Inter- and Intrarater Agreement of the AO Spine-DGOU Osteoporotic Fracture Classification System Using Radiography and Computed Tomography Imaging

STUDY DESIGN: Cross-sectional rater agreement.

OBJECTIVES: This study assessed whether inter- and intra-rater agreement for the AO spine-DGOU Osteoporotic Fracture (OF) Classification system differs between classifications based on computed tomography (CT) or radiography.

METHODS: Radiography and CT were retrieved for 64 consecutive patients with a single osteoporotic vertebral fracture (OVF) treated at our institution having both imaging modalities performed on the same date. Four spine surgeons independently classified the fractures twice 1 week apart after training on 10 cases. Classifications were based on radiography and CT separately. Crude (%) and Fleiss' Kappa (k) agreements were calculated for radiography and CT, respectively, and for intrarater agreement between radiography and CT within each rater. The proportion of cases with complete agreement among raters (concordant cases) was compared between radiography and CT using McNemar's test. Fracture-type distribution differences were analyzed using McNemar's test as well.

RESULTS: Intra-rater agreement ranged 75%-81% (k = 0.49-0.64) for radiography and 64%-77% (k = 0.42-0.63) for CT. Inter-rater agreement was 56% (k = 0.49) with radiography and 50% (k = 0.51) with CT. No significant difference in the proportion of concordant cases was found between imaging modalities (P = 0.37). Fractures were classified as higher-order OF types with CT (P < 0.001).

CONCLUSIONS: We found moderate to substantial reliability of the OF system for radiography and CT. Using CT over radiography for classification did not increase the rater agreement. However, fractures are likely to be categorized as a higher-order OF type when the classification is based on CT compared to radiography.