Multidrug-resistant organism carriage in Vietnamese district hospitals before and after an antimicrobial stewardship intervention Authors: Jaslyn Doshi, Thai Duy Pham, Yen Ngoc Pham, Thuy Thu Ma, Linh Thuy Duong, Qingbin Li, Thanh Ha Nguyen, Matthew Watts, Jon Iredell, Thu Anh Nguyen, Hoang Huy Tran, Greg J. Fox, Annaleise R. Howard-Jones Published: 2026 ABSTRACT IntroductionAntimicrobial resistance is a critical global health challenge, with Southeast Asia identified as a hotspot of multidrug-resistant organism (MRO) carriage in hospitalised patients and the general population. This study aimed to estimate the prevalence of gastrointestinal and nasal MRO carriage among district hospital inpatients in Vietnam and assess changes following an antimicrobial stewardship (AMS) intervention.MethodsTwo repeated cross-sectional prevalence surveys were conducted among inpatients at four district hospitals in Vietnam (two urban hospitals in Hanoi and two rural hospitals in Ca Mau). The baseline survey was undertaken immediately prior to implementation of a 6-month AMS intervention (March 2022), and the post-intervention survey was conducted during the final 2 weeks of the intervention (August 2022). The intervention bundle comprised establishing hospital AMS committees and teams, implementing antimicrobial prescribing guidelines, healthcare worker education, and regular antimicrobial prescribing audits with feedback to clinicians. Carriage of extended spectrum β-lactamase (ESBL)-producing Enterobacterales and carbapenemase-producing Enterobacterales (CPE) was determined from stool or rectal swabs, and methicillin-resistant Staphylococcus aureus (MRSA) carriage using nasal swabs. MROs were characterised using phenotypic antimicrobial susceptibility testing and molecular analyses. A participant questionnaire assessed demographic, behavioural and healthcare-related risk factors. Pre- and post-intervention prevalence estimates were compared using two-proportion tests. Mixed-effects binary regression models were used to identify risk factors for ESBL-producing Enterobacterales carriage.ResultsA total of 402 participants (ages 18–99 years, M:F 176:226) provided nasal swabs and rectal or stool samples across both surveys. Participant demographics differed modestly between surveys. Overall, 76.1% of participants carried Enterobacterales isolates harbouring blaCTX-M (ESBL) genes, while carriage of representative CPE (2.7%) and MRSA (3.0%) were low. Risk factors for ESBL-producing Enterobacterales carriage included female sex, older age, recent antibiotic use, smoking and admission to an urban hospital. The prevalence of ESBL-producing Enterobacterales and CPE were unchanged following the AMS intervention. MRSA carriage decreased from 5 to 1% following the intervention [− 4.0%, 95% CI − 7.2%, − 0.7%).ConclusionESBL-producing Enterobacterales carriage was highly prevalent among inpatients in four Vietnamese district hospitals, while CPE and MRSA carriage remained low. A 6-month AMS intervention was associated with a modest reduction in MRSA carriage but no change in ESBL-producing Enterobacterales or CPE carriage.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13756-026-01752-5. Introduction Antimicrobial resistance (AMR) is a major global health threat, responsible for an estimated 1.27 million deaths annually and contributing to nearly 5 million deaths worldwide, with the burden falling disproportionately on low- and middle-income countries (LMICs) [1]. Older adults, patients with frequent healthcare exposure, and those admitted to hospitals are particularly vulnerable to infections caused by multidrug-resistant organisms (MROs), which are associated with increased morbidity, mortality, length of hospital stay, and healthcare costs [1, 2]. Resistance to existing antibiotics continues to outpace the development of new agents, raising concerns about the growing incidence of untreatable infections. Vietnam, a middle-income country in Southeast Asia, is a recognised AMR “hotspot” [3]. Antibiotic overuse and misuse across hospitals, community, agricultural and animal health sectors have accelerated the emergence and spread of MROs such as extended-spectrum β-lactamase (ESBL)- and carbapenemase-producing Enterobacterales (CPE) and methicillin-resistant Staphylococcus aureus (MRSA). ESBL–producing Enterobacterales are among the most prevalent MROs globally, with intestinal carriage of ESBL-producing Escherichia coli estimated at 16.5% in community settings worldwide and approximately 27% in Southeast Asia [4]. In Vietnam, community carriage rates are even higher, reported at 35–51% [5]. Carbapenemase-producing Enterobacterales (CPE) carriage is also of significant clinical concern, with prevalence estimated at 4.5–10% among predominantly hospitalised and community populations worldwide and higher rates reported in Asian settings [6–8]. In Vietnam, tertiary hospital studies have documented carbapenem-resistant Enterobacterales colonisation in up to 52% of inpatients [9], compared with only 0.6% reported in community populations [10]. Methicillin-resistant Staphylococcus aureus (MRSA) carriage also varies widely, ranging from 0 to 23.5% in community settings and from 0.7 to 10.4% in hospital settings across the Asia–Pacific region [11], with a prevalence of approximately 7.9% reported in community populations in Vietnam [12] and 25.3% in Vietnamese hospitals [13]. However, data on MRO carriage in district/regional hospitals in Vietnam—where most inpatient care occurs—remain limited. In Vietnam, access to inpatient care in public hospitals is largely mediated through a national social health insurance system covering approximately 87% of the population, with government subsidies for vulnerable groups and broad access to inpatient services at district hospitals [14]. Although out-of-pocket payments remain a substantial component of health expenditure, district hospital inpatients generally represent the adult population requiring hospital admission rather than a highly selected referral population. District level hospitals (many of which have recently been reclassified as regional hospitals) deliver acute inpatient medical, surgical and obstetric care to local communities. The extent of AMR in Vietnam highlights the critical need for coordinated strategies [15]. Antimicrobial stewardship (AMS) plays a pivotal role in addressing this challenge by promoting the responsible use of antibiotics to minimise resistance, improve patient outcomes, and preserve the effectiveness of existing treatments [16, 17]. This study aimed to understand prevalence of and risk factors for MRO carriage in district/regional hospital inpatients. The primary outcome was prevalence of ESBL-producing Enterobacterales (represented by blaCTX-M-carrying Enterobacterales) among inpatients at the four district hospitals in Vietnam. Secondary outcomes included carriage of other MROs—representative CPEs and MRSA—as well as the antibiotic susceptibility and genotypic profiles of the isolated organisms. Additionally, this study examined patient factors associated with MRO carriage. Lastly, we evaluated the impact of an inpatient AMS program on the prevalence of MRO colonisation. Methods Study design, setting and population This study used a before-and-after study design with repeated cross-sectional surveys. The study was conducted at four public district hospitals in two provinces in urban (Hanoi) and rural (Ca Mau) Vietnam (Fig. 1, Supplementary Table S1). Study sites were selected to reflect geographic diversity and to include sites with limited pre-existing AMS capacity, consistent with the study aim of evaluating MRO carriage and AMS interventions in lower-level healthcare settings.Fig. 1Schematic of healthcare structure in Vietnam. Intervention was carried out at the district hospital level as indicated by red arrow. District level hospitals deliver acute inpatient medical, surgical and obstetric care, and basic emergency and intensive care services. MoH: Ministry of Health. (From Ha et al. 2015) [18] Schematic of healthcare structure in Vietnam. Intervention was carried out at the district hospital level as indicated by red arrow. District level hospitals deliver acute inpatient medical, surgical and obstetric care, and basic emergency and intensive care services. MoH: Ministry of Health. (From Ha et al. 2015) [18] Data were collected at two timepoints in each hospital, prior to implementation of AMS interventions, (March 2022) and during the final two weeks of a 6-month intervention period (August 2022). At each timepoint, a different cohort of inpatients was sampled. Eligible participants were adults aged ≥ 18 years who had been admitted to any ward for over 48 h at the time of sampling, and who provided informed consent to submit a stool sample or rectal swab, as well as a nasal swab. Inclusion was limited to adults aged ≥ 18 years to ensure participants could provide informed consent without parental or guardian approval. Sample size The sample size calculations for comparing the prevalence of CPE, ESBL and MRSA were based upon Vietnamese studies in which 45% of non-ICU patients were colonised with CRE [9], and 18% of a community population carried MRSA in the anterior nasal cavity (Howard-Jones et al., unpublished). In the absence of published estimates of the effect of AMS interventions on MRO carriage at the time of study initiation, an a priori estimation of a 15% drop for CRE and 10% for MRSA was assumed, representing conservative, clinically meaningful effect sizes. Calculations used a two-sided alpha of 0.05 and 80% power, resulting in a required sample size of 196 participants per time point. Antimicrobial stewardship interventions The AMS intervention was implemented over a 6-month period at each participating hospital and comprised four core components: governance, guideline development, education, and audit and feedback. The AMS interventions were based upon a World Health Organisation (WHO) toolkit for AMS in resource-limited settings and an Australian national AMS program, in alignment with the Vietnam Ministry of Health policy [16, 19, 20]. At each site, an AMS committee was established, including senior hospital leadership, clinical department heads, and pharmacy leads, to meet every 1–2 months and oversee implementation. Facility-level AMS teams, consisting of at least one doctor (with Infectious Diseases training where available) and one pharmacist, were responsible for delivering the intervention. Locally adapted antimicrobial guidelines for common infections and surgical prophylaxis were developed based on national Ministry of Health guidance and regional antibiogram data, translated into Vietnamese, and distributed to all wards. Two structured education workshops were delivered to doctors and pharmacists at each site, focusing on appropriate antimicrobial prescribing, guideline use, and AMS principles. Prescribing practices were assessed through periodic audits of inpatient medical records using a standardised tool adapted from the Australian Hospital National Antimicrobial Prescribing Survey (NAPS) tool [21]. Verbal and written feedback was delivered following audits to prescribers highlighting prescribing patterns, examples of suboptimal use, and practical recommendations. Details of the AMS interventions are provided in Supplementary Table S2. Each hospital also had formal infection control policies [22], including recommendations for hand washing, separation of patients with contact precautions and room cleaning. These remained unchanged during the course of this study. Enrolment and data collection Participant data were collected by study staff using a structured questionnaire. Collected variables included age, biological sex, ward location at time of sample collection, smoking status, antibiotic use during present admission, self-reported antibiotic use in the previous 12 months and hospitalisation within the previous 12 months. Self-reported past history of ICU admission, dialysis and exposure to invasive devices such as nasogastric tube insertion, indwelling urinary catheter insertion or central venous catheters were also documented. Variables were selected based on established literature on risk factors for MRO carriage and infections [23]. Outcomes The primary outcome of this study was the proportion of inpatients with rectal carriage of Enterobacterales carrying blaCTX-M (as the indicator ESBL gene) in the four district hospitals in Vietnam. Secondary outcomes included (i) rectal carriage of Enterobacterales with blaNDM-1, blaKPC and blaOXA-48-like (as representative carbapenemase genes), (ii) nasal carriage of MRSA carriage (iii) organism-level antimicrobial susceptibility patterns (iv) comparisons of prevalence of MRO carriage before and after the AMS intervention and (v) patient-level risk factors associated with MRO carriage. Sample collection and transport Rectal swabs were collected directly with eSwabs (Copan, Italy). Stool samples were collected in a sterile container, mixed thoroughly after collection and sampled in three separate locations using an eSwab, before being stored in Amies transport medium. An anterior nasal swab was collected on each participant using an eSwab liquid Amies collection system (Copan, Italy), moistened with sterile 0.9% normal saline. Samples were kept at 2 to 8 °C and transported to the microbiology lab within 48 h of collection. Isolation and characterisation of Enterobacterales isolates eSwab rectal and stool-derived specimens were inoculated onto MacConkey agar supplemented with cefotaxime (2 µg/mL) and onto ChromAgar Orientation plates (ThermoFisher, Paris, France) supplemented with meropenem (2 µg/ml), the latter following the procedure of Yen et al. [10]. These media were incubated aerobically for 18 to 24 h at 37 °C and colonies were identified using Matrix-Assisted Laser Desorption/Ionization Time-of-Flight mass spectrometry (MALDI-TOF MS) [24]. Stool and rectal swab-derived isolates were combined into a single Enterobacterales dataset for downstream analysis. Non-Enterobacterales isolates were excluded from further analysis. Phenotypic and genotypic antimicrobial resistance testing of Enterobacterales Phenotypic antimicrobial susceptibility testing (AST) was performed on all Enterobacterales isolates using standardised Clinical and Laboratory Standards Institute (CLSI) agar dilution methodology [25]. Organisms were determined as susceptible (S), intermediate (I) or resistant (R) to cefotaxime, cefoxitin, cefepime, ciprofloxacin, gentamicin, and meropenem according to CLSI breakpoints. Isolates were classified as “non-susceptible” if testing in the I or R range. Isolates were also assessed phenotypically for production of ESBL using a CLSI combination disk method, with a cefotaxime (30 µg) disc with and without clavulanic acid (10 µg). Only Enterobacterales isolates demonstrating an ESBL phenotype by this method underwent end-point polymerase chain reaction (PCR) assay testing for presence of blaCTX-M gene (Supplementary Table S3). Enterobacterales isolates with a meropenem minimum inhibitory concentration (MIC) exceeding 0.25 mg/L were assessed by PCR for presence of blaNDM-1, blaKPC and blaOXA-48-like genes (Supplementary Table S3). Isolation and characterisation of methicillin-resistant Staphylococcus aureus (MRSA) Methicillin-resistant Staphylococcus aureus (MRSA) was isolated from individual nasal swab samples using mannitol salt agar (MSA) containing cefoxitin (8 µg/ml). Media were incubated aerobically for 24 h at 37 °C. Colonies were identified using MALDI-TOF MS analysis and phenotypic AST characterisation performed using standardised CLSI agar dilution techniques [25]. Organisms were determined as susceptible, intermediate or resistant to oxacillin, cefoxitin, ciprofloxacin, clindamycin, sulfamethoxazole/trimethoprim and vancomycin according to CLSI breakpoints. The isolates were then assessed by PCR to confirm presence of the nuc and mecA genes (Supplementary Table S3). Statistical analyses Descriptive statistics were used to analyse participant characteristics, as well as organism characteristics, with continuous variables reported as median (with interquartile range, IQR) and categorical variables as counts and percentages. The prevalence of MRO carriage per participant and the proportion of non-susceptible organisms before and after the AMS intervention were compared using mixed-effects linear probability model adjusted for covariates when data were not sparse, and exact methods for 2 × 2 contingency tables when data were sparse. A mixed-effects modified Poisson regression clustered at hospital level was performed to evaluate participant factors associated with carriage of ESBL-producing Enterobacterales for both univariable and multivariable analyses. Results were presented as risk ratios (RRs) with 95% confidence intervals (CIs). A Directed Acyclic Graph (DAG), developed using DAGitty [26], guided the model specification by visualizing hypothesized relationships between variables (Supplementary Figure S2). All variables identified by the DAG were included in the multivariable model to estimate their independent associations with ESBL carriage while controlling for other factors. Multicollinearity was assessed using variance inflation factors (VIFs) for continuous covariate and Cramér’s V for categorical covariates. For CPE and MRSA carriage, outcomes were rare and risk factor analyses were therefore limited to descriptive statistics. All statistical analyses were performed using Stata (version 18). Ethics Ethical approval for this study was provided by the Institutional Review Board of the National Institute of Hygiene and Epidemiology (NIHE) in Vietnam (IRB—VN01057/IORG 0008555—approved 26/02/2021), and the Human Research Ethics Committee at the University of Sydney (Project no:2021/193—approved 11/05/2021). Participants received an information sheet about the study and the samples being collected and provided written informed consent. Data were de-identified to protect participant confidentiality. Results A total of 202 participants were recruited in the pre-intervention period (March 2022), and 200 participants following the AMS intervention period (August 2022). Participant characteristics are presented in Table 1.Table 1Descriptive analysis of participant demographics before and after antimicrobial stewardship interventionDemographicsAt baselineN (%)n = 202After the 6-month AMS interventionN (%)n = 200Total, N (%)n = 402Number of participantsTotal participants202 (100%)200 (100%)402 (100%)Nasal swabs collected202 (100%)200 (100%)402 (100%)Rectal swabs collected202 (100%)200 (100%)402 (100%)AgeMedian (IQR)65 (55–73)64 (51–73)64 (53–73)GenderMale99 (49.0%)77 (38.5%)176 (43.8%)Female103 (51.0%)123 (61.5%)226 (56.2%)Clinical departmentMedicine*159 (78.7%)138 (69.0%)297 (73.9%)Infectious Diseases3 (1.5%)30 (15.0%)33 (8.2%)Surgery**21 (10.4%)13 (6.5%)34 (8.4%)Obstetrics & Gynaecology6 (3.0%)10 (5.0%)16 (4.0%)Intensive care &Emergency13 (6.4%)9 (4.5%)22 (5.5%)Smoking statusCurrent smoker10 (5.0%)25 (12.5%)35 (8.7%)Previous smoker53 (26.2%)41 (20.5%)94 (23.4%)Never smoked139 (68.8%)134 (67.0%)273 (67.9%)Antibiotic use (current or in the previous 12 months)Yes124 (61.4%)113 (56.5%)187 (46.5%)No72 (35.6%)82 (41.0%)203 (50.5%)Unsure6 (3.0%)5 (2.5%)12 (3.0%)Other potential risk factorsHospital admission in the past 12 months76 (37.6%)53 (26.5%)129 (32.1%)Intensive care this admission3 (1.5%)2 (1.0%)5 (1.2%)Intensive care last hospital admission4 (2.0%)1 (0.5%)5 (1.2%)Dialysis0 (0)0 (0)0 (0)Invasive devices#2 (1.0%)1 (0.5%)3 (0.7%)AMS: antimicrobial stewardship, IQR: interquartile range*Medicine: Includes General Medical, Oncology, Cardiology and Traditional Medicine**Surgery: General Surgery and Orthopaedics#Invasive devices included: nasogastric tubes, indwelling urinary catheters or central venous catheters Table 1Descriptive analysis of participant demographics before and after antimicrobial stewardship interventionDemographicsAt baselineN (%)n = 202After the 6-month AMS interventionN (%)n = 200Total, N (%)n = 402Number of participantsTotal participants202 (100%)200 (100%)402 (100%)Nasal swabs collected202 (100%)200 (100%)402 (100%)Rectal swabs collected202 (100%)200 (100%)402 (100%)AgeMedian (IQR)65 (55–73)64 (51–73)64 (53–73)GenderMale99 (49.0%)77 (38.5%)176 (43.8%)Female103 (51.0%)123 (61.5%)226 (56.2%)Clinical departmentMedicine*159 (78.7%)138 (69.0%)297 (73.9%)Infectious Diseases3 (1.5%)30 (15.0%)33 (8.2%)Surgery**21 (10.4%)13 (6.5%)34 (8.4%)Obstetrics & Gynaecology6 (3.0%)10 (5.0%)16 (4.0%)Intensive care &Emergency13 (6.4%)9 (4.5%)22 (5.5%)Smoking statusCurrent smoker10 (5.0%)25 (12.5%)35 (8.7%)Previous smoker53 (26.2%)41 (20.5%)94 (23.4%)Never smoked139 (68.8%)134 (67.0%)273 (67.9%)Antibiotic use (current or in the previous 12 months)Yes124 (61.4%)113 (56.5%)187 (46.5%)No72 (35.6%)82 (41.0%)203 (50.5%)Unsure6 (3.0%)5 (2.5%)12 (3.0%)Other potential risk factorsHospital admission in the past 12 months76 (37.6%)53 (26.5%)129 (32.1%)Intensive care this admission3 (1.5%)2 (1.0%)5 (1.2%)Intensive care last hospital admission4 (2.0%)1 (0.5%)5 (1.2%)Dialysis0 (0)0 (0)0 (0)Invasive devices#2 (1.0%)1 (0.5%)3 (0.7%)AMS: antimicrobial stewardship, IQR: interquartile range*Medicine: Includes General Medical, Oncology, Cardiology and Traditional Medicine**Surgery: General Surgery and Orthopaedics#Invasive devices included: nasogastric tubes, indwelling urinary catheters or central venous catheters Descriptive analysis of participant demographics before and after antimicrobial stewardship intervention AMS: antimicrobial stewardship, IQR: interquartile range *Medicine: Includes General Medical, Oncology, Cardiology and Traditional Medicine **Surgery: General Surgery and Orthopaedics #Invasive devices included: nasogastric tubes, indwelling urinary catheters or central venous catheters Each participant provided one enteric specimen (stool sample or rectal swab) and one nasal swab, resulting in 202 enteric and 202 nasal specimens before the AMS interventions, and 200 enteric and 200 nasal specimens after the AMS interventions. One batch containing five samples from Ca Mau province was impacted by suboptimal transport conditions (8–10 °C, rather than the pre-specified 2–8° C, for 6 h) but were still included in the analysis and valid culture results obtained. From the 402 enteric specimens, 365 Enterobacterales isolates were recovered from selective media; where multiple Enterobacterales were cultured from a single specimen, one representative isolate was selected for analysis. From the 402 nasal specimens, 12 MRSA isolates were identified. Primary outcome Enterobacterales isolates identified from stool or rectal swabs included 299 Escherichia coli, 54 Klebsiella pneumoniae complex, 8 Enterobacter cloacae complex, 3 Klebsiella oxytoca and 1 Klebsiella aerogenes isolates (Table 2). Disk diffusion testing for ESBL production was performed in 361/365 isolates, with 274 (75.9%) testing positive (Table 5). Within the study population, 76.1% of participants were colonised with blaCTX-M-carrying Enterobacterales. (Table 3) At the isolate level, 86.5% of Escherichia coli isolates tested (256/297), 76.9% Klebsiella pneumoniae complex (40/52 isolates), 50% Enterobacter cloacae complex (4/8 isolates), 100% Klebsiella oxytoca (3/3 isolates) and 100% Klebsiella aerogenes (1/1 isolates) carried the blaCTX-M gene (Table 5).Table 2Phenotypic antibiotic sensitivities of all Enterobacterales isolates from enteric samplesOrganismsTotal isolates, nCefotaximeCefoxitinCefepimeCiprofloxacinGentamicinMeropenemSNSNSNSNSNSNEscherichia coli2990(0%)299(100%)124(41.5%)175(58.5%)22(7.4%)277(92.3%)73(24.4%)226(75.6%)136(45.5%)163 (54.5%)291(97.3%)8(2.7%)Klebsiella pneumoniae complex540(0%)54(100%)22(40.7%)32(59.3%)10(18.5%)44(81.5%)9(16.7%)45(83.3%)24(44.4%)30(55.6%)52(96.3%)2(3.7%)Enterobacter cloacae complex80(0%)8(100%)0(0)8(100%)3(37.5%)5(62.5%)1(12.5%)7(87.5%)4(50.0%)4(50.0%)8(100%)0(0)Klebsiella oxytoca30(0%)3(100%)3(100%)0(0)2(66.7%)1(33.3%)1(33.3%)2(66.7%)2(66.7%)1(33.3%)3(100%)0(0)Klebsiella aerogenes10(0%)1(100%)0(0)1(100%)0(0)1(100%)0(0)1(100%)0(0)1(100%)1(100%)0(0)Total, N (%)3650(0%)365(100%)149/365(40.8%)216/365(59.2%)37/365(10.1%)328/365 (89.9%)84/365 (23.0%)281/365 (77.0%)166/365 (45.5%)199/365(54.5%)355/365 (97.3%)10/365 (2.7%)AMS: Antimicrobial stewardshipS: Susceptible N: Non-susceptible, encompassing isolates testing either intermediate or resistant to the relevant antibioticEnteric samples include stool and rectal swab samplesTable 3Prevalence of blaCTX-M-producing Enterobacterales, representative carbapenemase-producing Enterobacterales and methicillin-resistant Staphylococcus aureus carriage at the participant levelHospitalsHospital A, n (%)(Hanoi)Hospital B, n (%)(Hanoi)Hospital C, n (%)(Ca Mau)Hospital D, n (%)(Ca Mau)Total BeforeTotal AfterTotalDifference in proportions statistical analysis (95% CI)BeforeAfterBeforeAfterBeforeAfterBeforeAfterTotal number of patients100100505027252525202200402NAExtended spectrum β-lactamase (ESBL) producing EnterobacteralesPrevalence of blaCTX-M—producing Enterobacterales carriage†82/100(82.0%)80/100(80.0%)39/50(78.0%)38/50(76.0%)18/27(66.7%)21/25(84.0%)18/25(72.0%)10/25(40.0%)157/202(77.7%)149/200(74.5%)306/402(76.1%)-2.5%(95% CI − 9.0, 4.1)**Representative Carbapenemase-producing Enterobacterales (CPE)§blaNDM-114*2–––1044*8*NAblaKPC11––––––112NAblaOXA-+483*–––––––3*3*NAPrevalence of CPE(represented by 3 carbapenemase genes tested) carriage†2/100 (2.0%)6/100(8.0%)2/50(4.0%)0001/25(8.0%)0.0%5/2022.5%6/200(3.0%)11/402(2.7%)0.5%(95% CI − 2.7, 3.7)Methicillin-resistant Staphylococcus aureus (MRSA)Prevalence of MRSA carriage†2/100 (2.0%)001/50 (2.0%)3/27 (11.1%)05/25 (20%)1/25 (4.0%)10/202 (5.0%)2/200 (1.0%)12/402 (3.0%)− 4.0%(95% CI − 7.2, − 0.7)*Two participants in Hospital A after interventions were carrying two CPE genes each—blaNDM-1 and blaOXA-48 in the same organism. Hence, although total CPE genes isolated were 8, the prevalence per participant was 6/100. Bolded text indicates confidence limits do not cross the null**Estimated using a mixed-effects linear probability model adjusted for age, gender, region, department, smoking status, prior antibiotic use, and prior hospital admissions. Remaining results were estimated using exact methods for 2 × 2 contingency tables§As represented by detection of organisms carrying blaNDM-1, blaKPC or blaOXA-48-like genes†Prevalence calculated as number of participants colonised/total number of participants Table 2Phenotypic antibiotic sensitivities of all Enterobacterales isolates from enteric samplesOrganismsTotal isolates, nCefotaximeCefoxitinCefepimeCiprofloxacinGentamicinMeropenemSNSNSNSNSNSNEscherichia coli2990(0%)299(100%)124(41.5%)175(58.5%)22(7.4%)277(92.3%)73(24.4%)226(75.6%)136(45.5%)163 (54.5%)291(97.3%)8(2.7%)Klebsiella pneumoniae complex540(0%)54(100%)22(40.7%)32(59.3%)10(18.5%)44(81.5%)9(16.7%)45(83.3%)24(44.4%)30(55.6%)52(96.3%)2(3.7%)Enterobacter cloacae complex80(0%)8(100%)0(0)8(100%)3(37.5%)5(62.5%)1(12.5%)7(87.5%)4(50.0%)4(50.0%)8(100%)0(0)Klebsiella oxytoca30(0%)3(100%)3(100%)0(0)2(66.7%)1(33.3%)1(33.3%)2(66.7%)2(66.7%)1(33.3%)3(100%)0(0)Klebsiella aerogenes10(0%)1(100%)0(0)1(100%)0(0)1(100%)0(0)1(100%)0(0)1(100%)1(100%)0(0)Total, N (%)3650(0%)365(100%)149/365(40.8%)216/365(59.2%)37/365(10.1%)328/365 (89.9%)84/365 (23.0%)281/365 (77.0%)166/365 (45.5%)199/365(54.5%)355/365 (97.3%)10/365 (2.7%)AMS: Antimicrobial stewardshipS: Susceptible N: Non-susceptible, encompassing isolates testing either intermediate or resistant to the relevant antibioticEnteric samples include stool and rectal swab samples Phenotypic antibiotic sensitivities of all Enterobacterales isolates from enteric samples 0 (0%) 299 (100%) 124 (41.5%) 175 (58.5%) 22 (7.4%) 277 (92.3%) 73 (24.4%) 226 (75.6%) 136 (45.5%) 291 (97.3%) 8 (2.7%) 0 (0%) 54 (100%) 22 (40.7%) 32 (59.3%) 10 (18.5%) 44 (81.5%) 9 (16.7%) 45 (83.3%) 24 (44.4%) 30 (55.6%) 52 (96.3%) 2 (3.7%) 0 (0%) 8 (100%) 0 (0) 8 (100%) 3 (37.5%) 5 (62.5%) 1 (12.5%) 7 (87.5%) 4 (50.0%) 4 (50.0%) 8 (100%) 0 (0) 0 (0%) 3 (100%) 3 (100%) 0 (0) 2 (66.7%) 1 (33.3%) 1 (33.3%) 2 (66.7%) 2 (66.7%) 1 (33.3%) 3 (100%) 0 (0) 0 (0%) 1 (100%) 0 (0) 1 (100%) 0 (0) 1 (100%) 0 (0) 1 (100%) 0 (0) 1 (100%) 1 (100%) 0 (0) 0 (0%) 365 (100%) 149/365 (40.8%) 216/365 (59.2%) 37/365 (10.1%) 199/365 (54.5%) AMS: Antimicrobial stewardship S: Susceptible N: Non-susceptible, encompassing isolates testing either intermediate or resistant to the relevant antibiotic Enteric samples include stool and rectal swab samples Table 3Prevalence of blaCTX-M-producing Enterobacterales, representative carbapenemase-producing Enterobacterales and methicillin-resistant Staphylococcus aureus carriage at the participant levelHospitalsHospital A, n (%)(Hanoi)Hospital B, n (%)(Hanoi)Hospital C, n (%)(Ca Mau)Hospital D, n (%)(Ca Mau)Total BeforeTotal AfterTotalDifference in proportions statistical analysis (95% CI)BeforeAfterBeforeAfterBeforeAfterBeforeAfterTotal number of patients100100505027252525202200402NAExtended spectrum β-lactamase (ESBL) producing EnterobacteralesPrevalence of blaCTX-M—producing Enterobacterales carriage†82/100(82.0%)80/100(80.0%)39/50(78.0%)38/50(76.0%)18/27(66.7%)21/25(84.0%)18/25(72.0%)10/25(40.0%)157/202(77.7%)149/200(74.5%)306/402(76.1%)-2.5%(95% CI − 9.0, 4.1)**Representative Carbapenemase-producing Enterobacterales (CPE)§blaNDM-114*2–––1044*8*NAblaKPC11––––––112NAblaOXA-+483*–––––––3*3*NAPrevalence of CPE(represented by 3 carbapenemase genes tested) carriage†2/100 (2.0%)6/100(8.0%)2/50(4.0%)0001/25(8.0%)0.0%5/2022.5%6/200(3.0%)11/402(2.7%)0.5%(95% CI − 2.7, 3.7)Methicillin-resistant Staphylococcus aureus (MRSA)Prevalence of MRSA carriage†2/100 (2.0%)001/50 (2.0%)3/27 (11.1%)05/25 (20%)1/25 (4.0%)10/202 (5.0%)2/200 (1.0%)12/402 (3.0%)− 4.0%(95% CI − 7.2, − 0.7)*Two participants in Hospital A after interventions were carrying two CPE genes each—blaNDM-1 and blaOXA-48 in the same organism. Hence, although total CPE genes isolated were 8, the prevalence per participant was 6/100. Bolded text indicates confidence limits do not cross the null**Estimated using a mixed-effects linear probability model adjusted for age, gender, region, department, smoking status, prior antibiotic use, and prior hospital admissions. Remaining results were estimated using exact methods for 2 × 2 contingency tables§As represented by detection of organisms carrying blaNDM-1, blaKPC or blaOXA-48-like genes†Prevalence calculated as number of participants colonised/total number of participants Prevalence of blaCTX-M-producing Enterobacterales, representative carbapenemase-producing Enterobacterales and methicillin-resistant Staphylococcus aureus carriage at the participant level 82/100 (82.0%) 80/100 (80.0%) 39/50 (78.0%) 38/50 (76.0%) 18/27 (66.7%) 21/25 (84.0%) 18/25 (72.0%) 10/25 (40.0%) 157/202 (77.7%) 149/200 (74.5%) 306/402 (76.1%) -2.5% (95% CI − 9.0, 4.1)** 2/100 (2.0%) 6/100 (8.0%) 2/50 (4.0%) 1/25 (8.0%) 5/202 2.5% 6/200 (3.0%) 11/402 (2.7%) 0.5% (95% CI − 2.7, 3.7) 2/100 (2.0%) 1/50 (2.0%) 3/27 (11.1%) 5/25 (20%) 1/25 (4.0%) 10/202 (5.0%) 2/200 (1.0%) 12/402 (3.0%) − 4.0% (95% CI − 7.2, − 0.7) *Two participants in Hospital A after interventions were carrying two CPE genes each—blaNDM-1 and blaOXA-48 in the same organism. Hence, although total CPE genes isolated were 8, the prevalence per participant was 6/100. Bolded text indicates confidence limits do not cross the null **Estimated using a mixed-effects linear probability model adjusted for age, gender, region, department, smoking status, prior antibiotic use, and prior hospital admissions. Remaining results were estimated using exact methods for 2 × 2 contingency tables §As represented by detection of organisms carrying blaNDM-1, blaKPC or blaOXA-48-like genes †Prevalence calculated as number of participants colonised/total number of participants Secondary outcomes Prevalence of CPE and MRSA genes Within the study population. 2.7% of participants were colonised with representative CPE genes (blaNDM-1, blaKPC and blaOXA-48-like) and 3.0% with MRSA (Table 3). At an isolate level, representative CPE genes were detected in only 2 out of 299 Escherichia coli isolates, and 1 out of 54 Klebsiella pneumoniae complex isolates (1.9% overall) (Supplementary Table S4). Antimicrobial susceptibility profile All isolates showed a high percentage of non-susceptibility to cefotaxime (100%, 365/365 isolates), cefoxitin (59.2%, 216/365 isolates), cefepime (90%, 328/365 isolates), ciprofloxacin (77%, 281/365 isolates), and gentamicin (54.5%, 199/365 isolates). Only a small percentage were non-susceptible to meropenem (3%, 10/365 isolates) (Table 2). A high percentage of MRSA isolates were resistant to clindamycin (92%, 11/12 isolates), few to trimethoprim/sulfamethoxazole (8.3%, 1/12) and none were resistant to vancomycin (0%, 0/12 isolates) (Supplementary Table S5); all isolates carried the mecA gene. Comparison before and after the antimicrobial stewardship intervention A total of 136/182 Enterobacterales isolates (74.7%) displayed phenotypic ESBL production before the AMS intervention, compared with 138/179 isolates (77.1%) after the intervention. No major difference was seen in the proportion of Enterobacterales isolates non-susceptible to antibiotics or prevalence of AMR genes per isolate before and after the implementation of AMS interventions (Table 4, Supplementary Table S6). Similarly, at the participant level, no difference was observed in the carriage of Enterobacterales with blaCTX-M [77.7% vs. 74.5%; − 2.5% (95% CI − 9.0, 4.1)] and representative carbapenemase genes—blaNDM-1, blaOXA-48-like and blaKPC [2.5% vs. 3.0%; 0.5% (95% CI − 2.7, 3.7)]—before and after the intervention (Table 3). The proportion of participants with MRSA carriage reduced significantly from 10 of 202 isolates (5%) to 2 of 200 isolates (1%), a reduction of 4.0% with the AMS intervention (95% CI − 7.2, − 0.7) (Tables 3, 5).Table 4Prevalence of blaCTX-M- and carbapenemase-producing Enterobacterales isolated in rectal carriage per isolate before and after AMS interventionIsolatesTotal isolates (Before Intervention)Total isolates (After Intervention)Proportion of isolates with blaCTX-M gene† (%)Difference in proportions in blaCTX-M before vs after intervention %, (95% CI)Proportion of isolates with carbapenemase genes† (%)Difference in proportions in carbapenemase before vs after intervention %, (95% CI)***Before InterventionAfter InterventionBefore InterventionAfter InterventionEscherichia coli152147135/151*(89.4%)122/146*(83.6%)− 5.8%(95% CI − 13.6, 1.9)3/152 (2.0%)4/147 (2.7%)**0.7%(95% CI − 2.7, 4.2)Klebsiella pneumoniae complex243017/22* (77.3%)23/30 (76.7%)− 0.6%(95% CI − 23.8, 22.5)2/24 (8.3%)2/30 (6.6%)**− 1.7%(95% CI − 15.9, 12.5)Enterobacter cloacae complex622/6 (33.3%)2/2 (100%)66.7%(95% CI − 28.9, 104.4)0/6 (0)0/0 (0)N/AKlebsiella oxytoca303/3 (100%)0/0(0)N/A0/3 (0)0/0 (0)N/AKlebsiella aerogenes010/0 (0)1/1 (100%)N/A0/0 (0)0/0 (0)N/ATotal185180157/182*(84.9%)148/179*(82.7%)− 2.6%(95% CI − 10.2, 5.0)5/185(2.7%)6/180(3.3%)0.6%(95% CI − 2.9, 4.1)*Denominators indicate the number of isolates where ceftriaxone minimum inhibitory concentration (MIC) and/or extended spectrum β-lactamase (ESBL) screen was used to rule-out ESBL, upon which blaCTX-M was performed**1 Escherichia coli and 1 Klebsiella pneumoniae complex isolate carried both blaNDM-1 and blaOXA-48 genes each. Klebsiella oxytoca and Klebsiella aerogenes were not analysed due to small numbers in blaCTX-M carriage. No carbapenemase genes detected in Klebsiella oxytoca, Klebsiella aerogenes and Enterobacter cloacae complexCarbapenemase genes tested: blaNDM-1, blaKPC and blaOXA-48-like genes***Results were estimated using exact methods for 2 × 2 contingency tables†Prevalence calculated as number of isolates with relevant gene/total number of isolatesTable 5Phenotypic and genotypic (blaCTX-M) characterisation of extended-spectrum beta-lactamase producing Enterobacterales isolatedOrganismNumber of isolatesNumber of isolates undergoing ESBL testingPhenotypic ESBL productionblaCTX-M detectionEscherichia coli299297230/297(77.4%)257/297(86.5%)Klebsiella pneumoniae complex545237/52(71.1%)40/52(76.9%)Enterobacter cloacae complex884/8(50.0%)4/8(50%)Klebsiella oxytoca332/3(66.7%)3/3(100%)Klebsiella aerogenes111/1(100%)1/1(100%)Total365361274/361(75.9%)305/361(84.5%) Table 4Prevalence of blaCTX-M- and carbapenemase-producing Enterobacterales isolated in rectal carriage per isolate before and after AMS interventionIsolatesTotal isolates (Before Intervention)Total isolates (After Intervention)Proportion of isolates with blaCTX-M gene† (%)Difference in proportions in blaCTX-M before vs after intervention %, (95% CI)Proportion of isolates with carbapenemase genes† (%)Difference in proportions in carbapenemase before vs after intervention %, (95% CI)***Before InterventionAfter InterventionBefore InterventionAfter InterventionEscherichia coli152147135/151*(89.4%)122/146*(83.6%)− 5.8%(95% CI − 13.6, 1.9)3/152 (2.0%)4/147 (2.7%)**0.7%(95% CI − 2.7, 4.2)Klebsiella pneumoniae complex243017/22* (77.3%)23/30 (76.7%)− 0.6%(95% CI − 23.8, 22.5)2/24 (8.3%)2/30 (6.6%)**− 1.7%(95% CI − 15.9, 12.5)Enterobacter cloacae complex622/6 (33.3%)2/2 (100%)66.7%(95% CI − 28.9, 104.4)0/6 (0)0/0 (0)N/AKlebsiella oxytoca303/3 (100%)0/0(0)N/A0/3 (0)0/0 (0)N/AKlebsiella aerogenes010/0 (0)1/1 (100%)N/A0/0 (0)0/0 (0)N/ATotal185180157/182*(84.9%)148/179*(82.7%)− 2.6%(95% CI − 10.2, 5.0)5/185(2.7%)6/180(3.3%)0.6%(95% CI − 2.9, 4.1)*Denominators indicate the number of isolates where ceftriaxone minimum inhibitory concentration (MIC) and/or extended spectrum β-lactamase (ESBL) screen was used to rule-out ESBL, upon which blaCTX-M was performed**1 Escherichia coli and 1 Klebsiella pneumoniae complex isolate carried both blaNDM-1 and blaOXA-48 genes each. Klebsiella oxytoca and Klebsiella aerogenes were not analysed due to small numbers in blaCTX-M carriage. No carbapenemase genes detected in Klebsiella oxytoca, Klebsiella aerogenes and Enterobacter cloacae complexCarbapenemase genes tested: blaNDM-1, blaKPC and blaOXA-48-like genes***Results were estimated using exact methods for 2 × 2 contingency tables†Prevalence calculated as number of isolates with relevant gene/total number of isolates Prevalence of blaCTX-M- and carbapenemase-producing Enterobacterales isolated in rectal carriage per isolate before and after AMS intervention 135/151* (89.4%) 122/146* (83.6%) − 5.8% (95% CI − 13.6, 1.9) 0.7% (95% CI − 2.7, 4.2) − 0.6% (95% CI − 23.8, 22.5) − 1.7% (95% CI − 15.9, 12.5) 66.7% (95% CI − 28.9, 104.4) 157/182* (84.9%) 148/179* (82.7%) − 2.6% (95% CI − 10.2, 5.0) 5/185 (2.7%) 6/180 (3.3%) 0.6% (95% CI − 2.9, 4.1) *Denominators indicate the number of isolates where ceftriaxone minimum inhibitory concentration (MIC) and/or extended spectrum β-lactamase (ESBL) screen was used to rule-out ESBL, upon which blaCTX-M was performed **1 Escherichia coli and 1 Klebsiella pneumoniae complex isolate carried both blaNDM-1 and blaOXA-48 genes each. Klebsiella oxytoca and Klebsiella aerogenes were not analysed due to small numbers in blaCTX-M carriage. No carbapenemase genes detected in Klebsiella oxytoca, Klebsiella aerogenes and Enterobacter cloacae complex Carbapenemase genes tested: blaNDM-1, blaKPC and blaOXA-48-like genes ***Results were estimated using exact methods for 2 × 2 contingency tables †Prevalence calculated as number of isolates with relevant gene/total number of isolates Table 5Phenotypic and genotypic (blaCTX-M) characterisation of extended-spectrum beta-lactamase producing Enterobacterales isolatedOrganismNumber of isolatesNumber of isolates undergoing ESBL testingPhenotypic ESBL productionblaCTX-M detectionEscherichia coli299297230/297(77.4%)257/297(86.5%)Klebsiella pneumoniae complex545237/52(71.1%)40/52(76.9%)Enterobacter cloacae complex884/8(50.0%)4/8(50%)Klebsiella oxytoca332/3(66.7%)3/3(100%)Klebsiella aerogenes111/1(100%)1/1(100%)Total365361274/361(75.9%)305/361(84.5%) Phenotypic and genotypic (blaCTX-M) characterisation of extended-spectrum beta-lactamase producing Enterobacterales isolated 230/297 (77.4%) 257/297 (86.5%) 37/52 (71.1%) 40/52 (76.9%) 4/8 (50.0%) 4/8 (50%) 2/3 (66.7%) 3/3 (100%) 1/1 (100%) 1/1 (100%) 274/361 (75.9%) 305/361 (84.5%) Factors associated with blaCTX-M-producing Enterobacterales carriage The univariable and multivariable risk factors for blaCTX-M-producing Enterobacterales carriage in the overall cohort are presented in Table 6. Female sex (aRR = 1.26, 95% CI 1.13, 1.41), older age (per 10 years, aRR = 1.03, 95% CI 1.01, 1.05), antibiotic use (current or previous, aRR = 1.10, 95% CI 1.01, 1.21), being a smoker (current or previous, aRR = 1.19, 95% CI 1.15, 1.24), being in the obstetrics & gynaecology ward (aRR = 1.19, 95% CI 1.05, 1.35) and urban hospital location (aRR = 1.10, 95% CI 1.05, 1.15) were identified as significant risk factors for for blaCTX-M-producing Enterobacterales carriage. The number of isolates with CPE and MRSA carriage were insufficient for inferential analyses (descriptive analysis is provided in Supplementary Table S7). Table 6Risk factors for blaCTX-M-producing Enterobacterales carriage by univariable and multivariable analysisVariableParticipants with blaCTX-M producing Enterobacterales carriage (n)Total participants (n)Proportion (%)Univariable modelMultivariable modelcRR (95% CI) p aRR (95% CI) p Total30640276.1–––– Time period Before intervention15720277.7RefRefRefRefAfter intervention14920074.50.96 (0.84, 1.10)0.5470.97 (0.88, 1.06)0.475 Age (per 10 years) Age (per 10 years)30640276.1 1.02 (1.01, 1.03) < 0.001 1.03 (1.01, 1.05) 0.001 Gender Male12517671.0RefRefRefRefFemale18122680.1 1.13 (1.03, 1.24) 0.010 1.26 (1.13, 1.41) < 0.001 Department Medicine22829776.8RefRefRefRefSurgery233467.70.88 (0.69, 1.13)0.3180.95 (0.84, 1.08)0.455Emergency Department/Intensive Care172277.31.01 (0.88, 1.15)0.9230.98 (0.88, 1.08)0.637Infectious Diseases233369.70.91 (0.74, 1.11)0.3390.90 (0.73, 1.10)0.294Obstetrics and Gynaecology151693.8 1.22 (1.00, 1.48) 0.045 1.19 (1.11, 1.28) < 0.001 Smoking status Never smoked20627375.5RefRefRefRefCurrent or previous smoker10012977.51.03 (0.99, 1.07)0.168 1.19 (1.14, 1.24) < 0.001 Antibiotic use (current or in the previous 12 months) No11415474.0RefRefRefRefYes18723778.91.07 (0.95, 1.19)0.274 1.11 (1.01, 1.22) 0.024 Recent hospital admission in 12 months No20125678.5RefRefRefRefYes10112978.31.00 (0.88, 1.13)0.9650.95 (0.86, 1.04)0.257 Region Ca Mau (rural)6710265.7RefRefRefRefHanoi (urban)23930079.71.21 (0.96, 1.53)0.106 1.10 (1.05, 1.15) < 0.001 cRR: Crude Risk Ratio aRR: Adjusted Risk RatioAnalysis accounts for clustering at a hospital levelWe used a complete-case approach for both univariable and multivariable analysis, where “don’t know” responses were excluded for analysisBolded text indicates confidence limits do not cross the null Table 6Risk factors for blaCTX-M-producing Enterobacterales carriage by univariable and multivariable analysisVariableParticipants with blaCTX-M producing Enterobacterales carriage (n)Total participants (n)Proportion (%)Univariable modelMultivariable modelcRR (95% CI) p aRR (95% CI) p Total30640276.1–––– Time period Before intervention15720277.7RefRefRefRefAfter intervention14920074.50.96 (0.84, 1.10)0.5470.97 (0.88, 1.06)0.475 Age (per 10 years) Age (per 10 years)30640276.1 1.02 (1.01, 1.03) < 0.001 1.03 (1.01, 1.05) 0.001 Gender Male12517671.0RefRefRefRefFemale18122680.1 1.13 (1.03, 1.24) 0.010 1.26 (1.13, 1.41) < 0.001 Department Medicine22829776.8RefRefRefRefSurgery233467.70.88 (0.69, 1.13)0.3180.95 (0.84, 1.08)0.455Emergency Department/Intensive Care172277.31.01 (0.88, 1.15)0.9230.98 (0.88, 1.08)0.637Infectious Diseases233369.70.91 (0.74, 1.11)0.3390.90 (0.73, 1.10)0.294Obstetrics and Gynaecology151693.8 1.22 (1.00, 1.48) 0.045 1.19 (1.11, 1.28) < 0.001 Smoking status Never smoked20627375.5RefRefRefRefCurrent or previous smoker10012977.51.03 (0.99, 1.07)0.168 1.19 (1.14, 1.24) < 0.001 Antibiotic use (current or in the previous 12 months) No11415474.0RefRefRefRefYes18723778.91.07 (0.95, 1.19)0.274 1.11 (1.01, 1.22) 0.024 Recent hospital admission in 12 months No20125678.5RefRefRefRefYes10112978.31.00 (0.88, 1.13)0.9650.95 (0.86, 1.04)0.257 Region Ca Mau (rural)6710265.7RefRefRefRefHanoi (urban)23930079.71.21 (0.96, 1.53)0.106 1.10 (1.05, 1.15) < 0.001 cRR: Crude Risk Ratio aRR: Adjusted Risk RatioAnalysis accounts for clustering at a hospital levelWe used a complete-case approach for both univariable and multivariable analysis, where “don’t know” responses were excluded for analysisBolded text indicates confidence limits do not cross the null Risk factors for blaCTX-M-producing Enterobacterales carriage by univariable and multivariable analysis cRR: Crude Risk Ratio aRR: Adjusted Risk Ratio Analysis accounts for clustering at a hospital level We used a complete-case approach for both univariable and multivariable analysis, where “don’t know” responses were excluded for analysis Bolded text indicates confidence limits do not cross the null Discussion This before-and-after study found high prevalence of rectal carriage of for blaCTX-M-producing Enterobacterales with relatively low prevalence of representative CPE and MRSA carriage in the four district/regional hospitals in Vietnam. Implementation of AMS interventions decreased the nasal carriage of MRSA but had no significant effect on prevalence of multidrug-resistant Enterobacterales carriage. The high overall population prevalence of rectal carriage of blaCTX-M-producing Enterobacterales in this study (76%) is comparable with the steady and concerning rise of ESBL-producing organism carriage reported in Vietnam [27]. Studies from predominantly tertiary hospitals in Vietnam have documented phenotypic ESBL prevalences of clinical isolates of 25% in 2010–12 [28], 59% in 2016–17 [29] and 65% in 2020–21 [30]. In the community, studies of healthy individuals have shown a prevalence ranging from 35 to 51% in 2013 [5]. Although the literature reporting criteria of ESBL carriage are variable—person-level or organism-level prevalence; phenotypic or genotypic ESBL production analysis; healthy participants or hospitalised patients—reported data collectively indicate widespread ESBL-producing Enterobacterales prevalence in both hospital and community settings in Vietnam [27]. Data from lower levels of the healthcare system, however, remains limited to date, and our results extend existing evidence to district and regional health facilities. Conversely, the prevalence of CPE and MRSA carriage observed in this study were lower than reported in previous research, where 52% of tertiary hospital inpatients were CRE-colonised [9], and 25% of tertiary hospital inpatients were MRSA-colonised [13]. Our study thus adds district-level data to augment current understanding of MRO risk at all levels of the health care system in Vietnam. Persons colonised with MROs not only are at high risk of developing MRO infections over time but also serve as an important source of transmission to other people [31]. Our data thus contribute to the imperative for comprehensive cross-sector surveillance programs to monitor these important differences in MRO colonisation prevalences between tertiary inpatient, community and (as in our study) district hospital inpatients. A high proportion of Enterobacterales isolates were not susceptible to a wide range of commonly-used antimicrobials, including broad-spectrum antibiotics such as ciprofloxacin and cefepime, as demonstrated in previous studies [32]. This likely reflects the long-standing overuse of broad-spectrum antibiotics in the community and in hospitals [33]. These findings are particularly concerning as treatment decisions within lower-level hospitals, with limited or absent access to microbiological services (Table S1), typically rely on empiric Ministry of Health-endorsed treatment regimens, encompassing many of these agents. For instance, while oral ciprofloxacin is the recommended antibiotic for patients presenting with urinary tract infections (UTIs) [34], findings from this study indicate that 77% of Enterobacterales, a common pathogen in UTIs, were non-susceptible to ciprofloxacin. This raises concerns that patients are receiving empiric therapy that may be ineffective while also potentially driving AMR escalation as a result of alterations in the target enzymes (DNA gyrase and topoisomerase IV), entry and efflux, or plasmid-mediated resistance mechanisms [35], induced by antibiotic selection pressure. When faced with clinical failures in such scenarios, clinicians may increasingly resort to using broader-spectrum empirical antibiotics, further complicating efforts to adhere to proper stewardship principles in daily practice. Carbapenems are one example of broader spectrum use that warrants caution. Although meropenem resistance was low in our study, increased use—driven by resistance to other antibiotics—could drive increased carbapenem resistance rates, as observed with ESBL-producing organisms. To mitigate these risks, avoiding the use of antibiotics if not indicated, as well as optimising dose, duration, and choice of antibiotics, ideally supported by laboratory data, is paramount. Qualitative research would also be beneficial in elucidating drivers of broad-spectrum antibiotic use within hospital settings in district hospitals in Vietnam. The prevalence of MRSA carriage showed a significant reduction with AMS interventions across the four hospitals. Some studies have shown a similar reduction in MRSA prevalence with AMS interventions [36, 37]. However, this needs to be interpreted with caution in the context of a low overall prevalence and differences in patient characteristics in the baseline and post-intervention groups (Table 1), which may have resulted in a biased effect estimate. In contrast, no significant changes were seen in the prevalence of ESBL-producing Enterobacterales and CPE carriage after the implementation of the AMS interventions in this study. Phenotypic susceptibility profiles were also largely unaffected by the intervention. The lack of effect observed with an AMS intervention on participant-level carriage of ESBL-producing Enterobacterales and CPE in this study most likely reflects the modest but significant reduction in antimicrobial consumption achieved during the intervention period [38]. It is also important to note that many of the drivers of MRO carriage are not likely to be affected by AMS interventions in healthcare facilities. These include easy access to antibiotics in the community without a prescription, as well as agricultural and animal use of antibiotics with impact on environmental reservoirs [39] which lead to the presence and persistence of MROs in the environment. Persistent hospital environmental sources such as sinks, drains, and sluices—unmeasured in this study—may also contribute to ongoing colonisation. Together, these factors suggest that multifaceted, One Health–aligned strategies extending beyond hospital AMS may be required to meaningfully reduce MRO carriage in similar settings. These differences between MRSA and Gram-negative MRO carriage effects may reflect distinct transmission dynamics and ecological niches. MRSA colonisation is thought to be more closely associated with healthcare-associated transmission and antibiotic selection pressure [40], which may allow more rapid response to stewardship interventions. In contrast, ESBL-producing Enterobacterales and CPE carriage are likely influenced by community, environmental, and agricultural reservoirs, as well as persistent gut colonisation [41], and may therefore be less responsive to short-term facility-based AMS alone. MRSA, ESBL-producing Enterobacterales and CPE carriage in hospitals are all thought to be influenced by infection prevention and control interventions such as handwashing and targeted personal protective equipment use, but these factors were not studied as part of the bundled AMS intervention in this work. Evidence regarding the impact of AMS on MRO prevalence is mixed, with some studies showing a reduction in antibiotic resistance and MRO prevalence, and others not [42, 43]. The most recent Cochrane review on effectiveness of AMS [17] was unable to draw definitive conclusions about its impact on reducing AMR due to heterogenous study designs, as well as concurrent changes in infection control practices that may confounding findings. Individual studies, however, have shown mixed effects [36, 44, 45]. Risk factor analysis showed an increase in blaCTX-M -producing Enterobacterales carriage among patients with current or recent antibiotic exposure, consistent with prior literature [46]. Females, increasing age, urban hospital location and smoking were also risk factors identified in this study, although published evidence for these factors is mixed [47, 48]. Findings should be interpreted cautiously given the potential for residual confounding, reliance on self-reported exposure variables, and broad exposure definitions such as combining current and prior antibiotic use. Higher blaCTX-M carriage in Obstetrics and Gynaecology wards may reflect confounding by sex distribution or the small sample size. Nonetheless, these observations highlight potential targets for AMS and IPC strategies in peripartum populations, warranting confirmation in larger studies. Smoking could also plausibly influence ESBL carriage via immune and gut microbiota alterations and related antibiotic exposure [49, 50], though epidemiological evidence for a direct association is limited. This study had several strengths. Our analyses focus on MRO carriage in local district health facilities at the lower level of the health system, a setting that is under-represented in the literature from Vietnam and other LMICs [17]. Even though the before-and-after cross-sectional study design incurred aspects of temporal bias, it provided valuable observational data on the effect of a bundle of AMS interventions on MRO carriage, an intervention which has been shown to have mixed results in the literature [17]. This study also had several limitations. As this was a repeated cross-sectional before–after study, reductions in MRSA carriage could not be causally attributed to the AMS intervention alone and may have reflected other temporal factors or case-mix differences. Our analysis focused on a representative subset of ESBL and carbapenemase genes—blaCTX-M for ESBLs, and blaNDM-1, blaOXA-48-like and blaKPC for carbapenemase genes. However, the prevalence of other circulating AMR genes was not specifically examined, such as extended-spectrum blaSHV and blaTEM, or blaIMP-4-like, blaNDM-4, blaNDM-5 and blaNDM-7 carbapenemases, leading to under-estimation of overall ESBL and CPE carriage prevalence. Additionally, non-beta-lactamase mechanisms causing resistance such as reduced permeability, efflux pumps or porin defects were not interrogated [51]. The observation that the number of isolates in which a carbapanemase gene was detected slightly exceeded the number of phenotypically meropenem-non-susceptible isolates highlights the potential for carriage of carbapenemase genes without a marked elevation in meropenem MIC, a known phenomenon particularly for blaOXA-48-like or blaOXA-181-like carrying Enterobacterales, underscoring the importance of molecular surveillance alongside phenotypic testing in district hospital settings. Additionally, the small sample size and lower-than-expected prevalence of CPE and MRSA carriage limited the power of analysis of the effect of AMS interventions. Sample size assumptions were based partly on unpublished data and estimated intervention effects. Sensitivity analyses excluding samples with transport temperature deviation were not able to be conducted and so it is unclear if this factor may have affected results. The duration of the AMS intervention may also have been insufficient to show a difference in the prevalence of rectal carriage of Enterobacterales. Ad hoc sampling of inpatients as well as the nature of self-reporting in the participant questionnaire could also have introduced bias to the risk factor analysis. The study was conducted in four district hospitals across two provinces, which may limit the representativeness of findings across all district hospitals in Vietnam. However, data from district and regional hospitals in Vietnam are scarce, and these findings provide important initial evidence from lower-level healthcare settings. This study provides a framework for future MRO surveillance and the design and implementation of AMS programs in this understudied population. Detailed genomic characterisation of MRO isolates would enable a more granular understanding of these factors to inform MRO carriage risk at the district hospital level. Future research should explore the effect of AMS programs on MRO prevalence using a cluster-based randomised controlled study design. Further investigation into the interplay between AMS and IPC practices and their effect on MRO carriage in resource-limited settings would also be valuable. Additionally, applying AMS principles to address out-of-hospital antibiotic access and community-level drivers of resistance from a One Health perspective could provide further insights into developing a holistic framework for combating AMR in LMIC. Given the high observed ESBL-producing Enterobacterales carriage rates at the district hospital level, this study also provides further evidence for the importance of access to routine bacteriology services in such centres. Such microbiological capacity could facilitate accurate susceptibility testing and targeted therapy for MRO-driven infection and monitoring of resistance trends in pathogens causing invasive disease, but would require dedicated resourcing. In conclusion, this before-and-after repeated cross-sectional study in four Vietnamese hospitals found a high prevalence of blaCTX-M-producing Enterobacterales carriage, but relatively low prevalence of representative CPE (blaNDM-1, blaKPC and blaOXA-48-like) and MRSA carriage. Observed reductions in MRO carriage should be interpreted as temporal associations, with further randomised prospective studies warranted to shed light on causal effects These findings inform AMR surveillance priorities in lower-level healthcare settings in Vietnam and other LMICs. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Supplementary Material 1.