Bircan Kayaaslan, Halime Araz, Hatice Yağmur Zengin, Rukiyye Bulut, Zeynep Oktay, Meltem Tasbakan, Sumeyra Istek, Elif Doyuk Kartal, Ozlem Bayrak, Ayse Kaya, Fatma Eser, Pınar Yuruk Atasoy, Hatice Esranur Kıratlı, Berfin Cirkin Doruk, Mustafa Uguz, Sema Sarı, Bahar Busra Sivrikaya, Yasemin Akkoyunlu, Lutfiye Nilsun Altunal, Ayse Albayrak, Handan Alay, Gulsen Iskender, Elkholy Amani, Taylan Onder, Dilek Ozcengiz, Mediha Turktan, Mihalj Mirko, Ilad Alavi Darazam, Mohammad Mehdi Shadravan, Reza Yahyavi Sahzabi, Ahmet Cem Yardımcı, Ruveyda Korkmazer, Gunes Senol, Tuba Kuruoglu, Fatih Temocin, Ozlem Alıcı, Meral Sonmezoglu, Aliye Esmaoglu, Zeynep Ture, Andrea Cortegiani, Ozay Akyıldız, Hatice Rahmet Guner, Hakan Erdem
Journal: Mycoses 2026;69(8):e70213
PMID: 42626934
BACKGROUND
Candidaemia remains a common, life-threatening infection among intensive care unit (ICU) patients, with high mortality, particularly in patients with delayed diagnosis and treatment.
OBJECTIVES
This international, multicentre, prospective, observational study aimed to identify risk factors for candidaemia in ICU patients and to develop an easy-to-use predictive tool, the CanDi-Score, for early initiation of empirical antifungal treatment.
PATIENTS/METHODS
All adults over 18 years hospitalised for more than 48 h in the ICUs were included in the study and followed for 30 days. Data on demographics, comorbidities, clinical severity scores including APACHE II and SOFA scores, Charlson comorbidity index (CCI), established risk factors for candidaemia and their time-dependent effects such as total parenteral nutrition (TPN), mechanical ventilation (MV), and central venous catheter (CVC) were collected. Univariate and multivariate logistic regression analyses were used to identify factors associated with candidaemia development, and fast backwards variable selection was used. To account for between-centre heterogeneity, sensitivity analyses were performed using mixed-effects logistic regression models including centre-specific random intercepts. A nomogram model was constructed using independent risk factors identified in multivariable analysis, and its predictive performance was evaluated with Area Under the Curve (AUC), sensitivity, and specificity.
RESULTS
A total of 2,704 ICU patients, including 204 with candidaemia, were enrolled. Patients with candidaemia had higher APACHE II (19 vs. 13) and CCI scores (6 vs. 5; both p < 0.001), and more frequently had chronic renal failure, haemodialysis, malignancy, recent gastrointestinal surgery, and neutropenia (all p ≤ 0.05). Patients were randomly split into development (80%) and validation (20%) cohorts. Multivariate analysis identified higher CCI, concurrent infection, neutropenia, ICU stay > 14 days, MV > 3 days, CVC duration and TPN > 14 days as independent risk factors. The risk associated with CVC increased substantially with prolonged use (OR: 0-7 days: 3.69; 8-21 days: 8.58; > 21 days: 17.21). The model demonstrated good performance in both the development and validation cohorts, with sensitivity, specificity, and AUC values of 79.6%, 76.5%, and 0.86 (95% CI: 0.83-0.88) in the development cohort, respectively, and comparable predictive performance in the validation cohort. The score model was demonstrated by a nomogram.
CONCLUSIONS
Given the high mortality of candidaemia in ICU patients, the CanDi-Score may serve as a valuable tool for early prediction and timely empirical antifungal treatment. Key predictors include CCI, presence of concurrent infections, the duration of CVC use, and TPN administration. CanDi-Score provides clinicians with a practical approach to identify high-risk patients and guides decision-making for the early initiation of antifungal treatment.
© 2026 The Author(s). Mycoses published by Wiley‐VCH GmbH.
© Copyright 2026, Nutrition Evidence
NED wishes to thank the following organisations for their support:
We use cookies to improve your experience and analyze site traffic with Google Analytics. By continuing to use our site, you agree to our use of cookies. Learn more.