Understanding the emergence of linezolid-resistance in Staphylococci has been allowed in the past years through the discovery of the clonal dissemination of a chromosomal cassette carrying a modified crf gene. New mutations have even been described. Though, clinical evidences are still lacking, especially concerning the factors associated to this emergence. It could seriously become quite problematic to eliminate one of the last therapeutic weapon at our disposal for the treatment of severe or complicated infections caused by resistant strains of Staphylococci and Enterococci.

We aim to describe the mechanisms that permitted to this resistance to become clinically significant, concerning meticillin-resistant Staphyloccocus epidermidis strains causing blood stream infections in ICU patients, and show the clinical risk factors associated with it through a case-control study on patients hospitalized in two ICUs of our hospital between 2011 and 2016.

Resistance to linezolid has been increasing in the past years, suggesting growing therapeutic difficulties while narrowing the options for treating severe or complicated infections involving Staphyloccoci or Enterococci strains. This emergence has been partially explained after the discovery of chromosomal dissemination of a particular gene, named "crf", conferring high level resistance to oxazolidinones.

Moreover, this mechanism has been showed as being prevalent in several countries across North America, Asia, and Europe.

But still the link with clinical evidences has not yet been very well established.

In particular the role of S. epidermidis has become clinically significant concerning blood stream infections and catheter-linked infections. The reality of the pathogenicity is more widely accepted, especially regarding prosthetic joint infections, and immunocompromised patient, as in hematology and oncology.

Thus, it appears necessary to provide more solid informations concerning the risk of using this antibiotic in a high burden setting of resistance, where it might be prescribed intensively on critical situations, and thus responsible for a high selection pressure of resistance.

We aim to describe the emergence of the resistance to linezolid in meticillin-resistant S. epidermidis strains in the two 20-beds ICUs of our Teaching Hospital, in the 5 past years, between april 2011 and october 2016.

We conduct a case-control study between two populations of patients hospitalized in one of these ICU.

They all presented a blood stream infection caused by a strain of S. epidermidis that was resistant to meticillin, and diagnosed prospectively or retrospectively being resistant as well to linezolid, or not. Indeed, microbiological data was quite easily available as all the strains isolated from blood samples are conserved after congelation for many years. Nevertheless, we lack the profile toward linezolid susceptibility as this antibiotic was not routinely tested before 2014, corresponding to the moment we became aware of the problematic rising. A previous work thus consists to establish the incidence of the resistance.

Comparison of these two populations through a case-control design study, accepting 2 control-patients for one case (matched on the period of hospitalization, mortality and severity score at admission), might allow understanding risk factors of being infected with a linezolid-resistant and meticillin-resistant S.epidermidis.

So to go further, we first conduct a phenotypic analysis in order to establish the incidence of this resistance, in accordance with the 2016 EUCAST criteria. Association with genotypic analysis using Pulsed-field gel electrophoresis and Multilocus-sequence typing will allow underlining the impact of the presence of the crf gene, and mutations described as being involved in high level resistance (mutations in domaine V of 23S ARN or ribosomal proteins L3 and L4).

All adult patient admitted to one of these two units can be included and evaluated if he presented at least two blood cultures sampled at less than 48hours of interval, being positive for the same strain (in accordance with the 2007 CTINILS definition). We considered the catheter-linked infections as being apart if it was associated with positive blood cultures.

We also search for clinical criteria supposed to be linked to an infection with such a resistant strain; we check for the demographic data, conditions at admission, antibiotic treatment before or at admission, procedures considered at high risk for health-care infections such as surgery, especially in the period of great instability, also endovascular procedures, like catheter implantation or extra-corporal assistance (Extra-corporal membrane for oxygenation, Renal Replacement Therapy).

We examine the therapeutic options once the diagnosis of blood stream infections has been made, the choice of antibiotic as treatment, and the outcome defined as cure or survival at 28 days, which is the period of matching patients.

Inclusion Criteria:

• all adult patient admitted in ICU between april 2011 and october 2016, diagnosed for a blood stream infection during the ICU stay, whatever the antibiotic therapy conducted before or after the diagnosis

Exclusion Criteria:

• minor-aged patients
• less than 2 blood cultures positive
• blood cultures distant from more than 48 hours
• cultures positive to distinct pathogens

critically ill patients admitted at one of the two 20-beds ICU of our teaching hospital, with no regard to the reason for admission, neither to conditions at admission or severity of illness, nor to prognosis of affection leading to ICU