Introduction There are currently hundreds of total hip replacement (THR) prosthesis designs in use. In practice the decision to use a THR prosthesis should be based on published long-term outcomes. This study provides a critical analysis of the literature to determine what prostheses have successful published results.

Methods The Medline (Ovid) database was searched for English text publications on primary THR published between 1975 and 2002. The search produced 12219 publications with 1400 (11.5%) reporting an incidence of revision or survivorship. Criteria for potentially successful designs were applied to the percentage of hips unrevised or survivorship data. Success was defined as less than 1% revised for any cause or less than 0.5% revised due to aseptic loosening. Using reported loss to follow-up, worst case analysis was performed to divide successful prostheses into definitely (DS), probably (PB) or possibly successful (PS).

Results Only 404 (23%) of the 1404 publications reviewed were methodologically adequate. The cemented acetabular prostheses had 39 (26.5%) reports of success (46%DS, 26%PB, 28%PS) at greater than 10 years and 80 (54%) reports of failure. The cement-less acetabular prostheses had 12 (10%) reports of success (42%DS, 42%PB, 16%PS), and 55 (46%) reports of failure. For cemented femoral prostheses 50 (30%) (62%DS, 18%PB, 20%PS) reports were successful, and 86 (51%) were failures. For cementless femoral prostheses 16 (18%) (69%DS, 25%PB, 6%PS) reports were successful and 38 (42%) were failures. Nineteen different cemented acetabular designs, and 34 different cementless acetabular designs had reports of success. For the femoral prostheses there were 27 different cemented designs and 26 different uncemented designs with reports of success. Thirty-two (18.2%) prostheses designs had reports of both success and failure, 71 (40%) had reports of failure only and 73 (41.5%) had reports of success only. It was not possible to identify what design alterations, if any, were attributed to a prosthesis being classified as both a success and a failure, because the technical details of the prosthesis were often not reported.

Conclusions Only 22% of prostheses reported were successful at greater than 10 years. High loss to follow-up meant only 56% of the potentially successful prostheses were definitely successful. The reporting of prosthesis manufacturer and catalogue numbers, and the publication of good and poor results are recommended to better gauge the success of designs.

In relation to the conduct of this study, one or more of the authors is in receipt of a research grant from a non-commercial source.