Moderate accuracy of the AO/OTA Forearm Shaft Fractures Classification as used in the Swedish Fracture Register.

Abstract

This study evaluated the accuracy of classification of forearm shaft fractures (FSFs) in adult patients registered in the Swedish Fracture Register (SFR). A secondary aim was to examine intra- and inter-rater reliability for the modified 2007 AO/OTA classification within the SFR. FSFs were classified into three types (A-C) based on fracture pattern and further grouped (1-3) according to involvement of the ulna, radius, or both. Of 130 randomly selected radiograph sets, 122 were available and suitable for validation. Three orthopaedic consultant trauma surgeons, blinded to patient and SFR data, independently classified the radiographs on two occasions, 6 weeks apart. A consensus gold-standard classification was established and used to assess the accuracy of SFR-recorded classifications. Intra- and inter-rater reliability among the experts was evaluated using kappa coefficients. The accuracy of SFR classifications compared with the gold standard was moderate (kappa = 0.56 for type and group). Classification accuracy was lowest for B1-3 fractures (kappa 0.24) and highest for A1-3 fractures (kappa 0.73). Expert raters consistently classified oblique fractures with an undisplaced wedge as B types, whereas SFR users often recorded them as A types. Intra-rater reliability among experts showed substantial to almost perfect agreement (kappa 0.70-0.82), and inter-rater reliability was similarly substantial to almost perfect agreement (kappa 0.78-0.92). Registration of FSFs in the SFR demonstrated moderate accuracy compared with the gold standard. Intra- and inter-rater reliability among experts was substantial to almost perfect, indicating that the classification system is comprehensible. However, lower accuracy for B-type fractures highlights the need to update the SFR classification guidelines for interpreting undisplaced wedge fractures.