Empirical Micafungin Treatment and Survival Without Invasive Fungal Infection in Adults With ICU-Acquired Sepsis, Candida Colonization, and Multiple Organ Failure: The EMPIRICUS Randomized Clinical Trial

Key Points

Question  Does empirical antifungal therapy increase invasive fungal infection–free survival at day 28 in nonneutropenic critically ill patients with sepsis, multiple Candida colonization, and multiple organ failure exposed to broad-spectrum antibacterials?

Findings  In this randomized clinical trial of 260 adults, there was no significant difference in the rate of survivors without any fungal infection at day 28 between micafungin-treated (87/128 [68%]) and placebo-treated (74/123 [60.2%]) groups.

Meaning  The use of micafungin as a routine empirical treatment in critically ill patients with suspected fungal infection did not improve fungal infection–free survival at 28 days.

Importance  Although frequently used in treating intensive care unit (ICU) patients with sepsis, empirical antifungal therapy, initiated for suspected fungal infection, has not been shown to improve outcome.

Objective  To determine whether empirical micafungin reduces invasive fungal infection (IFI)–free survival at day 28.

Design, Setting, and Participants  Multicenter double-blind placebo-controlled study of 260 nonneutropenic, nontransplanted, critically ill patients with ICU-acquired sepsis, multiple Candida colonization, multiple organ failure, exposed to broad-spectrum antibacterial agents, and enrolled between July 2012 and February 2015 in 19 French ICUs.

Interventions  Empirical treatment with micafungin (100 mg, once daily, for 14 days) (n = 131) vs placebo (n = 129).

Main Outcomes and Measures  The primary end point was survival without proven IFI 28 days after randomization. Key secondary end points included new proven fungal infections, survival at day 28 and day 90, organ failure, serum (1-3)-β-D-glucan level evolution, and incidence of ventilator-associated bacterial pneumonia.

Results  Among 260 patients (mean age 63 years; 91 [35%] women), 251 (128, micafungin group; 123, placebo group) were included in the modified intent-to-treat analysis. Median values were 8 for Sequential Organ Failure Assessment (SOFA) score, 3 for number of Candida-colonized sites, and 99 pg/mL for level of (1-3)-β-D-glucan. On day 28, there were 82 (68%) patients in the micafungin group vs 79 (60.2%) in the placebo group who were alive and IFI free (hazard ratio [HR], 1.35 [95% CI, 0.87-2.08]). Results were similar among patients with a (1-3)-β-D-glucan level of greater than 80 pg/mL (n = 175; HR, 1.41 [95% CI, 0.85-2.33]). Day-28 IFI–free survival in patients with a high SOFA score (>8) was not significantly different when compared between the micafungin vs placebo groups (HR, 1.69 [95% CI, 0.96-2.94]). Use of empirical micafungin decreased the rate of new invasive fungal infection in 4 of 128 patients (3%) in the micafungin group vs placebo (15/123 patients [12%]) (P = .008).

Conclusions and Relevance  Among nonneutropenic critically ill patients with ICU-acquired sepsis, Candida species colonization at multiple sites, and multiple organ failure, empirical treatment with micafungin, compared with placebo, did not increase fungal infection–free survival at day 28.

Trial Registration  clinicaltrials.gov Idenitfier: NCT01773876

Quiz Ref IDDespite the development of effective and safer drugs, invasive candidiasis and candidemia remain associated with high and increasing mortality,¹ particularly when complicated by septic shock.² The optimal management of Candida species infections includes early awareness of patients at risk, control of the infection source, and timely administration of appropriate antifungal agents.^2-5 Consequently, antifungal agents have been widely used as empirical therapy, ie, for treating suspected fungal infection in patients at risk for invasive candidiasis or patients with unresolved sepsis.^6-9

Two multicenter randomized clinical trials evaluated empirical antifungal therapy for fungal infection suspicion in patients with a central catheter and persistent fever despite treatment with broad-spectrum antibacterial agents. One study demonstrated that empirical fluconazole did not improve clinical outcomes vs placebo in patients at high risk for invasive candidiasis.⁵ Another trial evaluated antifungal prophylaxis using caspofungin among intensive care unit (ICU) patients with at least 2 risk factors for candidemia.¹⁰ Caspofungin failed to significantly improve the primary end point, as proven or probable invasive candidiasis occurred in 16.7% of the placebo recipients vs 9.8% of the caspofungin recipients. There was no difference in mortality across groups. Subsequently, empirical antifungal therapy was incorporated into guidelines for nonimmunocompromised critically ill patients with unresolved ICU-acquired sepsis.¹¹ Despite lack of evidence-based data, as much as 8% of ICU patients without documented Candida infection receive antifungal agents.^12,13Quiz Ref ID The number of organ system failures and incidences of Candida colonization at multiple sites and high serum (1-3)-β-D-glucan levels have been well established as risk factors for candidemia.^14,15 To our knowledge, no randomized clinical trial of colonization-driven empirical therapy has been performed in critically ill patients at risk for invasive candidiasis.

The multicenter, double-blind, placebo-controlled EMPIRICUS (Empirical Antifungal Treatment in ICUS) trial was designed to evaluate whether micafungin, as compared with placebo, increases 28-day invasive fungal infection–free survival among patients with ICU-acquired sepsis, Candida colonization at multiple sites, and multiple organ failure.

The study design has been published elsewhere¹⁶ and the trial protocol is reported in Supplement 1.

EMPIRICUS, a multicenter, randomized, double-blind, and parallel-group study, compared the benefit from a 14-day empirical treatment with micafungin (100 mg administered intraveneously, 1×/d) vs placebo associated with day-28 survival without invasive fungal infection among adult patients with suspected invasive candidiasis.

Empirical treatment was defined as an antifungal treatment for suspected nondocumented invasive fungal infection in patients with unresolved sepsis despite broad-spectrum antibacterial therapy for at least 4 days and multiple sites colonized with Candida species.

The study involved 19 ICUs in France and was approved by an authorized ethics committee (Comité de Protection des Personnes CPP Sud Est V; December 7, 2011; see the trial protocol in Supplement 1) and the French Health Authorities (AFSSAPS; December 2, 2011).

Written informed consent was obtained from all participants or their proxies (in cases of impaired decision-making capacity) at the time of enrollment.

Quiz Ref IDCritically ill adult patients were eligible for the study if they met the following criteria: (1) mechanically ventilated at least 5 days; (2) with at least 1 colonization site (other than rectal swab or stool) positive for Candida species using traditional culture methods; (3) at least 1 additional organ dysfunction; (4) previous treatment for more than 4 days using broad-spectrum antibacterial agents within the last 7 days; (5) 1 arterial or central vein catheter, and (6) 1 new finding of ICU-acquired sepsis of unknown origin (eBox in Supplement 2).

Main exclusion criteria were as follows: (1) neutrophil count of less than 500/mm³; (2) previous bone marrow or solid organ transplantation; (3) ongoing systemic immunosuppressant agent therapy other than corticosteroids at doses lower than 2 mg/kg/d of prednisolone or equivalent; and (4) antifungal treatment with an echinocandin agent for more than 1 day or with any other antifungal agent for more than 72 hours during the week prior to inclusion¹⁶ (see trial protocol in Supplement 1 and the statistical analysis plan in Supplement 3).

Permuted-block randomization with varying block sizes between groups used a web-based system programmed by an independent statistician. Immediately after randomization and for 14 days, the research pharmacists prepared reconstituted opaque bags of micafungin or placebo according to the randomization list and provided it to the site for infusion. ¹⁶

A set for blood culture inoculated with 10 mL of blood in aerobic, anaerobic, and selective milieu were drawn at inclusion before administration of the study drug. During the opening visit of each center, investigators were instructed to perform blood cultures, puncture or evacuation of possible infected sites, funduscopy, and echocardiography to confirm the fungal nature of any subsequent episodes of sepsis during the follow-up.

If the invasive candidiasis at inclusion was evidenced after randomization by the analysis of baseline samples (ie, results not available at randomization), or if the investigator started another antifungal treatment, the study treatment was withdrawn and the antifungal treatment usually prescribed at the investigation site was administered to the patient. However, blinding was not compromised, and the patient remained in the modified intent-to-treat (ITT) analysis. The end point was judged as the occurrence of a new invasive fungal infection or death within 28 days of inclusion.

Data collection and study management are detailed elsewhere (Supplement 1).¹⁶ Database lock and adjudication of all suspected or proven invasive candidiasis were performed before unblinding of the study. The independent adjudication committee reviewed records of all patients with new antifungal treatment and with positive culture from blood, operative room, or direct percutaneous puncture of sterile sites. Additionally, the committee reviewed records of patients with suspicion of documented infections and interviewed investigators by phone when questions were not solved by e-mail. Final judgment was unanimous in all cases.

The primary end point was 28-day survival free of proven invasive fungal infection, as defined according to adapted version of Tissot et al of the EORTC/MSG (European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group) definitions from 2008.¹⁷

Prespecified secondary end points included new proven invasive fungal infections during the follow-up, survival at day 28 and at day 90 (3 months after randomization), antifungal-free survival at day 28, incidence of ventilator-associated bacterial pneumonia, and evolution throughout the 28-day study period of the Sequential Organ Failure Assessment (SOFA) score (range, 0-24 with higher scores indicating worse outcome) and of the serum level of (1-3)-β-D-glucan (a fungal cell antigen identified in blood of patients with fungal infection). The primary end point was assessed also in prespecified patient subgroups at increased risk of fungal infection (medical vs surgical, low vs high SOFA score, low vs high (1-3)-β-D-glucan level, low vs high colonization index, Candida score <3 vs ≥3) and the pharmacokinetic and safety profiles of micafungin. The pharmacokinetics of micafungin were assessed after the first intravenous administration through the evaluation of the plasmatic peak (Cmax) and plasmatic trough (Cmin), which enabled calculation of parameters such as the area under the curve (AUC) of the plasmatic concentrations. Other additional outcomes not reported in the text were hospital survival, mechanical ventilation–free days, and colonization index during follow-up. Molecular biomarkers and molecular markers of resistance of recovered strains from blood cultures will need further analyses.

As previously published,¹⁶ it was estimated that (1) the mortality of patients fulfilling the selection criteria would be between 30% and 37%; (2) the candidemia-related mortality in case of early treatment would be 12% instead of 35% when the treatment is delayed (current practice); (3) according to Schuster et al,⁵ invasive fungal infection would be diagnosed in 7.1% of patients receiving antifungal therapy and 20.8% of those receiving placebo (absolute difference 13.7%); and (4) the sensitivity of conventional diagnostic tests (blood cultures, culture of sterile site) of invasive fungal infection diagnosis would be 60%.¹⁸ Therefore, in the micafungin group, the actual incidence of invasive fungal infections would be estimated at 11.8% (7.1%/0.6), the rate of candidemia-related mortality at 1.4% (11.8% × 12%), and the rate of overall events between 31.4% and 38.4%. In the placebo group, the rate of candidemia-related mortality would be estimated at 4.13% (11.8% × 35%), the number of additional invasive fungal infections diagnosed after randomization at 13.7%,⁵ and the rate of overall events between 49.4% and 56.4%. A difference of 18%, considering the lower and upper estimations of overall event rates in both groups, was therefore hypothesized.

A 2-sided log-rank test with an overall sample size of 235 patients (118 in the micafungin group and 117 in the placebo group) would achieve an ability to detect a difference of 18% in the primary end point with an 80% power at a 0.05 significance level. The hypothesis used was then to increase the proportion of patients surviving free of proven invasive fungal infection from 37% in the placebo group to 55% in the micafungin group. To account for secondary dropouts, 260 patients (130 in each group) were needed.

Analyses were calculated using SAS 9.4 (SAS, Inc) and R (R Foundation for Statistical Computing) software. Comparisons were performed with a modified ITT population. A 2-sided P value of .05 or less was considered statistically significant.

All patients who received at least 1 dose of study treatment were included in the modified ITT analysis.

Missing, unused, or outlying data were checked with investigators via queries. For instances in which missing values were confirmed, data concerning the independent variables were replaced using multiple-imputation methods. Data were reported as numbers (percentages) or medians (interquartile ranges [IQRs]). Continuous variables were compared using the Wilcoxon rank-sum test, and the Fisher exact test was used for proportions. Death or proven invasive fungal infection (primary end point) were evaluated at day 28 and analyzed using survival methods and the Kaplan-Meier estimate (stratified by center). A Cox model was used for adjustment of parameters imbalanced between groups. Proportionality assumption was tested using cumulative sums of martingale-based residuals. Analyses were 2-tailed and stratified by center. A generalized estimating equation, stratified by centers, was used to estimate the effect of the study drug on (1-3)-β-D-glucan. The statistical analysis plan was previously published¹⁶ (Supplement 3). Pharmacokinetic assessment used a population approach to obtain individual Bayesian estimates of micafungin clearance (used to calculate the AUC of micafungin for each patient pie, AUC = dose/clearance]).^19,20

From July 20, 2012, to February 7, 2015, a total of 260 patients in 19 ICUs were randomized. After database lock (September 30, 2015), 251 of them were included in the modified ITT analysis (Figure 1). Patients’ characteristics were well-balanced between groups, except for diabetes and body mass index (Table 1). The study patients were severely ill, as reflected by their overall Simplified Acute Physiology Score (SAPS II) (range, 0-124 with higher scores indicating worse outcome), with a median score of 48 (IQR, 39-57) and an overall median SOFA score of 8 (IQR, 6-11) at randomization. Vasopressors were administered to more than 50% of the patients, and renal replacement therapy to 1 of 3. All patients had multiple risk factors for invasive fungal infection. The median number of sites colonized at inclusion were 3 (IQR, 2-4 [range, 1-7]; eTable 1 in Supplement 2). The (1-3)-β-D-glucan level was greater than 80 pg/mL in 175 (70%) patients (Figure 2).

Eighty-seven (68%) patients in the micafungin group vs 74 (60.2%) patients in the placebo group were alive and free from invasive fungal infection at day 28 (hazard ratio [HR], 1.35 [95% CI, 0.87-2.08]; Figure 2). Results of the primary end point regarding various predefined subgroups of interest are reported in Figure 2 for the modified ITT population (with HRs substantially favoring the micafungin group for patients with [1-3]-β-D-glucan levels >80 pg/mL, [1-3]-β-D-glucan levels of 250 pg/mL, Candida scores at ≥3, and colonization index ≥50%). Unadjusted analyses provided similar results (eFigures 1 and 2; eTable 2 in Supplement 2). A posthoc analysis, not taking into account the 12 patients with invasive fungal infection at inclusion, had similar results (HR, 1.39 [95% CI, 0.88-2.22]; P = .15).

Day-28 survival was not significantly different between micafungin and placebo groups (Figure 3; eFigures 3 and 4 in Supplement 2). Similar results were observed for day-90 survival (eFigure 5 and eTable 3 in Supplement 2) and for the antifungal therapy–free survival rate (eFigure 6 in Supplement 2).

After inclusion, during the study follow-up, 15 (12%) patients in the placebo group and 4 (3%) patients in the micafungin group developed at least 1 new proven invasive fungal infection (P = .008) (Table 2). Of these 19 patients, 1 out of 4 (25%) in the micafungin group and 3 out of 15 (20%) in the placebo group died before day 28.

Other secondary end points, such as the number of organ failure–free days and the rate of ventilator-acquired pneumonia, were not significantly different between both groups (eTable 3 in Supplement 2).

After the first dose of micafungin, the mean (SD) Cmax level was 7.26 (2.43) mg/L (median, 7.4 [IQR, 5.4-9.2]), the mean (SD) Cmin level was 1.6 (0.54) mg/L (median, 2.1 [IQR, 1.4-3.1]), and the mean (SD) AUC was 78.2 (33.2) mg.h/L.

The drug was well tolerated with few adverse events; especially, liver enzymes variations were similar between micafungin and control groups (eTables 4 and 5; eFigures 7 and 8 in Supplement 2).

Quiz Ref IDIn this multicenter, double-blind, placebo-controlled trial in critically ill nonimmunocompromised patients with ICU-acquired severe sepsis, Candida colonization at multiple sites, and multiple organ failure, micafungin did not significantly improve the primary outcome of 28-day invasive fungal infection–free survival. There were no significant differences in the mortality rates, patient severity of illness following randomization, or in ICU or hospital lengths of stay. However, micafungin-treated patients had a significant reduction in the number of ICU-acquired invasive fungal infections following randomization.

The study failed to demonstrate that an empirical antifungal therapy with micafungin is able to improve, by at least 18%, the rate of survival free from proven fungal infection at day 28. This finding is unlikely to be from a lack of statistical power because the event rates were within the expected ranges, in line with the high severity of illness at admission or inclusion, also reflected by the number of patients receiving life-sustaining therapies. Furthermore, the intervention failed to improve outcomes overall, as well as in specific patient subsets such as those with high colonization index, high Candida score, or high (1-3)-β-D-glucan concentrations. The nonsignificant improvement of day-28 survival without invasive fungal infections among patients with high SOFA scores deserve further discussion. Because inclusion criteria for this trial comprised ICU-acquired sepsis, multiple organ dysfunctions, and other risk factors for candidemia, this finding suggests that among these selected patients, those who are most ill may benefit from antifungal agents. On one hand, this decreases the biological plausibility that supports the present intervention, limiting its benefit to a super niche. Conversely, because single-target interventions failed in this population, it can be assumed that reducing the incidence of invasive fungal infection could be seen as a therapeutic intervention that will ultimately improve survival in patients with established multiple organ dysfunctions. However the effect of this intervention on mortality is probably lower than suggested by previous literature.

This trial on empirical antifungals in ICU patients with Candida extra-intestinal colonization, unresolved sepsis, and multiple organ failure adds to the 2 previously published studies regarding 3 aspects. First, it shows that sepsis occurring in patients with multiple organ dysfunction and multiple-sites colonization in patients receiving broad-spectrum antibacterials agents is rarely due to invasive fungal infection. Second, it sheds light on the discrimination power of Candida colonization. Indeed, in the present trial, which included heavily colonized patients, questions remain about the relevance of sampling patients for Candida colonization when such sampling leads to financial burden from laboratory testing and also excessive antifungal consumption²² without any apparent clinical benefits. A study by Throughton et al reported that Candida colonization failed to guide empirical therapy,²³ and a study by Barenfanger et al demonstrated significantly reduced antifungal consumption when clinicians were not provided with Candida colonization results.²⁴Quiz Ref ID Altogether, these results call into question the routine use of systematic surveillance for Candida colonization. Besides sparing unnecessary use of health care resources, it may also avoid inducing resistances to antifungals.^25-27 Whether this trial closes 3 decades of clinical research on Candida colonization deserves consideration. Furthermore, the observation that the intervention failed, irrespective of the patients' (1-3)-β-D-glucan levels, is in line with previous publications showing that (1-3)-β-D-glucan was not significantly different between patients with candidemia vs those with multiple colonization.¹⁷ As for documented infections, (1-3)-β-D-glucan kinetics was not influenced by micafungin therapy, which did not support its use for guiding antifungal de-escalation.^28,29

In addition, this trial adds to the 2 others by reporting micafungin plasma concentrations. The observed median AUC is strictly similar to the value of 78.6 mg.h/L that was previously observed in ICU patients.³⁰ It confirms a decreased exposure by approximately 50% compared with healthy patients and by approximately 25% compared with non-ICU patients, suggesting that higher doses may be necessary in critically ill patients. The assumption that the present intervention would have proven benefits by using higher micafungin dosages is not supported by the significant reduction of ICU-acquired candidemia, unless considering that micafungin may have decreased blood culture sensitivity without clinical benefit.

Strengths of this study include the multicenter design and high adherence to the intervention started immediately after randomization. The proportion of ICU-acquired invasive candidiasis is within previously published ranges,^2,13,31-33 as is mortality.¹ Also, no patient was lost to follow-up. The risk of bias was also minimized by the blinded nature of the design, use of central randomization, concealment of study-group assignments before randomization to avoid selection bias, and a robust primary outcome that could not be influenced by observer bias. Because the centers belonged to a large study group that included university and non-university hospitals, the study may have external validity.

This study has a number of limitations. The first is its low rate of patients with a very high risk of invasive candidiasis, such as patients with postoperative gastrointestinal leakage of acute necrotizing pancreatitis.¹⁷ Also, micafungin underdosing cannot be ruled out because therapeutic drug monitoring was not performed after day 1.

Although maximal efforts were made to homogenize the diagnosis of invasive fungal infection, the procedure that was used daily in each center to diagnose fungal infections during the follow-up period might have slightly varied. However, the consequence of this information bias is limited by the stratification of the random process and the statistical analyses. The adapted EORTC definition for documented infection used in this study was previously used in studies by Tissot et al¹⁷ and Ostrosky-Zeichner et al¹⁰; however, this definition might possibly miss true fungal infections.

Besides having Candida species colonization, the inclusion criteria were similar to those used in previous trials of empirical antifungal use in critically ill patients^5,10,13 (in whom illness severity at randomization and mortality rates were similar). There was no evidence that micafungin influenced mortality estimates or was beneficial in treating specific subgroups. However, empirical treatment should be further evaluated in similar patients with a SOFA score greater than 8 at randomization.

Among nonneutropenic critically ill patients with ICU-acquired sepsis, Candida species colonization at multiple sites, and multiple organ failure, empirical treatment with micafungin, compared with placebo, did not increase fungal infection–free survival at day 28.

Corresponding Author: Jean-François Timsit, MD, PhD, Service de Réanimation Médicale et des Maladies Infectieuses, Hôpital Bichat–Claude-Bernard, 46 Rue Henri-Huchard, 75877 Paris Cedex 18, France (jean-francois.timsit@bch.aphp.fr).

Published Online: October 5, 2016. doi:10.1001/jama.2016.14655

Author Contributions: Dr Timsit had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Timsit, Azoulay, Charles, Cornet, Souweine, Gangneux, Jullien, Wolff.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Timsit, Azoulay, Souweine, Paris, Jullien, Wolff.

Critical revision of the manuscript for important intellectual content: Timsit, Azoulay, Schwebel, Charles, Cornet, Souweine, Klouche, Jaber, Trouillet, Bruneel, Argaud, Cousson, Meziani, Gruson, Darmon, Garrouste-Orgeas, Navellou, Foucrier, Allaouchiche, Das, Gangneux, Ruckly, Maubon, Jullien, Wolff.

Statistical analysis: Timsit, Azoulay, Ruckly, Jullien.

Administrative, technical, or material support: Azoulay, Charles, Cornet, Klouche, Trouillet, Paris, Das, Maubon, Wolff.

Study supervision: Timsit, Azoulay, Cornet, Souweine, Trouillet, Cousson, Paris, Garrouste-Orgeas, Jullien, Wolff.

No additional contributions: Schwebel, Argaud, Gruson, Darmon, Navellou, Foucrier, Allaouchiche.

Biological analysis: Cornet, Gangneux, Maubon, Jullien.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Timsit reports receipt of lecture fees from Gilead, Pfizer, Merck, Astellas; research grants to his university and research organization from Astellas, Gilead, Merck and Pfizer companies; a consultancy honorarium from Bayer; personal fees from Abbott for scientific board participation; and participation on a scientific committee of epidemiological studies organized by Astellas and Merck companies outside the submitted work. Dr Azoulay reports receipt of personal fees from Gilead, Astellas, and Alexion and grants from Cubist and Alexion during the conduct of the study. Dr Charles reports receipt of grants from Astellas, Merck, Thermofisher, and Pfizer and personal fees from Merck and Thermofisher outside the submitted work. Dr Cornet reports receipt of research grants from Pfizer. Dr Jaber reports receipt of personal fees from Drager, Maquet, Hamilton, and Fisher-Paykel outside the submitted work. Dr Darmon reports receipt of a grant from Merck; personal fees from Merck, Astellas, Bristol-Myers Squibb; and nonfinancial support from Merck, Astute Medical, and Jazz Pharmaceutical outside the submitted work. Dr Navellou reports receipt of grants from Astellas outside the submitted work. Dr Gangneux reports receipt of lecture fees and grants from Astellas, Gilead, Merck and Pfizer during the conduct of the study; and participation on a scientific committee of epidemiological studies organized by Astellas and Merck. Dr Jullien reports membership on a scientific committee for Astellas; receipt of lecture fees from Tibotec, Astellas, and BMS; consultancy fees from Basilea; and receipt of research grants from Biocodex and Sanofi outside the submitted work. Dr Wolff reports receipt of lecture fees from Astellas, Gilead, Merck, Pfizer, Aventis, Cubist and Astra-Zeneca; personal fees for board participation from MSD, sanofi, Gilead; and grant for scientific meetings from Astellas outside the submitted work. No other disclosures were reported.

Members of the Empiricus Study Group (named with permission):Investigators: (Medical ICU, Albert Michallon University Hospital, Grenoble): Jean François Timsit MD, PhD; Rebecca Hamidfar-Roy, MD; (CHI Andre Gregoire, Montreuil): Magalie Ciroldi, MD; (University Hospital Beaujon): Clichy Catherine Paugam-Burtz MD, PhD; Arnaud Foucrier, MD; (University Hospital J Minjoz, Besancon): Jean Christophe Navellou, MD; (Medical and Infectious Diseases ICU, Bichat-Claude Bernard University Hospital, Paris): Michel Wolff, MD; Jean-Francois Timsit, MD, PhD; Lila Bouadma, MD, PhD; Bruno Mourvillier, MD; Romain Sonneville, MD, PhD; Sarah Chemam, MD; (Medical ICU, Bordeaux University Hospital, France): Didier Gruson MD, PhD; (Medical ICU, University Hospital G Montpied Clermont-Ferrand): Bertrand Souweine, MD, PhD; Alexandre Lautrette, MD, PhD; (Medical ICU, François Mitterrand University Hospital, Dijon): Pierre Emmanuel Charles, MD, PhD; Rémi Bruyere, MD; Maël Hamet, MD; (Surgical ICU, Edouard Herriot University Hospital, Lyon): Bernard Allaouchiche, MD, PhD; Christian Guillaume, MD; Charles-Eric Ber, MD; Johanne Prothet, MD; Thomas Rimmele, MD; Medical ICU, Edouard Herriot University Hospital, Lyon): Laurent Argaud, MD, PhD; Marie Simon, MD; Martin Cour, MD; Romain Hernu, MD; (Surgical ICU, Intensive Care Unit, Department of Anesthesia and Critical Care Medicine, University of Montpellier, Saint Eloi Teaching Hospital): Samir Jaber, MD, PhD; Boris Jung, MD, PhD; Mathieu Conseil, MD; Yannael Coisel, MD; Fouad Belafia, MD; (Medical ICU, Lapeyronie University Hospital, Montpellier 2): Kada Klouche, MD, PhD; Laurent Amigues, MD; Sonia Machado, MD; Marianne Serveaux, MD; (Medical ICU, Institut de Cardiologie, Hôpital de la Pitié–Salpêtrière, Assistance Publique–Hôpitaux de Paris, Paris): Jean Chastre, MD, PhD; Jean-Louis Trouillet, MD; (CHU Reims): Joël Cousson, MD; Pascal Raclot, MD; Thierry Floch, MD; (Medical ICU, Saint-Etienne University Hospital, Saint-Priest en Jarez): Fabrice Zeni, MD, PhD; Michael Darmon, MD, PhD; Matthias Pichon, MD; Maud Coudrot, MD; Sebastien Ninet, MD; Eric Diconne, MD; (Saint-Louis University Hospital, Medical ICU, Paris): Benoit Schlemmer, MD; Elie Azoulay, MD, PhD; Virginie Lemiale, MD; Nicolas Maziers, MD; (Service de Réanimation Médicale, Nouvel Hôpital Civil, Hôpitaux Universitaires de Strasbourg, Strasbourg France): Ferhat Meziani, MD, PhD; David Schnell, MD; Julie Boisrame-Helms, MD; Raluca Neagu-Anca, MD; Xavier Delabranche, MD; Olivier Martinet, MD; (Medical-Surgical ICU, Saint-Joseph Hospital Network, Paris): Maité Garrouste-Orgeas, MD, PhD; Benoit Misset, MD, PhD; (Medical ICU, André Mignot Hospital, Versailles, France): Fabrice Brunee, MDL; Virginie Laurent, MD; Guillaume Lacave, MD; Jean-Pierre Bedos, MD. Study Monitors, Research Nurses, and Biohygiene Technicians: (Medical ICU, Albert Michallon University Hospital, Grenoble): Khadija Hammi, RM; Lenka Styfalova, MSc, RM; (Medical Surgical ICU, CHU J Minjoz, Besancon): Joelle Fritzsch, RM; (Medical and Infectious Diseases ICU, Bichat-Claude Bernard University Hospital, Paris): Sophie Letrou, RM; (Medical ICU, CHU Bordeaux): Lucie Estevez, RM; (Medical ICU, CHU Gabriel Montpied Clermont-Ferrand): Mireille, ADDA, RM; (Medical ICU, CHU Dijon): Therese Devaux, RM; (Surgical ICU, Hospices Civil de Lyon, Edouard Herriot Hospital): Celine Dubien, RM; Soumia Bayarassou, RM; Catherine Jouvene Faure, RM; (Medical ICU, Hospices Civil de Lyon, Edouard Herriot Hospital): Sylvie de La Salle, RM; (Surgical ICU, Montpellier Hospital): Albert Prades, RN; (Surgical ICU, Montpellier Hospital): Annie Rodriguez, RM; (Medical Surgical ICU CHU Reims): Pierre Meur, RM; Magda Warchol, RM; (Medical ICU, CHU Saint-Etienne): Hanane El Haouari, RM; (Saint-Louis University Hospital, Medical ICU, Paris): Igor Theodose, RM; (Medical-Surgical ICU Fondation Hospital Saint Joseph, Paris): Julien Fournier, RM; (Medical-Surgical ICU CH Versailles-Le Chesnay): Sebastien Cavelot, RM. Pharmacists: (Albert Michallon University Hospital, Grenoble): Lilia Bakir Kodja, PharmD; Marie Joyeux Faure, Pharm D; (CHI André Gregoire, Montreuil): Frédéric Tacco, PharmD; Sonia Roos, PharmD; Karima Dupre, PharmD; (CHU Beaujon, Clichy): Malek Abazid, PharmD; (CHU J Minjoz, Besancon): Michele Essert, PharmD; (Bichat-Claude Bernard University Hospital, Paris): Philippe Arnaud, PharmD; Emmanuelle Papy, Pharm D; (CHU Bordeaux): Bellabes Ghezzoul, PharmD; Olivier Gerbouin, PharmD; (CHU Gabriel Montpied Clermont-Ferrand): Sandrine Corny Peccoux, PharmD; (CHU Dijon): Philippe Fagnoni, PharmD; (Hospices Civil de Lyon, Edouard Herriot Hospital): Anne Millaret, PharmD; Christine Pivot, PharmD; Cecile Gerard, PharmD; (CHU Montpellier): Cyril Breuker, PharmD; Audrey Castet, PharmD; Fanny Charbonnier, PharmD; (CHU Reims): Maryline Legrand, PharmD; (CHU Saint Etienne): Julia Mordini, PharmD; (APHP Hopital Saint Louis): Isabelle Madeleine Chambrin, PharmD; (CHU de Strasbourg): Anne Hutt Clauss, PharmD; (ICU Fondation Hospital Saint Joseph, Paris): Mohamed Cherifi, PharmD; (CH Versailles-Le Chesnay): Anne Pattyn, PharmD. Mycologists: (Albert Michallon University Hospital, Grenoble): Murielle Cornet, PharmD, PhD; Danièle Maubon, MD, PhD; (CHI André Gregoire, Montreuil): Eliane Benveniste, PharmD; (CHU Besancon): Fréderic Grenouillet, PharmD; (CHU Beaujon, Clichy and CHU Bichat Paris): Christian Chochillon, PharmD, PhD; (CHU Bordeaux): Isabelle Accoberry, PharmD; (CHU Gabriel Montpied): Denis Pons, PharmD; Natacha Mrozek, MD; (CHU Dijon): Frédéric Dalle, PharmD; (Hospices Civils de Lyon, Edouard Heriot Hospital): Stephane Picot, PharmD; Francoise Beyerle, PharmD; Anne-Lise Bienvenu, PharmD; (CHU Montpellier): Nathalie Bourgeois, PharmD; (APHP Hopital La Pitie, Paris): Arnaud Fekkar, PharmD; (CHU Reims): Dominique Toubas, PharmD; (CHU Saint-Etienne): Hélène Raberin, PharmD; (CHU de Strasbourg): Ermanno Candolfi, PharmD; Valérie Bru, PharmD; (Fondation Hopital Saint Joseph, Paris): Marie Dominique Kitzis, PharmD; Yaye Senghor, PharmD; (CH Versailles Le Cheynay): Catherine Palette, PharmD. Statistics: (Outcomerea Research Network, Paris, France and Icuresearch Company, Paris, France), Stéphane Ruckly, MSc; (Delta Consultant Company, Grenoble, France), Aurélien Vesin, MSc. Safety Monitoring Board: (Medical ICU, APHP Hopital Cochin Paris), Jean-Paul Mira, MD, PhD; (Infection Control, CHU Angers), Jean-Ralph Zahar, MD, PhD; (Pharmacy and Vigilance, CHU Grenoble), Edith SHIR, PharmD; (Hepatology-Gastro-enterology, CHU Montpellier), Dominique Larrey, MD, PhD; (Hepatology-Gastro-enterology, CHU Grenoble), Jean-Pierre Zarski, MD, PhD. Independent Adjudication Committee: (Angers), Jean-Ralph Zahar, MD, PhD; (Paris Cochin), Jean-Paul Mira, MD, PhD; (Montpellier), Dominique Larrey, MD, PhD; (Grenoble), Jean-Pierre Zarski, MD; and Edith Schir, PharmD.

Funding/Support:Astellas provided a research grant to the Grenoble Alpes University Hospital based on the final study protocol. The study was sponsored by the University of Grenoble 1/Albert Michallon University Hospital. The University of Grenoble provided compensation to the participating hospitals and universities for extra costs associated with the study.

Role of the Funder/Sponsor: Astellas had no role in study design and conduct, and did not intervene in data collection, data management, or in data analysis and interpretation.The manuscript was prepared, reviewed, approved by coauthors without any intervention from the sponsor, as was the decision to submit it for publication.

Additional Contributions: The methods and full statistical analysis were performed at the University Paris Diderot UMR 1127 IAME-Team 5 Decision Science in Infectious Diseases, Paris, France by Stéphane Ruckly, MScs (ICUREsearch) and Aurélien Vesin, MSc (Delta Consultant) under the supervision of Dr Timsit. The authors thank Celine Feger, MD (EMIBiotech) for her editorial support. None were compensated in association with their contributions to this article.