Abstract
OBJECTIVE. To determine the diagnostic performance of blood culture positivity times for distinguishing true bacteremia from contaminants in the automated "BACT/ALERT®" system. MATERIALS AND METHODS. A cross-sectional, diagnostic test-type study was conducted from a database of blood culture samples processed between January 2016 and August 2021. All blood culture samples from patients with suspected bacteremia were included; blood culture samples were entered into the "BACT/ALERT®" system to differentiate true bacteremia from contaminants. RESULTS. We obtained 33,951 blood cultures samples, of which 3875 were positive. Of the total number of positive blood cultures, 75.2% (n=2913) were true bacteremia and 24.8% (n=962) were contaminants. The median time to positivity in blood cultures with true bacteremia was significantly shorter (16.3 hours; IQR: 11.2 - 24.9) than the median time to positivity of blood cultures with contaminants (22.5 hours; IQR: 18.4 - 31.8; p<0.001). The positivity time showed the capacity to differentiate true bacteremia from contaminants, with an AUC-ROC of 0.73 (95%CI: 0.71 - 0.75), with 85% and 63% sensitivity and specificity respectively for the diagnosis of contaminants when the positivity time exceeds 16.5 hours. The use of antibiotics prior to sampling delayed the time to positivity, while having fever before sampling shortened the time to positivity. CONCLUSIONS. Our results show good diagnostic performance of blood culture positivity times to differentiate true bacteremia from contaminants using the "BACT/ALERT®" system when the positivity time was longer than 16.5 hours.