Although commendable, Winston and colleagues' evaluation of the utility of fluconazole prophylaxis among patients with acute leukemia seems overly enthusiastic [1]. Perfect [2] has suggested six criteria for justifying antifungal prophylaxis: safety, efficacy, cost, consequence, prevalence, and resistance. Based on these criteria, fluconazole does not show clear benefits in this study. No difference was observed in the incidence of invasive fungal infections, the most important measure of efficacy. There was a decrease in the rate of superficial fungal infections among fluconazole recipients (6%; 95% CI, 2% to 17%) when compared with placebo recipients (15%); fluconazole recipients also experienced a delay in the time before initiation of empiric amphotericin. Oropharyngeal and cutaneous candidal infections, however, respond rapidly to treatment. Moreover, because clotrimazole or nystatin were not routinely used, the study may not have been effectively blinded, as the presence of thrush may have influenced the decision to begin empiric amphotericin B. Determining whether the cost of fluconazole prophylaxis was justified by other savings is not possible without data on length of hospitalization. Consequences were not influenced, as shown by the 90-day mortality rate of 21% in fluconazole recipients compared with 18% in placebo recipients.

The inability to show a major benefit may be due to the relatively low prevalence of invasive fungal infections among patients with acute leukemia (8% in this study) compared with bone marrow transplant recipients [3], among whom fluconazole prophylaxis led to a clinically significant decrease in invasive fungal infections. Colonization with resistant fungi and yeasts did not emerge during this study; however, the study involved 256 patients at 18 institutions. Reports of the emergence of C. krusei infections associated with fluconazole use have come from prolonged observation at single institutions [4, 5] and may reflect selection pressure on the hospital flora.

The routine use of prophylactic fluconazole in patients with acute leukemia of average risk is not supported by this study. It is possible that a subgroup of patients with leukemia, such as those with very prolonged neutropenia, might benefit from fluconazole prophylaxis. Physicians who use fluconazole prophylaxis should continue to use early empiric amphotericin for unexplained fevers in patients with neutropenia.