Session VII - Polytrauma

Sat, 10/9/04 Polytrauma, Paper #39, 10:58 am

Validation of the New Injury Severity Score for Outcome Prediction:

A Study in 13,301 Patients from the German Trauma Registry

Paul J. Harwood, MD1; Christian Probst, MD2; Peter V. Giannoudis, MD1; Martijn van Griensven, MD2; Christian Krettek, MD2;
Hans-Christoph Pape, MD2
(all authors a-AO International)
1Academic Unit Orthopaedic Trauma Surgery, St James's University Hospital, Leeds, United Kingdom
2Hannover Medical School, Hannover, Germany

Purpose: Systems based on the Abbreviated Injury Scale (AIS), the Injury Severity Score (ISS), New Injury Severity Score (NISS), and AISmax (worst single AIS score) continue to find widespread application in research and patient care. The merits of each in predicting outcome are controversial. The NISS is a recent modification of the ISS, calculated from the sum of the squares of AIS for the three worst injuries, regardless of body region. Previous work suggests that this new system may offer improved accuracy in determining outcome. A recent publication, based on a very large American database, has suggested that the worst AIS score alone may, however, be superior to the NISS in predicting mortality. The aim of this study was to investigate these findings further in our usual population of patients with a predominantly blunt-injury mechanism.

Methods: The German Trauma Registry, a collaborative, multi-center database, was the source of data for this investigation. All patients who come to emergency departments with acute traumatic injuries requiring intensive care-unit admission are included. The predictive capacity of the ISS, NISS, and AISmax scores were analyzed for outcome, including sepsis, multiple organ failure (MOF), length of stay (LOS), and mortality by using receiver operator characteristic (ROC) curves. Significance was assumed at the P <0.05 level.

Results: A total of 13,301 patients met the inclusion criteria; the mean age was 33.81 years and 76.8% were men. Blunt trauma was the cause of injury in 95.3% of the patients. Mortality was 15.9%; 11.1% had sepsis, and 19.3% had MOF. The ISS was 23.1 and NISS, 29.2.

Area under ROC curve

 Score  Sepsis  MOF  Mortality  LOS ICU stay 
 ISS   0.660




 NISS   0.658  0.710


  0.685  0.770
 AISmax  0.599*  0.663*




Area under ROC curves, *Significant differences between scoring
systems in ability to predict that outcome, inferior system marked
(P <0.0001)

All systems were significant outcome predictors for sepsis, MOF, LOS, length of ICU admission, and mortality (P <0.0001). NISS was a significantly better predictor than the ISS for mortality (P <0.0001). NISS was equivalent to the AISmax for mortality prediction and on sub-group analysis was superior for patients with blunt and orthopaedic injuries. The AISmax was superior for patients with penetrating trauma. The NISS was significantly better for sepsis, MOF, ICU stay, and total LOS (P <0.0001). Analysis of the ROC curves showed that traditional cut-off of ISS of 16, 25, and 50 for severity of injury should be increased to 20, 30, and 55 to provide groups with equivalent outcome if this is required.

Conclusion/Significance: The NISS was found to be superior or equivalent to the ISS and AISmax for prediction of all investigated outcomes in a population of typical European trauma patients. This was particularly the case for those with blunt trauma and orthopaedic injuries. Our disparity with findings of previous North American studies may be partly due to differences in the distribution of injury mechanisms. As NISS is easier to calculate, its use is recommended to stratify patients for clinical and research purposes.