An algorithm was implemented in the clinical microbiology laboratory to assess the clinical significance of organisms that are often considered contaminants (coagulase-negative staphylococci, aerobic and anaerobic diphtheroids, Micrococcus spp., Bacillus spp., and viridans group streptococci) when isolated from blood cultures. From 25 August 1999 through 30 April 2000, 12,374 blood cultures were submitted to the University of Iowa Clinical Microbiology Laboratory. Potential contaminants were recovered from 495 of 1,040 positive blood cultures. If one or more additional blood cultures were obtained within ؎48 h and all were negative, the isolate was considered a contaminant. Antimicrobial susceptibility testing (AST) of these probable contaminants was not performed unless requested. If no additional blood cultures were submitted or there were additional positive blood cultures (within ؎48 h), a pathology resident gathered patient clinical information and made a judgment regarding the isolate's significance. To evaluate the accuracy of these algorithm-based assignments, a nurse epidemiologist in approximately 60% of the cases performed a retrospective chart review. Agreement between the findings of the retrospective chart review and the automatic classification of the isolates with additional negative blood cultures as probable contaminants occurred among 85.8% of 225 isolates. In response to physician requests, AST had been performed on 15 of the 32 isolates with additional negative cultures considered significant by retrospective chart review. Agreement of pathology resident assignment with the retrospective chart review occurred among 74.6% of 71 isolates. The laboratory-based algorithm provided an acceptably accurate means for assessing the clinical significance of potential contaminants recovered from blood cultures.False-positive blood cultures lead to additional laboratory tests, unnecessary antibiotic use, and longer hospitalizations that increase patient care costs (2, 16). Assessing the clinical significance of blood culture isolates can be difficult, but actions based on the accurate identification of contaminants may minimize the associated costs and lower future contaminant rates. An algorithm was implemented in the University of Iowa Clinical Microbiology Laboratory to assess the clinical significance of organisms that are often considered contaminants when isolated from blood cultures. An independent retrospective chart review was used to assess the accuracy of automatic assignments made by the algorithm, determinations of clinical significance by pathology residents, and the reclassification of probable contaminant isolates as pathogens after physicians requested antimicrobial susceptibility testing (AST).(This report was presented in part at the 101st General Meeting of the American Society for Microbiology, 21 May 2001 [abstr. C-65].)
MATERIALS AND METHODSFrom 25 August 1999 through 30 April 2000, 12,374 blood culture sets were submitted to the University of Iowa Clinical Microbiology Laboratory....