Observer variation in the detection of acetabular bone deficiencies

To determine observer variation in the detection of acetabular bone deficiencies, 42 pairs of frontal (AP) and lateral hip radiographs and CT studies for total hip arthroplasty patients obtained within an average of 4 weeks of each other were reviewed separately by five radiologists and one orthopedic surgeon. Interobserver variations were calculated for each individual reading the films using kappa values. The individual film readings were then compared with a consensus reading of the CT data. When separate observers were analyzed, agreement on plain film readings was slight to fair (av. kappa = 0.1440 +/- 0.1047). The individual observers were not able to give readings which were very consistent with the CT consensus reading, resulting in a low sensitivity (65%) and specificity (74%) for acetabular defect classification with plain radiographs. The identification of acetabular bone defects from the AP and lateral views of the hip is highly subjective and variable from observer to observer.