|Home | About | Journals | Submit | Contact Us | Français|
Coagulase‐negative staphylococci (CoNS) are the most common bacteria isolated from blood cultures, but a great proportion of them are most likely contaminants.1,2 On the other hand, one single positive blood culture for Staphylococcus aureus is usually sufficient to diagnose bacteraemia. Therefore, a rapid method of differentiating CoNS from S aureus would assist in the clinical decision on starting antimicrobial treatment in a timely manner. As many laboratories routinely process blood culture bottles in automated systems that record time to positivity (TTP) and this information is readily available along with the Gram‐smear results, we assessed the predictive value of TTP for the exclusion of S aureus in the setting of a positive result for clustered Gram‐positive cocci.1,3,4
For this purpose, we retrospectively analysed all blood cultures collected from patients between January 2005 and February 2006 in a large university teaching hospital. In our hospital, we routinely collected two blood cultures for every patient, each one consisting of a single BacT/ALERT FA aerobic bottle. The sets were incubated in the BacT/ALERT 240 system (Organon Teknika, Boxtel, The Netherlands). A routine Gram‐stain microscopic examination was performed whenever bacterial growth was detected. The TTP was recorded in the software system. The definitive identification of staphylococci relied on the tube coagulase test, which is usually available only overnight. We included only the first positive bottle from each patient in the analysis, because this is the moment when the clinical decision to initiate antibiotic treatment is made. TTP was grouped in the following time intervals: 13 h, from 13.1 to 14.9 h, from 15 to 17.9 h and >18 h (table 11).). The positive predictive value (PPV) and the likelihood ratios (LRs) for the presence of S aureus or CoNS for each time interval were then calculated (table 22).
A total of 18479 blood culture samples were collected between January 2005 and February 2006. We found 1030 isolates of staphylococci, of which 140 (13.6%) were identified as S aureus and 890 (86.4%) as CoNS (table 11).). The PPV for S aureus was 71% for isolates growing in 13.0 h of incubation (LR 15.8). In the same time interval, however, only 28% of positive isolates were characterised as CoNS (LR 0.063). At the opposite extreme (18 h), the PPV for S aureus fell to 5.6% (LR 0.37), whereas for CoNS, the PPV was 94% (LR 2.6; table 22).
The demonstration of clustered Gram‐positive cocci in the direct blood examination constitutes a common clinical dilemma.1,4 As the majority of the isolates will eventually be revealed to be CoNS, systematic initiation of early empirical anti‐Gram‐positive treatment would not be warranted because of cost, toxicity and selective pressure issues.5 On the other hand, even minimal delays in starting appropriate treatment could be deleterious in the case of true bacteraemia due to S aureus. Therefore, a rapid method of accurate differentiation between S aureus and CoNS would be of great clinical importance.1,3,4,6 Oliveira et al3 described a method of direct identification using fluorescence in situ hybridisation, which was both sensitive and specific for species differentiation. However, the test required up to 2.5 h of preparation and was not cost free. Murdoch and Greenlees6 described a method for identification based on direct Gram‐stain characteristics, such as the size of the bacterial cells and the number of cells in a typical cluster. However, the method requires expertise and may not be feasible in a large tertiary care centre that processes a great number of samples. Ruimy et al4 recently showed that the TTP, which is automatically recorded by automated systems, could be used for rapid identification. Using the BACTEC 9240 System (Becton Dickinson Diagnostic Instrument Systems, Sparks, Maryland, USA), the authors found a PPV of 83% for S aureus when the TTP was from 2 to 9 h and a PPV of 91% for CoNS when positivity occurred beyond 18 h of incubation. We theorised that this also would apply for the BacT/ALERT 240, as this system is associated with a higher mean detection time than the BACTEC 9240.7,8
In summary, to our knowledge, there are no previous studies analysing the performance of TTP using BacT/ALERT 240. Our results support the notion that the TTP is also a useful tool while using the BacT/ALERT 240. In particular, we suggest that clinicians should not routinely initiate empirical anti‐Gram‐positive treatment when the first bottle displays clustered Gram‐positive cocci after >18 h of incubation. In this setting, the very high PPV for CoNS permits deferring treatment until the results of the second bottle are known. Because of the low virulence of CoNS, the delay to starting antimicrobial treatment does not jeopardise the clinical outcome in cases of true bacteraemia. In addition, a great deal of unnecessary empirical treatment would be avoided in the hospital setting.
Competing interests: None declared.