Outcome Prediction of TBI: Are There Parameters That Affect the IMPACT and CRASH Models?

Abstract

Abnormal hematologic parameters associated with unfavorable neurological outcomes in traumatic brain injury (TBI) have been studied in isolation. We aimed to study whether there are any additional parameters that improve standard prognostic models in TBI. This prospective observational study conducted in a tertiary neurological care center included adult patients with moderate and severe isolated head injury. Laboratory and clinical parameters were noted at admission, and the Glasgow Outcome Score-Extended of patients was assessed after 6 months. Multiple logistic regression was conducted using fixed coefficients of IMPACT (International Mission for Prognosis and Analysis of Clinical Trials) and CRASH (Corticosteroid Randomisation After Significant Head Injury) prognostic models. The new composite models were compared with the original models. The study comprised 96 patients. Parameters with relatively good predictability for mortality were elevated international normalized ratio (area under the curve [AUC] 0.69, odds ratio 13.2), total leukocyte count (AUC 0.68, odds ratio 1.15), and transfusion of blood products (AUC 0.72, odds ratio 6.43). Addition of these led to a statistically small improvement in predictions of IMPACT and CRASH. Neutrophil-to-lymphocyte ratio was not a good predictor of mortality or morbidity (AUC 0.58 and 0.47, respectively). International normalized ratio, total leukocyte count, and blood transfusion were found to be predictors of mortality and unfavorable neurological outcome in TBI at 6 months. Their addition to the IMPACT and CRASH prognostic models resulted in a modest improvement in the prediction of outcome in TBI.