Inter- and intraobserver study of radiographic assessment of cemented total hip arthroplasties

Radiographs are commonly used to identify loosened total hip prostheses. Interobserver and intraobserver variation was studied in an attempt to interpret 60 random pairs of radiographs with a time interval of 6 months to 12 years between the radiographs. Each of the four observers evaluated all pairs of radiographs independently, and the evaluations were repeated after 2 to 4 weeks. Each observer classified the femoral stem and the acetabular cup separately as stable or loose. Radiolucent lines of more than 2 mm were recorded. For the acetabular components, any change in inclination or migration was noted, and for the femoral components, subsidence or any change in varus or valgus of the stem was noted. There was considerable disagreement in observations of the acetabular component; agreement by all four observers was obtained in only 34 of 60 cases (57%). For the femoral component, agreement was obtained in 37 cases (62%). The observers agreed two and two (pairwise) on the acetabular component in 70 to 83% of cases and on the femoral component, in 72 to 82% of cases. Intraobserver variation was great with regard to the acetabular component, with kappa values ranging from 0.489 to 0.633. As to the femoral component, kappa values ranged from 0.737 to 0.800. Interpretation of radiographs of artificial hip arthroplasty is difficult, and comparisons between different series of implants should be made by the same person.