Abstract
Category:
Basic Sciences/Biologics; Trauma
Introduction/Purpose:
There is still no consensus regarding which is the best classification system for the management of tibial plafond fractures. The goal of this study is to perform a independent agreement evaluation to compare two recently published systems: Leonetti/Tigani and the new AO classification.
Methods:
Seventy-five patients with tibial plafond fracture and preoperative CT scan were included. Six raters with different level of expertise (two foot and ankle surgeons and three orthopedic surgery residents) classified the fractures using CT scans according to the morphological grading of both systems. The Leonetti/Tigani classification system considers four types (I, II, III and IV) and six subtypes (I, IIF, IIS, IIIF, IIIS, IV). The AO system considers three types (A, B, C) and nine subtypes (A1, A2, A3, B1, B2, B3, C1, C2, C3). After six weeks all cases were randomly re-evaluated by the same raters. The kappa coefficient (κ) was used to determine the degree of reliability.
Results:
Inter-observer reliability: strong using the Leonetti/Tigani classification system considering types, with a κ of 0,65 (0,60 - 0,69), and subtypes, with a κ of 0,62 (0,58 - 0,66). Reliability for the AO system was strong considering types with a κ of 0,72 (0,66 - 0,78), but moderate when including subtypes with a κ of 0,54 (0,50 - 0,57). Intra-rater reliability: Almost perfect using the Leonetti/Tigani classification considering both types and subtypes with a κ of 0,94 (0,88- 1,01) and 0,94 (0,89- 0,96), respectively. Reliability for the AO system was almost perfect considering types with a κ of 0,83 (0,75- 0,92), but strong when including subtypes with a κ of 0,61 (0,57- 0,66). No statistically significant difference between different levels of expertise.
Conclusion:
The system proposed by Leonetti/Tigani demonstrated a strong and almost perfect inter and intra-rater reliability, respectively. Although the new AO classification has a strong inter-rater reliability when including the main categories, it only reached a moderate reliability when including subtypes.