Inter- and intra-observer reproducibility of posterior malleolus fracture classifications

Abstract

Objectives: To analyze whether PMF classifications can influence treatment choice and surgical approach. In addition, verify the intraand interobserver reproducibility of the three main classifications, stratified by observers’ experience. Methods: Ankle computed tomography of 50 patients was evaluated by ten observers, four orthopedists specialized in foot and ankle surgery, and six non specialist orthopedists, with an interval of two weeks between evaluations. The evaluators classified PMF according to the Mason, Haraguchi, and Bartoníček/Rammelt classifications and determined whether to treat PMF conservatively or surgically (in this case, by access route). In addition, the reproducibility of the classifications was evaluated. The association between decision-making and access route was analyzed using the Chi-Square Test (χ2). Cronbach’s alpha was used to assess intraobserver agreement, and the kappa statistic was used to evaluate interobserver agreement. Results: In analyses of decision-making and the access route, all classifications showed large effect sizes (V > 0.50). Intraobserver reproducibility across the entire sample ranged from 0.53 to 0.95 (0.78 ± 0.12) for the Haraguchi classification, from 0.47 to 0.95 (0.74 ± 0.17) for Mason, and from 0.53 to 0.94 (0.72 ± 0.12) for Bartoníček/Rammelt, indicating adequate agreement across the three classifications. For the specialist orthopedists, the mean ratings for the Haraguchi, Mason, and Bartoníček/Rammelt classifications were 0.86, 0.84, and 0.75, corresponding to good, good, and adequate, respectively. For the group of non-specialists, the means were 0.72, 0.68, and 0.70, indicating adequate, average, and adequate, respectively. Interobserver reproducibility was considered reasonable for Haraguchi (0.38) and moderate for Mason (0.42) and Bartoníček/Rammelt (0.43). Conclusion: All three classifications had large effects on treatment choice and access route decisions, with the Bartoníček/Rammelt classification showing the highest effect. All three PMF classifications were considered adequately reproducible by intraobserver assessment. Interobserver reproducibility was considered reasonable for Haraguchi and moderate for the others. Level of evidence IV; Therapeutic studies; Case series.