Between 1978 and 1986, 59 patients received a GSB III elbow prosthesis, six of them in both elbows. Rheumatoid arthritis (RA) was the underlying cause in 51 of the patients and post-traumatic osteoarthritis (PTOA) in eight. Of these, 24 patients (28 prostheses) have since died; two, both operated on bilaterally, had had their implants for more than ten years and had already been assessed for inclusion in the long-term follow-up. Two patients, each with one elbow prosthesis, have been lost to follow-up and three males who are still living (two with PTOA, one with juvenile RA) had their prosthesis removed before ten years had elapsed.

The remaining 32 patients (28 RA, 4 PTOA) with 36 GSB III elbows were examined clinically and radiologically after a mean period of 13.5 years. Pain was considerably reduced in 91.6%. Mobility was increased by 37° in those with RA and by 67° in those with PTOA.

There were three cases of aseptic loosening and three of deep infection. The main complication was disassembly of the prosthetic component in nine elbows (13.8%). This last group included two patients with postoperative fractures unrelated to the operative technique and one with neuropathic arthritis. Ulnar neuritis occurred in two patients.

Since 87.7% of all the GSB III prostheses implanted in this period remained in situ, our results are comparable with those for hip and knee arthroplasty.

Loosening is a serious problem in total arthroplasty and early detection of bone loss in the vicinity of an implant would help in its investigation. We present a method for the objective evaluation of bone adjacent to metallic implants in which a modified technique of quantitative computed tomography (QCT) is used to reconstruct cross-sectional images with few artefacts. We have used this technique in 19 patients with knee arthroplasties to monitor the changes in bone density around the tibial stem of the prosthesis. In the first weeks after operation all patients showed a decrease in bone density ranging from 0.4% to 3.6% per month. One year after arthroplasty bone density had stabilised and only minor changes were observed. Our work indicates that modified QCT is a sensitive method for the long-term monitoring of the anchorage of implants and allows the early detection of osteolytic changes.