Ankle Fracture Stability-Based Classification: A Study of Reproducibility and Clinical Prognostic Ability

To compare the reproducibility and prognostic capacity of 2 commonly used ankle fracture classifications to the stability-based classification.

One hundred ninety-three consecutive rotational-type ankle fractures treated during a year at our institution in patients older than 18 years were retrospectively analyzed. Pilon and pathologic fractures were excluded. The fractures were treated by attending physicians who were unaware of the stability-based classification system. Three observers classified injury radiographs using the Lauge-Hansen, Weber/AO, and stability-based classifications systems. Reproducibility (interobserver variation) of each classification system was calculated using kappa statistics. Prognostic values were evaluated by calculating the area under the curve for the receiver-operating characteristic curves (using surgery as the positive outcome).

The stability-based and Weber/AO classifications showed better reproducibility [kappa 0.938 (95% confidence interval 0.921-0.952), kappa 0.97 (0.961-0.976)], respectively, than the Lauge-Hansen [kappa 0.74 (0.664-0.795); P < 0.05]. The stability-based classification was more accurate (P < 0.001) in predicting surgical treatment [area under the curve 0.883 (95% confidence interval 0.852-0.914)] compared with the other 2 classifications [0.626 (0.576-0.675) and 0.698 (0.641-0.755)], respectively.

The stability-based classification was both highly reproducible (kappa 0.938) and had superior prognostic capacity to identify patients who needed surgical intervention compared with both the Lauge-Hansen and AO/Weber classification systems. Importantly, there were no patients who were classified as stable who failed nonoperative treatment. This extends earlier studies by directly demonstrating its prognostic advantage to other classification systems.