BLOOD CULTURE CONTAMINATION - A COLLEGE-OF-AMERICAN-PATHOLOGISTS Q-PROBES STUDY INVOLVING 640 INSTITUTIONS AND 497 134 SPECIMENS FROM ADULTPATIENTS

Objective.-To examine clinical and laboratory practices associated wit h contamination of blood culture specimens from adults. Design and Set ting-A College of American Pathologists Q-Probes quality improvement s tudy involving prospective evaluation of adult blood culture contamina tion rates in 640 institutions. Main Outcome Measure.-Proportion of co ntaminated blood cultures. Results.-A total of 497 134 blood cultures were studied. The median adult inpatient blood culture contamination r ate was 2.5% (central 80th percentile = 0.9%-5.4%) by laboratory asses sment. There was no significant difference in contamination rates betw een inpatient and outpatient cultures (P = .273). The median contamina tion rate by clinical assessment (2.1%) was significantly lower (P = . 005), primarily because of a lower proportion of cultures with coagula se-negative Staphylococcus that were interpreted as contaminants when only one of multiple specimens was positive. Specimen collection varia bles associated with significantly lower contamination rates included use of a dedicated phlebotomy service (P = .039), use of tincture of i odine for skin disinfection (P = .036), and application of an antisept ic to the top of the collection device before inoculation (P = .018). Teaching institutions and high numbers of occupied beds were demograph ic factors associated with higher contamination rates for inpatients b ut not for outpatients. Culture parameters associated with higher cont amination rates included microbial growth from a single specimen, isol ation of certain microbial species (eg, coagulase-negative Staphylococ cus), and longer time to detect growth in culture. Contamination rates were not significantly affected by the type of blood culture method u sed, specimen volume, or use of a double-needle collection procedure. Conclusions.-There is wide variation in blood culture contamination ra tes among institutions. Three specimen collection factors and three cu lture variables were identified as having a significant effect on bloo d culture contamination.