Citation

Miller PR, Chang MC, Hoth JJ, Hildreth AN, Wolfe SQ, Gross JL, Martin RS, Carter JE, Meredith JW, D'Agostino RB. J. Am. Coll. Surg. 2017; 224(4): 680-685.

Affiliation

Wake Forest Health Science Department of Biostatistical Sciences, Wake Forest University, Winston-Salem, North Carolina.

Copyright

(Copyright © 2017, American College of Surgeons, Publisher Elsevier Publishing)

DOI

10.1016/j.jamcollsurg.2016.12.053

PMID

28263858

Abstract

INTRODUCTION: Aging worsens outcome in traumatic brain injury (TBI), but available studies may not provide accurate outcome predictions due to confounding associated injuries. Our goal was to develop a predictive tool using variables available at admission to predict outcome related to severity of brain injury in aging patients.

METHODS: Characteristics and outcomes of blunt trauma patients with isolated TBI of ages ≥ 50 in National Trauma Data Bank (NTDB) were evaluated. Equations predicting survival and independence at DC (IDC) were developed and validated using patients from our trauma registry, comparing predicted to actual outcomes.

RESULTS: Logistic regression for survival and IDC was performed in 57,588 patients using age, gender, Glasgow Coma Scale score (GCS), and revised trauma score (RTS). All variables were independent predictors of outcome. Two models were developed using these data. The first included age, gender, and GCS. The second substituted RTS for GCS. C statistics from the models for survival and IDC were 0.90 and 0.82 in the GCS model. In the RTS model, C statistics were 0.80 and 0.67. The use of GCS provided better discrimination, and was chosen for further examination. Using the predictive equation derived from the logistic regression model, outcome probabilities were calculated for 894 similar patients from our trauma registry (1/12-3/16). The survival and IDC models both showed excellent discrimination (p<0.0001). IDC and survival generally decreased by decade: Age 50-59 (80% IDC, 6.5% mortality), 60-69 (82%, 7.0%), 70-79 (76%, 8.9%), and 80-89 (67%, 13.4%).

CONCLUSION: These models can assist in predicting the probability of survival and IDC for aging patients with TBI. This provides important data for loved ones of these patients when addressing goals of care.

Copyright © 2017. Published by Elsevier Inc.

Language: en