Purpose: Despite the development of new imaging techniques (MRI, CT scan) longitudinal studies of total hip arthroplasty (THA) are still conducted with conventional radiographs. New techniques for conventional radiograpy such as luminous screens with memory raise the question of longitudinal study in patients with THA where the new screen-film might produce artefacts.

Material and method: This prospective study examined intermethod and interobserver agreement. Thirty-seven patients were included in the series from July 1st, 1998 and September 30, 1998. Each patient had a double radiography series: three plain films using the conventional technique (C) and three digitalized screen films (D). The C were taken with a 1/1 ratio on a Philips Diagnost 90 table and developed using the Kodak M6 method using a 36 x 43 cm cassette for the pelvic x-ray and a 24 x 30 cm cassette for the x-ray of the prosthetic hip. The D were made on the Philips Diagnost 90 table and developed with the Agfa ADC70 procedure on a memory screen with a 5 pl/mm spatial resolution for 36 x 43 cm for the pelvic x-ray and 28 x 35 for the prosthetic hip. The same operator performed the complete radiography series in the same patient (C and D). The images were read examining the cement/bone interfaces and the prosthesis/cement interfaces looking for the classical radio-lucent lines in the De Lee and Charnley sectors. The presence and the thickness of the radiolucent line were classed in three groups: no line, line less than 2 mm, line greater than 2 mm. For each patient, the films were placed in anonymous folders and two subgroups were selected at random for the readers (a radiologist and an orthopaedic surgeon) who did not read successively the same films for the same patient. The results were recorded with an Excel data sheet and the statistical analysis was done with the BMDP software.

Results: Thirty-seven patients were included (22 women and 15 men) with 40 THA. Mean age was 64 years (42–86). Mean follow-up of the THA was 25 months (2–248). Four patients had mechanical pain or deceased joint amplitude and 33 patients had no clinical sign. Only one lucent line was found measuring less than 2 mm in the 1st quadrant of the cup and the 3, 4, and 5 zones on the AP view and the 10, 11, 12 zones on the lateral view on the tail of the pros-thesis. In this study, the kappa value was less than 0.5 for the mean concordance according to the Landis and Koch classification. The Kappa was higher for the intermethod analysis irrespective of the reader, than for the interobserver analysis.

Discussion, conclusion: Independent readers of the two types of images (C and D) did not demonstrate any difference for cemented or noncemented prosthesis in a longitudinal study of THA. The reproducibility between the C and D techniques was small. However, our study only analysed a few of the numerous radiographic signs considered to favour loosening (stress shielding, lucent lines etc.…). However, the analysis of the Kappa results demonstrated mean concordance between the techniques better than mean concordance between observers.