Epidemiological and clinical features of prosthetic joint infections caused by gram-negative bacteria.

OBJECTIVE: To review the clinical literature focusing on epidemiology. clinical presentation and outcomes of prosthetic joint infections (PJIs) due to gram-negative bacteria (GNB) and to report the experience of a multicentric cohort. PATIENTS AND METHODS: A retrospective, observational, cohort study was performed in three Italian hospitals. All consecutive PJIs caused by GNB over a 12-year period (from May 2007 to March 2018) were enrolled. Epidemiological, clinical, microbiological and therapeutic features were described. Factors related to treatment failure (defined as the occurrence of death, amputation or starting long-term antimicrobial suppression therapy) were analysed with a Cox regression model. RESULTS: A total of 82 PJIs due to GNB (42.7% men; median age 73 years) were studied. The implants included 65 (79.3%) hip, 16 (19.5%) knee and one (1.2%) shoulder. An early PJI was diagnosed in 16.2% of patients. a delayed PJI in 29.4% and a late PJI in 54.4%. The most common isolated organisms were Escherichia coli (21.7%) and Pseudomonas spp. (20.9%). 13.4% of the isolates were carbapenem-resistant bacteria (CRB). In 53.8% of cases a two-stage exchange arthroplasty was performed and in 32.5% a Girdlestone excision arthroplasty. The average therapeutic failure occurred in 17.7% of cases. The therapeutic failure rate of the two-stage was 10%. PJI due to CRB was identified as a potential risk factor for failure (aHR 4.90; IC 95%, 0.96-25.08; p=0.05). The therapeutic failure rate in the CRB group was 50%. CONCLUSIONS: The treatment with the two-stage procedure for PJIs caused by GNB seems to be associated with a low rate of failure, while PJI due to CRB seems to be related to the worst outcome.