Patients sustaining multisystem trauma are at risk for oral and maxillofacial fractures. Although the University of Wisconsin established criteria to help guide the clinician in obtaining additional cross-sectional imaging to evaluate possible facial fractures, it has not been externally validated. Our aim was to evaluate whether the University of Wisconsin's Criteria is generalizable to external institutions through validation and to report modern practice patterns at a level 1 trauma center.Methods:
A retrospective case study was performed of all patients who had computed tomography of the facial bones (CT face) at a tertiary, academic, Level 1 trauma center over the 6-month period ending on June 30, 2015. The electronic medical record was reviewed for the 5 University of Wisconsin criteria (bony step off or instability, periorbital ecchymosis, malocclusion, tooth absence, and glasgow coma scale). Final interpretation of CT face findings by board-certified radiologists (facial fractures, intracranial hemorrhage, and cervical spine injury) were also captured. Our modeling was similar to that described by the reference study, the internal validation study. Sensitivity, specificity, negative, and positive predictive values with 95% confidence intervals were evaluated. A P < 0.05 was considered significant.Results:
The presence of any ≥1 of the 5 criteria identified on physical examination resulted in 81% sensitivity for any facial fracture, which is lower than the sensitivity initially described (98%) and subsequently internally validated (97%). The absence of all 5 physical examination criteria had a negative predictive value of 60%, again lower than that initially described (87%) and then internally validated (81%).Conclusion:
We were unable to validate the University of Wisconsin criteria for predicting facial fractures. These criteria may be institutionally specific and not generalizable to other trauma centers. Further research to refine the criteria for CT of the face is needed to improve resource allocation.