Nosocomial candidemia: an ounce of prevention is better than a pound of cure.

Abstract

This issue of Infection Control and Hospital Epidemiology contains an article by Puzniak et al.1 entitled “Has the Epidemiology of Nosocomial Candidemia Changed?” There is no short or simple answer to this question. A review of progress since the early 1980s (prior to the dawn of the “fluconazole era”) reveals that several advances have been made in the diagnosis and management of candidemia. Modern, automated blood culture methods have improved our ability to detect candidemia, a thoughtful and rigorous effort has led to the development of standardized methods for susceptibility testing of Candida (and to the demonstration of clinical correlation between in vitro test results and outcome),2-4 and several new antifungal agents provide equivalent therapeutic results with less toxicity than amphotericin B deoxycholate.5 In addition, the introduction of fluconazole prophylaxis in selected high-risk patient populations (eg, patients receiving a bone marrow transplant and high-risk patients receiving a liver transplant) has resulted in decreased infection rates in these groups6,7 and may be associated with an overall decline in mortality due to invasive candidiasis.8 But in keeping with the maxim that “no good deed goes unpunished,” widespread azole use may have facilitated a less encouraging change in the epidemiology of nosocomial candidemia: the emergence of C. glabrata as a more frequent nosocomial pathogen.9 Although the decreased susceptibility of C. glabrata to azoles and amphotericin B makes this development disturbing, the other species commonly causing nosocomial candidemia (C. albicans, C. tropicalis, and C. parapsilosis) remain susceptible to fluconazole,10 and the feared emergence of C. krusei as a more common cause of nosocomial candidemia has not occurred. What has not changed is that hospital acquisition of Candida bloodstream infection (BSI) remains a devastating complication of healthcare delivery. The study by Puzniak et al. is the latest in a series of recent articles ascribing an extremely high crude (35% to 61%) and attributable (24% to 49%) mortality to nosocomial candidemia.1,11-14 Acquiring candidemia in the hospital carries no less risk of death during hospitalization today than it did in the 1980s and early 1990s.15-20 Could it be that this unchanging, high crude mortality means that nosocomial candidemia is merely a marker for severe underlying illness, but doesn’t itself contribute significantly to mortality? Prospective clinical trials generally estimate attributable mortality to be much lower than do retrospective cohort designs,21,22 and we have previously outlined reasons why a retrospective cohort design might overestimate attributable mortality.14 However, we believe that estimates based on the presence of Candida in sterile sites within 48 hours of death or at autopsy21,22 grossly underestimate attributable mortality by not including deaths among patients who, although they may clear their infection, die of downstream effects of the physiologic insult sustained during infection. There is evidence supporting a substantial contribution of nosocomial candidemia to mortality, including (1) the high attributable mortality estimates from several retrospective cohort studies,13,14,20 (2) the independent association of Candida BSI with mortality in large studies using multivariate models to examine microbiologic risk factors for mortality among patients with BSI,23,24 and (3) the mortality benefit documented in association with reduced rates of candidiasis in patients receiving a bone marrow transplant with the use of fluconazole prophylaxis.6 So will advances in treatment lead to reductions in Candida-associated mortality? As Puzniak et al. point out,