and their coefficients (i.e., Revised Trauma Score [RTS], Injury Severity Score [ISS] and age index) in a regional trauma population and derived a new model in a Level I trauma population. The model was checked for its quality and internally validated using bootstrapping methods.

RESULTS Regional data set included 10,235 patients, 9,903 (96.8%) of whom with blunt and 332 (3.2%) with penetrating trauma. Level I data set included 4,649 patients, of whom 4,373 (94.1%) with blunt and 276 (5.9%) with penetrating trauma. In the regional data set, the external validation showed an R2 of 0.293, a good fit (p = 0.168), and an area under the curve of 0.851. The new model in the Level I data set resulted in Glasgow Coma Scale (GCS) score, ISS, age index, and systolic blood pressure (SBP) in the model (R2 = 0.516; a good fit test, p = 0.104; and AUC, 0.939). The Wald test of the SBP in this model was 6.46.

CONCLUSION We conclude that the current trauma predictors are applicable in a regional mixed trauma population, but not in a Level I trauma population. The physiologic parameters, SBP and RR, must be excluded from the current model, and new coefficients should be calculated to maintain accurate predictions in a Level I trauma population.

LEVEL OF EVIDENCE Prognostic study, level III.

From the Department of Trauma Surgery, University Medical Center Utrecht, Utrecht, the Netherlands.

Submitted: March 26, 2014, Revised: June 30, 2014, Accepted: June 30, 2014.

Address for reprints: Amy C. Gunning, MD, MSc, Department of Surgery, University Medical Center Utrecht, Suite: G04.228, Heidelberglaan 100, 3584 CX Utrecht, the Netherlands; email:

© 2014 Lippincott Williams & Wilkins, Inc.