High gastrointestinal carriage rates of extended-spectrum-β-lactamase-producing enterobacterales and associated factors among hospitalized and nonhospitalized children in Kenya

Abstract

Gastrointestinal carriage of extended-spectrum β-lactamase-producing Enterobacterales (ESBL-E) presents a critical public health threat globally. However, in resource-constrained countries with poor sanitation, inadequate drinking water, and limited microbiology laboratories like Kenya, epidemiological data of these strains is limited. This study assessed the gastrointestinal carriage of ESBL-E and the risk factors for colonization among children (≤ 5 years) in the inpatient department (IPD) and outpatient department (OPD). This was a hospital-based cross-sectional study at Thika Level 5 Hospital, Kenya, from February to June 2023. In total, 540 participants (OPD: 270, IPD: 270) were recruited, using systematic random sampling and consecutive sampling in OPD and IPD, respectively. Children admitted for less than 48 h in the paediatrics ward and those with a prior history of hospitalization (≤ 3 months) in OPD were excluded. Demographic data were collected using a well-structured questionnaire. Following the standard microbiology methods, stool or rectal swab samples were cultured, with the identity and antimicrobial susceptibility of isolates elucidated by automated platforms. The overall ESBL-E gastrointestinal carriage rate was 35.4% (191/540), and was highest among outpatients at 40.4% (109/270). Isolates demonstrated co-resistance to aminoglycosides (43–52%), quinolones (52–62%), carbapenems (44–50%), and sulfonamides (92–97%). They were more susceptible to piperacillin/tazobactam (67–95%) and colistin (96–99%). Carbapenemase-producing Enterobacterales (CPE) co-carriage rate was 17.6% (16/91), with similar rates for inpatients (50%, 8/16) and outpatients (50%, 8/16). Escherichia coli was the predominant ESBL-E overall (82.2%, 157/191), among outpatients (83.5%, 91/109), and inpatients (80.5%, 66/82), and was also the main CPE (overall: 81.3%, 13/16; OPD: 75%, 6/8; IPD: 87.5%, 7/8). Independent predictors of colonization included child age (adjusted odds ratio (OR): 1.60, p = 0.045) and a history of antimicrobial use from retail pharmacies without a clinician’s prescription (adjusted OR: 0.18, p = 0.047). This study demonstrates a substantial burden of gastrointestinal carriage of ESBL-E and CPE co-carriage among children (≤ 5 years), with E. coli being the predominant organism. Age less than two years and a history of exposure to non-prescribed antimicrobials were independent factors for colonization. Efforts to limit exposure to contaminated environments and targeted antimicrobial stewardship initiatives are required to mitigate AMR in the current study setting.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-43265-6.

Background

Extended-spectrum β-lactamase (ESBL)-producing Enterobacterales (ESBL-E) pose a severe global health threat. The World Health Organization (WHO) designated these strains as “critical priority pathogens,” highlighting the urgent need for new antimicrobial agents and effective surveillance to prevent their spread in clinical settings¹. ESBLs are potent, plasmid-encoded enzymes that inactivate β-lactam antimicrobials with oxyimino groups, such as oxyimino-cephalosporins and oxyimino-monobactams, but not cephamycins and carbapenems². Infections caused by Enterobacterales that produce ESBLs are associated with severe clinical outcomes, increased mortality rates, prolonged hospitalization, and higher healthcare costs^3,4.

The global gastrointestinal carriage of ESBL-E is around 14%, with variation ranging from 2% to 46% among individuals based on geographical regions⁵. A high carriage of 59% and 50.4% of ESBL-E in children residing in rural Africa, characterized by high poverty and suboptimal hygienic conditions, is documented^6,7. Plasmid-borne-ESBL genes are frequently transmitted, facilitating swift spread among bacterial populations in hospital and community settings⁸. ESBL genetic determinants are often found in mobile genetic elements, which carry genes that confer resistance to many antimicrobials, such as quinolones, aminoglycosides, tetracyclines, and trimethoprim/sulfamethoxazole. Consequently, ESBL-E frequently exhibit multidrug resistance phenotypes⁹.

ESBL-E causes septicemia, intra-abdominal abscesses, urinary tract infections (UTIs), brain abscesses, and pneumonia, especially those acquired within hospital environments¹⁰. Infections caused by bacteria carrying ESBLs are treated using carbapenem antimicrobials; however, the widespread use of this antimicrobial class has expedited the spread of another significant group of β-lactamases known as carbapenemases. Improper and irrational utilization of antimicrobial drugs, inadequate sanitation, and suboptimal infection control practices, especially in developing countries, create favourable conditions for the emergence and dissemination of resistant microorganisms¹¹.

Enterobacterales naturally resides in the intestinal tracts of humans and animals; therefore, the gastrointestinal tract is the primary reservoir for ESBLs, allowing horizontal exchange among colonizing bacterial populations¹². The colonized individuals are at significant risk for subsequent infections and disseminating these ‘superbugs’ through environmental faecal contamination¹². In Kenya, as in many other resource-constrained countries, epidemiological data on ESBL-E faecal carriage is limited. This study investigated the gastrointestinal carriage rates and risk factors for ESBL-E colonization among children under five years seeking treatment in the outpatient department and those hospitalized at Thika Level 5 Hospital in Kenya.

Materials and methods

Study setting, design, and population

This study was conducted at Thika Level 5 Hospital, a 467-bed public health facility with approximately 800 to 1000 daily outpatient workload. The facility serves a large population catchment area, including Kiambu, Nairobi City, Muranga, Kirinyaga and Machakos counties.

A cross-sectional study design was adopted among children (≤ 5 years) admitted to the hospital’s pediatric ward and those seeking healthcare services in the outpatient department (OPD). A total of 540 children, equally drawn from inpatient (hospitalized for at least 48 h) and outpatient departments were recruited between February and June 2023. A systematic random sampling method was utilized in the OPD, where hospital records indicated that approximately 3,000 children (≤ 5 years) were anticipated during the 5-month study period. Therefore, every 11th child seeking treatment at the OPD was eligible for enrollment. In the inpatient department, consecutive sampling was adopted due to low number of children admitted (≥ 48 h) in the pediatric ward during the study period.

The study excluded children admitted for less than 48 h and outpatients with a prior history of hospitalization within three months of sampling to avoid those with healthcare-associated ESBL-E colonization. Informed consent was obtained from the parent or guardian of each participating child. Those who declined to give informed consent were excluded from the study.

Samples collection

A structured questionnaire was administered during face-to-face interviews to obtain demographic and clinical data from participants. Collected information included the child’s age, gender, history, and sources of antimicrobial use, drinking water sources, and clinical presentation at the time of sample collection. Diarrhoea was defined as the passage of three or more loose or liquid stools per day¹³. Data regarding antimicrobial use within the previous three months, drinking water sources, sources of antimicrobials, and dose completion behaviors were obtained from parents or guardians.

Stool samples were collected in clean, dry, wide-necked containers. For a child unable to provide a stool sample, a rectal swab was used as an alternative. Stool was prioritized as the preferred sample type due to its higher bacterial yield and reliability in detecting intestinal carriage. Rectal swabs served as a substitute when the selected child was unable to provide a stool sample; thus, maximizing participant inclusion and minimizing data loss. Eleven per cent (58/540) of the study samples were stool, while 89.3% (482/540) were swabs. All samples were placed in Cary-Blair transport medium (Hopebio, Qingdao, China) and transported in a cool box to the hospital’s microbiology laboratory for processing within two hours.

Screening for ESBL-E gastrointestinal carriage

Stool samples and rectal swabs were inoculated onto MacConkey agar (Oxoid Limited, Basingstoke, Hampshire, UK) supplemented with ceftazidime (2 mg/mL) as previously described by Reddy and colleagues¹⁴, and incubated aerobically at 37 °C for 18 to 24 h. The resultant colonies were screened for ESBL production using the Double Disk Test. A bacterial suspension equivalent to 0.5 McFarland-equivalent standard was prepared and inoculated on Mueller-Hinton Agar (MHA), air-dried for 3 min, added cefotaxime (30 µg), ceftazidime (30 µg), and amoxicillin/clavulanic acid (20 µg/10 µg) discs at a 30 mm radius to radius distance. The plates were incubated overnight in ambient air at 37 ℃. Klebsiella pneumoniae ( American Culture Type Collection (ATCC) 700603 and Escherichia coli ATCC 25922 were used as quality control organisms. ESBL-producing isolates showed an inhibition zone surrounding the cefotaxime and/or ceftazidime that increased towards the β-lactam inhibitor¹⁵.

Confirmation of ESBL production among isolates

The Phenotypic Confirmatory Disk Diffusion Test was used to confirm the ESBL-E following the 2023 Clinical and Laboratory Standards Institute (CLSI) guidelines¹⁶. Bacterial suspensions equivalent to 0.5 McFarland standard were prepared and inoculated on MHA. Ceftazidime (30 µg) disk alone and in combination with clavulanic acid (30/10 µg), as well as cefotaxime (30 µg) disk alone and in combination with clavulanic acid (30/10 µg), were added, and the plates were incubated in ambient air for 16–18 h at 37 °C. Isolates that exhibited an increase of ≥ 5 mm in the zone of inhibition for either antimicrobial agent tested in combination with clavulanic acid compared to the zone diameter of the agent when tested alone were considered and confirmed as ESBL producers. This study used E. coli ATCC 25,922 and K. pneumoniae ATCC 700,603 as the negative and positive control organisms.

Isolates identification and antimicrobial susceptibility testing

The study identified ESBL-E isolates using the matrix-assisted laser desorption ionization-time of flight mass spectrometry ((MALDI-TOF MS) (Bruker Daltonik GmbH, Bremen, Germany)) and tested the isolates for antimicrobial susceptibility using VITEK2 (bioMK2 bi S.A., NS. A, Nu, Germany). CLSI (2023) guidelines informed the choice and interpretation of antimicrobial tested, including cefepime, ceftriaxone, cefuroxime, ceftazidime, cefotaxime, meropenem, ciprofloxacin, gentamicin, ampicillin, colistin, trimethoprim-sulfamethoxazole, piperacillin-tazobactam, and amikacin. Pseudomonas aeruginosa (ATCC 27853) and E. coli (ATCC 25922) served as the reference organisms for quality control. Multidrug resistance was defined as resistance to at least one antimicrobial in at least three classes¹⁷.

Colistin susceptibility testing

Colistin susceptibility was tested using broth microdilution testing as recommended by the CLSI 2023¹⁶. Tubes containing calcium magnesium Muller Hinton Broth for each isolate were labelled 1,2, 4 µg/ml and 0 µg/ml (control), corresponding to colistin discs added. The tubes were gently vortexed for 30 min to allow the colistin to elute from the disc. A 0.5 MacFarland standard suspension for the test organism was prepared, 50 µl added to the control tube (0 µg/ml) and the tubes labelled 1,2 and 4 µg/ml. The original inoculum was inoculated on a blood agar plate using a ten¹⁰ µl loop for purity check. The plates and tubes were incubated at 35 °C for 16–20 h. The purity plates were examined to ensure the inoculum was pure with the control tube (0 µg/ml) checked for turbidity, demonstrating the test validity. The minimum inhibition concentration (MIC) of ≤ 2 µg/ml was interpreted as intermediate, whereas MIC ≥ 4 µg/ml was resistant as per the CLSI, 2023¹⁶. The CLSI guideline only provides the “intermediate” and “resistance” but not “susceptible” interpretation for colistin susceptibility testing because the breakpoints are challenging to define due to variability in testing methods and clinical outcomes. Further, the lack of a clear “susceptible” category prevents misleading interpretations and ensures cautious use of colistin. E. coli ATCC 25,922 and P. aeruginosa ATCC 27,853 were used as control organisms.

Screening for carbapenemase production among carbapenem-resistant ESBL-E isolates

Carbapenem-resistant ESBL-E were screened for carbapenemase production using the modified carbapenem inactivation method (mCIM)¹⁸. Each isolate was inoculated in 2 ml of Trypticase Soy Broth (TSB) (HiMedia, India) that contained 10 µg of meropenem, and the cultures were incubated at 30–35 °C for 4 h. After incubation, a suspension of E. coli (ATCC 25922) was prepared in sterile saline to achieve a suspension density equivalent to the 0.5 McFarland standard. The suspension was then spread onto Mueller-Hinton agar (MHA) and the meropenem disk was retrieved from the TSB culture containing the test organism and placed on the agar. The plates were incubated for 18–24 h at 35 °C. Results were interpreted based on the inhibition zone diameters as outlined by the Clinical and Laboratory Standards Institute (CLSI)¹⁶. An inhibition zone diameter of 6–15 mm was considered a positive result, while a diameter of ≥ 19 mm was classified as negative. For quality control purposes, E. coli (ATCC 25922) served as the negative control, and K. pneumoniae (ATCC 1705) served as the positive control.

Statistical analysis

The Stata version 13 (STATA Corporation, College Station, TX, USA) was used for data analysis. Categorical data is presented in frequencies and percentages, whereas continuous data in means and medians in tables. The spectrum of ESBL isolates is illustrated in the figures, while the antimicrobial susceptibility profiles are presented in tables. Binary logistic regression (BLR) was used to determine association between gastrointestinal carriage of ESBL-E and patient type (inpatient vs. outpatient). Additionally, BLR was conducted to determine the association between the gastrointestinal carriage of ESBL-E and the demographic and clinical characteristics of the participants. Associations with p-values of ≤ 0.2 were further analyzed using multivariate regression to assess the independence of these associations. Statistically significant results were defined as having a p-value of ≤ 0.05 within a 95% confidence interval (95% CI). Carbapenem resistance was defined as resistance to meropenem (≥ 4 µg/ml).

Results

Demographic and clinical characteristics of study participants

This study sampled 540 children (≤ 5 years), equally drawn from outpatient department (OPD: 50%, 270/540) and inpatient departments (IPD: 50%, 270/540). The majority of the participants were males (51.9%, 280/540) and aged ≤ 24 months (75%, 405/540). Only 25.7% (139/540) of the participants presented with diarrhoea, Table 1. Most of the participants had a history of antimicrobial use with 3 months of this study (62.4%, 338/540) and relied on municipal water for domestic use (91.7%, 495/540).

Table 1.

Demographic and clinical characteristics of inpatient and outpatient children (≤ 5 years) at Thika Level 5 Hospital, February – June 2023.

Variable	Category	All participants N = 540 (%)	Inpatients, n = 270 (%)	Outpatients, n = 270 (%)
Age {IQR: 12(6–24)} months
	≤ 24 months	405(75.0)	207(76.7)	198(73.3)
	> 24 months	135(25.0)	63(23.3)	72(26.7)
Gender
	Male	280(51.9)	136(50.4)	144(53.3)
	Female	260(48.1)	134(49.6)	126(46.7)
Clinical presentation
	Vomiting	148(27.4)	49(18.1)	99(36.7)
	Chills	16(3.0)	3(1.1)	13(4.8)
	Diarrhoea	139(25.7)	42(15.6)	97(35.9)
	Headache	5(0.9)	3(1.1)	2(0.7)
	Respiratory illness	220(40.7)	76(28.1)	144(53.3)
	Fever	204(37.8)	68(25.2)	136(50.4)
History of antimicrobials use in the past 3 months		338(62.4)	146(54.1)	192(71.1)
Participants administered with clinician’s prescribed antimicrobials		75(13.9)	38(14.1)	37(13.7)
Source of antimicrobials used
	Retail pharmacy	72(13.3)	37(25.3)	35(13.0)
	Healthcare facility pharmacy	266(49.3)	109(40.4)	157(58.1)
Sourced antimicrobials from retail pharmacy with clinician’s prescription		72(12.6)	37(13.7)	35 (13.0)
Completed the last antimicrobial dose		304(56.3)	196(72.6)	108(40.0)
Source of domestic water
	Municipal water	495(91.7)	254(94.1)	241(89.3)
	Borehole water	45(8.3)	16(5.9)	29(10.7)

IQR: interquartile range.

Gastrointestinal carriage of extended-spectrum beta-lactamase- producing enterobacterales

In this study, the overall gastrointestinal carriage of ESBL-E was 35.4%, 95% CI: 31.3–39.4% (191/540). There was no co-carriage of ESBL-E among the study participants. Among outpatients, the prevalence of ESBL-E carriage was 40.4% (109/270) compared with 30.4% (82/270) among inpatients. The odds of ESBL-E colonization were 1.55-fold higher among outpatients than inpatients. Among outpatients, the rate was 40.4% (95% CI: 34.5–46.2% (109/270)), 1.55 times higher than 30.4% among the inpatient (95% CI: 24.9–35.9% (82/270). This difference was statistically significant (odds ratio = 1.55, 95% CI: 1.09–2.21, p = 0.015 (Table 2).

Table 2.

Gastrointestinal carriage of ESBL-E among inpatient and outpatient children (≤ 5 years) at Thika Level 5 Hospital, February – June 2023.

Patient type	ESBL-E carriage		Odds ratio (95%CI)	P- value
	Present n(%)	Absent n(%)
Outpatients	109 (40.4)	161 (59.6)	1.55(1.09–2.21)	0.015
Inpatients	82 (30.4)	188 (69.6%)	Ref
Overall	191(35.4)	349(64.6)

A total of four ESBL-E species were recovered from the inpatients, with E. coli (66/270, 24.4%) as the predominant isolate (Fig. 1). Seven ESBL-E species were recovered, including four species (Citrobacter freundii, Citrobacter farmeri, Citrobacter yongae, and Enterobacter bugandensis) that were distinct from those isolated from inpatients. ESBL-producing Klebsiella variicola was not found among outpatients. As observed among inpatients, E. coli was the most common ESBL-E (91/270, 33.7%), Fig. 1.

Fig. 1.

Gastrointestinal carriage of ESBL-E among inpatient and outpatient children (≤ 5 years) at Thika Level 5 Hospital, February – June 2023. ESBL-E: extended spectrum beta-lactamase-producing Enterobacterales, OPD outpatient department, IPD inpatient department, %: percentage.

Antimicrobial susceptibility profiles of ESBL-E isolates

Due to the limited number of isolates, the antimicrobial resistance (AMR) for C. freundii (n = 4), C. farmeri (n = 1), Citrobacter youngae (n = 1), E. bugandensis (n = 1), K. variicola (n = 1), and Morganella morganii (n = 2) was not determined; however, their antimicrobial susceptibility results are shown in Supplementary Table S1. Co-resistance levels among extended-spectrum β-lactamase-producing K. pneumoniae (ESBL-Kp) and extended-spectrum β-lactamase-producing E. coli (ESBL-Ec) were high for aminoglycosides (43–52%), carbapenem (44–50%), quinolones (52–62%), and sulfonamides (92–97%), as detailed in Table 3. In contrast, the lowest co-resistance was seen with colistin (1–4%) and the β-lactam/β-lactamase inhibitor piperacillin/tazobactam (TZP), which had resistance rates of 10–24%. Notably, TZP resistance was highest among ESBL-Kp isolates from inpatients (24%), Table 3.

Table 3.

Antimicrobial co-resistance of ESBL- producing E. coli and K. pneumoniae isolates from inpatient and outpatient children (≤ 5 years) at Thika Level 5 Hospital, February – June 2023.

Antimicrobial class	Antimicrobial agent	E. coli			K. pneumoniae
		All n = 157 (%)	OPD n = 91 (%)	IPD n = 66 (%)	All n = 25 (%)	OPD n = 15(%)	IPD n = 10 (%)
Penicillin	Ampicillin	152(97)	88(56)	64(41)	25(100)	10(40)	15 (60)
	TZP	16(10)	5 (3)	11(7)	6(24)	1(4)	5(20)
2GC	Cefuroxime	155(99)	89(57)	66(42)	25(100)	10(40)	15(60)
3GC	Cefotaxime	154(98)	88(56)	66(42)	25(100)	10(40)	15(60)
4GC	Cefepime	130(83)	78(50)	52(33)	21(84)	8(32)	13(52)
AMI	Amikacin	68(43)	50(32)	18(11)	13(52)	6(24)	7(28)
	Gentamicin	73(46)	51(32)	22(14)	12(48)	5(20)	7(28)
Quinolones	Ciprofloxacin	97(62)	65(41)	32(20)	13(52)	0(0)	13(52)
Sulfonamides	SXT	153(97)	91(58)	62(39)	23(92)	9(36)	14 (56)
Carbapenem	Meropenem	79(50)	43(27.4)	36(22.9)	11(44)	4(16)	7(28)
Polymyxins	Colistin	1(1)	1(1)	0(0)	1(4)	1(4)	0(0)

OPD outpatient department, IPD inpatient department, 2GC second-generation cephalosporin, 3GC third-generation cephalosporin, 4GC fourth-generation cephalosporin, AMI aminoglycosides, TZP piperacillin/tazobactam, SXT Trimethoprim/sulfamethoxazole.

Co-carriage of ESBL-E and CPE

Overall, 47.6% (91/191) of ESBL-E in this study were carbapenem-resistant (CR). Of these, sixteen isolates were carbapenemase-producing Enterobacterales (CPE), translating into a co-carriage rate of 17.6% (16/91). The CPE/ESBL-E co-carriages were similarly distributed among inpatients (50.0%, 7/16) and outpatients (50.0%, 8/16), Fig. 2. E. coli was the most common ESBL-E/CPE co-carrying isolate overall (81.3%, 13/16) among (OPD: 75%, 6/8), and among IPD (87.5%, 7/8), Fig. 2.

Fig. 2.

Co-carriage of ESBL-E and CPE among inpatient and outpatient children (≤ 5 years) at Thika Level 5 Hospital, February – June 2023. ESBL-E: extended spectrum beta-lactamase-producing Enterobacterales, CPE: carbapenemase-producing Enterobacterales, %: percentage, OPD: outpatient department, IPD: inpatient department.

Risk factors associated with gastrointestinal carriage of ESBL-E

Children aged 24 months or younger were found to be twice as likely to have gastrointestinal colonization with ESBL-E compared to those older than 24 months, with an adjusted odds ratio (OR) of 1.60 (95% CI: 1.01–2.38, p = 0.045), Table 4. Additionally, outpatients who purchased antimicrobials from retail pharmacies with a clinician’s prescription were 82% less likely to harbor ESBL-E compared to those who bought the antimicrobials from such drug stores without a prescription (adjusted OR of 0.18 (95% CI: 0.03–0.97, p = 0.047), Supplementary Table S2.

Table 4.

Factors associated with the gastrointestinal carriage of the ESBL-E among children (≤ 5 years) at Thika Level 5 Hospital, February – June 2023.

Characteristics	Total, n(%)	ESBL-E		Crude OR (95%CI)	P-value	Adjusted OR (95%CI)	P-value
		Yes n(%)	No n(%)
Age
≤ 24 months	405(75.0)	149(80.1)	256(72.3)	1.54(1.01–2.37)	0.048*	1.60(1.01–2.38)	0.045*
> 24 months	135(25.0)	37(19.9)	98(27.7)	Ref
Gender
Male	280(51.7)	93(50.0)	186(52.7)	0.90(0.63–1.28)	0.584
Female	260(48.3)	93(50.0)	168(47.3)	Ref
Clinical presentation
Vomiting
Yes	148(27.4)	49(26.3)	99(28.0)	0.92(0.62–1.38)	0.761
No	392(72.6)	137(73.7)	255(72.0)	Ref
Chills
Yes	16(3.0)	5(2.7)	11(3.1)	0.86(0.29–2.51)	0.506
No	524(97.0)	181(97.3)	343(96.9)	Ref
Diarrhoea
Yes	139(25.8)	46(24.7)	93(26.3)	0.92(0.61–1.38)	0.756
No	401(74.2)	140(75.3)	261(73.7)	Ref
Headache
Yes	5(0.9)	1(0.5)	4(1.1)	0.47(0.05–4.24)	0.664
No	535(99.1)	186(99.5)	349(98.9)	Ref
Respiratory illness
Yes	220(40.7)	77(41.4)	143(40.6)	1.03(0.72–1.48)	0.927
No	320(59.3)	110(58.8)	210(59.5)	Ref
Fever
Yes	204(38.0)	63(33.9)	141(40.2)	0.76(0.53–1.11)	0.162	0.77(0.53–1.12)	0.165
No	333(62.0)	123(66.1)	210(59.5)	Ref
Antibiotics use history in the past 3 months
Yes	338(62.5)	112(60.2)	226(63.8)	0.86(0.60–1.24)	0.454
No	202(37.5)	74(39.8)	128(36.2)	Ref
Source of antimicrobials
Retail pharmacy	72(21.3)	22(19.8)	50(22.0)	0.99(0.58–1.71)	0.55
Healthcare facility pharmacy	266(78.7)	89(80.2)	177(78.0)	Ref
Participants that administered clinician’s prescribed antimicrobials
Yes	34(45.3)	11(44.0)	23(46.0)	0.92(0.35–2.42)	0.534
No	41(54.7)	14(56.0)	27(54.0)	Ref
Completed the last antimicrobial dose
Yes	304(56.3)	109(56.8)	195(56.0)	0.94(0.67–1.36)	0.855
No	236(43.7)	83(43.2)	153(44.0)	Ref
Source of water
Municipal water
Yes	495(91.7)	168(90.3)	327(92.4)	0.77(0.41–1.44)	0.417
No	45(8.3)	18(9.7)	27(7.6)	Ref
Borehole water
Yes	45(8.3)	18(9.7)	27(7.6)	1.30(0.70–2.42)	0.417
No	495(91.7)	168(90.3)	327(92.4)	Ref

Discussion

In this study, the overall gastrointestinal carriage rate of ESBL-E was 35.4% (191/540). This rate is comparable to 34.3% (207/603) reported among children in Dar es Salaam, Tanzania⁷ and 37% among children under five years of age from hospitals and community settings in the Gaza Strip, Palestine¹⁹. Our observed ESBL-E carriage rate is lower than the 40.3% faecal carriage observed in rural Tanzania among patients with urinary tract infections²⁰. Among inpatients in our study, the ESBL-E carriage rate was 30.4%, 95% CI: 24.9–35.9% (82/270), meaning that one in every three hospitalized patients harbored ESBL-producing organisms. Our inpatient carriage rate is notably higher than the 10% (59/569) reported among neonates (0–28 days) at the point of admission in Kilifi, Kenya²¹, and the 21.1% rate for inpatients reported in a systematic review and meta-analysis of global prevalence and trends of human intestinal carriage of ESBL-Ec²². However, our inpatient rate is lower than 39.1% reported in Gaza Strip in Palestine¹⁹ and 50.4% in Dar es Salaam, Tanzania⁷ among children (< 2 years). Additionally, it was lower than 44% carriage rate for ESBL-producing E. coli among children (1–59 months) in Kisii and Homa Bay hospitals, Kenya, as reported by Tornberg-Belanger et al.²³, and the 62.9% (154/245) rate for Klebsiella spp. isolates at discharge in the same Kenyan hospitals²⁴. Our study indicates a high rate of community transmission of multidrug-resistant bacteria. Additional research is needed to identify the sources of these resistant bacterial strains within the community in order to inform AMR control programs.

Further, the ESBL-E carriage rate observed in our study was lower than reported by Edwards et al. (50%, 9/18) among neonates sampled between 1 and 46 days post admission in Kisumu, Kenya²⁵, and lower than the post-acquisition levels (55%) reported among neonates (1–59 months) admitted to a neonatal ward in Kilifi, Kenya²¹. ESBL-E colonization rates have been associated with longer hospital stays and antimicrobial use²⁴. The differences in hospital stays in the current study and others in Kenya could have impacted the carriage rate levels. Hospital environments are known reservoirs of MDR organisms, with prolonged hospitalization likely to increase the risk of colonization and infection.

The ESBL-E carriage rate among outpatients was 40.4% (95% CI: 34.5–46.2%; 109/270), which is similar to the 43% (159/366) found in rural children aged 2 and 5 years in the Somali region of Ethiopia²⁶. This rate in our study is higher than the 10% (59/569) reported among neonates (0–28 days) at admission in Kilifi, Kenya²¹, 11.6% among children under 2 years in Dar es Salaam, Tanzania⁷, 17.1% (46/269) for ESBL-producing E. coli and K. pneumoniae among children under five in Addis Ababa, Ethiopia²⁸, and 35.1% among children under five in Gaza Strip, Palestine¹⁹. Even though colonization by MDR organisms is expectedly higher in hospital settings²⁹, outpatients in our study exhibited a higher ESBL-E carriage rate than inpatients. Notably, a recent systematic review and meta-analysis highlighted a sharper annual increase in intestinal carriage of ESBL-producing E. coli within the community (1.5% per year) compared to healthcare settings (1.3% per year)²². Globally, community carriage rates escalated dramatically, rising from just 2.6% in 2001 to 26.4% in 2016—a tenfold surge²⁷. Meanwhile, in healthcare environments, prevalence climbed from 7% to 25.7% during the same period²². These patterns underscore an accelerating burden of ESBL-E in community settings, which provides important context for the findings of the present study. The high community-level carriage suggests community dissemination of ESBL genes, likely driven by antimicrobial misuse, poor sanitation, and unregulated access to antimicrobials. In general, the observed discrepancies in the ESBL carriage rates in the current study and previous ones suggest differences in the study population, variations in adherence to antimicrobials stewardship and infection prevention and control programs, sanitary conditions, and ESBL detection methods^30,31. The high ESBL-E faecal carriage in developing countries may be attributable to substandard hygiene conditions and the inappropriate use of antimicrobials, particularly when antimicrobials are sold over the counter without a prescription by clinicians³².

In the in- and outpatients, E. coli (inpatients: 66/83, 79.5%); outpatients: 91/109, 83.5%) was the predominant ESBL-E. These results are comparable with the previous studies in Ethiopia³³, , Tanzania²⁰, Burkina Faso³⁴, Netherlands³⁵, Uganda³⁶, and Turkey³⁷ but contradict other study reports in Ethiopia and Uganda³⁸, where K. pneumoniae was the predominant ESBL producer compared to E. coli. E. coli is the most abundant facultative anaerobe of the human gastrointestinal tract (GIT) microflora that causes disease in immunocompromised individuals and those with breached GIT integrity, except clones that have acquired specific virulence attributes that allow them to cause a broad spectrum of disease³⁹. The high gastrointestinal carriage of ESBL-producing E. coli and K. pneumoniae observed in the current study presents a critical public health threat, especially in settings with poor adherence to infection prevention and control measures and inadequate microbiology laboratory capacity and sanitation. GIT colonization with ESBL-producing may precede infections, with the colonized patients serving as silent reservoirs for transmission of AMR traits⁴⁰, and infections are frequently associated with increased morbidity, mortality, length of hospitalization, and healthcare costs^4,3.

ESBL-producing E. coli and K. pneumoniae isolates showed high co-resistance to ampicillin (97%) and increased susceptibility (90%) to the tested β-lactam/β-lactamase inhibitor, piperacillin/tazobactam (TZP). Inhibitor resistance was highest among inpatient isolates (7–20%) compared to outpatients (3–4%). As expected, resistance to β-lactam/β-lactamase inhibitors was lower than to β-lactams alone, as these inhibitors target serine β-lactamase enzymes⁴¹. However, our findings indicate notable tazobactam resistance in both E. coli and K. pneumoniae. Like other first-generation inhibitors such as clavulanic acid and sulbactam, tazobactam inhibits Ambler class A enzymes⁴¹. This may explain the observed resistance and suggests possible over-reliance on β-lactam/β-lactamase inhibitors in this setting.

Nearly half (47.6%, 91/191) of all ESBL-E isolates in this study were carbapenem-resistant (CR). This is higher than the 31.1% prevalence reported among ESBL-E isolates from children (aged < 1 year) in the Gaza Strip¹⁹. In our study, 16.5% (15/91) of CR isolates were carbapenemase producers, with a similar distribution among inpatients (16.3%, 7/43) and outpatients (16.7%, 8/48). The co-carriage rate of ESBL-E and carbapenemase-producing Enterobacterales (CPE) among inpatients was lower than what was observed among neonates at discharge in Kisumu, Kenya (33.3%, 6/18) and Ibadan, Nigeria (79.2%, 18/24)²⁵. However, the co-carriage in our study was higher than the 2.9% reported among inpatients and outpatients in five hospitals in Burkina Faso, where over 54.9% of the study population were children under 15 years old, including 78% of infants⁴². These findings suggest a substantial reservoir of carbapenem-resistant ESBL-E in both hospital and community settings. This poses significant infection control and antimicrobial stewardship challenges, such as an increased risk of hard-to-treat infections and the potential for transmission. Notably, the CR ESBL-E isolates in this study showed low resistance to colistin, with susceptibility rates between 96% and 99%. Kanwalpreet et al. also reported a similarly low colistin resistance (5.6%) in Gram-negative bacteria⁴³. These results suggest that colistin retains clinical value for treating CR ESBL-E infections in our setting. Use of newer agents for infections caused by CR bacteria—including cefiderocol, imipenem-cilastatin-sulbactam, ceftazidime-avibactam, meropenem-vaborbactam, ceftolozane-tazobactam, eravacycline, and plazomicin—is limited by high cost, restricted availability, and a lack of comprehensive clinical data⁴⁴.

This study found that children aged 24 months or younger had significantly higher odds of gastrointestinal colonization with ESBL-E. This aligns with previous research demonstrating that age is a major risk factor for ESBL-E carriage, with the highest vulnerability observed among infants under one year^7,45. During this early developmental stage, frequent hand-to-mouth behavior and poor hygiene practices—particularly in crawling children who come into contact with contaminated surfaces—create ideal conditions for acquiring drug-resistant bacteria⁴⁶. Other factors such as healthcare exposure within the previous six months further could have increased the likelihood of colonization, consistent with earlier studies showing that repeated contact with healthcare settings facilitates the acquisition and transmission of ESBL-E^21,44. Additional factors identified in other studies, such as the use of indwelling devices and consumption of untreated water sources like boreholes, similarly highlight how environmental and clinical exposures compound the risk of colonization⁴⁵.

Among outpatients in the present study, a history of non-prescribed antimicrobial use was associated with an 82% higher likelihood of ESBL-E carriage. This finding is consistent with global evidence indicating that inappropriate antimicrobial access and use are major drivers of antimicrobial resistance. A recent scoping review estimated that 62% of antimicrobial use worldwide occurs without a prescription⁴⁷. In regions with limited regulatory oversight and weak antimicrobial stewardship—such as many parts of sub-Saharan Africa—non-prescription antimicrobial use is particularly pervasive⁴⁸. For example, one review reported that 69% of requests for antimicrobials were dispensed without prescription or consultation⁴⁹. Such practices promote selective pressure favouring resistant organisms and accelerate community spread of AMR⁵⁰. Taken together, the findings of this study reinforce existing evidence that younger age and inappropriate antimicrobial use all contribute meaningfully to ESBL-E colonization in children.

Study limitations

This study has several limitations. It was conducted in a single hospital, which may limit the generalizability of the findings to other settings. Molecular characterization of isolates was not performed due to limited laboratory capacity, preventing identification of specific ESBL genes and clonal relationships. Reliance solely on phenotypic methods may have reduced diagnostic sensitivity compared to molecular techniques. Additionally, variability in sample types-rectal swabs may underestimate colonization compared with stool samples- and the cross-sectional design may have influenced the accuracy of colonization estimates and limited the ability to assess temporal trends. Despite these limitations, the findings provide valuable baseline data on ESBL-E carriage and resistance profiles, contributing to antimicrobial stewardship efforts and highlighting the need for larger, multicentric, and molecular-based studies to better understand the epidemiology of ESBL-producing bacteria.

Conclusion

This study demonstrates a substantial burden of gastrointestinal carriage of extended-spectrum β-lactamase–producing Enterobacterales and co-carriage with carbapenemase-producing Enterobacterales among children aged five years and below, with E. coli being the predominant organism. Younger children—particularly those aged 24 months or less—and those with a history of exposure to non-prescribed antimicrobials were significantly more likely to be colonized.

To mitigate the spread of multidrug-resistant organisms and protect young children from preventable colonization and subsequent infections, interventions should prioritize strengthening hygiene practices in early childhood settings, households, and hospitals, limiting children’s exposure to contaminated environments. At the same time, targeted antimicrobial stewardship initiatives are essential, including community education on the risks of over-the-counter antimicrobial use, improved regulation of antimicrobial access, and enhanced counselling by healthcare providers.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

WHO

World Health Organization

AST

Antimicrobial Susceptibility Testing

CLSI

Clinical and Laboratory Standards Institute

MALDI-TOF MS

Matrix assisted laser desorption ionization-time of flight mass spectrometry

MDR

Multidrug-drug resistant

ESBL

Extended-spectrum beta-lactamases

ESBL-E

Extended-spectrum beta-lactamases producing Enterobacterales

OPD

Outpatient Department

IPD

Inpatient Department

Author contributions

SG, JM and AM conceived and developed the study, interpreted the data, and drafted the manuscript; CN assisted in laboratory work. All authors read and approved the final manuscript.

Funding

This study did not receive financial support from any funding agencies.

Data availability

The datasets utilized in this study can be acquired from the corresponding author upon request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was approved by Kenyatta University’s ethical research committee (Protocol Approval No. PKU/2620/E1745) as well as the National Commission of Science, Technology, and Innovation (NACOSTI). Approval for collection of samples was obtained from Ethical Research Committee, Thika Level 5 hospital. An informed consent was obtained from parent or guardian of each child before participating in this study. The research project was conducted in accordance with the Declaration of Helsinki, prioritizing participants’ well-being and ensuring timely communication of critical findings to the doctors responsible for the patients. Each patient was assigned a unique Personal Identification Number (PIN) for anonymity, and all information collected was kept confidential. Participants in the study received no monetary compensation, and there were no penalties for those who chose not to participate.

Consent for publication

Not applicable.