Incidence and outcome of invasive candidiasis in intensive care units (ICUs) in Europe: results of the EUCANDICU project

1Invasive candidiasis (IC) can develop in adult patients admitted in intensive care units (ICUs), with a significant impact on morbidity, mortality, and healthcare costs. The most frequent clinical forms of IC in critically ill patients are candidemia and intra-abdominal candidiasis (IAC), which affect up to 5% of all ICU admissions.

2In the last 10 years, other series addressed specifically the cumulative incidence of IC in the ICU setting. The majority of those studies was limited to candidemia, some were limited to one or a few countries, and some occasionally included cases not representing true infections but colonization.

3The aim of the present multinational study was to expand our knowledge of the cumulative incidence of ICU-acquired IC in Europe.

4The present multicenter, retrospective study was conducted from January 2015 to December 2016 in 23 ICUs in 22 large tertiary care European hospitals (9 in Italy, 4 in France, 2 in Greece, 1 in Belgium, 1 in Czech Republic, 1 in Germany, 1 in Ireland, 1 in Portugal, 1 in Spain, 1 in The Netherlands, and 1 in the UK). All patients who developed an episode of candidemia or a microbiologically documented IAC during their stay in the ICU (at least 48 h after admission) were included in the study.

5The primary objective of the study was to assess the cumulative incidence of ICU-acquired IC in European ICUs. Secondary objectives were (i) to assess the independent impact of center-level factors on the cumulative incidence of ICU-acquired IC and (ii) to assess factors associated with crude 30-day mortality in patients with ICU-acquired IC.

6The entire EUCANDICU project consists of two phases: (i) a first study describing the cumulative incidence of ICU-acquired IC in Europe, reported in the present paper; (ii) a second, case-control study to assess patient-level predictors of ICU-acquired IC, which will be conducted and published subsequently.

7ICU-acquired IC was defined as candidemia or IAC with signs and symptoms of infection developing at least 48 h after ICU admission. Candidemia and IAC were defined according to previously published definitions. More in detail, candidemia was defined as the presence of at least one positive blood culture for Candida spp. in patients with signs and symptoms of infection. IAC was defined as the presence of at least one of the following: (i) Candida detection by direct microscopy or growth in culture from necrotic or purulent intra-abdominal specimens obtained by percutaneous aspiration or during surgery; (ii) growth of Candida from the bile or intra-biliary duct devices, plus biopsy of intra-abdominal organs; (iii) growth of Candida from blood cultures in the presence of secondary or tertiary peritonitis in the presence of no other pathogens; (iv) growth of Candida from drainage tubes inserted less than 24 h before culture sampling. In case of multiple episodes of IC in the same patient, a subsequent event was considered as independent if developing at least 30 days after the last positive culture related to the previous episode. Crude 30-day mortality was defined as death within 30 days from the onset of signs and symptoms of ICU-acquired IC. All patients suitable for inclusion were identified starting from the laboratory databases of the participating hospitals, and subsequent review of clinical records.

8The cumulative incidence of ICU-acquired IC was measured as the number of episodes per 1000 ICU admissions. The impact of center-level factors on the cumulative incidence of ICU-acquired IC was assessed by means of a multivariable generalized Poisson mixed model with center as a random intercept, after having verified the absence of overdispersion in count data. Subgroup analyses were also conducted according to the type of ICU-acquired IC (candidemia and IAC).

9Demographic and clinical characteristics of patients are presented with number and percentage for categorical variables and median and interquartile range (IQR) for continuous variables. Their possible association with crude 30-day mortality was firstly tested through univariable logistic regression. Then, factors potentially associated with the outcome in univariable comparisons (p < 0.10) were included in an initial multivariable regression model and further selected for the final multivariable model (model A) by means of a stepwise backward procedure based on the Akaike information criterion. Only the first IC episode per patient was considered for this analysis. Variables included in model A were also included in an additional multivariable mixed logistic regression model (model B), with center as a random intercept.

10All the analyses were conducted using R Statistical Software 3.5.2 (R Foundation for Statistical Computing, Vienna, Austria). Mixed models were built using the glmer function in the lme4 package.

11During the study period, the 23 ICUs (median number of beds 18, interquartile range 14–43) had 80,645 admissions and 570 episodes of ICU-acquired IC, corresponding to an incidence of 7.07 episodes per 1000 ICU admissions (6.67 and 7.47 in 2015 and 2016, respectively). Separated, non-mutually exclusive incidences of candidemia and IAC were 5.52 and 1.84 episodes per 1000 ICU admissions, respectively. As shown in Table 1, in subgroup comparisons, admission to a surgical ICU was associated with lower incidence when compared with a medical ICU (cumulative incidence ratio 0.10, 95% CI 0.01–0.76 for surgical vs. medical, p 0.022). The observed random effects variance testified to the presence of between-center variability in the cumulative incidence of IC that was unexplained by the explored, fixed center-level predictors (Table 1 legend). A graphic representation of the cumulative incidence of IC is also available in Fig. 1.

12An additional, secondary analysis of predictors of mortality was conducted in the subgroup of patients for whom patient-level clinical data was registered (mostly cases of IC meeting the following inclusion criterion for the future case-control study: availability of controls with the same length of ICU stay without IC). In this regard, clinical data were available for 330/570 cases of IC (58%). The median age of patients was 66 (IQR 55–75), and 60% were males (198/330). Candidemia, IAC, and IAC plus candidemia accounted for 65% (215/330), 29% (97/330), and 5% (18/330) of episodes, respectively. C. albicans was the most frequently isolated species (189/330, 57%), followed by C. glabrata (68/330, 21%) and C. parapsilosis (42/330, 13%). Crude 30-day mortality was 42% (137/330). Table 2 shows univariable and multivariable analyses of predictors of crude 30-day mortality. In the final multivariable model (model A), age (odds ratio [OR] 1.04 per year, 95% CI 1.02–1.06, p < 0.001), severe hepatic failure (OR 3.25, 95% 1.31–8.08, p 0.011), SOFA score at the onset of IC (OR 1.11 per point, 95% CI 1.04–1.17, p 0.001), and septic shock (OR 2.12, 95% CI 1.24–3.36, p 0.006) were associated with increased 30-day mortality. As reported in the legend of Table 2, the additional multivariable model with center as random intercept (model B) confirmed the results obtained in model A.

13This is the largest multinational study assessing the cumulative incidence of ICU-acquired IC in Europe to date. Based on our report, the incidence of IC in European ICUs was 7.07 episodes per 1000 ICU admissions, with crude 30-day mortality of 42%.

14Our results are similar to those of the large point-prevalence EPIC II study, which previously reported a prevalence of 6.87 episodes of candidemia per 1000 ICU patients, although we recorded a slightly lower cumulative incidence of candidemia (5.52 episodes per 1000 ICU admissions). In addition, our analysis indicated two center-level factors that may influence local incidences: (i) the type of ICU, with medical ICUs being associated with increased incidence vs. surgical ICUs in subgroup comparisons, and (ii) the presence of random between-center variability, although likely relying in part on unexplored but exhaustive center-level factors. It is also worth noting that some unexplored center-level factors (e.g., cumulative days at risk, antifungal prophylaxis policies), together with the low number (and thus low generalizability) of ICU wards included in the subgroup comparison surgical vs. medical (3 vs. 5, respectively), might have influenced the unexpected but significant observation of a lower cumulative incidence of IC in surgical ICUs, which calls for future confirmatory studies specifically addressing this aspect.

15In a secondary analysis, we confirmed the importance of baseline factors (age, severe hepatic failure) and severity of disease/clinical presentation (SOFA score, septic shock) in unfavorably influencing the prognosis of ICU-acquired IC. However, we could not explore the impact on mortality of both source control and antifungal therapy performed/administered beyond 24 h after the onset of IC. Dedicated studies and models aimed at primarily evaluating these variables are necessary to comprehensively assess their effect.

16An important limitation of our study is the limited number of assessed, potential center-level predictors of cumulative incidence. Further studies with a higher number of participating ICUs are necessary for allowing inclusion of more variables in the model. Other main limitations of our study are the lack of a control group (patients without IC) for assessing patient-level predictors of IC and the inherent possibility of having considered some contamination as IAC. As regards the former, it should be noted that the main goal in which we were interested in this first phase was to provide a center-level perspective on the cumulative incidence of ICU-acquired IC, by describing its burden and assessing the possible effect of fixed and random center-level predictors. Although it will ultimately add to the discussion, the assessment of patient-level predictors is an independent and different analysis, which will be the core of the future case-control study. As regards the possibility of including contaminations, it is worth noting that we used a consensus definition for minimizing the number of contaminations misidentified as IAC. Another limitation is that most data came from a few countries (e.g., 9/23 participating ICUs are in Italy), thus preventing us from reliably assessing the existence of any possible between-country variability impacting on the cumulative incidence of ICU-acquired IC. Finally, it should be acknowledged that the impact on mortality of some covariates (e.g., COPD) may depend on their degree of severity, which was not registered.

17Global incidence of IC in 23 European ICUs was 7.07 episodes per 1000 ICU admissions. Future in-depth analyses will allow explaining part of the observed between-center variability, with the ultimate aim of helping to improve local infection control and antifungal stewardship interventions.