Reproducibility of Two Novel Bone Defect Classification Systems

To determine the reproducibility of two novel bone defect classification systems, the Tetsworth classification system and the Universal Long Bone Defect (ULBD) classification system.

Two independent reproducibility studies were undertaken. A series of cases with bone defects was collected and sent to a panel of orthopaedic surgeons who were members of the OTA or LLRS who routinely manage bone defects. The surgeons were provided with a measurement calibration scale for defect length assessment. Analysis of reproducibility was performed using a Fleiss's multi rater kappa model, where a Fleiss's kappa of 0 indicated agreement no better than random chance, and 1 indicated perfect agreement. Surgeons also rated their confidence in their rating from 1 - 3, with 3 as "very confident."

For the Tetsworth classification, Fleiss' kappa was 0.264 for the overall classification system (fair agreement) (95% CI 0.224,0.305). The kappa for the D1,D2,D3 part of the code related to defect completeness (incomplete, subcritical, or critical) was 0.557 (moderate agreement) (95%CI 0.446, 0.648). The kappa for the A,B,C part of the code related to defect shape was 0.402 (moderate agreement) (95% CI 0.323, 0.480). The mean confidence for all cases was 2.46 (range 1.86 - 3). For the ULBD classification, the Fleiss' kappa for the entire classification was 0.133 (slight agreement) (95% CI 0.123,0.144). The kappa for the location and segment was 0.568 (moderate agreement) (95% CI 0.526,0.610). The kappa for the morphology component was 0.124 (slight agreement) (95% CI 0.114, 0.135). The mean confidence for all cases was 2.0 (range 1.0 - 3.0).

The bone defect classification system proposed by Tetsworth et al was reproducible by surgeons familiar with bone defects. The system can be considered for future research use. The ULBD classification system had low overall reproducibility.