ABSTRACT
BACKGROUND: Neonatal early onset sepsis (EOS) remains an important etiology of neonatal morbidity and mortality. Diagnosis is difficult due to a lack of sensitivity and specificity markers. In France, the management of newborn infants suspected of infection includes the analysis of gastric suction. The objective of the study was to identify early clinical signs in newborn infants with suspected neonatal sepsis to differentiate a likely infection with pathogen bacteria in the gastric suction culture (Streptococcus agalactiae or Escherichia coli) from a possible infection without such pathogen bacteria. METHODS: We conducted a retrospective study in the Amiens University Hospital. All term newborn infants born between 1 January and 31 December 2013 and hospitalized for suspected EOS were included. Suspicion of EOS was considered when there were arguments to treat by antibiotics for a period of at least 5 days. RESULTS: Fifty-eight newborn infants were included, 25 had a likely EOS and 33 a possible EOS. Newborn infants with a likely EOS were less mature (P<0.01) with more clinical signs at birth (P<0.01). The most common clinical signs were: hyperthermia (P=0.01), somnolence (P<0.01), and hypotonia (P=0.01). After adjusting for the term, the presence of hyperthermia was no longer significantly different between the two groups (P=0.059), the other clinical signs remained significantly different. CONCLUSION: The presence of neonatal symptoms at birth appears to be a useful clinical marker of probable neonatal EOS.
Subject(s)
Bacterial Infections/diagnosis , Bacterial Infections/microbiology , Neonatal Sepsis/diagnosis , Neonatal Sepsis/microbiology , Bacteria/pathogenicity , Female , Humans , Infant, Newborn , Male , Retrospective StudiesABSTRACT
The aim of the paper is to identify the key physiological variables and ventilator settings involved in ventilation management, and required for an appropriate Clinical Decision Support System (CDSS). Based on the results of a questionnaire designed for the purpose of the research, 70 hours of physiological and ventilation data were recorded. Recorded data were classified by clinicians into three major lung pathologies and were further statistically analyzed for identifying strong relationships between monitored and controlled ventilator parameters. Correlation analysis was evaluated by Intensive Care Unit (ICU) clinicians. Based on the evaluators' majority voting the number and type of participating variables in a CDSS was drastically decreased. The number and type of monitored variables ranged from a single one to six, depending on the patient's lung pathology, and the controlled ventilator setting. Evaluation results were successfully applied to Neural Network models for providing suggestions on Tidal Volume and the Fraction of inspired Oxygen.