The emergence and rapid worldwide spread of Extended- Spectrum Beta-Lactamase-producing enterobacteriaceae (ESBLE) both in hospital and community, led physicians, and notably intensivists, to prescribe more carbapenems, particularly in the most fragile patients such as ICU patients. Unfortunately, the increased carbapenem consumption favored the emergence of carbapenem resistance mechanisms. Moreover, several preliminary results suggest that carbapenem could markedly impact the human intestinal microbiota, Thus, reduction of carbapenem exposure is widely desired both by national and international antibiotic plans. Therefore, the use of rapid diagnostic tests evaluating bacterial resistance to reduce inappropriate exposure to carbapenems could be a relevant solution. Due to its good diagnostic performance, the betaLACTA® test could meet these objectives.

Experimental plan :

Randomized, open-labeled non-inferiority clinical trial involving an in vitro diagnostic medical device (close to a phase III study), comparing two parallel groups:

• Experimental group: early carbapenems de-escalation since the second dose, guided by results of the betaLACTA® test performed directly on the bacterial pellet from the microbiological sample positive on direct examination.
• Control group: carbapenems de-escalation guided by definitive results of the antibiotic susceptibility test obtained 48 to 72h after microbiological sampling (reference strategy).

This study is conducted on ICU patients with a suspected pneumonia, primary blood-stream infection (BSI), and/or urinary tract infection (UTI).

The primary objective of the study is to demonstrate that in ICU infections treated empirically by carbapenems and documented with GNB on direct examination of a respiratory, urinary and/or blood sample(s), the early de-escalation guided by the results of the betaLACTA® test is not inferior to the reference strategy de-escalating on antibiotic susceptibility test (AST) results obtained 48-72h after sampling, in terms of mortality at D90 and infection recurrence in ICU.

The secondary objectives are to compare the early de-escalation guided by the betaLACTA® test results to the reference strategy de-escalating on the AST results on:

• The exposure to carbapenems.
• The total use of ICU and hospital resources and the cost-effectiveness.
• The occurrence of other infections.
• The colonization of the digestive tractus of patients with 3rd generation cephalosporins (3rdGC) resistant Gram-negative bacteria (GNB).

In addition, an ancillary study will be performed (only in participating centers from the Ile de France region) to compare :

• The composition of the intestinal microbiota among patients with an early de-escalation guided by the betaLACTA® test results and a standard de-escalation on AST results at 48-72h.
• The evolution of intestinal microbiota of patients after exposure to different beta-lactam antibiotics, from carbapenems or cefepim/ceftazidim during the empirical treatment, to the definitive beta-lactam antibiotic chosen to cure the infection after antibiotic susceptibility test results.
• The outcomes of ICU patients (mortality at D90, occurrence of infection, ICU length of stay, etc.) according to the composition of their intestinal microbiota and to its evolution during antibiotic treatment.

To meet these objectives, 646 patients will be enrolled.

Inclusion Criteria:

• ICU patients ≥18 years.
• With a suspected pneumonia (according to CPIS definition), primary blood-stream infection (according to CDC definition), and/or urinary tract infection (according to IDSA Guidelines).
• And presence of ≥2 GNB/field on direct examination of a respiratory sample (quantitative bronchial aspirate with an available volume ≥ 1mL), urinary sample or blood culture.
• Leading to an empirical carbapenem prescription, not administered for more than 6 hours when considering the inclusion of the patient.
• Written informed consent signed by the patient / the trustworthy person / the next-of-kin / close relative, or inclusion in case of emergency and written informed consent will been signed by the patient as soon as possible.
• Patients affiliated to French social security.

Exclusion Criteria:

• Pregnancy.
• Allergy to beta-lactams.
• Patients already treated with ongoing carbapenems for another documented infection, blocking carbapenem de-escalation.
• Patients included in another interventional study.
• Patients in whom a procedure of withdrawing life-sustaining treatment was decided before inclusion.
• Moribund patients.
• Patients with aplasia.
• Patients under tutorship/curatorship or patient deprived of freedom