Simple jQuery Dropdowns
Please use this identifier to cite or link to this item:
Title: Traumatic optic neuropathy prediction after blunt facial trauma: Derivation of a risk score based on facial CT findings at admission
Authors: Uttam K. Bodanapally
Giulia Van Der Byl
Kathirkamanathan Shanmuganathan
Lee Katzman
Elena Geraymovych
Nitima Saksobhavivat
Stuart E. Mirvis
Kuladeep R. Sudini
Jaroslaw Krejza
Robert Kang Shin
University of Maryland R Adams Cowley Shock Trauma Center
University of Maryland Medical Center
Universita degli Studi di Pavia
Ohio State University
Mahidol University
Johns Hopkins Bloomberg School of Public Health
Keywords: Medicine
Issue Date: 1-Jan-2014
Citation: Radiology. Vol.272, No.3 (2014), 824-831
Abstract: Purpose: To determine the specific facial computed tomographic (CT) findings that can be used to predict traumatic optic neuropathy (TON) in patients with blunt craniofacial trauma and propose a scoring system to identify patients at highest risk of TON. Matrial and Methods: This study was compliant with HIPAA, and permission was obtained from the institutional review board. Facial CT examination findings in 637 consecutive patients with a history of blunt facial trauma were evaluated retrospectively. The following CT variables were evaluated: midfacial fractures, extraconal hematoma, intraconal hematoma, hematoma along the optic nerve, hematoma along the posterior globe, optic canal fracture, nerve impingement by optic canal fracture fragment, extraconal emphysema, and intraconal emphysema. A prediction model was derived by using regression analysis, followed by receiver operating characteristic analysis to assess the diagnostic performance. To examine the degree of overfitting of the prediction model, a k-fold cross-validation procedure (k = 5) was performed. The ability of the cross-validated model to allow prediction of TON was examined by comparing the mean area under the receiver operating characteristic curve (AUC) from cross-validations with that obtained from the observations used to create the model. Results: The five CT variables with significance as predictors were intraconal hematoma (odds ratio, 12.73; 95% confidence interval [CI]: 5.16, 31.42; P < .001), intraconal emphysema (odds ratio, 5.21; 95% CI: 2.03, 13.36; P = .001), optic canal fracture (odds ratio, 4.45; 95% CI: 1.91, 10.35; P = .001), hematoma along the posterior globe (odds ratio, 0.326; 95% CI: 0.111, 0.958; P = .041), and extraconal hematoma (odds ratio, 2.36; 95% CI: 1.03, 5.41; P = .042). The AUC was 0.818 (95% CI: 0.734, 0.902) for the proposed model based on the observations used to create the model and 0.812 (95% CI: 0.723, 0.9) after crossvalidation, excluding substantial overfitting of the model. Conclusion: The risk model developed may help radiologists suggest the possibility of TON and prioritize ophthalmology consults. However, future external validation of this prediction model is necessary. © RSNA, 2014.
ISSN: 15271315
Appears in Collections:Scopus 2011-2015

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.