Objectives: To propose a previous implant fractures (PIFs) classification system with good interobserver reliability.
Design: Retrospective classification.
Setting: Four academic medical centers.
Patients/participants: A retrospective review of PIFs treated at 4 academic medical centers over 10 years was performed. Data collected included initial implant and PIF radiographs. There were 103 PIFs in 96 patients during the study period. Seventy-three (70.9%) were about plate/screw (PS) constructs and 30 (29.1%) were about intramedullary (IM) devices.
Intervention: Assignment of PIF classification.
Main outcome measurements: PIFs were classified based on initial implant (PS or IM) and fracture location with respect to the initial implant (proximal or distal to the implant, at the tip of the construct, or within the construct). Reliability of this scheme was assessed among 5 observers using Fleiss' kappa tests.
Results: Of PIFs about plate/screw constructs, 26.0% were proximal/distal to the implant (classification: PS1), 57.5% involved bone between the most proximal/distal screw and the same end of the plate (classification: PS2), and 16.4% involved only bone between the most proximal and distal screws (classification: PS3). Of PIFs about IM, 43.3% were distal to the device (classification: IM1), 46.7% involved bone between the most proximal/distal locking bolt and the same end of the device (classification: IM2), and 10.0% involved only bone between locking bolts (classification: IM3). Interobserver reliability for the classification system was excellent between observers, κ = 0.839, P < 0.0005.
Conclusions: The proposed system offers a simple method to classify and describe fractures that occur about a previously implanted fracture device. Development of a classification system will allow for comparison of treatment modalities between injury types.