Objectives: This study was conducted to determine interrater and intrarater reliabilities on the healing assessment of femoral neck fractures between orthopedic surgeons and radiologists and to test the performance of a checklist system for hip fracture healing.
Methods: We developed and used a scoring system [radiographic union score in hip fracture (RUSH) score] to determine the validity of quantifying fracture healing. A panel of 6 reviewers (3 orthopedic surgeons and 3 radiologists) independently assessed fracture healing with the RUSH system using radiographs of 150 femoral neck fractures at various stages in healing on 2 occasions 4 weeks apart.
Results: Using subjective assessment, the interrater agreement between reviewer groups for fracture healing was fair [intraclass coefficient = 0.22, 95% confidence interval (CI): 0.01-0.41] with no significant difference in agreement within the orthopedic surgeon and radiologist groups (0.17 vs. 0.21). There was higher agreement for fracture healing using the RUSH score (intraclass coefficient = 0.53, 95%CI: 0.30-0.69) compared with physician impression of healing, highlighting the difficulties with plain radiographic assessments of healing. Intrarater agreement was consistently high across all measures for both surgeons and radiologists. The RUSH score and medial cortex bridging correlated well with overall assessment of healing (r = 0.868 and 0.643, respectively).
Conclusions: The level of agreement between and within orthopedic surgeon and radiologist reviewers in the assessment of fracture healing is low, though intrarater agreement is high. The RUSH score shows promise as a tool to improve agreement on fracture healing. Studies evaluating reliability and accuracy of healing with clinical information and temporal evaluation are needed and may further improve agreement.