Background: Understanding the causative microorganisms and initiating appropriate empirical antibiotics early are important in the management of knee periprosthetic joint infections (PJIs). The aim of this study was to identify trends in PJI microorganisms and antibiotic resistance profiles over 24 years to guide empirical antibiotic selection. Methods: This study included 487 first-episode PJIs identified between 2000 and 2023 following primary total knee arthroplasty (TKA) at 3 large tertiary hospitals. PJIs were classified using the Tsukayama classification, which is based on the timing from the primary TKA and the source of infection. Multivariable logistic regression was used to analyze risk factors for polymicrobial and resistant infections. Results: A total of 487 PJI cases with 608 culture specimens were identified. The mean patient age (and standard deviation) was 70 ± 11 years, with 65% male patients and 35% female patients. All ethnicity data were self-reported. Of the patients in this study, 57% were New Zealand European, 14% were other European, 14% were Pacific Islander, 10% were New Zealand Māori, and 6% were Asian. The most common pathogen for PJIs was Staphylococcus aureus . The proportion of resistant cases (19% to 24%) was consistent across the 24-year period. A prosthesis in situ for 90%. Conclusions: Despite the known emergence of resistant organisms in health-care settings, the primary causative microorganisms remained the same in knee PJIs, with no notable increase in resistant cases, over 24 years. Based on the findings of this study, vancomycin with gram-negative coverage is recommended as the empirical treatment of choice in early PJIs, and beta-lactams, such as flucloxacillin and a first-generation cephalosporin (e.g., cefazolin), were found to still be effective for late hematogenous PJIs. For septic PJI, dual therapy with vancomycin and a gram-negative agent is recommended, regardless of infection timing. Level of Evidence: Prognostic Level III . See Instructions for Authors for a complete description of levels of evidence.
Guo et al. (2026) studied this question.