Background
The Injury Severity Score (ISS) is the most commonly used injury scoring system in trauma research and benchmarking. An ISS>15 conventionally defines severe injury; however, no studies evaluate whether ISS performs similarly between adults and children. Our objective was to evaluate ISS and AIS to predict mortality and define optimal thresholds of severe injury in pediatric trauma.
Methods
Patients from the Pennsylvania trauma registry 2000–2013 were included. Children were defined as age<16years. Logistic regression predicted mortality from ISS for children and adults. The optimal ISS cut-off for mortality that maximized diagnostic characteristics was determined in children. Regression also evaluated association between mortality and maximum AIS in each body-region, controlling for age, mechanism, and non-accidental trauma. Analysis was performed in single and multisystem injuries. Sensitivity analyses with alternative outcomes were performed.
Results
There were 352,127 adults and 50,579 children included. Children had similar predicted mortality at ISS of 25 as adults at ISS of 15 (5%). The optimal ISS cut-off in children was ISS>25 and had a positive predictive value (PPV) of 19% and negative predictive value (NPV) of 99% compared to PPV of 7% and NPV of 99% for ISS>15 to predict mortality. In single-system injured children, mortality was associated with head (OR 4.80; 95%CI 2.61–8.84, p<0.01) and chest AIS (OR 3.55; 95%CI 1.81–6.97, p<0.01), but not abdomen, face, neck, spine, or extremity AIS (p>0.05). For multisystem injury, all body region AIS were associated with mortality except extremities. Sensitivity analysis demonstrated ISS>23 to predict need for full trauma activation, and ISS>26 to predict impaired functional independence were optimal.
Conclusions
ISS>25 may be a more appropriate definition of severe injury in children. Pattern of injury is important, as only head and chest injury drive mortality in single-system injured children. These findings should be considered in benchmarking and performance improvement efforts.
Level of Evidence
III, epidemiologic