Failure of an Automated Blood Culture System To Detect Nonfermentative Gram-Negative Bacteria

Dr. Klaerner and colleagues (1) reported interesting data about the automated BacT/Alert blood culture system. A total of 605 out of 8,107 blood culture bottles yielded relevant microorganisms. The system failed to detect 15 nonfermentative species out of 605 bottles. The authors speculated that this was due to growth of microorganisms during preincubation at 36°C, and they suggested keeping the bottles at room temperature until loading. Preincubation at room temperature may reduce the failure to detect nonfermentative species; however, detection of other clinically relevant microorganisms may be delayed.

Automated blood culture systems (BACTEC 9240 and BacT/Alert) detect growth of microorganisms by continuous monitoring of CO₂ without manipulations of the bottles, which reduces the contamination risk and the workload (5). The microbiological yield is determined by an adequate filling of the blood culture bottles (3, 4). Therefore, the remark of Dr. Klaerner et al. that a smaller inoculum can reduce the number of false negatives appears not to be clinically useful.

Rapid detection and identification of bacteremia by a clinical laboratory add to prompt and adequate antibiotic therapy. Hence, the reported failure in detection of bacteremia is of great concern. Fast transportation of the bottles to the laboratory and immediate loading are not always feasible, and the growth of microorganisms in the bottles may already have reached steady state. The algorithm for detection of growth estimates both the initial CO₂ level (reflection units) and changes in this level but proved to be not sensitive enough. Adaptation of the software may be indicated; however, this should not cause an increase in the number of false-positive readings (2).

The results of Dr. Klaerner et al. (1) indicate the need for continuous quality improvement to ensure optimal use of the benefits of the automated blood culture systems, including short transportation time, adequate filling of the bottles, and evaluation of the detection algorithm of the blood culture system. The authors suggest subculturing of all bottles with a transportation time longer than 4 h prior to loading. This causes a substantial increase in the workload. In order to cope with the advanced growth in blood culture bottles, in our laboratory the technicians screen the color at the bottom of the bottles prior to loading. From “suspect” bottles with yellow color indicators a sample is drawn for Gram staining and subculture. In our experience the BacT/Alert system (software version BacT/Link E.00) detects growth of microorganisms in such bottles soon after loading, though we did not record this routinely. Subculturing on indication seems reliable and is less laborious than subculturing all bottles with a transportation time longer than 4 h; moreover, often the transportation time is unknown.

Measurement of the absolute number of reflection units by the system is more accurate than evaluation of the color by the human eye; however, a plot of reflection units can be generated only after six measurements (60 min). If there is a correlation between the number of bacteria in the bottles at arrival in the laboratory and the degree of yellowness (or the absolute number of reflection units), this can be used to detect preincubational growth and it would be worthwhile to investigate whether the software of the blood culture system can select bottles which should be subcultured.