Beadmill processing combined with automated blood culture bottle methods (BACTEC™) has a greater sensitivity and specificity, and a shorter time to positivity compared with primary plates (PP) for prosthetic joint infection (PJI) diagnosis but the clinical impact of Bactec on antimicrobial therapy has not yet been evaluated. We compared time-to-positivity of Columbia agar with horse blood plates (BA) and chocolatized horse blood plates (CHOC) versus anaerobic (ANA) and aerobic blood culture bottles (02) in patients with PJI. We compared the contributions of the two methods to the commencement of effective and targeted antimicrobial therapy.

Retrospective observational study from June 2013 to March 2014. Inclusion criteria were confirmed PJI (IDSA criteria) with at least 2 perioperative samples. After beadmill processing BA and CHOC plates were incubated for 2 days and discarded if negative, BactecTM bottles were incubated for 14 days and sub-cultured if positive. MALDI-TOF (Microflex, Brucker) was used for identification and all isolates had sensitivities performed (Phoenix, BD). Standard empirical antibiotic treatment was teicoplanin, piperacillin/tazobactam and amikacin. We defined time to switch as difference between date of sample collection and date of commencing targeted or effective therapy; prior antibiotic therapy was defined as the use of antibiotics within 14 days before samples collection.

Fifty cases were identified during the study period. 330 microbiological isolates were included: 24 (7.3%) were considered contaminants; 153 isolates (50.0%) were detected both from BactecTM and PP; 152 (49.7%) from BactecTM only; 1 isolate (0.3%) from PP only. 17 (34%) diagnoses of PJI was made exclusively by BactecTM. The majority of isolates on BA and CHOC plates grew in the first 24 hours (81.2% and 77.5% respectively). 293/305 isolates from BactecTM (96.1%) grew in the first 2 days. Antibiograms were available after 2.5 days from PP versus 4 days from BACTEC (p<0.0001). When we compared time to switch from empiric to targeted therapy, no difference was seen between patients with positive BACTEC cultures only (median 4 days, range 2–15) versus patients with positive PP cultures, (median 5 days, range 2–9) (p=0.984). Where organisms were resistant to empirical therapy, PP results did not contribute to switching to effective therapy. Prior antibiotic therapy had no impact on time-to-positivity for both methods (R=−0.005, p=0.936).

Compared to BACTEC cultures for the diagnosis of PJIs, primary plate cultures did not provide additional diagnostic information and did not significantly reduce the time to effective or targeted antimicrobial therapy.