Terminal-stage prognostic analysis in candidemia.

Background: Candidemia has an extremely high mortality rate but is not always the direct cause of death. Therefore, determining the effect of candidemia on death is extremely difficult.

Methods: We investigated prognostic factors in patients with culture-proven candidemia at 2 Japanese university teaching hospitals from April 2009 through May 2013. To examine the effects of comorbid conditions, the Charlson comorbidity index was determined, and patients were subjectively classified into 3 clinical prognostic stages (terminal [death expected within 1 month], semiterminal [death expected within 6 months], and nonterminal [expected to live more than 6 months]). The Cox proportional hazard model was used for univariate and multivariate analyses of factors possibly affecting survival.

Results: On univariate analysis, factors identified as associated with an increased mortality rate were: admission to an internal medicine department, Candida glabrata, immunosuppression, hypotension, hypoxemia, and a terminal prognostic stage. Factors associated with a decreased mortality rate were: serum albumin, endophthalmitis investigation, and nonterminal prognostic stage. The mortality rate was significantly related to the prognostic stage on multivariate analysis (P < 0.0001), was increased by age (P = 0.0014), and was decreased by a delayed start of antifungal therapy (P = 0.0374).

Conclusion: In contrast to earlier studies, the present study has found that later antifungal usage is associated with a decreased mortality rate in cases of candidemia. More important than candidemia in causing the deaths of patients with candidemia were the patients' background and comorbidity status. Therefore, rigorous methods should be used when investigating causes of death in terminally ill patients with candidemia.