Reliability of Current Classification Systems for Periprosthetic Fractures of the Humerus

Numerous classification systems exist to evaluate periprosthetic humerus fractures, although most are based on limited patient numbers. The Wright and Cofield system is the most widely used classification system. We sought to evaluate the interobserver and intraobserver reliabilities of this system compared with the Unified Classification System (UCS) using the largest patient sample to date.

This retrospective study identified patients between December 2011 and January 2021 with a periprosthetic fracture of the humerus around the stem of a shoulder arthroplasty component. Three upper extremity fellowship-trained surgeons evaluated all radiographs for stem stability, evidence of preinjury stem loosening, Wright and Cofield classification, UCS classification, and recommended treatment for each case at 2 timepoints separated by 2 months. The kappa statistic for interobserver and intraobserver reliability was calculated.

Seventy-six patients were included. There was moderate interobserver (kappa 0.53) and substantial intraobserver (kappa 0.69) agreement when classifying stem stability after fracture. There was moderate interobserver (kappa 0.48) and intraobserver (kappa 0.60) agreement when evaluating for stem loosening before fracture. There was fair interobserver (kappa 0.29) and moderate intraobserver (kappa 0.51) agreement regarding the UCS class. There was moderate interobserver (kappa 0.41) and intraobserver (kappa 0.57) agreement regarding the proposed treatment. There was slight interobserver (kappa 0.04) and moderate intraobserver (kappa 0.44) agreement regarding the Wright and Cofield classification.

The Wright and Cofield system is less reliable than the UCS classification. A more reliable and clinically relevant classification system is needed to standardize discussion of periprosthetic proximal humerus fractures.