Study design and eligibility criteria
This monocentric retrospective cohort study was conducted in the tertiary NICU of the Hôpital Femme Mère Enfant, Hospices Civils de Lyon, Lyon, France. All neonates hospitalized in this NICU between 1st January 2010 and 31st December 2019 for whom at least one administration of linezolid was encoded in the medical record were eligible for inclusion. Patients for whom linezolid was previously initiated in another NICU before referral to the study NICU were not included.
Collection of clinical data
The list of patients eligible for inclusion and the clinical data were retrieved using the software ICCA (Philips®, Suresne, France) which is used to prospectively record medical information for NICU patients. For each included patient the following items were collected: date of hospitalization in the study NICU, delivery mode, weight and gestational age at birth, sex, Apgar score, death during hospitalization in the NICU, data about the infection leading to the use of linezolid including age at the onset of the sepsis, infective agent, infection site, need for central line removal, and finally data about the antibiotic treatment including first-line treatment, second-line treatment, delay until the introduction of linezolid, duration and route of linezolid administration, reason to stop it, reported side effects notably thrombopenia, anemia, hyperlactacidemia. Moreover, data about the strains involved in these infections were obtained from the laboratory record, including antimicrobial susceptibility tests. According to EUCAST recommendations, resistance to vancomycin was defined as a minimal inhibitory concentration > 2 mg/L for S. aureus or > 4 mg/L for CoNS and Enterococcus. Resistance to linezolid was defined as a minimal inhibitory concentration > 4 mg/L.
Classification of adequate/inadequate indication of linezolid
The administration of linezolid was secondly classified as adequate or inadequate independently by two pediatricians, following previously published Australian guidelines (12). Briefly, indication was considered as adequate if linezolid was prescribed for a Gram positive infection with either resistance to other antibiotics (including beta-lactams and vancomycin), or if there was a persistent bacteremia despite vancomycin therapy with appropriate serum level (15-20 mg/L) and removal of central venous line (or impossibility to remove it), or also if there was a contraindication to vancomycin (renal failure). In other situations, linezolid was considered as inadequate.
Data about other antibiotics
To have an overview of all the antibiotics used during the study period in this NICU and especially antibiotics targeting Gram positive bacteria, we extracted from the software ICCA the number of patients that have received at least one course of vancomycin, rifampicin, fosfomycin, oxacillin and cefazolin. The use of each antibiotics was expressed as number of patients receiving at least one course of each antibiotic for 1000 patient-days.
Microbiological and epidemiological data
To confront our data with the microbial epidemiology of infections during the study period, all bacterial bloodstream infections that occurred in this NICU during the study period were retrieved from the epidemiological monitoring routinely performed by the Department of Infection Control of the hospital and were expressed as number of septic episode for 1000 patient-days. Of note in this NICU, clinical reports of all patients with a positive blood culture are routinely analyzed by both a NICU practitioner and a medical officer of health, to classify positive blood cultures as bloodstream infections or contaminant. Bloodstream infection was defined as the association of clinical signs with at least one positive blood culture. Pneumonia was defined as the combination of physical exam findings, radiographic evidence, and a positive culture of suctioned tracheal sputum. Persistent bloodstream infection was defined as more than 2 blood cultures positive for a same bacterium and spaced out of at least 48 hours.
Ethics
This monocentric retrospective cohort study was approved by the local institutional ethics committee (Hospices Civils de Lyon) under the number 20-62 and was registered to the French data protection authority under the number 20-137. A written information about this study was delivered to the families of the included patients to give them opportunity to decline participation to this study. In accordance with French legislation, a written consent was not required owing to the observational nature of the study. All clinical data were anonymized.
Statistics
A descriptive analysis was performed. Patient characteristics were described as median, interquartile range (IQR) or as percentage. No statistical test was performed for this observational study.