Do traction–internal rotation radiographs facilitate the classification of trochanteric region and basicervical femoral neck fractures? An Inter- and intra-observer reliability assessment

Plain radiographs are often sufficient for diagnosis of hip fractures. Most hip fractures in older adults are treated surgically. Accurate assessment of fracture stability is important for treatment planning and appropriate implant selection. The aim of this study was to determine whether adding traction–internal rotation radiography to standard AP radiography changes inter- and intra-observer reliability and concordance with an expert reference classification when classifying trochanteric region and basicervical femoral neck fracture patterns. Patients over 65 years of age with displaced trochanteric region fractures (with or without subtrochanteric extension) and basicervical femoral neck fractures who had appropriate preoperative anteroposterior and traction-internal rotation hip radiographs were included in the study. The radiographs were evaluated by 13 orthopedic residents stratified according to residency year, which was used as a pragmatic proxy for cumulative clinical training. Evaluators first classified each fracture using the AP radiograph alone and then repeated the classification after the traction–internal rotation radiograph was added. Intra- and inter-observer reliability and concordance with the expert reference classification were assessed. Seventy-one patients were included. Overall inter-observer reliability was moderate for both the AP-only assessment (κ = 0.488, 95% CI 0.397–0.569) and the AP plus traction–internal rotation assessment (κ = 0.441, 95% CI 0.344–0.524), with no significant between-assessment difference (Δκ=−0.047, 95% CI − 0.109 to 0.017; paired permutation p = 0.170). Overall intra-observer reliability was substantial for the AP-only assessment (κ = 0.761, 95% CI 0.712–0.802) and almost perfect for the AP plus traction–internal rotation assessment (κ = 0.855, 95% CI 0.821–0.886). Concordance with the expert reference classification was 67.9% (627/923) for both assessments, with no net improvement after the additional radiograph (exact McNemar p = 1.000). The addition of traction–internal rotation radiographs to standard AP radiographs was not associated with a statistically significant improvement in overall inter-observer reliability or concordance with the expert reference classification. Although overall intra-observer reliability was higher after the additional radiograph, the inter-observer estimates changed in both directions across residency-year groups and did not demonstrate a consistent pattern of improvement. These subgroup findings should be considered exploratory. The results are limited to radiographic classification performance and do not establish whether the additional radiograph affects operative decision-making, implant selection, surgical technique, or clinical outcomes.