European Prospective Cohort Study on Enterobacteriaceae Showing Resistance to Carbapenems

HYPOTHESIS:

H1: 5 independent predictors for cure and mortality can be identified, including active empirical therapy, early targeted optimized therapy and early source management if needed.

H2: For pneumonia, cIAI (complicated intrabdominal infection) and BSI (bloodstream infection), combination therapy with two active drugs, one of them being (if available) an "active" beta-lactam (such as meropenem or imipenem if minimum inhibitory concentration [MIC] <16 mg/L, aztreonam if isolate is susceptible as in many metallo-beta-lactamase producers, or cephalosporin if isolate is susceptible as in some OXA-48 producers). For cUTI (complicated urinary tract infection), monotherapy with an "active" beta-lactam as above, colistin or an aminoglycoside (if active in vitro) is as effective as combination therapy.

H3. Clinical cure rate at test of cure (TOC) will be 50% with BAT. H4: Specific carbapenemase types do not independently influence cure rate or mortality.

H5: CRE infections caused by isolates showing a carbapenem MIC <16 mg/L are associated with higher probability of cure and lower mortality.

H6: five significant independent predictors (risk factors) for CRE infection can be found.

H7: the targeted infections due to CRE are significantly and independently associated with higher mortality, hospital stay and hospital costs than infections caused by carbapenem-susceptible Enterobacteriaceae (CSE) or than other diseases causing hospitalisation.

SPECIFIC OBJECTIVES. O1. To characterise the features, clinical management and outcomes of hospitalised patients with cIAI, pneumonia, cUTI and BSI caused by CRE and CRAB.

O1A. To provide cohorts of patients with the targeted infections caused by CRE and CRAB that would eventually be used as historical cohorts for comparison of efficacy and safety of newer drugs against these organisms.

O1B. To identify the outcome predictors of patients with cIAI, pneumonia, cUTI and BSI caused by CRE and CRAB, with identification of the best alternative therapy (BAT).

O1C. To exploratively investigate the importance of the specific carbapenemase and carbapenem-MIC in the outcome of CRE and CRAB infections.

O2. To identify the risk factors for target infections caused by CRE to inform a more efficient design of future randomized clinical trials for these infections.

O3. To assess the mortality, length of hospital stay and hospital costs associated with target infections caused by CRE.

DESIGN AND STUDIES To answer the above objectives, a prospective, multinational, multicentre, observational and analytic project including 3 studies was design.

Study 1. For the analysis of outcome predictors of CRE and CRAB infections (objective 1), a prospective cohort study of patients with the target infections due to CRE and CRAB will be performed.

Study 2. For the analysis of risk factors for target infections caused by CRE (objective 2), a nested case-control-control will be performed. The first group of controls will be formed by matched patients with CSE infections, and the second groups of controls will be formed by admitted patients non-infected patients by CRE or CSE.

Study 3. For the analysis of cost, outcome impact and length of stay associated to target infections caused by CRE (objective 3), a matched cohorts study will be performed. The cohorts will be formed by selected patients with infections due to CRE and the patients with infections due to CSE (identical to the CSE control group above). Additionally, a control group of admitted patients not infected by CRE or CSE will be studied (identical to the admitted control group above).

SETTING This study will be performed in 50 European hospitals from Spain, Italy, Greece, Turkey, Serbia, Croatia, Montenegro, Kosovo, Albania, Bulgaria and Romania.

STUDY PERIOD The recruitment period of the study is planned from February 2016 to June 2017.

ENDPOINTS (MAIN OUTCOME VARIABLES) See below

STUDY VARIABLES AND DEFINITIONS

• CRE: any isolate identified as an Enterobacteriaceae showing a minimum inhibitory concentration (MIC) ≥1 mg/L if using any dilution method and/or ≤22 mm if using a disc-diffusion method for imipenem or meropenem (10 µg disks); all others will be considered carbapenem susceptible (CSE), but meropenem and imipenem susceptible isolates showing resistance to ertapenem will be excluded.
• CRAB: Any isolate identified as Acinetobacter baumannii showing a minimum inhibitory concentration (MIC) ≥16 mg/L for imipenem or meropenem if using any dilution method and/or ≤17 mm for meropenem and/or ≤15 for imipenem if using a disc-diffusion method.
• Independent variables for Study 1: demographics, comorbidities in adults (Charlson's index), comorbidities in children (ARPEC PPS definitions), type of acquisition, systemic inflammatory response syndrome severity in adults, sepsis criteria in children, Pitt score, SOFA score, APACHE-II score, PIM2, invasive procedures, neutropenia, microbiological variables (poly or monomicrobial infection, carbapenemase producer, carbapenemases type, susceptibility profile, carbapenem MIC), clinical management (source control and support therapy), antimicrobial therapy.
• Independent variables for Study 2: demographics, length of hospital stay, epidemiological variables (travels, previous contact with persons colonised by CRE, previous hospitalization, nursing home or other long term-care facility residency, previous colonisation by CRE), comorbidities, acquisition type, SIRS severity in adults, sepsis in children, Pitt score, SOFA score, APACHE-II score, PIM2, invasive procedures, neutropenia, type of infection, microbiological variables, antibiotics received in the last 3 months.
• Independent variables for Study 3: all those included in Study 2.

DATA COLLECTION AND FOLLOW-UP Data will be collected by trained local investigators. Patients will be followed for 30 days from day 0. Data prior to study entry will be collected by reviewing medical records or interviewing the patient, his/her family or the attending healthcare staff. After that, the patients will be followed prospectively. If the patients had been discharged before assessement, outcome must be assessed by an outpatient visit or phone call according to a pre-design questionnaire.

MICROBIOLOGICAL STUDIES. All procedures will be performed locally using accepted, standard microbiological protocols. Isolates identified as CRE or CRAB according to above criteria will be locally studied for carbapenemase production using the CARBA-NP test. Susceptibility tests to key antimicrobial agents will be collected. Isolates preservation CRE and CRAB isolates will be preserved locally at least at -20ºC.

SAFETY ASSESMENT No investigation drugs will be used in this study. Adverse event will be collected during follow-up as one variable of interest for analysis purposes.

SAMPLE SIZE :

• Study 1 The sample size for the CRE and CRAB cohorts were calculated so that the cohorts may serve as 'historical' cohorts for future comparison with new drugs for CRE and CRAB. To do so, and because the estimates for the outcome variable of the new drug is unknown, we seek to estimate the clinical cure rate of BAT with 95% confidence interval and 8% precision. For an estimated cure rate of 50% based on data from previous studies, 151 patients for each of the five types of infection are needed (due to CRE: cUTI, pneumonia, cIAI, and BSI; due to CRAB: BSI). However, because around 25% of patients will not receive BAT, we will need 201 patients per type of infection except for BSI due to A. baumannii, for which we will need 221 because some blood isolates identified as Acinetobacter spp in sites not specifying the species will not be A. baumannii (total, 1025 patients). Such sample size will be enough also to investigate the best available therapy with the merged CRE cohorts.
• Study 2: The CRE case group will be formed by 248 patients with CRE infections selected from the study 1 cohort. Per each CRE case, one CSE control patient and 3 non-infected controls will be selected. Because of possible heterogeneity due to the different infection types, we have chosen to over sample by a ratio of 248/201*100%. The number of matched controls has been chosen to be four which is well known to well approximate the power of full cohort data.
• Study 3 The CRE, CSE and non-infected cohorts will be formed by the same patients as in Study 2. Therefore, the sample size rationale is as for Study 2, however with the refinement that the nested case control matching is now viewed as an exposure density sampling.

STATISTICAL ANALYSIS

• Study 1. The outcomes associated with exposure to different variables will be compared; the targeted exposures will be empirical active antimicrobial therapy, early targeted optimized therapy, and early source control. Antimicrobial regimens will be analysed as empirical (administered before the susceptibility testing is available) and targeted (thereafter) therapy. The primary endpoint "mortality from any cause until day 30" will be analysed using survival methods (Kaplan-Meier, Cox regression). The other primary endpoint "clinical response at TOC" will be analysed as a dichotomous outcome, regression analyses will use the logistic regression model. The analysis of the secondary endpoints will be analogous. Microbiological response at TOC is a polychotomous outcome and will be analysed using multinomial logistic regression. Goodness of fit will be assessed throughout. Variable selection will be based on Akaike's information criterion.
• Study 2. Exposure to potential risk factors of patients will be compared between CRE cases and CSE controls, and between CRE cases and admitted controls. A stratified and weighted Cox analysis will be performed. Multilevel hospital data (local rate of CRE, antimicrobial consumption, infection control measures) will also be considered in the analyses above.
• Study 3. The impact of CRE infection on mortality, length of stay, cost, length of hospital and ICU stay, and length of mechanical ventilation of patients will be assessed by comparison with those of CSE infection and admitted patients. All time-to-event outcomes (for costs and extra hospital/ICU days see below) will be compared using survival techniques. Because the outcome infection due to CRE is subject to the competing risks of death in hospital w/o CRE infection and discharged alive from hospital with or without CRE infection, the analyses shall be supplemented by those of the competing outcomes. Goodness of fit will be assessed throughout. Extra hospital/ICU days will be estimated using the multistate approach of Beyersmann et al. Variable selection will be based on Akaike's information criterion. Goodness of fit will be assessed throughout.

ETHICAL CONSIDERATIONS. Prior to initiation of a study site, approval will be sought from the appropriate regulatory agency and local Ethics Committees of Research or IRBs to conduct the study in accordance with regulatory requirements. This is an observational study and therefore no intervention is performed on behalf of the investigation. Management of all patients including all antibiotic regimens prescribed will be decided by the physician doctor/team in charge without any interference. The processing of the patients' personal data collected in this study shall comply with the Data Protection Act 1998 and with the European Directive on the Privacy of Data.

MONITORING The study will be monitored for quality and consistency of data.