Validation of the CRASH and IMPACT Prognostic Models in Predicting Outcomes in Moderate and Severe Traumatic Brain Injury Patients Presenting to the Emergency Department in India: An Observational Study (Code: T0021)

Introduction: Traumatic brain injury (TBI) is a leading cause of death and disability worldwide, with a higher burden in low- and middle-income countries. Accurate prognostic tools, such as the CRASH and IMPACT models, are crucial for guiding clinical decision-making and predicting both short- and long-term outcomes in TBI patients. These models help in stratifying patients based on their risk, thereby improving patient management and resource allocation.

Aims & Objective:

Aim: To evaluate the validity of the CRASH and IMPACT prognostic models in predicting outcomes in moderate and severe traumatic brain injury patients presenting to the emergency department.

Primary Objective: To assess the accuracy of the CRASH model in predicting 16-day mortality in patients with moderate and severe traumatic brain injury.

Secondary Objective: To evaluate the ability of both the CRASH and IMPACT models in predicting unfavorable outcomes at 6 months in patients with moderate and severe traumatic brain injury.

Methods: This prospective observational study included 361 patients with moderate to severe TBI (Glasgow Coma Scale <14) admitted to a tertiary care hospital in South India. The CRASH and IMPACT scores were calculated upon admission, and patient outcomes were assessed using the Glasgow Outcome Scale at both 15 days and 6 months post-injury. Statistical analyses, including sensitivity, specificity, and area under the curve (AUC), were performed to assess the models' performance.

Results: The 14-day and 6-month mortality rates were 21.8% and 31%, respectively. The CRASH (CT) model demonstrated an AUC of 0.90 for predicting 14-day mortality, while the IMPACT (Core + CT + Lab) model achieved an AUC of 0.92 for predicting 6-month unfavorable outcomes. Significant predictors of poor outcomes included non-reactive pupils, hypoxia, hypotension, and low Glasgow Coma Scores.

Conclusion: Both the CRASH and IMPACT models showed good predictive accuracy, with the IMPACT (Core + CT + Lab) model exhibiting superior long-term prognostic capability in moderate and severe TBI patients.

Keywords: Traumatic Brain Injury, CRASH Model, IMPACT Model, Mortality Prediction, Unfavorable Outcome, Prognostic Validation