Abstract
Candida species bloodstream isolates were collected from institutions participating in an active, population-based surveillance for candidemia. Species identifications were performed locally and then confirmed at the Centers for Disease Control and Prevention (CDC) by phenotype-based methods. Discrepancies in species identification between the referring institution and the CDC were noted for 43 of 935 isolates (4.6%). A DNA probe-based species identification system (PCR-enzyme immunoassay [EIA]) was then used to resolve these discrepancies. The PCR-EIA result was identical to the CDC phenotypic identification method for 98% of the isolates tested. The most frequently misidentified species was Candida glabrata (37% of all discrepant identifications). Such misidentifications could lead to the administration of inappropriate therapy given the propensity of C. glabrata to develop resistance to azole antifungal drugs.