We reviewed a consecutive series of 527 uninfected hip replacements in patients resident in the UK which had been implanted from 1981 to 1993. All had the same basic design of femoral prosthesis, but four fixation techniques had been used: two press-fit, one HA-coated and one cemented. Review and radiography were planned prospectively. For assessment the components were retrospectively placed into two groups: those which had failed from two years onwards by aseptic femoral loosening and those in which the femoral component had survived without revision or recommendation for revision.

All available radiographs in both groups were measured to determine vertical migration and examined by two observers to agree the presence of radiolucent lines (RLLs), lytic lesions, resorption of the neck, proximal osteopenia and distal intramedullary and distal subperiosteal formation of new bone. We then related the presence or absence of these features and the rate of migration at two years to the outcome with regard to aseptic loosening and determined the predictive value of each of these variables.

Migration of ≥2 mm at two years, the presence of an RLL of 2 mm occupying one-third of any one zone, and subperiosteal formation of new bone at the tip of the stem were predictors of aseptic loosening after two years. There were too few lytic lesions to assess at two years, but at five years a lytic lesion ≥2 mm also predicted failure. We discuss the use of these variables as predictors of femoral aseptic loosening for groups of hips and for individual hips.

We conclude that if a group of about 50 total hip replacements, perhaps with a new design of femoral stem, were studied in this way at two years, a mean migration of < 0.4 mm and an incidence of < 10% of RLLs of 2 mm in any one zone would predict 95% survival at ten years.

For an individual prosthesis, migration of < 2 mm and the absence of an RLL of ≤2 mm at two years predict a 6% chance of revision over approximately ten years. If either 2 mm of migration or an RLL of 2 mm is present, the chances of revision rise to 27%, and if both radiological signs are present they are 50%. If at five years a lytic lesion has developed, whatever the situation at two years, there is approximately a 50% chance of failure in the following five years.

Our findings suggest that replacements using a limited number of any new design of femoral prosthesis should be screened radiologically at two years before they are generally introduced. We also suggest that radiographs of individual patients at two years and perhaps at five years should be studied to help to decide whether or not the patient should remain under close review or be discharged from specialist follow-up.