Transmission of extended-spectrum β-lactamase-producing Klebsiella pneumoniae in a Malawian neonatal unit: a clinical and genomic analysis of a prospective cohort study

Background Klebsiella pneumoniae is an important cause of neonatal sepsis in sub-Saharan Africa and is frequently antimicrobial resistant. In this study, we investigated key transmission routes of extended-spectrum β-lactamase-producing K pneumoniae (ESBL- K pneumoniae ) within a single neonatal unit to inform development of contextually appropriate infection prevention and control practices. Methods We recruited a prospective mother and neonate cohort of babies admitted for less than 48 h and expected to stay for 24 h or more, where the mother or guardian was aged 18 years or older. Participants were followed up until day 7 of admission or discharge, with samples taken on day 1 and 3, and at the final follow-up (maternal stool, neonatal stool, maternal hands, cots, and swaddling cloth [day 1 only]). We sampled the ward environment weekly and included contemporaneous invasive ESBL- K pneumoniae isolates in our analysis. The primary outcome was ESBL- K pneumoniae colonisation . K pneumoniae single-colony isolates were subjected to single-colony whole-genome sequencing, and plate-sweeps had post-enrichment metagenomics performed, before evaluation of multilocus sequence types and single-nucleotide polymorphisms. Statistical analyses (Kaplan–Meier, state-transition models, source-attribution models, transmission analysis combining single-nucleotide polymorphisms and timing of sample collection, and exponential random graph models) were used to infer whether transmission occurred and from which sources. Findings We recruited 94 neonate–mother pairs (from July 23, 2019, to March 30, 2020). Median gestational age was 37 (IQR 35–38) weeks, 41 (44%) of the 94 neonates were girls, all were of Black ethnicity, and 69 (73%) of 94 neonates were antibiotic exposed. ESBL- K pneumoniae rapidly colonised neonates (Kaplan–Meier analyses indicated median 5 days [95% CI 4–6]), with 44 (49%) of 90 of neonates being colonised at least once. State-transition model indicated females had increased colonisation risk (hazard ratio 2·33 [95% credible interval 1·07–5·40]). Invasive isolates (13 [76%] of 17) clustered with stool and environmental isolates. Results of transmission analysis using single-colony whole-genome sequencing and post-enrichment metagenomics were consistent with ESBL- K pneumoniae transmission from cots (13–29%), ward surfaces (37–65%; especially sinks and oxygen delivery equipment), mother's hands (6–13%), and other neonates (8–13%). The results of network analysis (exponential random graph models) were consistent with shared cots increasing transmission risk from 5·7 × 10 −5 (95% CI 1·4 × 10 −5 to 2·3 × 10 −4 ) to 5·9 × 10 −4 (3·1 × 10 −4 to 1·1 × 10 −3 ). Interpretation The hospital environment is implicated in colonisation and invasive infection with ESBL- K pneumoniae . Infection prevention and control interventions should focus on containing neonatal stool, maternal hand hygiene, cot decontamination, single-use oxygen delivery equipment, and surface cleaning, particularly sinks. Funding Medical Research Council (UK); National Institute for Health Research (UK); Gates Foundation (USA); Wellcome (UK).