We report 60 synovectomies of the knee in 51 children with juvenile chronic arthritis. Synovitis had been present for an average of 5 years and the average age at operation was 13 years. Results were evaluated in terms of pain, knee movement, relapse of synovitis and radiological change during a follow-up averaging 7.5 years. The relief of pain was rewarding and there was a slight postoperative gain in range of knee movement in most cases. The older the patient at onset of disease, the greater the risk of pain during follow-up. Progressive joint destruction was more common in younger patients, those with systemic or polyarticular disease, and those with highly active disease at the time of operation. Recurrence of synovitis was more frequent in patients who had their operation in a phase of high disease activity; this occurred most often in patients with polyarticular disease.

Twenty consecutive patients treated by arthrodesis for failed knee arthroplasty are reviewed. Eight hinged, five stabilised and seven compartmental prostheses were removed, for infection (15 cases), loosening (4) and instability (1). One patient refused a second-stage operation but the remainder gained sound fusion. Infected knees had a two-stage procedure with temporary insertion of gentamicin-loaded beads after removal of the prosthesis; all infections healed. Six arthrodeses using a Hoffmann-Vidal external fixator resulted in two temporary failures. One Ace-Fischer external fixation was successful. Of 10 primary attempts at arthrodesis with an intramedullary Kuntscher nail, nine were successful; the tenth fused after two more attempts by the same method. The two failures of external fixation and two failures after Charnley single-frame compression done elsewhere were successfully fused with intramedullary nails. Delayed union in three cases fused after prolonged fixation and repeated bone grafts. The indications for and methods of arthrodesis after failed knee arthroplasty are discussed.

Fifteen patients with rheumatoid arthritis had 19 elbow arthroplasties with the Wadsworth type of surface replacement prosthesis. After a follow-up period averaging 30 months, 11 patients with 15 elbow arthroplasties were entirely satisfied with their freedom from pain and range of movement. Radiologically, however, the humeral component was loose in 10 of the 19 elbows and the ulnar component in 5. Two patients had reoperation, one to remove a prosthesis for early deep infection and one to exchange a humeral component which was loose. The risk of mechanical loosening is reduced by accurate positioning of the humeral component, but there is a high potential for failure. Changes in the design of the prosthesis and better instrumentation for alignment of the components are desirable. Prosthetic replacement of the elbow should still be regarded as experimental.