Improved Interobserver Reliability of the Sanders Classification in Calcaneal Fractures Using Segmented Three-Dimensional Prints

We examined the added value of 3-dimensional (3D) prints in improving the interobserver reliability of the Sanders classification of displaced intraarticular calcaneal fractures. Twenty-four observers (radiologists, trainees, and foot surgeons) were asked to rate 2-dimensional (2D) computed tomography images and 3D prints of a series of 11 fractures, selected from cases treatment at our level I trauma center between 2014 and 2016. The interobserver reliability for the Sanders classification was assessed using kappa coefficients. Three versions of the Sanders classification were considered: Sanders classification with subclasses, Sanders classification without subclasses, and the combination of Sanders types III and IV because of the high incidence of comminution in both types. The reference standard for classification was the perioperative findings by a single surgeon. The 3D print always yielded higher values for agreement and chance-corrected agreement. The Brennan-Prediger-weighted kappa equaled 0.35 for the 2D views and 0.63 for the 3D prints for the Sanders classification with subclasses (p = .004), 0.55 (2D) and 0.76 (3D) for the classification without subclasses (p = .003), and 0.58 (2D) and 0.78 (3D) for the fusion of Sanders types III and IV (p = .027). Greater agreement was also found between the perioperative evaluation and the 3D prints (88% versus 65% for the 2D views; p  < .0001). However, a greater percentage of Sanders type III-IV were classified with 2D than with 3D (56% versus 32%; p  < .0001). The interobserver agreement for the evaluation of calcaneal fractures was improved with the use of 3D prints after "digital disarticulation."