Abstract
Identifying fungi to the genus or species level provides valuable insight for guiding antifungal therapy. We evaluated the performance and clinical utility of the Roche cobas eplex blood culture identification fungal pathogen panel (BCID-FP). Although used in fewer than 1% of patients with blood cultures drawn, BCID-FP identified fungal pathogens in more than 96% of tested samples within 72 hours. Over a 3-year period, BCID-FP demonstrated a positive percent agreement (PPA) of 91.7% compared to culture (n = 327). PPA was 100% for Candida auris, Candida kefyr, Candida krusei, Candida tropicalis, Candida parapsilosis, and Candida neoformans; slightly lower for Candida albicans (90.3%) and Candida glabrata (91.1%); and lowest for Candida lusitaniae (55.6%) and Fusarium spp. (0%). To assess clinical impact, we compared patient records from preimplementation (results withheld from providers; n = 31) and postimplementation (results released; n = 68). Compared to the T2Dx Candida panel, BCID-FP showed superior accuracy (PPA 98.0% vs 61.7%). It also reduced time to fungal identification by 1.36 days relative to culture (P < .0001). Postimplementation, there was also a significant increase in infectious disease physician recommendations for antifungal deescalation (39.7% vs 16.1%; P = .0219), typically shifting from micafungin to fluconazole in C. albicans and C. parapsilosis cases. However, no significant differences were observed in time to antifungal optimization, empiric therapy duration, length of stay, or mortality. Although BCID-FP offers clear diagnostic advantages, timely implementation of expert recommendations remains essential to improving outcomes.