Hospital-acquired infections are common complications in preterm infants. The diagnosis has to be fast and accurate. Indeed, the early identification of a suspected infection is very important, since the early administration of antibiotics lowers the risk of septic shock and improves long term outcome in the infected newborns who survive. Besides, a high specificity in the diagnosis of infection allows for the reduction of inappropriate treatment and thus prevents the emergence of antibiotic resistance.
The aim of this study is to develop a computer-assisted diagnosis tool, based on the real time analysis of cardio-respiratory signals, to aid the neonatologist in the diagnosis of infection of the preterm infant, at the bedside.
Hospital-acquired infections increase morbidity and mortality in the preterm infants. Early diagnosis of infection is difficult mainly due to the poor performance of clinical signs and to the need for invasive procedure to get blood tests. However, early administration of antibiotics lowers the risk of septic shock and improves long term outcome in the infected newborns who survive. Many clinical features have been described, associated with an ongoing infection but they are inconsistent, variable and nonspecific. Similarly, many invasive laboratory tests have been proposed for the diagnosis of infection in the newborn but they all need blood sampling and none has a good predictive value.
The combined analysis of the heart rate and respiratory characteristics appears to be a promising tool for the diagnosis of infection in the preterm infants. These signals are non-invasively recorded and their computerized real time analyses would allow for a continuous assessment of the risk of infection.
The main objective is to test the hypothesis that the analyses of the variability of the cardiac cycle duration, the variability of the respiratory cycle amplitude and duration, and their relationships, can significantly improve the performance of the diagnosis of late onset infection in the preterm infant at the bedside in neonatal units.
For this purpose, 500 preterm infants less than 32 weeks’ gestational age, are being recruited in 3 level 3 neonatal intensive care units (Rennes Lille Angers, France) for this prospective observational cohort study. Heart rate and respiratory characteristics are being collected from day 3 to 34 wks post-mentrual age, stored and processed by a dedicated system developed by LTSI/INSERM 1099 and then analyzed in regard to the occurence of late onset sepsis with time dependent multidimensional analysis to take in account potential confounders.