Clinical Characteristics, Microbiological Sources, and Outcomes of Candida-Positive ICU Cultures in Critically Ill...

Background:Candida isolation is common in critically ill patients, but its clinical interpretation depends strongly on microbiological source, host factors, and clinical context. Bloodstream isolation, candiduria, respiratory tract isolation, surveillance cultures, catheter-tip cultures, and wound/skin cultures have different clinical implications. We aimed to evaluate clinical characteristics, microbiological sources, species distribution, antifungal treatment patterns, and outcomes among adult ICU patients with Candida-positive ICU cultures. Methods: This single-center retrospective observational cohort study was conducted in the medical intensive care unit of Marmara University Faculty of Medicine between 1 October 2022 and 5 September 2025. Adult ICU patients with at least one Candida-positive ICU culture were included. Non-Candida fungal isolates and duplicate patient-level records were excluded. The primary outcome was all-cause 28-day mortality. ICU mortality was defined as all-cause death during ICU stay. Source-stratified analyses and expanded multivariable logistic regression models were performed to evaluate factors associated with mortality. Results: A total of 349 adult ICU patients were included. Median age was 71 years [IQR, 62-82], and 185 patients were male (53.0%). Overall, 28-day mortality was 59.0% (206/349), and ICU mortality was 65.9% (230/349). Candida colonization was identified in 247 patients (70.8%), whereas Candida infection was identified in 102 patients (29.2%). The most common species were Candida albicans (48.4%), Candida glabrata (13.8%), and Candida auris (12.9%). The most frequent microbiological sources were urine (42.4%), lower respiratory tract samples (26.4%), and blood cultures (14.9%). Blood/sterile-site isolation was associated with higher ICU mortality than non-blood/non-sterile-site isolation (79.2% vs. 63.5%, p = 0.026), whereas the difference in 28-day mortality was not statistically significant (66.0% vs. 57.8%, p = 0.260). Antifungal treatment was more frequent among patients with blood/sterile-site isolation (94.3% vs. 16.9%, p < 0.001). In the expanded 28-day mortality model, lactate, NLR, and carbapenem exposure were independently associated with mortality. In the expanded ICU mortality model, lactate and CRRT/hemodialysis were independently associated with mortality. Candida score was not independently associated with either 28-day mortality or ICU mortality after broader adjustment. Conclusions:Candida-positive ICU cultures represent a heterogeneous clinical and microbiological spectrum. Source-specific interpretation is essential, particularly when distinguishing bloodstream or sterile-site isolation from non-sterile-site colonization. Candida score may reflect a higher-risk clinical phenotype, but it should not be interpreted as a stand-alone mortality prediction tool.