A pilot agreement study of a new classification system for Peri-implant femoral fractures.

BACKGROUND

Peri-implant fractures occur in association with an implant, used to treat a previous injury that is still attached to the bone. We recently published a proposal for a classification system for peri-implant femoral fractures [Videla-Cés, Injury,2019].

AIM

To assess the agreement of our classification system for peri-implant femoral fractures among a group of future users of said classification system.

METHODS

A prospective, multicentre, international agreement pilot study was conducted among a group of independent traumatologists/orthopaedic surgeons (evaluators): senior (a minimum of 10 years' experience) and junior (in fourth or fifth training year). A set of 30 radiographs of peri-implant femoral fractures were selected (stratified into 3 levels of difficulty: low, medium and high). Each evaluator interpreted the radiographs on 2 different occasions separated by a period of one month and in a different order each time. The level of difficulty was masked from the evaluators and they had one week to classify the radiographs each time. Logistic regression and Cohen's kappa coefficient (κ) and its 95% confidence interval (95%CI) were used to assess the accuracy and agreement, both intra- and inter-evaluator. Between senior and junior surgeons was performed an exploratory analysis.

RESULTS

35 traumatologists/orthopaedic surgeons (17 senior and 18 junior) from 16 tertiary hospitals from 4 countries acted as evaluators. The accuracy, percentage of correct classifications (2 digits and 3 letters), in the first and second evaluation was: 56% (95%CI: 53-59%) and 55% (95%CI: 51-57%), respectively. (insert space) Negligible differences were found between junior and senior evaluators (first evaluation, OR: 1.46, 95%CI: 0.82-2.61, p-value: 0.199; second evaluation, OR: 1.06, 95%CI: 0.56-2.00, p-value: 0.860). (insert space) Both medium and low radiograph difficulty were associated with a lower probability of an incorrect classification compared with those of high difficulty (first evaluation, OR: 7.60, 95%CI: 5.24-11.05, p-value: <0.001; OR: 14.15, 95%CI: 9.12-21.96, p-value: <0.001, respectively; second evaluation, OR: 7.11, 95%CI: 4.88-10.38, p-value: <0.001, OR: 15.28, 95%CI: 9.77-23.89, p-value: <0.001). (insert space) The kappa for intra-observer agreement between the first and second evaluation was: 0.66, 95%CI: 0.63, 0.69. The kappa for inter-observer agreement considering all 30 radiographs was: 0.40, 95%CI: 0.40, 0.41 in the first evaluation and 0.39, 95%CI: 0.39, 0.40 in the second evaluation.

CONCLUSION

The proposed classification for peri-implant femoral fractures may be useful and user-friendly. Future studies are needed to assess the how clinically useful this classification system may be (the third phase in the validation process).