[Revision in non-infected total knee arthroplasty: an analysis of 69 consecutive cases]

We reviewed 69 consecutive cases of total knee arthroplasty revisions to analyze the causes of failure.

Sixty-nine total knee arthroplasty revisions were required between 1990 and 1997 for non-septic failure. Five categories of failures were identified: 30 loosenings including 11 with an initial malposition (varus position of the tibial component in 8 cases), 14 laxities (medial in 5, lateral in 5 and anteroposterior in 4), 11 stiff knees with no other clinical or radiological anomaly, 6 patellar failures (2 dislocations, 2 cases of excessive wear, 2 painful knees with a Freeman prosthesis), and 8 cases of painful knees with no other detectable anomaly.

A three-phase reconstruction procedure was used after removing the failing TKA: 1) reconstruction of the tibia with replacement of lost bone, 2) reconstruction of the femur with balanced flexion determining the size of the implant, 3) balanced extension determining the distal/proximal position of the femoral component. A "simple" sliding prosthesis was used in 16 cases, a modular reconstruction prosthesis in 40 cases and a hinge prosthesis in 13 cases. Mean follow-up for functional and radiographic assessment after revision surgery was 37 months (59 cases) with a minimum follow-up of 1 year. The best outcome was observed in the "loosening", "laxity", and "stiffness" patients. Outcome was less favorable for the group "isolated pain" with IKS functional scores of 35.5 +/- 16 and 52.5 +/- 21.

In 36 p. 100 of cases, TKA failure was related to a technical mistake (component malposition, poor ligament alignment). In 33 p. 100, failure was patient related (multiple procedures, congenital hip dysplasia, rheumatoid arthritis.). Outcome after revision TKA was less favorable than after primary TKA, particularly in case of painful knees with no other detectable anomaly.

Surgical revision of TKA must follow a rigorous procedure with a detailed preoperative work-up. The decision for revision must not be made unless a precise anomaly has been identified.