3.4 Culture-guided antibiotics
Three retrospective cohort studies investigated the use of antibiotics in the context of tracheal aspirate cultures.
Two studies aimed to establish the usefulness of endotracheal aspirate cultures in guiding choice of antimicrobial therapy. Prinzi et al. looked at the association between over-reporting of such cultures in tracheostomised paediatric patients and its subsequent effect on antimicrobial prescribing.37Over-reporting was defined according to the American Society of Microbiology guidelines as reporting organisms not known to be respiratory pathogens. During the one-year study period, 826 endotracheal aspirate cultures were collected from 448 children. From these cultures, 310 isolates were identified in tracheostomised children. Of which, 25 (8%) organisms were over-reported, resulting in 48 days of excess antimicrobial therapy. Cline et al. aimed to assess the utility of surveillance cultures (routine tracheal aspirate cultures) in children with tracheostomies in guiding antimicrobial selection for subsequent LRTIs.38The study concluded that due to the dynamic nature of the tracheal microbiome on serial cultures, historical cultures are of little value to dictate antimicrobial choice in subsequent infections. Indeed, they report that in over half of cases (n = 36), limiting empirical antimicrobials to a previous culture result would not cover organisms isolated on subsequent cultures.
Yalamanchi et al. explored whether microscopic purulence, which is the quantitative assessment of neutrophils, in positive tracheal aspirate cultures could be used to predict subsequent antimicrobial use in a single centre retrospective review39. In their study cohort of 231 children admitted to intensive care units (81 tracheostomised), there were 361 positive tracheal aspirate cultures, of which a fifth (22%, n=81) were treated with antibiotics. Using multivariate logistic regression, they showed microscopic purulence to be an independent predictor of antimicrobial use, alongside pyrexia and respiratory failure. However, microscopic purulence was not associated with clinical symptoms of infection (hypotension, fever, or increased respiratory support). It should also be noted that this regression model aimed to predict current antibiotic prescribing practice rather than a “gold-standard” benchmark of confirmed bacterial infections. As such it may represent a useful metric in aiding the decision to initiate antimicrobial therapy, but only in the clinical context of suspected infection.