Molecular epidemiology and antimicrobial resistance of extended-spectrum beta-lactamases-producing Enterobacter cloacae complex among mothers, neonates, healthcare workers and hospital environments in Tanga, Tanzania

Abstract

Background

The Enterobacter cloacae complex (ECC) is an emerging healthcare-associated pathogen and an increasing cause of infections in neonatal intensive care units. Its clinical relevance is driven by the rapid acquisition of multidrug resistance (MDR), particularly to extended-spectrum β-lactams (ESBLs) and carbapenems. This study investigated the distribution patterns and genomic diversity of MDR ECC in two hospitals in Tanzania.

Methods

Between April 2022 and March 2023, samples from neonates, mothers, healthcare workers and hospital environments were analysed. ECC isolates underwent European Committee on Antimicrobial Susceptibility Testing–guided antimicrobial susceptibility testing and whole-genome sequencing to determine sequence types (STs), resistance determinants, plasmids and phylogenetic relationships.

Results

A total of 132 ESBL-producing ECC isolates were recovered, with the majority originating from environmental sources (65.9%, n/N = 87/132). Neonatal colonization after ≥48 h of hospitalization was 10.2% (n/N = 18/176). Overall, 74.2% (n/N = 98/132) of isolates were MDR, while 10.6% (n/N = 14/132) showed resistance to carbapenems. Whole-genome sequencing of 120 isolates revealed predominance of Enterobacter hormaechei (93.3%, n/N = 112/120) and widespread ESBL genes, mainly bla_CTX-M-15 (87.5%, n/N = 105/120). Carbapenemase genes bla_NDM-1 and bla_NDM-5 were identified in 10% (n/N = 12/120), frequently associated with IncX3 plasmids, while the colistin resistance gene mcr-10.1 was chromosomal. Forty-one distinct STs were detected, including 12 novel STs. ST346 was the most prevalent lineage. Repeated temporal clusters of identical STs, including ST346, were detected across environmental and human isolates.

Conclusions

These findings demonstrate substantial MDR ECC colonization in neonatal wards, suggesting hospital environments as reservoirs for resistant ECC. Temporally clustered isolates across environmental and patient samples underscore the need for strengthened infection prevention and genomic surveillance.

Introduction

Enterobacter cloacae complex (ECC) comprises several closely related Enterobacter species, with Enterobacter hormaechei being the most prevalent in the clinical setting.¹ ECC are common gut colonizers but have emerged as important causes of multidrug-resistant (MDR) healthcare-associated infections worldwide.^2–4 ECC pose a therapeutic challenge due to intrinsic resistance mediated by inducible chromosomal AmpC beta-lactamases, and the acquisition of extended-spectrum beta-lactamase (ESBL) genes, particularly bla_CTX-M-15, further limits the use of third-generation cephalosporins.^5,6 Treatment options are further restricted by rising resistance to carbapenems,⁷ polymyxins mediated by mobile colistin resistance (mcr) genes⁸ and fluoroquinolones.⁹ In addition to antimicrobial resistance (AMR), ECC harbour virulence-associated genes involved in adhesion, iron acquisition and environmental persistence, facilitating survival in hospital environments.^10,11

MDR Enterobacterales, including ECC, are associated with high morbidity and mortality, particularly among neonates and other vulnerable hospitalized populations.^12–14 In sub-Saharan Africa (SSA), increasing detection of MDR ECC in neonatal units is alarming, given the region’s limited infection prevention and control (IPC) measures.^15,16 Colonization plays a critical role in the transmission and subsequent infections, with hospital environments and healthcare workers (HCWs) acting as important reservoirs.^17,18

Molecular epidemiological studies have revealed substantial ECC diversity, with sequence types (STs) such as ST171, ST114 and ST78 associated with healthcare outbreaks and MDR phenotypes.^19,20 However, the epidemiology and transmission dynamics of ECC remain poorly characterized in many African healthcare settings, including Tanzania.

This study investigates the colonization, AMR and molecular epidemiology of ECC among mothers, neonates, HCWs and hospital environments in two hospitals in the Tanga Region of Tanzania to better understand distribution patterns, potentially facilitating transmission, and inform IPC strategies.

Methods

Study design and study area

This cross-sectional hospital-based study was conducted in the obstetric and neonatal wards of Tanga Regional Referral Hospital (TRRH) and Korogwe Township Council Hospital (KTCH), Tanzania. TRRH is a tertiary healthcare facility with a capacity of 600 beds. The facility provides healthcare services to a population of approximately two million people. KTCH is a secondary healthcare facility situated in Korogwe District of the Tanga Region, approximately 100 km from Tanga city. The facility has a capacity of 142 beds and serves a population of approximately 300 000 people.

Ethics

Ethical approval for this study was obtained from the National Institute for Medical Research (NIMR), Tanzania (Reference No: NIMR/HQ/R.8a/Vol.IX/3966). The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. Permission to conduct the study and collect samples within hospital facilities was obtained from the respective hospital administrations.

Sample collection, ECC isolation, identification and antimicrobial susceptibility testing

Pregnant women admitted for delivery or presenting in labour were informed about the study objectives by trained study nurses, and demographic and clinical information was collected from consenting women. Neonates included infants from birth to 28 days of life with gestational age >34 weeks. Neonates with congenital malformations, stillbirth, or preterm were excluded.

Peri-anal samples were collected from mothers at admission (pre-delivery) using sterile swabs moistened with Liquid Amies medium (eSwabs 490C, Copan Diagnostics, USA). Rectal samples were obtained from neonates shortly after birth and again at discharge after ≥48 h of hospitalization. Hospital environmental samples were collected weekly using Amies swabs (Copan Diagnostics Inc.) from frequently touched surfaces, medical devices and HCWs’ hands (entire palm, including fingertips). Sampled surfaces included medication trolleys, bedside rails, keyboards and computer mice, baby warmers, cots, weighing scales, stethoscopes, thermometers, incubators and oximeters. All swabs were transported to the NIMR, Tanga Research Centre and Korogwe Research Station microbiology laboratories. Samples were enriched in 1 mL of Brain Heart Infusion Broth (Oxoid, Basingstoke, UK) and incubated for 18–24 h at 35°C–37°C and then cultured on CHROMagar^TM ESBL (Paris, France) and then further incubated for 18–24 h at 35°C–37°C. Suspected ESBL-producing Enterobacterales were identified using API 20E (bioMérieux, France) and confirmed using the VITEK 2 automate (VITEK GN-ID, bioMérieux). Antimicrobial susceptibility testing (AST) and ESBL confirmation were performed using the VITEK 2 Compact system (AST-N214 cards) and disc diffusion following the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines (Version 12, 2023). MDR was defined as resistance to greater than or equal to three antibiotic classes.²¹ EUCAST clinical breakpoints were used for the interpretation of AST results. Quality control was performed using the ECC strain ATCC 13047.

Antibiotics were classified according to the World Health Organization (WHO) Access, Watch, Reserve (AWaRe) framework using the 2023 classification.²²

Genome sequencing and bioinformatic analysis

Sequencing of bacterial isolates

DNA was extracted from pure colonies using the Master Pure Complete DNA and RNA Purification Kit (Lucigen, USA) as per the manufacturer’s instructions. DNA concentration was determined using a Qubit^TM 4 Fluorometer (Thermo Fisher Scientific, Singapore). Whole-genome sequencing (WGS) was performed at Beijing Genomics Institute (Hong Kong, China) using the DNBSEQ-G400 platform.

Assembly and sequence typing

High-quality reads were assembled in Ridom SeqSphere+ v10²³ using the Strategic k-mer Extension for Scrupulous Assembler v2.3.0.²⁴ Assembly quality was assessed using MASH (MinHash-based genome distance estimation tool) and Benchmarking Universal Single-Copy Orthologs.²⁵ STs were assigned with Ridom SeqSphere+, and novel STs were deposited in PubMLST.²⁶

Phylogenetic analysis

Phylogenetic trees were generated for E. hormaechei genomes using PhyloPhlAn v3.1.1²⁷ with UniRef90 CORE proteins as reference (generated 4 December 2024). Trees were annotated with AMR, virulence and plasmid data and visualized in R version 4.4.2. Phylogenetic groupings are described descriptively based on ST and tree topology, without applying a predefined single-nucleotide polymorphism (SNP) threshold.

Temporal cluster analysis

Temporal and spatial patterns of ECC STs were assessed to identify potential transmission events. A temporal cluster was defined as greater than or equal to two isolates of the same ST recovered from the same ward within 30 days, consistent with reported survival times of Enterobacterales on hospital surfaces.²⁸ Clusters were interpreted as potential persistence or transmission events; directionality could not be determined due to a lack of spatial or clinical linkage data.

Definitions

Colonization was defined as the presence of bacteria on body surfaces without tissue invasion or inflammatory response, whereas infection was defined as bacteria invasion into viable tissue with an inflammatory host response and clinical symptoms.

Data analysis

Data were collected and managed using REDCap (Vanderbilt University, USA) and Excel and analysed using Stata 18.0 (StataCorp, USA) and R 4.2.1 (R Core Team). Categorical variables were summarized as frequencies and percentages, and continuous variables as median and interquartile range (IQR). Differences between ESBL-ECC positive and negative samples were assessed using χ² tests for categorical variables and Kruskal–Wallis tests for continuous variables. Logistic regression was used to identify factors associated with MDR. A P-value of <0.05 was considered statistically significant.

Data availability

Raw sequencing data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject PRJNA1328667 (SRA accessions SRR35403460–SRR35403583). Genome assemblies are available in GenBank (accessions JBRQVH000000000–JBRQZZ000000000).

Results

This study was conducted between May 2022 and March 2023. A total of 3584 samples were taken: (25%, n/N = 899/3584) from pregnant women, (23%, n/N = 826/3584) from neonates at delivery, (5%, n/N = 176/3584) from neonates at discharge, (8%, n/N, 286/3584) from the hands of HCWs and (39%, n/N = 1397/3584) from hospital surfaces including HCWs’ personal devices. Background characteristics of the samples are shown in Table 1. Significantly greater proportions of ESBL-ECC positive samples were observed from neonatal wards and from environmental samples, compared with those from labour wards and mothers and baby pairs, respectively.

Table 1.

Description of sample sources by presence of ESBL-producing E. cloacae complex

	ESBL-ECC positive	Total	P-value
All samples
N	132 (3.7%)	3584 (100.0%)
Hospital
TRRH	84 (4.0%)	2105 (58.7%)	0.244
KTCH	48 (3.2%)	1479 (41.3%)
Ward
Labour	83 (3.0%)	2740 (76.5%)	<0.001
Neonatal	49 (5.8%)	844 (23.5%)
Hospital/ward
TRRH/labour	50 (3.3%)	1532 (42.7%)	0.002
TRRH/neonatal	34 (5.9%)	573 (16.0%)
KTCH/labour	33 (2.7%)	1208 (33.7%)
KTCH/neonatal	15 (5.5%)	271 (7.6%)
Source of sample
Mothers and neonates	45 (2.4%)	1901 (53.0%)	<0.001
Environment	87 (5.2%)	1683 (47.0%)
Mother samples
N	11 (1.2%)	899 (100.0%)
Age (years) (IQR)	27 (22–33)	27 (22–32)	0.793
No. of pregnancies (including current) (IQR)	3 (1–4)	2 (1–3)	0.464
Antibiotic treatment in current pregnancy
No	3 (0.7%)	423 (47.1%)	0.130
Yes	7 (1.5%)	458 (50.9%)
Unknown	1 (5.6%)	18 (2.0%)
Neonate (delivery) samples
N	16 (1.9%)	826 (100.0%)
Birth weight (kg) (IQR)	3.0 (2.8–3.5)	3.0 (2.7–3.4)	0.781
Sex of neonate female	10 (2.6%)	380 (46.0%)	0.181
Caesarean section performed	7 (2.1%)	336 (40.7%)	0.801
Neonate (discharge) samples
N	18 (10.2%)	176 (100.0%)
Birth weight (kg) (IQR)	3.1 (2.7–3.6)	3.0 (2.7–3.4)	0.581
Sex of neonate Female	7 (9.0%)	78 (44.3%)	0.625
Hospital stay >48 h	18 (10.2%)	176 (100.0%)
Caesarean section performed	17 (12.5%)	136 (77.3%)	0.067
Environmental samples
N	87 (5.2%)	1683 (100.0%)
Source of environmental samples
HCW hands	3 (1.0%)	286 (17.0%)	<0.001
HCW devices	1 (0.4%)	271 (16.1%)
Bedside rail	26 (9.4%)	277 (16.5%)
Trolley	14 (8.6%)	162 (9.6%)
Baby warmer/incubator/cot	15 (7.5%)	201 (11.9%)
Computer keyboard/mouse	7 (4.1%)	169 (10.0%)
Weighing scale	6 (6.1%)	98 (5.8%)
Medical device	10 (6.0%)	166 (9.9%)
Obstetrics theatre: any surface	5 (9.4%)	53 (3.1%)

Categorical variable cells are n (%), P-values from χ² tests. Continuous variable cells are median (IQR), P-values from Kruskal–Wallis rank tests.

ECC isolated from mothers, neonates and the hospital environments

ESBL-producing ECC strains were isolated from 132 of 3584 samples (3.7%). Among human samples, 1.2% of mothers (n/N = 11/899), 1.9% of neonates at delivery (n/N = 16/826) and 10.2% of neonates at discharge after >48 h of hospitalization (n/N = 18/176) were positive. Among environmental samples, including HCWs’ hands, 5.2% were positive (n/N = 87/1683). The lowest detection rates were observed on HCWs’ hands (1.0%, n/N = 3/286) and personal devices (0.4%, n/N = 1/271). In contrast, hospital surfaces showed higher positivity, ranging from 4.1% on computer keyboards and mice (n/N = 7/169) to 9.4% on bedside rails (n/N = 26/277) and obstetric theatre surfaces (n/N = 5/53).

AMR profiles of ESBL-producing ECC isolates from human and environmental sources were compared (Figure 1). Resistance in environmental isolates was generally higher or comparable to that in human isolates, reaching statistical significance only for trimethoprim–sulfamethoxazole (χ²  P = 0.017). Resistance to second- and third-generation cephalosporins (ceftazidime, cefpodoxime, cefotaxime and cefuroxime-axetil) exceeded 89.4% (n/N = 118/132–132/132). No resistance was detected to tigecycline, while lower resistance levels were observed for carbapenems (10.6%, n/N = 14/132) and ciprofloxacin (15.9%, n/N = 21/132).

Figure 1.

AMR for ESBL-producing ECC isolated from mothers and neonates, and the hospital environment, sorted by Access and Watch agents. CN, gentamicin; SXT, trimethoprim–sulfamethoxazole; CAZ, ceftazidime; CPD, cefpodoxime; CTX, cefotaxime; CXA, cefuroxime-axetil; CIP, ciprofloxacin; IPM, imipenem; MEM, meropenem; TGC; tigecycline; TZP, piperacillin–tazobactam. Access & Watch, AWaRe categories of antibiotics as per the WHO 2023 classification²²; Access and Watch shown; reserve not tested.

MDR levels across sample sources and wards

Overall, 74.2% (n/N = 98/132) of ESBL-producing ECC isolates were MDR (Figure 2). MDR was more frequent in neonatal wards (81.6%, n/N = 40/49) than in labour wards (69.9%, n/N = 58/83), a pattern observed in both hospitals. Similarly, MDR prevalence was higher in environmental samples (79.3%, n/N = 69/87) than in samples from mothers and neonates (64.4%, n/N = 29/45).

Figure 2.

MDR levels (Levels 3–5) across sample source and ward; percentages based on the total number of ESBL-ECC samples in each source/ward category in both hospitals.

Adjusted logistic regression analysis showed increased odds of MDR in isolates from neonatal wards [odds ratio (OR) = 1.50, 95% confidence interval (CI): 0.58–3.90] and from the hospital environment (OR = 1.79, 95% CI: 0.74–4.34). However, these associations were not statistically significant (Table S2a, available as Supplementary data at JAC-AMR Online).

Species distribution and genome assembly characteristics

Of 132 sequenced genomes, 12 were excluded due to low coverage, contamination, or incomplete assemblies, leaving 120 high-quality ECC genomes. These were dominated by E. hormaechei (n = 112), with E. cloacae (n = 5) and single isolates of Enterobacter hoffmannii, Enterobacter quasihormaechei and Enterobacter vonholyi.

Across the five species, sequencing depth was high (mean coverage ×180 to ×210). Assemblies showed moderate fragmentation (28–189 contigs; N50 78–537 kb), with conserved guanine-cytosine (GC) content (54.7%–55.9%). Genome sizes ranged from 4.5 to 5.1 Mb, with E. quasihormaechei having the smallest and E. cloacae the largest genomes (Table S1).

Distribution of antimicrobial genes

A total of 22 beta-lactamase genes were identified. The bla_CTX-M group was most common (89.2%, n/N = 107/120), dominated by bla_CTX-M-15 (87.5%, n/N = 105/120). Other frequently detected genes included blaTEM-1 (60.8%, n/N = 73/120) and bla_OXA-1 (49.2%, n/N = 59/120). Chromosomal bla_ACT genes were present in 95.8% of isolates. Carbapenem resistance genes were detected in 10.8%, including bla_NDM-1, bla_NDM-5 and bla_OXA-181 (Table S2a).

Other resistance genes

Resistance determinants for multiple antimicrobial classes were identified. Common genes included fosA (91.7%, n/N = 110/120), aminoglycoside resistance genes aph(6)-Id and aph(3'‘)-Ib (68.3%, n/N = 82/120), sulphonamide genes sul2 (69.2%, n/N = 83/120) and fluoroquinolone resistance genes oqxAB (90%, n/N = 108/120) and qnrB1 (74.2%, n/N = 90/120). Additional genes conferring resistance to tetracycline, trimethoprim, colistin (mcr-10.1), macrolides, rifamycin and bleomycin were also detected. Furthermore, the qacE and qacE delta 1 resistance genes for quaternary ammonium compound-related disinfectants were also detected (12.1%, n/N = 15/124) (Table S2b).

Localization of resistance genes

A total of 1492 AMR genes were identified across the genomes. Several genes showed distinct genomic locations: bla_TEM-1 and sulphonamide genes were predominantly plasmid-borne, whereas bla_ACT, bla_CMH and fosA were chromosomal. bla_CTX-M-15 was mainly plasmid-associated, while bla_OXA-1 occurred more frequently on chromosomes (Table S3).

Genotype–phenotype correlation of AMR

All 120 isolates carried β-lactam resistance genes and were also phenotypically resistant (100% concordance). Carbapenem resistance genes were identified in 12 isolates, all phenotypically resistant, although one additional isolate showed carbapenem resistance without a detectable gene. Fluoroquinolone resistance genes were common (118/120), but phenotypic resistance was observed in only 21 isolates (Table 2).

Table 2.

Comparison of genotypic resistance genes and phenotypic resistance

Antibiotic class tested	Resistance genes detected (genotypic)	Isolates carrying resistance genes	Phenotypically resistant among gene carriers	Phenotypically resistant isolates	Genotypically confirmed resistant isolates
Non-carbapenem β-lactams (penicillins and cephalosporins)	bla CTX-M-15 , bla CTX-M-14 , bla TEM-1 , bla OXA-1 , bla SHV-12 , bla ACT , bla CMH	120	120	120	120
Carbapenems	bla NDM-1 , bla NDM-5 , bla OXA-181	12	12	13	12
Aminoglycosides	aac(3)-IId, aac(3)-IIe, aac(6‘)-Ib-cr5 aph(3“)-Ib, aph(6)Id, aadA1, aadA2, aadA16	107	88	75	75
Fluoroquinolones	qnrS1, qnrB1, qnrB4, oqxA/oqxB	118	21	21	21
Sulphonamides + diaminopyrimines	sul1/sul2, dfrA	106	106	120	106

Phenotypically resistant among gene carriers = number of gene-carrying isolates that also showed phenotypic resistance. Genotypically confirmed resistant isolates = number of phenotypically resistant isolates that also had corresponding resistance genes. Comparison done with 120 sequenced isolates.

ST distribution

Among the 120 ECC isolates, 41 STs were identified, including twelve novel STs (10%, n/N = 12/120). ST346 was the most common (25%, n/N = 30/120), followed by ST2713 and ST109 (8.3%, n/N = 10/120 each), ST66 (7.5%, n/N = 9/120) and ST171 (4.1%, n/N = 5/120). ST346 predominated in both environmental (28.6%, n/N = 22/77) and human isolates (18.6%, n/N = 8/43). Other STs were more source-specific, with ST66 and ST171 mainly environmental, while ST2713 was more frequent in human isolates (11.6%, n/N = 5/43) (Table S4).

Phylogenetic analysis and associated genetic profiles of E. hormaechei isolates

Phylogenetic analysis showed that E. hormaechei isolates clustered according to ST, forming distinct phylogenetic groupings with isolates from different sources interspersed within clusters (Figure 3a and b).

Figure 3.

Phylogenetic relationship and genetic profiles of E. hormaechei isolates collected from mothers, neonates, the hospital environment and staff at KTCH (a) and TRRH (b). The dendrogram illustrates the phylogenetic relatedness of isolates, while the presence (black) and absence (white) of key β-lactamase genes, carbapenemase genes, virulence genes and plasmid sequences are displayed in corresponding columns. Isolate sources and STs are indicated by different shapes and colours, respectively.

At KTCH (Figure 3a), ST346 was the dominant lineage (42.9%; n/N = 18/42) and included isolates from all sources, including a closely related mother–infant pair. All ST346 isolates carried the ESBL determinant (bla_CTX-M-15) and consistently harboured virulence genes associated with biofilm formation and iron acquisition (csgG, iroABCD) as well as adhesion and colonization (ompA). Notably, carbapenem resistance genes (bla_NDM-5) were identified in isolates belonging to ST124 and ST200, both associated with IncX3 plasmids.

At TRRH (Figure 3b), a more diverse phylogenetic structure was observed. ST346 was also the dominant lineage (17.1%; n/N = 12/70) and showed a resistance and virulence profile similar to that observed at KTCH, including carriage of the ESBL and virulence-associated genes. Several additional ST-associated clusters displayed distinct resistance profiles. ST66 accounted for 11.4% (n/N = 8/70) of isolates and formed a well-defined cluster characterized by carriage of colistin resistance genes (mcr-10.1), ESBL determinants (bla_CTX-M-15), and virulence-associated genes involved in iron acquisition and adhesion (entB, ompA). Notably, a separate ST3379 cluster carried carbapenem resistance genes (bla_NDM-1) together with resistance determinants spanning eight additional antimicrobial classes (Table S5). ST1518 isolates formed a closely related cluster, with both isolates carrying carbapenem resistance genes (bla_NDM-1) and one additionally harbouring bla_OXA-181. Carbapenem resistance genes were also identified in additional STs, including ST200, ST916 and ST344.

Across both hospitals, ST2713 formed a distinct phylogenetic cluster, comprising 11.9% (n/N = 5/42) of isolates at KTCH and 7.1% (n/N = 5/70) at TRRH. All ST2713 isolates from TRRH carried virulence operons associated with capsule biosynthesis and export (tviBCDE and vexABCDE), whereas three of five corresponding isolates from KTCH harboured these operons.

Temporal clustering of ECC isolates

Temporal patterns of ECC recovery were assessed to identify potential transmission or persistence within wards at TRRH and KTCH (Figure 4). In the labour ward of TRRH, temporal clusters were observed for ST109 (two neonatal isolates) and ST200, which was detected first in a neonate and within 30 days in the environment. In the neonatal ward, clusters involved only environmental isolates (ST66, ST68 and ST114). ST346 also formed four distinct temporal clusters in this ward.

Figure 4.

Temporal distribution of STs in labour and neonatal wards of TRRH and KTCH. Phylogenetically closely related ST clusters occurring within a 30 day time frame (highlighted by ovals) were considered potential transmission events.

At KTCH, a large ST346 cluster in the labour ward involved 14 isolates, including a mother-neonate pair detected on the same day. Another cluster (ST3369) included two environmental isolates. In the neonatal ward, clusters were observed for ST344 (two neonatal isolates) and ST346 (environmental and neonatal isolates detected on the same day). Additional clusters included ST2713, first detected in the environment and within 30 days in a neonate, and ST3369 (both environmental isolates).

Virulence genes and plasmid incompatibility types

Several virulence-associated genes were detected among the 120 ECC isolates (Table S6). The adhesion gene ompA was present in all isolates (100%). Biofilm formation gene csgG (55.8%) and the iron acquisition operon iroBCDEN (42.5%) were the most frequent additional virulence determinants.

Seventeen distinct plasmid incompatibility (Inc) types were identified among the 120 ECC isolates (Table S7). The most frequent were IncR (37.5%), IncFIB(K) (27.5%) and Col440I (27.5%), followed by IncFIA(HI1) (25%) and IncFII(Yp) (20.8%). Other plasmid types, including IncHI2/IncHI2A (14.2%), IncX3 (6.6%) and IncFIB(pECLA) (5.8%), were detected at lower frequencies, while several additional types occurred only sporadically (<5%).

Discussion

This study provides insights into the genetic epidemiology and potential transmission dynamics of ESBL-producing ECC in neonatal and labour wards in two Tanzanian hospitals. We observed frequent colonization of neonates and hospital surfaces, accompanied by substantial genomic diversity, including 12 previously undescribed STs. Clonal clustering across wards, sources and time points suggests persistence and likely intra-hospital transmission of MDR ECC within these healthcare settings.

This study showed that neonatal colonization with ESBL-ECC was low at delivery (1.9%) but increased to 10.2% after 48 h of hospitalization, suggesting hospital-associated acquisition. Similar patterns have been observed in tertiary hospitals in Tanzania and South Africa,^16,29,30 highlighting the likely role of the hospital environment in neonatal colonization. While post-partum colonization with bacteria is a priori a physiological process, it bears a risk for neonates of developing subsequent infections in case of colonization with bacteria showing increased virulence or in case of reduced immunocompetence, including sepsis, which is associated with significant morbidity and mortality.²⁸ If bacteria causing such infections are associated with increased AMR, the therapeutic management of respective infections will become more challenging. Therefore, the detection of increased resistance rates calls for a strengthening of IPC measures in the studied facilities, including hand hygiene, environmental decontamination and targeted surveillance strategies in high-risk neonatal units.³¹

Environmental samples showed a higher occurrence of ESBL-ECC (5.2%) than the human samples (2.4%). This is consistent with previous reports from SSA^32–34 and supports the role of environmental reservoirs in pathogen persistence. Although HCW hand contamination was low (1.0%), previous studies highlight their role in pathogen transmission, underscoring the need for continued reinforcement of hand hygiene and infection prevention practices.^35–37

Interestingly, most neonates colonized at birth had mothers without detectable colonization, suggesting colonization below detection limits or alternative sources of early microbial acquisition. Previous studies link neonatal colonization to delivery mode, gestational age, feeding practices and environmental exposure.^38–40 These findings highlight the importance of rigorous environmental hygiene immediately after delivery.

AMR among ECC isolates was substantial, with most isolates resistant to third-generation cephalosporins due to ESBL production. Notably, 10.8% of isolates carried carbapenem resistance genes, consistent with previous observations in Tanzanian neonatal units.⁴¹ Environmental isolates showed higher MDR rates than human isolates, supporting the role of hospital surfaces as reservoirs for resistant strains, in line with reports from Tanzania, Kenya and Pakistan.^32,33,42 MDR was more frequent in neonatal wards than labour wards, possibly reflecting more intensive antibiotic use in high-risk neonates.^16,43

Genomic analysis revealed widespread resistance determinants across multiple antibiotic classes. The ESBL gene bla_CTX-M-15 and chromosomal ampC-type bla_ACT genes were highly prevalent, consistent with global ECC resistance patterns.⁴⁴ Notably, carbapenemase genes were detected in ECC isolates from both neonatal and environmental samples. While NewDelhi metallo-beta-lactamase (NDM)-type carbapenemases have previously been reported in Enterobacterales from Tanzanian neonatal units,³⁰ our findings demonstrate their occurrence in ECC and in hospital environmental isolates, suggesting the hospital environment as a possible reservoir for dissemination. Moreover, the detection of recently identified colistin resistance gene mcr-10.1 in a subset of isolates represents one of the first reports from Tanzania,⁴⁵ reflecting the potential emergence of resistance to last-resort antibiotics. In addition, resistance genes for aminoglycosides, sulphonamides and fluoroquinolones were widespread, highlighting the MDR nature of the isolates and the potential for horizontal gene transfer and dissemination within hospital environments.

The present study has also demonstrated the presence of qacE and qacE delta 1 genes. These genes are known to reduce susceptibility to quaternary ammonium compounds (QACs), which are commonly used disinfectants in healthcare settings, thereby potentially compromising the IPC measures.⁴⁶

Phenotype–genotype comparison showed full concordance for β-lactams and carbapenems, whereas aminoglycosides and fluoroquinolones displayed discordance, with resistance genes present but lacking phenotypic resistance. This likely reflects low-level plasmid-mediated mechanisms (e.g. qnr, oqx, aminoglycoside-modifying enzymes) that do not exceed clinical breakpoints without additional mutations or expression changes.^47–49 Such a discrepancy can lead to treatment failure as the low-level resistance mechanisms may evolve to induce high-level resistance during the course of treatment; furthermore, the genes may be transmissible, and strains are also able to spread, increasing the risk of AMR.⁵⁰

Molecular typing revealed substantial ST diversity, with ST346 emerging as the most dominant lineage (25%), across both hospitals and sample sources. Its repeated detection on frequently touched surfaces and in patients over time may indicate persistence and potential environmental transmission within hospital settings. Globally, ST346 has been rarely reported,⁵¹ indicating the presence of locally adapted hospital lineages. Notably, all ST346 isolates were MDR and carried ESBL genes, highlighting significant infection control challenges.

Another noteworthy lineage, ST2713, was uniquely associated with the virulence operons, tviBCDE and vexABCDE,⁵² implicated in immune evasion and increased virulence.⁵³ Temporal clustering of ST2713 isolates was also observed, suggesting enhanced pathogenic potential and persistence within the hospital environment.

Interestingly, detection of chromosomal mcr-10.1 exclusively occurred within ST66 isolates, suggesting a lineage-specific association. Although mcr-10.1 has been sporadically reported in ECC globally,⁵⁴ to the best of our knowledge, this study provides the first evidence of its lineage-specific association with ST66 in a Tanzanian hospital setting. These isolates also occurred within temporally clustered groups, highlighting the emergence of colistin resistance in circulating hospital lineages.

We detected carbapenem resistance genes in several ST-defined lineages, including STs previously associated with carbapenemase carriage (ST109, ST121, ST1518, ST124 and ST200), indicating the circulation of lineages known to harbour carbapenem resistance.^7,19 Notably, ST1518 lineage carried both bla_NDM-1 and bla_OXA-181, representing a rare dual-carbapenemase genotype in ECC. Intriguingly, all positive bla_NDM-5 isolates carried IncX3 plasmids, known globally for efficient horizontal transfer of bla_NDM and bla_OXA across Enterobacterales.⁵⁵ A newly identified ST3379 lineage at TRRH carried bla_NDM-1 together with resistance genes to eight additional antibiotic classes, indicating a highly resistant profile and raising concerns about the emergence of such strains in neonatal wards.

This study has several limitations. The WGS analysis has been performed to the level of ST, which is not sufficient to prove cross-contamination from the hospital environment. The use of higher-resolution analyses, such as the SNP, could have provided stronger evidence on potential transmission pathways. However, without clear-cut epidemiological evidence, transmission events remain speculative. In addition, the study was conducted in only two hospitals in one region of Tanzania, which may limit generalizability. Finally, colonized neonates were not followed to assess subsequent infections such as late-onset sepsis. Future studies should evaluate the clinical impact of intestinal colonization with resistant ECC and determine whether these strains contribute to infection or persist primarily as commensal colonizers.

Conclusions

This study provides molecular and epidemiological characterization of MDR ECC colonization across mothers, neonates, HCWs and hospital environments in two hospitals in Tanga, Tanzania. We identified 12 novel ECC STs, indicating substantial genomic diversification. The dominance and persistence of ST346 among patients and frequently touched environmental surfaces highlight the role of the hospital environment as a possible reservoir facilitating potential transmission events. Detection of ESBL and carbapenem resistance genes, including bla_CTX-M-15 and bla_NDM, as well as the colistin resistance gene mcr-10.1 within temporally and phylogenetically related isolates, underscores the risk of intra-hospital spread of MDR and extensively drug-resistant strains. These findings emphasize the need for strengthened IPC measures, targeted environmental decontamination and sustained genomic surveillance to limit the dissemination of MDR ECC in neonatal care settings.

Funding

This study was supported by the German Federal Ministry of Health (Bundesministerium fur Gesundheit, BMG) through the Global Health Protection Programme (GHPP) (grant nos. ZMI1-2520GHP705 and ZMI5-2523GHP005) as part of the TRINEO study.

Transparency declarations

Not applicable.

Author contributions

All authors read and approved the final manuscript. Conceptualization: A.D.M., R.S. and D.D.; methodology: A.D.M., R.S., N.A.K., H.F., D.S., R.K. and D.D.; investigation: A.D.M., L.B., J.R.M. and A.M.; formal analysis: R.K., D.S., N.A.K. and D.H.C.; data curation: A.J. and D.S.; visualization: R.K., D.S., N.A.K. and H.F.; writing—original draft: N.A.K. and A.D.M.; writing—review & editing: N.A.K, A.D.M, R.S., D.D., D.S., S.S., H.F., J.M. and M.S.; supervision: D.D. and J.P.A.L.; project administration: A.D.M., N.A.K., R.S., J.P.A.L., M.C. and M.S.; funding acquisition: D.D. and J.M.

Supplementary data

Tables S1–S7 are available as Supplementary data at JAC-AMR Online.