Staged Revision With Antibiotic Spacers for Shoulder Prosthetic Joint Infections Yields High Infection Control

BACKGROUND The treatment of periprosthetic joint infection (PJI) of the shoulder with two-stage revision arthroplasty using an antibiotic-loaded cement spacer is established strategy, but there is sparse information regarding the likelihood of infection control and restoration of shoulder. QUESTIONS/PURPOSE (1) What is the likelihood of infection control after two-stage revision using an antibiotic cement spacer for patients with PJI of the shoulder? (2) What are the improvements in Constant and Murley scores at 2 years after these staged revisions? PATIENTS AND METHODS Between 2000 and 2013, we treated 48 patients with PJI of the shoulder using two-stage revision including an antibiotic-containing cement spacer during the first stage. Of those, 38 (79%) were available for review at a minimum of 24 months (mean, 52 ± 34 months). Ten patients (21%) were excluded because they were deceased (n = 3), moved abroad (n = 4), or refused followup (n = 3), leaving 38 for analysis in this retrospective study. During the first stage, removal of the prosthesis, débridement, and implantation of a gentamicin and vancomycin-filled cement spacer were performed by four different surgeons followed by antibiotic therapy (2 weeks intravenous plus 10 weeks oral). For the second stage, we generally tried a reverse total shoulder arthroplasty (RTSA; n = 26). In case of severe glenoid destruction, hemiarthroplasty (HA; n = 8) was used as a salvage option. In 14 patients the cement spacer was left in place because the patients refused further surgery or were not operable owing to medical reasons. The primary outcome included the proportion of patients achieving infection control 2 years after the second-stage procedure after implantation of the cement spacer. Infection control was determined as the absence of the Musculoskeletal Infection Society PJI criteria. The clinical outcome assessed with the Constant and Murley scores served as the secondary outcome parameter. A subgroup (RTSA; HA, spacer retention) analysis of the Constant and Murley scores was performed. RESULTS Successful infection control was achieved in 36 of 38 patients (95%). Patients who underwent treatment with a cement spacer had increased Constant and Murley scores at latest followup compared with their pretreatment scores (mean ± SD, 27 ± 19 versus 43 ± 20; mean difference, 17; 95% CI, 10-24; p = 0.001). For patients who underwent staged treatment followed by second-stage RTSA (n = 23), the Constant and Murley scores increased (mean ± SD, 31 ± 20 versus 51 ± 20; mean difference, 20; 95% CI, 11-30; p = 0.001). The Constant and Murley scores did not improve in patients who underwent HA (mean ± SD, 22 ±15 versus 24 ± 90; mean difference, 3; 95% CI, -10 to 16; p = 0.509) or who retained the spacer (mean ± SD, 18 ±12 versus 35 ±10; mean difference, 19; 95% CI, -5 to 44; p = 0.093). CONCLUSION Revision arthroplasty using an antibiotic-loaded cement spacer provided successful infection control in patients with periprosthetic shoulder infections in this small, retrospective series. Functional improvement was obtained after reimplantation of a reverse total shoulder prosthesis but was not seen after HA and cement spacer; however, baseline differences among patient groups very likely contributed to these differences, and they should not be attributed to implant selection alone. LEVEL OF EVIDENCE Level III, therapeutic study