A Corresponding Point Measurement System Provides Reliable Measurement of Displacement for Medial Epicondyle Fractures

Abstract

Little consensus exists on the best method for evaluation and management of pediatric medial epicondyle fractures because of an inability to reliably evaluate fracture displacement with standard imaging techniques. This study aimed to determine the performance of various radiographic views in evaluating displaced medial epicondyle fractures when using a standardized measurement methodology.

Methods: Ten fellowship-trained pediatric orthopaedic surgeons assessed fracture displacement in 6 patients with displaced medial epicondyle fractures using radiographic views (anteroposterior, lateral, axial, internal oblique [IO], and external oblique [EO]) and computed tomographic (CT) views (axial, 3-dimensional [3D] horizontal, and 3D vertical). Raters used a corresponding point method for measuring displacement. For each image, raters measured the absolute displacement, categorized the percent of displacement relative to the size of the fragment and fracture bed, and indicated a treatment option. Interobserver reliability was calculated for each view. Bland-Altman plots were constructed to evaluate the bias between each radiograph and the mean of the CT methods.

Results: For absolute displacement, anteroposterior and EO views showed almost perfect interobserver reliability, with an interclass correlation coefficient (ICC) of 0.944 for the anteroposterior view and an ICC of 0.975 for the EO view. The axial view showed substantial reliability (ICC = 0.775). For the displacement category, almost perfect reliability was shown for the anteroposterior view (ICC = 0.821), the axial view (ICC = 0.911), the EO view (ICC = 0.869), and the IO view (ICC = 0.871). Displacement measurements from the anteroposterior, axial, and EO views corresponded to the measurements from the CT views with a mean bias of <1 mm for each view. However, the upper and lower limits of agreement were >5 mm for all views, indicating a substantial discrepancy between radiographic and CT assessments. Treatment recommendations based on CT changed relative to the recommendation made using the anteroposterior view 29% of the time, the EO view 41% of the time, and the axial view 47% of the time.

Conclusions: Using a corresponding point measurement system, surgeons can reliably measure and categorize fracture displacement using anteroposterior, EO, and axial radiographic views. CT-based measurements are also reliable. However, although the mean difference between the radiograph-based measurements and the CT-based measurements was only about 1 mm, the discrepancy between radiographic views and CT-based methods could be as large as 5 to 6 mm.

Level of evidence: Diagnostic Level II. See Instructions for Authors for a complete description of levels of evidence.

Fig. 1

Anteroposterior radiograph (Fig. 1-A), lateral radiograph (Fig. 1-B), axial radiograph (Fig. 1-C), IO radiograph (Fig. 1-D), and EO radiograph (Fig. 1-E) along with CT axial image (Fig. 1-F), CT horizontal image (Fig. 1-G), and CT vertical image (Fig. 1-H) for a 12-year-old boy who presented with a medial epicondyle fracture.

Fig. 2

Corresponding point measurement method. Sample images displaying a simple displaced fracture (top left), double-density fracture (>1 fragment) (top right), and rotated fracture (bottom left), along with 3D CT reconstruction (bottom right). To perform the corresponding point measurement, points on the fracture bed and the fracture fragment that are thought to have corresponded to one another prior to the fracture are identified. The corresponding points with the longest distance from the fracture bed to the fracture fragment should be measured. For instance, for rotated fractures, measure from the proximal upper corner of the donor site to the presumed portion of the fragment that was originally attached there. If in doubt, measure on the proximal corner. Double-density contours are ones in which 2 edges (1 lighter and 1 darker) can be visualized at the fracture edge. In these scenarios, usually 1 light line appears to correspond with 1 darker line. If in doubt on these fractures, use the proximal corner of the darker line as the proximal corner. AP = anteroposterior.

Fig. 3

Bland-Altman plots for analysis of the agreement between a mean of the CT measurements and anteroposterior (AP) radiographs (n = 59), axial radiographs (n = 52), and EO radiographs (n = 55). The mean bias (dashed line), limits of agreement (dotted lines), and simple linear regression fit have been applied to each plot.