Persistent fungemia--risk factors and outcome in 40 episodes.

The aim of this multicenter survey was to assess risk factors and mortality in patients with persistent fungemia (PF). Cases of persistent fungemia, defined as positive blood culture for at least 3 causative days of antifungal therapy were selected. Forty cases of persistent fungemia (lasting more than 3 days) were compared with 270 non-persistent fungemias appearing within the same period, and analyzed by univariate and multivariate analysis for risk factors and outcome. The median number of days of positive culture was 4.4 (3 - 20): 22 episodes were due to Candida albicans, 1 due to non-albicans Candida spp., 6 episodes due to non-Candida spp. Yeasts: 15 were catheter related, 16 patients had yeast-infected surgical wounds, 12 were neutropenic, 4 cases were caused by species resistant in vitro, 2 to amphotericin B (Trichosporon spp.) and 2 to fluconazole (C. laurentii, C. glabrata). Fifteen patients (37.5%) died, 7 of whom due to fungemia. Nineteen cases had one known risk factor (10 had infected wound, 4 infected vascular catheter, 3 were neutropenic and 2 had inappropriate therapy). Fourteen cases had two known risk factors (4 had wound and infected catheter, 4 neutropenia and infected catheter, 2 neutropenia and resistant organism, 4 other combinations. Two cases had 3 known risk factors and one had 4 risk factors for persistent fungemia. Artificial ventilation, C. glabrata etiology, non-Candida spp. yeasts such as Trichosporon spp. and Cryptococcus spp. and prior surgery were significantly associated with persistent fungemia in univariate, whereas only C. glabrata etiology in multivariate analysis. Breakthrough fungemia during empiric therapy with fluconazole was also observed more frequently in patients with persistent fungemia. However, there was no difference in both attributable and overall mortality between both groups.