Improved acetabular fracture diagnosis after training in a CT-based method.

Background: Acetabular fractures remain challenging to diagnose, particularly when they are complex. An accurate diagnosis is nevertheless crucial to select the best surgical strategy. None of the training methods described to date relies on the Letournel classification with a detailed analysis of each abnormality seen by computed tomography (CT). We therefore prospectively assessed a CT-based diagnostic method by (1) determining the rate of correct diagnoses by orthopaedic surgeons before and after training in the method, (2) comparing the times needed to read the CT images before and after training, (3) and assessing the repeatability of the method.

Hypothesis: Training in the CT-based diagnostic method significantly increases the rate of correct diagnoses.

Method: The CT-based diagnostic method involves analysing eight anatomical landmarks in the anterior, posterior, and no man's land zones. From our institutional database (450 cases between 2007 and 2016), we selected 35 acetabular fractures that replicated the overall distribution of fracture types. The images were reviewed by 10 inexperienced and 3 experienced readers before and after they received training in the CT-based diagnostic method. The rates of correct diagnoses and times needed to read the images were compared. Finally, an additional reading was performed to allow an assessment of reproducibility.

Results: After training, the rate of correct diagnoses by the unexperienced readers improved by 16.64% for all fractures combined (from 212/350, 60.5% [37-83%] to 270/350, 77.14% [63-86%]; P=0.001) and by 25.9% for associated fractures (from 90/180, 50% [11-89%] to 114/140, 75.6% [61-90%]; P=0.003). Mean time required by the inexperienced readers to interpret the 35 sets of images decreased after training, from 66.1 to 47.6min (i.e., a 1.22-minute decrease per patient, P=0.001). None of the study variables changed significantly after training of the experienced readers (P>0.05). Reproducibility among the inexperienced readers was 0.78.

Conclusion: Analysing the eight anatomical landmarks located in the anterior, posterior, and no man's land zones is a simple and reproducible method for diagnosing all fracture patterns defined by the Letournel classification.

Level Of Evidence: Level III, non-randomised prospective case-control diagnostic study.