This was a prospective study conducted in a newborn unit at KNH between January and April 2019. Rectal swabs from newborns, vaginal swabs from mothers and environmental samples from surfaces and medical equipment (on study days d55-57 and d89) were taken over a period of four months. These samples were analyzed for the presence of MDRO (Department of Medical Microbiology, University of Nairobi (UoN) and the Department of Laboratory Medicine, KNH). For a subset of MDRO, whole-genome-sequencing was performed, and phylogenetic relatedness of the bacterial isolates was assessed (Institute of Medical Microbiology and Infection Control, University Hospital Frankfurt am Main, Germany). Clusters were analyzed by integrated metadata analysis (sampling, location within the hospital).

In this study, 300 mother-newborn pairs over a period of 110 days were included. Inclusion criteria were (i) admission to the NBU and (ii) given informed consent. Exclusion criteria was any given medical or ethical contradiction to rectal swabs of newborns or vaginal swabs of mothers. Ethical approval was given by KNH – UoN Ethics & Research Committee (KNH/UoN-ERC: P208/04/2018; University of Nairobi, Kenya, College of Health Sciences, July 11th, 2018) and by the Ethics Committee of the Medical Faculty Goethe University Frankfurt am Main, Germany (FKZ 01KA1772; 15/05/2018).

At KNH NBU, 200 to 300 newborns per month receive medical care. NBU is organized into nine sub-units. At admission, newborns are examined in the admission room. Medical care is provided in newborn intensive care units (NICU1, NICU2, NICU3), in nurseries (nursery B1, nursery B2, nursery B3) or in an isolation room. Nursery D is reserved for newborns with improved health condition. The delivery ward is separated from NBU by two floors. Mothers stay in different post-natal wards and visit the NBU every three hours to (breast-)feed their babies.

Vaginal swabs (MK Plast, New Delhi, India) were collected from mothers once at the day their newborns were admitted (according to the ethics proposal KNH/UoN-ERC: P208/04/2018). Rectal swabs of newborns were taken at the day of admission, weekly during their stay at NBU and at discharge from the NBU. Due to ethical reasons, no sample was taken from deceased newborns. Environmental samples were categorized as (i) medical devices (e.g., ventilators, ultrasound transducer), (ii) near patient (e.g., cots, incubators), (iii) far from patient (e.g., desks, computer equipment) or (iv) unclean areas (e.g., surfaces, sinks), and were taken on d55-57 and d89.

Bacterial cultures were incubated for 24 hours at 35-37°C on selective chromogenic ESBL agar (CHROMagar, Mast, Paris, France). Identification (ID) and antimicrobial susceptibility testing (AST) was done via VITEK-2 (bioMérieux SA, Marcy-l’Étoile, France) using GN83 and P580 cards and imipenem E-test strips (Liofilchem, Roseto degli Abruzzi, Italy) according to Clinical Laboratory Standards Institute (CLSI) guidelines (Clinical and Laboratory Standards Institute (CLSI), 2019). Each ID and AST included a purity control on Columbia blood agar. All MDRO isolates were stored at -80°C in CRYOBANK™ medium (Mast).

Due to the agreements of the ethics proposal (KNH/UoN-ERC: P208/04/2018), 160 bacterial isolates (including 51/63 CRO) were selected from 288 detected MDRO for whole genome sequencing (WGS) prioritized by the following criteria: (i) carbapenem-resistant phenotype, (ii) culturable bacterial status upon arrival in Germany and (iii) likeliness of transmission onto or among newborns. Isolates were shipped on dry ice in CRYOBANK medium to Frankfurt am Main, Germany and were phenotypically re-assessed upon arrival using routine microbiology methods. Identification and AST were confirmed using Vitek-2, ID MALDI-ToF MS (bioMérieux SA, Nürtingen, Germany) according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines version 8.0 (accessible via https://www.eucast.org/clinical_breakpoints/). Lateral flow assays (Hardy, Santa Maria, USA) were used to detect the following carbapenemases: NDM, KPC, OXA-48, VIM and IMP. All laboratory testing was performed under strict quality control criteria (laboratory accreditation according to ISO 15189:2011 standards) at the Institute for Medical Microbiology and Infection Control, University Hospital Frankfurt am Main, Germany. Isolates with inconsistent AST, unclear documented origin and copy strains (meaning that the same pathogen was detected in the same newborn multiple times) were excluded from further analysis.

DNA of cultured bacteria was extracted using DNeasy UltraClean 96 Kit (Qiagen, Venlo, Netherlands). Library preparation and sequencing was performed by a commercial service provider (Novogene, Cambridge, UK) using Illumina chemistry. Sequencing was carried out on a NovaSeq 6000 flow cell using a paired-end sequencing strategy of 2x150 bp. Details for in silico-sequence analysis is described in the Supplementary Information .

Research Electronic Data Capture software (REDCap, Vanderbilt University, Nashville, USA) was used to capture and store sample metadata (e.g., sample type, sampling date and time, location) and resistance phenotype (including ID, AST and subcultures). Confidence intervals (CI) were calculated using Newcombe-Wilson (Newcombe, 1998) method and the Relative Risk (RR) was determined using the Armitage -Berry Methods (Armitage et al., 2008).

A total of 300 mother-newborn pairs were included in this study. The median age of the mothers was 27 years, and the range of newborn gestation age was between 25 weeks+6 days to 42 weeks+0 days. These newborns were admitted to the NBU at the day of delivery (or immediately after referral from other hospitals).

In total, 1,097 swabs were obtained, including 164 environmental samples ( Figure 1A ). Among the 288 detected MDRO, the most frequent species was K. pneumoniae (n=155) followed by E. coli (n=83) and A. baumannii (n=7). Multidrug-resistant P. aeruginosa isolates were not detected in any sample. Furthermore, 63 bacterial isolates were identified as CRO (K. pneumoniae: n=35; E. coli: n=13; A. baumannii: n=5).

At admission to NBU (after delivery or after referral from another hospital), MDRO were detected from 16% of newborns (n=48/300; 16%; CI 12-21%), including 3% CRO (n=8/300; 3%; CI 1-5%). Among mothers, a 15% MDRO rate (n=45/300; 15%; CI 11-19%), including 2% CRO (n=7/300; 2%; CI 1-5%) was observed. The rates for mothers and newborns at admission were similar (MDRO: RR 0.94; CI 0.65-1.36; CRO: RR 0.88; CI 0.32-2.38). For newborns, the rate of MDRO increased from admission to discharge from 16% to 44% (n=97/218; 44%; CI 38-51%), and for CRO from 3% to 14% (n=29/218; 14%; CI 9-18%) indicating that 49 newborns became colonized with MDRO and of these, 21 newborns with CRO. This represents a three-fold increase of MDRO and a five-fold increase of CRO (MDRO: RR 2.78; CI 2.06-3.75; CRO: RR 4.99 CI 2.33-10.70). The highest increase was observed in the first week after admission to NBU (see Figure 1B).

MDRO and CRO isolates were obtained from medical devices (n=9), unclean areas (n=6) and near patient areas (n=3). Rates in NBU-environmental samples were 11% (n=18/164; 11%; CI 7-17%) for MDRO and 2% (n=3/164; 2%; CI 1-5%) for CRO.

Of all sequenced Enterobacteriaceae, 89% (n=137/154) harbored an extended spectrum β‐lactamase (ESBL) type bla _CTX-M-15. For 42/51 sequenced CRO, a bla _NDM-type carbapenemase gene was identified. These were distributed among K. pneumoniae (n=25; bla _NDM-1: n=14, bla _NDM-5: n=10, bla _NDM-7: n=1), E. coli (n=11; bla _NDM-5: n=6, bla _NDM-7: n=5) and other Enterobacteriaceae (n=6; all harboring bla _NDM-1). For all four A. baumannii isolates, a bla _OXA‐23 carbapenemase was detected; one of those additionally harbored bla _OXA‐66, two others bla _OXA‐69 and one bla _OXA‐365. Other detected carbapenemases include bla _OXA‐232 and bla _OXA‐181 (each found in one K. pneumoniae isolate, respectively). Detailed information regarding sequence type and antimicrobial resistance genes is given in Supplementary Table 1 .

A phylogenetic analysis of the 160 selected isolates was carried out. Copy strains (n=17) were excluded once confirmed by sequence analysis. Results revealed 20 clusters of closely related isolates, including five CRO clusters. Of these, 19 were formed by K. pneumoniae (CRO clusters: n=4) and one cluster was formed by E. coli (see Figure 2 and Supplementary Table 2 ).

Among these clusters, cluster I and II as well as cluster VI, VII and VIII consist of isolates of the same sequence type. The median difference between isolates of cluster VI and VII is 132 SNVs (min: 130; max: 133), of cluster VI and VIII 182 SNVs (min: 175; max: 184) and of cluster VII and VIII 191 SNVs (min: 183; max: 193), respectively. Similarly, cluster I and II both belong to ST39, with a median difference between the isolates of 2,380 SNVs (min: 2,380, max: 2,384). These results demonstrate that these clusters VI, VII and VIII are distinguishable within ST 17 and clusters I and II within ST 39.

Cluster I, which consist of K. pneumoniae ST39 with bla _CTX-M-15, represents the largest cluster (n=15). The isolates spanned the complete investigated period (d14 until d106) and were obtained from 15 neonates in seven of the nine NBU-subunits (except isolation room and NICU3).

Cluster VIII, formed by K. pneumoniae isolates of ST17 carrying bla _NDM-5, was the largest CRO cluster ( Figure 3 ). The respective isolates derived from nine different newborns and were obtained from three different subunits ( Supplementary Figure 1 ). The initial isolate was sampled on Nursery B3 on d46. Six isolates were detected in Nursery D (on d74 (n=2), d82, d85 (n=2) and d96) and two in NICU2 on d90 and d95, respectively.

Clusters indicating transmissions among mothers and newborns were rarely found (only cluster XIII and XIV). Only in three cases, bacteria of the same species (K. pneumoniae with MDRO status) were detected in mothers and their respective newborns but none of these bacteria were phylogenetically closely related (pairwise SNP distance: 16,780, 16,583 and >133k SNPs, respectively) excluding vertical transmission from mother to child. Clusters consisting of isolates obtained from the NBU-environment and among newborn samples (cluster IX) as well as clusters consisting exclusively of samples from mothers (cluster XIX) were detected. Besides the already mentioned ST39 (n=26; cluster I: n=15, cluster II: n=9; no cluster: n=2), ST17 (n=17; including the NDM-5-positive cluster VIII: n=9) and ST348 (n=13; no carbapenemase detected) were the most frequently found K. pneumoniae sequence types.

Plasmid MLST analysis and genomic assessment of those regions flanking carbapenemase genes indicate that transmission of bacteria rather than plasmid hospitalism is the dominant mechanism for the spread of carbapenemases and the occurrence of CRO (see Supplementary Figure 2, Supplementary Figure 3 ).