Derivation of a clinical risk score for traumatic orbital fracture

BACKGROUND: Given that orbital fractures are found in only one of every eight patients receiving computed tomography for acute orbital trauma, we sought to prospectively identify clinical predictors of orbital fracture that may obviate the need for exposing low-risk patients to ionizing radiation. METHODS: Prospective cohort study conducted from July 2007 through October 2009 at two urban emergency departments. Consecutive patients undergoing computed tomography for acute blunt orbital trauma were evaluated on 15 clinical findings before imaging. The primary outcome of interest was presence of any acute orbital fracture. The secondary outcome was a fracture requiring emergent operative intervention. Multivariable logistic regression analysis with multiple imputation was used to derive a predictive risk score. RESULTS: A total of 2,262 patients with acute orbital trauma were enrolled. Median age was 38 years with male predominance (68.3%). Acute orbital fractures were found in 360 patients (15.9%). The derived risk score included orbital rim tenderness, periorbital emphysema, subconjunctival hemorrhage, pain with extraocular movement, impaired extraocular movement, and epistaxis. Across 10 multiply imputed data sets, a mean of 660 patients (29.2%) lacked all six equally weighted predictors, of which 6.3% (95% confidence interval, 4.3-8.2) experienced an acute orbital fracture and only 0.5% (95% confidence interval, 0.0-1.0) required emergent operative intervention. CONCLUSION: Six clinical predictors identify patients with blunt orbital trauma at increased risk for acute orbital fracture. A risk score of 0 identifies patients at very low risk for emergent operative intervention. Multicenter studies are needed to validate these findings and derive a clinical decision instrument to reduce orbital imaging without compromising patient safety. LEVEL OF EVIDENCE: Diagnostic study, level II.