Total hip arthroplasty after hip arthrodesis performed for septic arthritis

Revision total hip arthroplasty (THA) after hip arthrodesis is an uncommon and challenging operation. The task would appear to be even more difficult if the arthrodesis was performed because of septic arthritis due to the theoretical risk of recurrent infection. We report our fifteen-year experience. This retrospective study concerned 17 procedures performed in 17 patients (11 women, 6 men) between 1988 and 2003 on 5 right and 12 left hips. All of the patients had arthrodesis for sepsis: eight subsequent to tuberculosis and nine subsequent to septic arthritis (Staphylococcus aureus). We examined the impact of the initial arthrodesis (surgical technique, position, leg length) on neighboring joints and indications for de-fusion. Mean age was 53 years (range 32-74) and on average, the patients had a fixed hip for 36 years (range 7-59). Mean follow-up was six years (range 11 months to 15 years). Revision surgery was performed via a posterolateral approach for 12 hips (nine trochanterotomies) and via an anterolateral approach for five hips for implantation of nine cemented implants, six press fit implants, and two hybrid implants (cemented cup and press fit stem). Clinical assessment at last follow-up noted pain, walking capacity and joint motion. Leg length discrepancy was measured and complications were noted. The position of the original arthrodesis was considered satisfactory (flexion 20 degrees , adduction 0-10 degrees , external rotation 0-20 degrees ) for eight hips; leg length discrepancy was 4 cm (2-8 cm). Neighboring joints involved concerned the lumbar spine in 15 patients, the ipsilateral knee in ten patients, the contralateral knee in eight and the contralateral hip in six. The decision to remove the arthrodesis was based on functional needs related to lumbar pain (n=6), the homolateral knee (n=10, limping and leg length discrepancy), or an operation on the ipsilateral knee. After surgery, 14 hips (83%) were free of pain with improvement of the lumbar pain and pain of the homolateral knee. Six patients walked without support but 16 still had a limp. Flexion was 78 degrees . Leg length discrepancy was 2.5 cm on average and seven patients had balanced limbs. The postoperative period was uneventful for 14 of 17 patients (one paresia of the common fibular nerve, one femoral phlebitis, one early infection). Six late complications were noted: nonunion of the greater trochanter (n=2), recurrent ankylosis (n=1) and loosening (n=3). An earlier history of in