The reliability and validity of the modified Unified Classification System for periprosthetic femoral fractures after hip arthroplasty

Abstract

Background: This study assessed the reliability and validity of the modified Unified Classification System for femur fractures after hip arthroplasty.

Methods: Four hundred and two cases were evaluated by 6 observers, 3 experts and 3 trainee surgeons. Each observer read the radiographs on 2 separate occasions and classified each case as to its type. Reliability was assessed by looking at the intraobserver and interobserver agreement using the Kappa statistic. Validity was assessed within the B group by looking at the agreement between the radiographic classification and the intraoperative findings. Interobserver and intraobserver agreement and validity were analyzed, using weighted kappa statistics.

Results: The mean k value for interobserver agreement was found to be 0.882 (0.833-0.929) for consultants (almost perfect agreement) and 0.776 (0.706-0.836) for the trainees (substantial agreement). Intraobserver k values ranged from 0.701 to 0.972, showing substantial to almost perfect agreement. Validity analysis of 299 type B cases revealed 89.854% agreement with a mean k value of 0.849 (0.770-0.946) (almost perfect agreement).

Conclusions: This study has shown that the modified Unified Classification System is reliable and valid. We believe it is useful to improve the judgment of the implant stability, and establish the therapeutic strategy for periprosthetic femoral fracture.