Prognosticating Traumatic Brain Injury (TBI): Analysis of CRASH, IMPACT Models and Patient Age in a Cohort of TBI Patients

Abstract

Abstract INTRODUCTION Both CRASH and IMPACT models have been developed in recent years to predict the outcome of Traumatic Brain Injury (TBI). However, there is no clear evidence as to how these models perform in a modern cohort of UK-patients. There is also predictive uncertainty with regards to survival rates and functional outcome in elderly (>65 yr) patients. METHODS Patients referred to a tertiary neuroscience center from December 2014 to January 2016 with a suspected TBI were retrospectively examined. For each model, the predicted survival and overall outcome were compared to the actual outcome on admission and at 6 mo post injury, stratified by patient age (>65 yr vs ≤65 yr). RESULTS A total of 161 patients met the initial criteria; mean age 65 yr (SD = 21) and 110 male. Both CRASH and IMPACT correctly predicted 6-mo mortality rates and functional outcomes in most patients (range 61.7%-82.4%), with better predictive performance for patients not accepted to the center (range 84%-98%). There was no significant difference in the initial survival of elderly patients if accepted (78% [95% CI 50.6-104.0] vs 81% [95% CI 67.8-94.8] but were lower for those not accepted (24% [95% CI 4.2-43.7] vs 76% [95% CI 63.5-88.5], P = .027). CONCLUSION Patients >65 yr admitted to tertiary neuroscience center had good survival rates on admission and at 6 mo. The lesser ability of CRASH and IMPACT models to predict poorer outcomes when accepted suggests that acceptance to specialist centers may be able to improve outcome and suggests more optimistic treatment and acceptance of appropriate over 65 yr should be considered.