We retrospectively reviewed 21 patients (22 shoulders) who presented with deep infection after surgery to the shoulder, 17 having previously undergone hemiarthroplasty and five open repair of the rotator cuff. Nine shoulders had undergone previous surgical attempts to eradicate their infection. The diagnosis of infection was based on a combination of clinical suspicion (16 shoulders), positive frozen sections (> 5 polymorphonuclear leukocytes per high-power field) at the time of revision (15 shoulders), positive intra-operative cultures (18 shoulders) or the pre-operative radiological appearances. The patients were treated by an extensive debridement, intravenous antibiotics, and conversion to a reverse shoulder prosthesis in either a single- (10 shoulders) or a two-stage (12 shoulders) procedure.

At a mean follow-up of 43 months (25 to 66) there was no evidence of recurrent infection. All outcome measures showed statistically significant improvements. Mean abduction improved from 36.1° (sd 27.8) pre-operatively to 75.7° (sd 36.0) (p < 0.0001), the mean forward flexion from 43.1° (sd 33.5) to 79.5° (sd 43.2) (p = 0.0003), and mean external rotation from 10.2° (sd 18.7) to 25.4° (sd 23.5) (p = 0.0037). There was no statistically significant difference in any outcome between the single-stage and the two-stage group.

There are few reports in the literature of the diagnosis and treatment of the infected shoulder arthroplasty. Most deal with resection arthroplasty and two-stage exchange surgery. We present our results of one-stage exchange operation as treatment for the infected shoulder arthroplasty.

Our group comprised 16 patients (ten men, six women) with 16 infected arthroplasties. By the time of follow-up, two patients had died (mean 5.8 years), two could not be located and three had already undergone revision surgery. Nine patients were thus available for clinical examination and assessment.

The infections were largely caused by staphylococci, Propionibacterium species and streptococci. Two were early infections (within three months of surgery) and 14 were late infections. The mean follow-up was 5.8 years (13 months to 13.25 years) when the mean Constant-Murley score was 33.6 points and the mean University College of Los Angeles score 18.3 points.

Further revision was performed in three patients. One sustained a peri-prosthetic humeral fracture, another developed an acromial pseudarthrosis after transacromial surgery and the third suffered recurrent dislocations. No patient had a recurrence of infection.

A one-stage exchange procedure using antibiotic-loaded bone cement eradicated infection in all our patients and we suggest that such a procedure is at least as successful as either a resection arthroplasty or a two-stage exchange in the management of the infected shoulder arthroplasty.