Evaluation and Validation of Veterinary Trauma Scoring Systems
Rebecca A Packer, MS, DVM, DACVIM (Neurology/Neurosurgery)
Researchers evaluated the predictive value of the 2 most common trauma scoring systems in veterinary medicine.
Trauma scores are used often in human medicine to predict patient outcomes or dictate treatment algorithms. Although less common in veterinary medicine, the 2 most commonly used veterinary trauma scoring systems are the modified Glascow coma scale (mGCS) and the animal trauma triage (ATT) score. A study recently published in the Journal of Veterinary Emergency and Critical Care validated and compared the predictive value of these two 2 scoring systems in veterinary trauma patients.
Data were collected from the American College of Veterinary Emergency and Critical Care’s Veterinary Committee on Trauma (VetCOT) trauma registry. A cohort of 3599 dogs from 9 veterinary hospitals had complete data (including mGCS and ATT scores, signalment, type of trauma, and outcome data at the time of discharge) and were used for analysis. Descriptive and comparative analyses were performed, as well as the area under receiver operating characteristic (AUROC), for predictive assessments.
- AVMA 2017: Respiratory Complications of Trauma
- Evaluating Neurological Trauma in a Veterinary Setting
The ATT score is comprised of 6 categories: perfusion, cardiac, respiratory, eye/muscle/integument, skeletal, neurologic. For each category, a grade of 0 (least severe, slight or no injury) to 3 (most severe) is assigned based on a rubric of various physical and physiologic assessments. Where assessments fall into more than one grade, the more severe grade is assigned. Total ATT scores range from 0 to 18.
Unlike the ATT, the mGCS was developed specifically for head trauma patients. The mGCS includes 3 categories: motor activity, brain stem, and level of consciousness. For each category a score of 1 (normal) to 6 (severe) is assigned based on clinical assessments. Total mGCS scores range from 3 to 18.
Results and Discussion
The ATT score was associated directly with mortality (higher ATT scores were associated with higher mortality). The AUROC for the full ATT scoring system was 0.92 (95% CI, 0.91-0.94). For every 1-point increase in ATT score, the odds of mortality increased by 2.07 (P < .001).
Analyses were performed on the full ATT model as well as each subcomponent. Analyses were then performed by sequentially eliminating the least predictive subcomponents to determine whether a simplified scoring system could be used with similar discriminatory reliability. The 3 most predictive subcomponents were neurologic, perfusion, and respiratory. Elimination of the lowest 3 subscores (eye/muscle/integument, skeletal, cardiac) from the model did not affect the discriminatory performance of the ATT score, indicating that mortality can be predicted using only the neurologic, perfusion, and respiratory scores (AUROC, 0.91; 95% CI, 0.89-0.93). The most predictive ATT model eliminated the most poorly performing subcategory (eye/muscle/integument) and yielded an AUROC of 0.93 (95% CI, 0.91-0.94).
Similar analyses on the mGCS as a predictor of mortality yielded an AUROC of 0.82 (95% CI, 0.79-0.85). When restricted to only dogs with head trauma, the AUROC increased to 0.84 (95% CI, 0.79-0.90). Sequential elimination of subcategories showed a loss of discriminatory reliability; therefore, the full mGCS is superior to an abbreviated scoring system with the mGCS.
Overall, the ATT score showed superior predictive ability over the mGCS (AUROC 0.92 vs 0.82, respectively, P < .001). A more abbreviated ATT score using the neurologic, perfusion, and respiratory components was comparable to the full score. Despite the original purpose, the mGCS can be used in the general population of trauma patients with similar reliability as in those with head trauma; however, overall the ATT score was superior to the mGCS.