435 - Risk factors for antibiotic resistant bacterial colonisation in European Neonatal Units – results of the NeoIPC feasibility study

Publication Number: 2423.435

Chiara Minotti, University Children's Hospital Basel, Basel, Basel-Stadt, Switzerland; Aislinn Cook, City St George's, University of London, London, England, United Kingdom; Matilda Berkell, University of Antwerp, Antwerpen, Antwerpen, Belgium; Surbhi Malhotra, Universiteit Antwerpen, Wilrijk, Antwerpen, Belgium; Julia A. Bielicki, City St George's, University of London, Basel, Basel-Stadt, Switzerland

Background: Infants on neonatal units are at risk of acquiring bacterial infection, often preceded by colonisation. It is important to understand how individual- and unit-level factors contribute to the risk of colonisation, which drives ongoing transmission. This may inform infection prevention and control (IPC) strategies focusing on direct and indirect effects on admitted infants.

Objective: To assess the individual- and unit-level risk factors associated with infant gut colonisation by antibiotic-resistance genes .

Design/Methods: 4 cross-sectional surveys were conducted in 1 month across 24 neonatal units in 8 European Countries (EU-funded NeoIPC feasibility study, preceding NeoDeco trial). Stool samples were collected at each survey from all infants on the unit and were analysed by RT-qPCR for the presence of target antibiotic-resistance genes (Figure 1). Colonisation was defined as detection of at least one resistance gene in stool. A Bayesian logistic regression model was used to assess individual- and unit-level risk factors.

Results: 752 infants contributed data for 1338 infant-survey observations. Median gestational age was 32 weeks(IQR:28-36 weeks), median birthweight was 1690g(IQR:998-2600g) and median length of stay was 15 days(IQR:6-33 days). 22% (289/1338) of stool samples were positive for any antibiotic-resistance gene (ARG), of which the most common were ESBL; both preterm and term infants were colonised (Figure 2). There were high levels of intra-unit (SD: 1.9, 95%CrI: 1.04-3.14) and intra-infant (SD: 2.42, 95%CrI: 1.78-3.2) variation in infant resistant bacterial colonisation. Colonisation pressure (0.39, 95%CrI: 0.16-0.63) and length of stay (0.15, 95%CrI: 0.06-0.25) were associated with increased log-odds of colonisation. Higher gestational age had an 86% probability to be associated with a slight increase in log-odds of infant colonisation. Receipt of TPN was associated with a decrease in log-odds of colonisation(-1.25, 95%CrI:-2.27,-0.29). Previous surgery, increasing proportion of very preterm infants on the unit and presence of a central line had a >85% predictive probability of being negatively associated with colonisation. There was no impact of previous antibiotic exposure (Figure 3).

Conclusion(s): Colonisation pressure and length of stay on neonatal units were risk factors for individual colonisation. Receipt of TPN and clinical features associated with severe disease were associated with lower odds of colonisation. Unit-level IPC interventions are important to target all admitted infants, due to their direct and indirect effects on preventing colonisation and transmission.

Figure 1. Gene targets and assays


Figure 2. Stool resistant bacterial colonisation status of infants per survey for each neonatal unit showing colonisation by gestational age ( <32 weeks gestational age and >= 32 weeks gestational age).


Model results from primary analysis of infant stool colonisation model on log-odds scale.
A) Point estimates and 80% credible interval (thicker band) and 95% credible interval (thinner band) of posterior draws of model fit for fixed effect parameters. B) Point estimates and 80% credible interval (thicker band) and 95% credible interval (thinner band) of posterior draws of unit-level random effects log-odds deviation from the global intercept.