Fracture healing assessment comparing stiffness measurements using radiographs

Based on published reports, we presumed radiographs would be unreliable as a sole measure of fracture healing. To confirm this presumption we correlated radiographic fracture healing assessments with fracture stiffness measurements. We showed 100 plain radiographs of fractures with corresponding fracture stiffness measurements to 92 observers. The radiographs were shown twice to assess intraobserver variation. Observers were divided into three groups and asked to determine whether each fracture had healed (union corresponded to a fracture stiffness greater than 15 nm/degrees). Group 1 based fracture healing on the general appearance of healing. Groups 2 and 3 assessed fracture healing based on the number of cortices bridged by callus. In Group 2, the fracture was considered healed if two or more cortices were bridged on both radiographic views and in Group 3 if three or more cortices were bridged by callus. All groups performed poorly. There was no difference in terms of correct prediction of healing between methods, although there was a trend toward more reliability with cortical callus bridging assessment. We found substantial intraobserver variability, which improved using cortical bridging methods. Observers were less reliable at predicting healing when there was a metaphyseal extension to a diaphyseal fracture.