To assess concordance and reproducibility in the observation of radiolucent lines of the humeral prosthetic component in cemented shoulder hemiarthroplasties.
Five observers evaluated 128 X-rays obtained from 32 cemented shoulder hemiarthroplasties at 2 points of time Turing follow-up in anteroposterior and scapular profile views; a re-evaluation was made after 3 weeks. Radiolucent lines were studied in 7 areas around the implant at the cement-bone and cement-implant interfaces. Radiolucencies were classified into 4 categories according to their size in millimeters. Results were assessed with the kappa statistical method.
In the study of intraobserver reproducibility kappa indices of 0 to 0.6 were obtained for the implant-cement and cement-bone interfaces. In the study of interobserver concordance for each of the areas and interfaces, 65 % of kappa indices obtained range between 0 and 0.4. Intraobserver capacity to follow up one same patient in time was assessed on the basis of the percentage of possible responses by each observer, which ranged between 85 and 90 % .
There is a low rate of concordance and reproducibility when analyzing radiolucencies both at the bone-cement and at the cement-implant interfaces. There is also a low error index when analyzing the evolution of radioludencies in 2 X-rays of the same implant obtained at 2 different points in time.