Our study began with an outbreak investigation of ICU fungal HAI. Nosocomial fungal outbreaks induced by a common exogenous source have been reported in the literature
. These have included contamination of milk bottles
, parenteral nutrition
, and IV medication
. Outbreaks induced by yeast carried on the hands of healthcare workers have also been reported
[27–29]. There was no evidence of cross transmission or common source outbreak in our study. It was more likely a secular trend; however, yeasts were detected in the environment of more heavily infected patients. This finding still points out the importance of traditional infection control measures such as hand washing and environmental hygiene. Unfortunately, the fungal stocks of the environmental yeasts that were found failed to grow, so we cannot compare the species and genotypes of isolates yielded from the patients and their surroundings.
The second part of our study focused on an analysis of risk factors. Five independent risk factors were identified for all sites of ICU fungal HAI by the multivariable analysis, which included TPN use, sepsis, surgical patients, mechanical ventilation and an indwelling urinary catheter. These results are similar to those of previous studies,
[30–34] though the case definitions of our study were different from others. Most of the previous studies defined cases by fungal growth from sterile or non-sterile body sites, or histology finding, or the EORTC/MSG criteria
. However, active surveillance on HAIs is a common practice in the hospitals worldwide, especially in the ICUs. The clinical implications from our study which defined cases with standardized HAI definitions can be closer to the daily practice in the ICUs.
Systemic review was performed by the Fungal Infection Risk Evaluation (FIRE) Study group to identify and summarize the important risk factors from published studies with multivariable analyses, risk prediction models and clinical decision rules for invasive fungal diseases (IFDs) in critically ill, adult patients. Meta-analysis was not performed due to the heterogeneity of these studies
. In this systemic review, the following risk factors were found in multiple studies to be significantly associated with IFD: surgery, TPN, fungal colonization, renal replacement therapy, infection/sepsis, mechanical ventilation, diabetes, APACHE II or III scores
. Five (surgery, TPN, mechanical ventilation, infection/sepsis, APACHE II score) of the 8 “important” risk factors reported by the FIRE Study group were also independent risk factors identified in various infection types of our study. The significant risk factors in our study were TPN use, mechanical ventilation and an indwelling urinary catheter in fungal UTI; TPN use, sepsis, and higher APACHE II score in fungal BSI; TPN use, surgical patients in fungal pneumonia; surgical patients and TPN use in fungal SSI. Total parenteral nutrition was a significant risk factor for all types of ICU fungal HAIs; the odds ratios were 3.51 in UTI, 8.47 in BSI, 8.07 in pneumonia, 8.82 in SSI, 4.83 in all sites infections.
The independent risk factor, patients with an indwelling urinary catheter, identified in UTI of our study had been reported in other studies; though they are not mentioned in the FIRE Study. A urinary catheter was included in Leon’s report from the EPCAN study group, but it was not a significant risk factor for proven fungal infection in comparison with the fungal colonization group
. They performed weekly surveillance cultures of urine, tracheal, and gastric samples, and compared risk factors between patients with unifocal or multifocal colonization and proven candidal infection. This may be explained by the fact that Candida species are frequently isolated from urine specimens from urinary catheters, whether symptomatic or not; especially when the patient is medicated with antibacterial agents
. The utilization rate of urinary catheters was high in both groups (colonized 98.5% vs. proven infection 95.9%) of Leon’s study, with broad spectrum antibiotics administered to 98% of the colonized vs. 100% of the proven infection group. In this case, a urinary catheter should be considered a common risk factor for both groups instead of an independent risk factor for proven fungal infection. In our study, the 2 groups for comparison were patients with ICU fungal HAI vs no HAI. Asymptomatic bacteriuria was still included as a specific infection type of healthcare associated UTIs during the study period according to the case definitions.
Diagnoses of sepsis, admission to the surgical ICU service and mechanical ventilation were significant risk factors for both ICU fungal HAIs and ICU mortality; these factors, especially sepsis and surgical patients, cannot be avoided in daily ICU practice. Early treatment had been documented to improve the outcome of candidemia,
[35, 36] although early diagnostic tools are still inadequate. Empirical or preemptive antifungal therapy has generally been accepted as a strategy under such circumstances;
 therefore, when signs of ICU HAI appear; and more risk factors for fungal HAI exist, then stronger empirical antifungal therapy is indicated. Fungal ICU HAI was shown to be an independent predictor for ICU mortality in our study. Avoiding risk factors for ICU fungal HAIs, especially TPN use, may therefore improve ICU morbidity and mortality.
The mortality rate of patients with ICU fungal HAIs was noted to be over 2 times that of patients without ICU HAIs in our study. This was consistent with other reports in the literatures
[30, 38, 39]. Candidemia has resulted in mortality rates as high as 50-70%
[17, 39–41]. Candiduria was documented to precede candidemia in the ICU, and itself had a mortality rate of 31.3%
. Fungal wound infections have been reported to be independently associated with mortality in burn patients
. These studies support our finding that patients with ICU fungal HAIs have a higher mortality rates than patients without HAIs regardless of the affected site. The rates and species distribution of fungal HAIs have differed geographically and institutionally,
[9, 42] and strains of Candida species resistant to antifungals have been increasing
. It is important for each healthcare unit to have its own epidemiologic knowledge about fungal HAIs in order to monitor the trend and formulate a strategy to control it.
Some limitations of our study should be noted. Firstly, our study was limited by the pre-existing database. Some factors listed in other studies such as corticosteroid use and the presence of chronic diseases were lacking. We did not specify chronic diseases among the variables because they had been included in the calculation of the APACHE II score. Corticosteroid use and chronic diseases as variables were not found to be significant risk factors in other studies
[30, 31]. Overall, most of the important risk factors were included in our study
. Secondly, the effort of environmental cultures from ICU patients was incomplete; because, surveillance cultures were not performed in patients with no HAI to see whether there were any differences. Finally, the information about the prescription of antibiotics was inadequate. The policy regarding the use of antibiotics may be correlated with fungal HAIs, and should be reviewed more comprehensively. However, no major new classes of antibiotics were introduced into our hospital during the study period. The prescription of antibiotics is complicated in the ICU; this could be the subject of a future study.