Severe community-acquired and nosocomial pneumonia are associated with substantial morbidity and mortality. Early and appropriate antimicrobial therapy (AAT) is consistently the most effective intervention for reducing mortality. Cure is most likely when pharmacokinetic (PK) / pharmacodynamics (PD) targets associated with maximum antibiotic (ABX) activity are achieved. However, the process of optimizing antibiotic therapy for critically ill patients remains a complicated challenge.

A key issue is pathogen identification (ID) with subsequent antibiotic susceptibility testing (AST) results which allow for selection of AAT. Standard laboratory procedures typically require 2-3 days to provide ID and AST results. Optimal ABX dosing/dosing intervals depend in large part on PK properties in individual patients, and antibacterial effects on the infecting bacteria (PD). Alterations in the primary PK parameters, namely volume of distribution (Vd) and clearance (CL), are commonly observed, and are the most influential parameters in determining ABX dosing and exposure. ABX dosing/dosing intervals that do not account for these features are likely to lead to suboptimal ABX exposure and therapeutic failures. Because of 48-72-hours delays in ID/AST, initial treatment is frequently inappropriate in coverage, unnecessarily broad in spectrum, and/or suboptimal in dosing.

Methods for rapid bacterial growth, ID, AST and minimum inhibitory concentration (MIC) identification were developed and are capable of quantitative ID in 1-2 hours and major AST in 6-8 hours using clinical specimens. Rapid ID of the infecting pathogen and its individual AST could significantly impact the early selection of AAT and, combined with therapeutic drug monitoring data, could be used to calculate optimized dosing regimens that are personalized for the patient in order to achieve appropriate PK/PD targets.

Hypothesis: Application of these rapid ID/AST systems, together with prospective PK/PD monitoring of antibiotic plasma concentrations, will significantly shorten time from "sample to answer" for pathogen ID/AST, enhance personalized prescribing of antibiotics, optimize the time to targeted effective and AAT, and result in decreased treatment failure.

Inclusion Criteria:

• Patients hospitalized in ICU
• 18 years of age or older
• With a pulmonary sepsis defined a s documented or suspected acute pulmonary infection (nosocomial and community-acquired pneumonia) and a SOFA score >2.
• Written Informed consent from the patient whenever possible or written ascent from next of kin whenever present at inclusion. When a patient would not be capable of consenting prior to randomization, his/her deferred consent will be gotten.

Exclusion Criteria:

• COVID-19 patients
• Severe anaphylactic beta-lactam allergy
• First measurements of prescribed antibiotic concentration (TDM) not possible within 24 hr after randomization
• Pregnancy or lactation
• Any decision of limitation of care
• Pre-existing medical condition with a life expectancy of less than 3 months
• Absence of affiliation to social security
• Patient under guardianship, curatorship and deprived of liberty