M OSTHOFF¹, N GÜRTLER¹, S BASSETTI², GM BALESTRA³, S MARSCH³, H PARGGER⁴, M WEISSER¹, A EGLI<sup>5

1</sup>Infectious Diseases and Hospital Epidemiology, University Hospital Basel, Basel, ²Internal Medicine, University Hospital Basel, Basel, ³Medical Intensive Care Unit, University Hospital Basel, Basel, ⁴Surgical Intensive Care Unit, University Hospital Basel, Basel, ⁵Clincial Microbiology, University Hospital Basel, Basel

Background: Rapid identification of pathogens directly from positive blood cultures (BCs) has been associated with improved antibiotic treatment and patient outcomes in combination with an antimicrobial stewardship program (ASP). Hence, we aimed to quantify the impact of rapid identification in a hospital setting with low resistance rates and an already established ASP.
Material/methods: 425 patients with positive BCs during a one-year period underwent randomization by weekday into two arms: (i) rapid identification by MALDI-TOF (Microflex, Bruker) directly from positive BCs (BactAlert FA/FN plus, bioMérieux) using the Sepsityper Kit (Bruker) versus (ii) conventional processing including subculture and subsequent identification. 57 patients were excluded from analysis (lost to follow-up). Episodes of bloodstream infections (BSI) (n=242) and contaminants (n=126) were analyzed separately.

Results: Median time to organism identification was significantly shorter in the MALDI-TOF group (23.5h vs. 46.0h, p<0.001). We could not observe a significant difference in the duration of intravenous antimicrobial therapy (median 8.6d vs. 10.7d, p=0.7) or length of hospital stay (11d vs. 13d, p=0.5) between the groups. We observed a trend towards a reduced 30-day mortality in the MALDI-TOF group (HR 0.5, 95% CI 0.2-1.1, p=0.09). Importantly, effective treatment within 2h after identification was administered more frequently in the MALDI-TOF group (97.4 vs. 90.6%, p=0.03). Rapid identification resulted in improved optimal treatment at 48h only in the subgroup of BSIs (100/242 episodes) caused by AmpC-producing and non-fermenting organisms, S. aureus and Streptococcus spp. (57 vs. 38%, p=0.06), whereas it had no impact on E. coli or Klebsiella spp. BSI (98/242). In patients with contaminations rapid identification was associated with a shorter median duration of antimicrobial therapy (7.0d vs. 9.3d, p=0.04).

Conclusions: Rapid identification by MALDI-TOF directly from positive BCs provided fast and reliable microbiological results and improved treatment quality in case of contaminated BCs and BSIs not caused by E. coli and Klebsiella spp in a setting with an established ASP.