Survivorship analysis is a useful way of expressing the long-term results of joint replacement: it provides a means of comparing the life span of one type of prosthesis with another. However, such studies should include a full life-table and the confidence intervals for the survival rates at given times. At least 40 surviving prostheses are required to provide reliable results.

We have reviewed 15 patients with infected total knee replacements after removal of the prosthesis, rigorous debridement, antibiotic irrigation, and prolonged systemic antibiotics. Infection was permanently eradicated in all patients; they were left with a functioning limb, on which they could walk with either a caliper (8 patients), a simple splint (3), crutches, or sticks. Three were disappointed because of residual pain. We believe that, if exchange arthroplasty is inappropriate, this procedure is preferable to arthrodesis or amputation for persistent and disabling infection, particularly where constrained artificial joints have been used.

Between 1970 and 1982, 50 total elbow replacements were carried out for rheumatoid arthritis using the Stanmore prosthesis. A long-term follow-up of the 44 elbows available for review is presented. Thirty-four of these (77%) had good results, five were fair, and five were poor. The complications and limitations are discussed.

Between 1969 and 1971 20 prototypes of the Stanmore total knee prosthesis were used to replace severely arthritic knees in 18 patients. Seven patients died before the final follow-up and one had to have her leg amputated because of deep infection; in none of these patients had the prosthesis become loose. Ten patients (11 knees) were reviewed at least 10 years after operation. There was no significant clinical deterioration in 7 of these 11 surviving knees when the results at one year were compared with those at 10 years.

We review 210 Stanmore knee replacements in 163 patients to assess the survival of the prostheses and the long-term results. The annual rate of failure reached a maximum of 4.6% in the fourth year after operation; thereafter it declined to reach zero by the eighth year. Between two and eight years after operation, 66.3% of the surviving knees were completely free of pain and 30.2% had mild retropatellar pain. Fixed flexion deformities present before operation were completely corrected in 73% of the knees, and varus or valgus deformities were invariably corrected. Stability was always restored to unstable knees and 80.8% of knees flexed to 90 degrees or more after replacement. Aseptic loosening (8.1%), prosthetic infection (4.3%) and femoral fracture (2.9%) led to 8.5% of the prostheses being revised or removed over eight years. Modifications in prosthetic design and operative techniques have been introduced to minimise such complications in the future.

The Stanmore hinged total knee replacement was introduced in 1969 for severe destructive arthropathy of the knee, and the results of one hundred consecutive operations are presented after an average interval of two and a half years. Insertion of the prosthesis relieved pain in 94 per cent, improved the range of movement in 67 per cent, invariably restored stability and corrected valgus or varus deformity. Gross flexion contractures were improved but not always fully corrected. Serious complications were few, though of three cases of deep infection two came to amputation and one to fibrous ankylosis. There were no mechanical failures of components of the prosthesis.