The BacT/Alert blood culture system (bioMerieux, Marcy l'Etoile, France) was used throughout the study period.

Therefore, we conducted this study to evaluate the accuracy of the preliminary blood culture reports based on Gram stain and wet mount microscopy. Besides the direct implications for antibiotic treatment, the Gram stain result may also prompt further diagnostic and therapeutic interventions.ĭespite the acknowledged importance of the first notification, the accuracy of the Gram stain result has only been addressed sporadically, and the studies have mainly focused on distinction of either contaminants from true bacteremia ( 3, 13) or staphylococci from streptococci ( 1, 28). ( 12) reported that combining Gram stain results with information on whether the infection was hospital or community acquired could further improve the appropriateness of the antibiotic treatment. The Gram stain report has been shown to have a much greater impact on antimicrobial treatment than provision of cultural identification and antimicrobial susceptibility test results ( 17, 22), and recently Hautala et al. At this time, 12 to 20% of the patients may not have started antibiotic treatment, and in another 30 to 45% of patients, the Gram stain result is followed by a change in the empirical treatment ( 2, 7, 19, 22, 24). The first notification of a positive blood culture is typically based on the Gram stain result. Therefore, an important task for the microbiological laboratory is to provide expedient reports on positive blood cultures that may guide antibiotic therapy. Early administration of appropriate empirical antibiotic treatment has repeatedly been associated with improved survival in patients with bacteremia ( 5, 15, 27), yet up to 40% of all patients with bacteremia receive inadequate antibiotic treatment until the first notification of a positive blood culture ( 5, 6, 22, 24).

In conclusion, we demonstrated a high accuracy of Gram stain reports, whereas wet mount microscopy was generally less accurate.īacteremia is a serious condition with an overall in-hospital mortality above 20% ( 15, 19). Wet mount reports were less accurate (sensitivity of 30 to 70% for species with peritrichous motility), and Enterobacteriaceae (notably Salmonella spp.) accounted for 25% of the reports stating polar motility. The sensitivity for the most frequent species was in the range 91.3 to 100%, with nonhemolytic streptococci having the lowest value (sensitivity, 91.3% 95% confidence interval, 86.2 to 94.9%). We obtained 5,893 positive blood cultures and the following results for the defined groups: sensitivity, range of 91.3 to 99.7% specificity, 98.9 to 100% PPV, 94.6 to 100% and NPV, 99.0 to 100%. We further evaluated the Gram stain and wet mount findings for the most frequent bacterial species/groups. Using cultural identification as a reference, we estimated the sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively) for the following defined morphological groups: gram-positive cocci in clusters, gram-positive cocci in chains or diplococci, gram-positive rods, gram-negative cocci, gram-negative rods, and yeasts. The study was restricted to cultures with one morphological type. We used data from computer files and technicians' laboratory notes. We examined the accuracy of preliminary blood culture reports provided by a regional laboratory in an observational study including the years 1996, 2000 to 2001, and 2003. However, information on the accuracy of Gram staining is very limited. When blood cultures turn positive, the attending physicians are usually notified immediately about Gram stain findings.

3 gram negative rods