Simple computed tomography-based calculations of orbital floor fracture defect size are not sufficiently accurate for clinical use

Purpose The precise computed tomography-based calculation of the size of an orbital floor (OF) fracture defect is tedious and time-consuming. The aims of this study were to evaluate the accuracy of simple, rapid methods of calculating OF fracture defect size and to determine their suitability for clinical use.

Materials and Methods A retrospective review of the electronic medical records of patients with OF fractures presenting to Baylor Scott and White Hospital between October 2009 and April 2013 was performed. True OF defect sizes (the outcome variable) were calculated using a previously validated formula, on the basis of measurements obtained from coronally reformatted thin (<3-mm) axial computed tomographic images. Estimated OF defect sizes (the predictor variable) were calculated using geometric area formulas, assuming that the defect approximated the shape of an ellipse, circle, square, or rectangle on the basis of measurements obtained from coronal and sagittal computed tomographic images. Accuracy, sensitivity, specificity, and negative and positive predictive values in declaring a defect critical were determined for each method.

Results Ninety-nine patients with OF fractures were identified (69 men, 30 women; mean age = 46.9 years); 55 patients had a true OF defects of critical (≥2 cm²) or greater size. Geometric formulas showed ranges of accuracy (0.76 to 0.93), sensitivity (0.62 to 1.0), and specificity (0.63 to 0.91). The accuracy of defect size approximation using the area of an ellipse was highest.

Conclusions The geometric formulas estimated OF defect area with good but, in the authors' opinion, clinically unacceptable accuracy. Although highly sensitive, the formulas lacked specificity and tended to overestimate true defect sizes in most cases. Using rapid, simple geometric methods to assess the sizes of OF defects may lead to inappropriate surgical decisions. Thus, the most accurate estimation of OF defect size still requires the calculation of average defect length from coronal computed tomographic images, knowledge of slice thickness, and knowledge of the number of slices involved.