The present study aimed to investigate predictors of mortality among candidemic patients requiring source control in three university hospitals, representing two regions with distinct epidemiological profiles, susceptibility patterns, and clinical management practices. Our findings underscore the paramount importance of source control, which emerged as the most influential factor affecting patient outcomes.

Consistent with previous publications, our study underscores the benefits of timely source control interventions, particularly catheter removal, in enhancing patient survival [11,12,13,14,15, 20]. This aligns with guidelines recommending early source control, although the debate surrounding its efficacy continues, driven by varying study outcomes and constraints in conducting randomized controlled trials [7, 9, 11,12,13,14,15, 18, 25,26,27,28,29]. One notable challenge is that catheter removal is not always feasible or safe, especially in cases of severe thrombocytopenia, administration of vasoactive drugs, or continuous renal replacement therapy. For instance, in a randomized trial on candidemia, only 51% of patients underwent early catheter removal, despite it being protocol recommended [30]. Therefore, retrospective observational studies such as ours, influenced by patient-specific factors, maximal care versus palliative approaches, and infection severity, face limitations in drawing definitive conclusions [25, 31, 32]. When considering patients on maximal care versus those with care withdrawal, it becomes evident that the absence of source control due to care withdrawal plays a significant role [7, 19, 25, 31]. To address this, we performed Kaplan–Meier curves among patients who remained alive and under maximal care for 7 days after candidemia onset, reaffirming the significant impact of source control on survival. It is worth noting that the participating hospitals exhibited heterogeneous management strategies, with Greek hospitals (UGHP: 87% and UGHH: 80%) more frequently performing early source control compared to LUH (51%). This disparity may be attributed to higher rates of catheter-related bloodstream infections in Greek ICUs [33]. In addition, the prevalence of C. parapsilosis, commonly associated with catheter-related candidemia, was higher in Greek ICUs than in LUH. Although C. parapsilosis was previously associated with a better outcome, the species of Candida species did not exert any influence on mortality in the present study [34].

In the majority of patients (83%) an antifungal therapy was initiated early (within 72 h from candidemia onset), and was considered appropriate in 79% of them. Interestingly, contrary to previous findings, the initiation of appropriate antifungal therapy was not associated with survival in our study [8, 10,11,12,13]. This may be attributed to the more pronounced impact of source control on outcomes. Previous studies have highlighted the role of empiric antifungal choice in patient outcomes, with echinocandins being associated with reduced mortality, especially in patients with septic shock [8, 11, 35]. However, in our study, 24% of critically ill patients received initial antifungal therapy with non-echinocandin drugs (fluconazole or liposomal amphotericin B), and this did not lead to worse outcomes compared to the group initially receiving echinocandin therapy.

While most studies have demonstrated the favorable impact of either prompt source control or early appropriate antifungal treatment on outcomes, not all have evaluated the significance of early combined management and which component is more crucial [7, 11,12,13, 17, 20, 21]. Previous studies have indicated that a combination of early source control and early appropriate antifungal treatment is associated with improved outcomes. In contrast, considering each intervention separately (i.e., source control or early appropriate antifungal therapy) has not consistently shown significant associations [7, 14]. Bassetti et al. demonstrated that both inadequate source control and inadequate antifungal therapy were individual predictors of worse outcomes [21]. In the present study, patients receiving both early source control and appropriate antifungal treatment exhibited comparable survival to those with only early source control. In addition, early appropriate antifungal treatment was not associated with a better outcome when compared to those without both early source control and appropriate antifungal treatment. These findings underscore the paramount importance of prompt source control in managing critically ill candidemic patients.

In the present study, SARS-CoV-2 infection was associated with increased mortality, aligning with prior reports [4, 6]. We observed a lower 30-day mortality among COVID-19 candidemic patients (36%) compared to previously reported rates (60–88%) [4, 6, 36]. An increase in incidence of candidemia among critically ill COVID-19 patients has been reported in the literature [4,5,6], which was more prominent in UGHP among the participating hospitals. This increased incidence may be attributed to factors such as the higher administration of immunosuppressive treatments (e.g., corticosteroids and tocilizumab) and broad-spectrum antibiotics among COVID-19 patients [36, 37].

As previously demonstrated, infection severity, as indicated by the SOFA score or the development of septic shock, was associated with mortality [10, 11, 14, 16, 17, 27]. Early source control was significantly associated with better outcome in patients with septic shock and those without. This was also shown in two previous studies with ICU candidemic patients with septic shock [7, 21]. Unlike previous research, host-related factors such as advanced age or comorbidities did not influence outcome [7, 11, 13, 15, 17].

The present study has several limitations. First, it is a retrospective study; however, it included a high number of critically ill patients from three university centers each with its distinct incidence rates, epidemiology, and clinical management practices. Second, the use of a 72-h cutoff for defining early source control and antifungal treatment may appear arbitrary. As shown in a study from LUH, approximately 30% of candidemias became positive in blood cultures after 72 h [20]. Moreover, cultures positive before 72 h faced delays in pathogen identification due to working hours, subsequently impacting source control and antifungal treatment initiation. In addition, the two Greek ICUs did not have access to rapid diagnostic tests [38]. Although beta-d-glucan was available at LUH, its usage was infrequent and, as previously demonstrated, was employed to either refrain from or discontinue empirical antifungal therapy [39]. Furthermore, no data on hydroalcoholic consumption, site of intravascular catheter insertion (jugular, subclavian, or femoral), and type of disinfection were available. Lastly, no research was conducted on the virulence or biofilm formation of different Candida spp.

In conclusion, this multicenter study conducted in the ICU of three university centers with varying epidemiological and clinical practices underscores the critical importance of prompt source control, particularly catheter removal in cases of catheter-related candidemia or candidemia of unknown origin. Hence, for patients diagnosed with candidemia, in addition to promptly initiating appropriate antifungal treatment, it is imperative to expeditiously undertake source control procedures, an aspect that is often overlooked in clinical practice.