[Measurement of head penetration on digitalized radiographs: reproducibility and accuracy]

Wear of the acetabular component of total hip arthroplasty (THA) is incriminated as the cause of loosening and bone resorption. Consequently, an accurate evaluation of wear can contribute to the prediction of mechanical failure of the arthroplasty. Among the different methods proposed, digitalized imaging using a high-resolution scanner associated with data processing procedures appears to be a simple easily accessible technique. A system based on this concept has been introduced in orthopedics. To our knowledge, there has been no report on the system's reproducibility and accuracy.

Thirty-nine radiography series for THA served as the basis for the evaluation of intra- and interobserver reproducibility. We evaluated the error induced by digitalization, the error induced by digitalized measurement, the accuracy of the measurements as a function of the material constituting the bearing, the intra- and interobserver reproducibility for repeated measures of THA wear (six observers and two observations). All measurements were done after digitalization and analyzed with a specimen designed software.

The inter and intra-observer coefficients of concordance were 0.6 and 0.58 respectively, i.e. moderate reproducibility. Depending on the prosthetic material, the error and accuracy of the system varied from 0.112 to 0.44 mm and 0.28 to 1.29 mm respectively. To obtain valid inter-observer reproducibility, the number of observers had to be limited to three (coefficient = 0.82). The type of implant had an influence on measurement error. The error was 0.342 for polyethylene cups and 0.118 for press-fit metal back cups. Likewise, for a metallic head measuring 22.2 mm, the error was 0.138 mm while for a ceramic head or metal head measuring 28 mm, the error was 0.28 mm and 0.112 mm respectively.

The accuracy and error depend directly on the type of implant. The accuracy was better for metallic heads associated with metal-backed cups. To obtain satisfactory interclass correlation, the number of observers should be three. The proposed digitalized measurement system should be relatively accurate and reproducible. Its use can be recommended for the evaluation of wear after five years of follow-up, limiting the number of observers to three.