Prevalence and Antibiotic Resistance Pattern of Escherichia coli Isolated from Urine Samples of Pediatric Outpatients with Underlying Medical Conditions from Pakistan

Abstract

Background

Urinary tract infections (UTIs) caused by Escherichia coli pose a significant health burden in children, particularly in those with underlying medical conditions. The emergence of antimicrobial resistance, including extended-spectrum β-lactamase (ESBL) production and multidrug resistance (MDR), complicates treatment strategies. Data on resistance patterns in this vulnerable population remain limited in low-income and middle-income countries.

Methods

To determine the prevalence and antibiotic resistance patterns of E. coli isolated from urine samples of pediatric outpatients (1–12 years old) with underlying medical conditions. This hospital-based cross-sectional study was conducted at the Pakistan Institute of Medical Sciences (January 2022–January 2023). Urine samples were collected from 941 pediatric outpatients with documented underlying conditions. E. coli were identified biochemically and molecularly. Antimicrobial susceptibility testing was performed using Kirby-Bauer disc diffusion, with ESBL screening and confirmation.

Results

E. coli was detected in 30.08% (283/941) of samples, which was significantly higher in females than in males (38.1%, p <0.001). ESBL production was detected in 30.4% (86/283) of the isolates and MDR isolates in 18.7% (53/283). The highest rates of resistance were observed against ampicillin (70.3%), cephalexin (60.1%), and trimethoprim-sulfamethoxazole (59.4%). Low resistance was noted to nitrofurantoin (2.8%) and amikacin (1.8%). Age was a significant predictor, with 1-year old children showing 2.1-fold 95% CI: 1.01–4.55) and 3.6‑fold (95% CI: 1.60–8.15) increased odds of ESBL production and MDR, respectively, compared to children aged 6–12 years.

Conclusion

High rates of ESBL-producing and MDR E. coli in Pakistani children with underlying conditions underscore the critical need for antimicrobial stewardship and empirical therapy guided by local susceptibility patterns, particularly in younger children.

Introduction

Urinary tract infection (UTI) is defined as the detection of the pure growth of pathogenic organisms in the urine. In the case of bacteria, the presence of more than 10⁵ pure colony-forming units (CFU) in one milliliter (mL) of urine sample is considered as UTI.¹ UTI are also associated with a combination of signs and symptoms, including dysuria, frequency, urgency, suprapubic pain, fever, or pyuria.² It has been reported that more than 10% of girls and 3% of boys will have had at least one episode of UTI by the age of 16, which makes it one of the most common bacterial infections in the pediatric age group. Factors such as age, sex, race, and circumcision status in boys affect the frequency of UTI.^1,3 The most common cause of bacterial infection is Escherichia coli followed by other bacteria such as Klebsiella, Proteus, and Enterococcus species. In pediatric patients, 90% of UTI are associated with E. coli. Furthermore, kidney stones, inadequate bladder emptying, bladder and bowel dysfunction, and congenital kidney and urinary tract abnormalities are among the primary risk factors for UTI development.⁴ Constipation has also been recognized as a significant modifiable risk factor for pediatric UTIs. The pathophysiological mechanism may involve mechanical compression of the bladder by fecal impaction, leading to incomplete bladder emptying, urinary stasis, and increased susceptibility to bacterial colonization.^5–7

Antibiotic resistance and the rise of extended-spectrum β-lactamase (ESBL) infections are among the increasing issues in public health at a global scale. ESBL bacteria synthesize β-lactamases that they use to break the β lactam antibiotics excluding carbapenems and cephamycins and rendering these antibiotics ineffective. In 2019, the Centre for Disease Control and Prevention (CDC) declared ESBL Gram-negative bacteria as a “serious threat” pathogen in their report.⁸ These types of isolates have been reported to be resistant to many classes of antibiotics such as quinolones and aminoglycosides. This increasing trend has been significantly limiting available treatment options. As a result, in most cases the empirical therapy becomes ineffective, leading to many adverse outcomes including septicemia and renal scarring.⁹ ESBL-associated UTI are increasing worldwide, and a recent study reported that infections from ESBL E. coli increased from 30.18 in 2012 to 47.81 in 2021 per ten thousand discharges.^10,11 Furthermore, the highest cases of UTI have been reported from Asian countries.¹⁰ In the WHO Eastern Mediterranean Region, nearly 1 in 3 infections are resistant to antibiotics, with rates among the highest globally.¹² Importantly, over 41% of AMR-attributable deaths in the Eastern Mediterranean Region occur in Pakistan, where antimicrobial consumption exceeds both regional and global averages.^12,13 A recent WHO report estimates that AMR contributes to over 200,000 annual deaths in Pakistan, underscoring the urgent need for context-specific surveillance data to inform national and regional antimicrobial stewardship strategies.¹⁴

Pakistan faces one of the world’s highest burdens of antimicrobial resistance, with recent studies documenting alarming resistance rates in pediatric populations.¹⁵ A multicenter study from Lahore reported that 66% of E. coli isolates from pediatric UTI patients were MDR, with amoxicillin, ampicillin, and cefuroxime showing ≥90% resistance.⁸ Similarly, a study from Faisalabad found significant resistance to amoxicillin-clavulanate, ceftriaxone, ampicillin, and trimethoprim-sulfamethoxazole among ESBL-producing E. coli isolates from children.¹⁶ Concerningly, a survey of healthy infants in Pakistan revealed 43% were colonized with ESBL-producing Enterobacteriaceae, suggesting widespread community transmission of resistant organisms.⁴ In a tertiary care hospital in Rawalpindi, 67% of E. coli isolates from children under five years of age were resistant to three or more drug classes.¹⁵ These findings highlight that resistance is not confined to healthcare settings but is endemic in the community, posing a serious threat to empirical therapy.¹⁵

The problem is compounded by antibiotic prescribing practices in Pakistan. A point prevalence survey across 14 hospitals in Punjab found that 82% of hospitalized pediatric patients were receiving antibiotics, with 72% of prescriptions falling under the WHO’s “Watch” category.^3,17 Alarmingly, only 2% of antibiotic use was guided by culture sensitivity reports, indicating a near-total reliance on empirical therapy.^3,17 Similarly, in pediatric outpatient settings, Watch antibiotics comprised 69% of prescriptions, with azithromycin (19.5%) and amoxicillin-clavulanate (19.1%) being most commonly prescribed.¹⁰ In an outpatient clinic near Lahore, 40.5% of patients received antibiotics, with only 23.4% of prescriptions concordant with recommended first- or second-line therapies.^2,3 These patterns reflect systemic challenges: antibiotics are cheap and readily available over-the-counter, physicians face high patient loads, and stewardship programs remain inadequately implemented despite Pakistan’s National Action Plan against AMR being in effect since 2017.^2,3

Pediatric patients with underlying medical conditions, including congenital anomalies of the kidney and urinary tract (CAKUT), bowel and bladder dysfunction, spinal dysraphism, and immune deficiency, are at increased risk of developing UTIs. Furthermore, constipation has been identified as a prevalent and modifiable risk factor in this population.^18,19 A recent study of children with underlying urologic problems reported that 70% of UTIs are resistant to at least one antibiotic, while more than 40% have been reported as multidrug resistant (MDR).²⁰ These studies highlight antibiotic resistance as a potential hurdle in treating UTI in children with underlying medical conditions. In this vulnerable population, underlying comorbidities often necessitate broader-spectrum empirical therapy, increase the risk of treatment failure, and predispose patients to complications such as renal scarring and bacteremia.^18,21 However, to our knowledge, no study from Pakistan has specifically examined antibiotic resistance patterns of E. coli in pediatric outpatients with underlying medical conditions—a population that is particularly vulnerable due to frequent healthcare exposure, recurrent infections, and limited therapeutic options.^22–24

Furthermore, age-specific resistance‑risk stratification in this subgroup remains unexplored. Our study addresses these gaps by providing comprehensive resistance data, including ESBL and MDR prevalence, stratified by age, sex, underlying condition, and weight status, thereby offering actionable data for empirical therapy selection in this high-risk pediatric population. Appropriate empirical antibiotic therapy is critical in this population to prevent complications such as renal scarring, bacteremia, and prolonged hospitalization. However, empirical choices are often made without local resistance data, leading to treatment failures and further resistance selection. Given the high resistance rates reported in Pakistan and the lack of comorbidity-specific data, empirical guidelines derived from general populations or developed countries may not be applicable to Pakistani children with underlying conditions. Therefore, population and region-specific data, particularly those stratified by underlying medical conditions, are urgently needed to inform effective clinical decision making and antimicrobial stewardship. Therefore, the current study was designed to investigate the prevalence of E. coli and its antibiotic resistance profile, including ESBL production and MDR patterns, in urine samples from pediatric outpatients with underlying medical conditions. By stratifying resistance data by age, sex, comorbidity type, and weight status, we aimed to identify high-risk subgroups and provide evidence to guide empirical antibiotic selection in this understudied population.

Methodology and Methods

Study Design, Population, and Sampling

This hospital-based, cross-sectional study was conducted between January 2022 and January 2023 at the Pakistan Institute of Medical Sciences (PIMS) Hospital Islamabad. Pediatric outpatients aged 1–12 years with documented underlying medical conditions and clinically suspected UTI were enrolled in the study.

To ensure that the enrolled sample was sufficient for the primary objective of estimating E. coli prevalence with acceptable precision, the sample size was calculated using the following formula:

Where

Z = 1.96 (95% confidence level)

p = 0.25 (estimated prevalence based on prior regional studies)^25,26

d = 0.035 (margin of error).

n = 588

This yielded a minimum of 588 participants, adjusted to 653 to account for 10% sample loss. The final enrollment of 941 patients exceeds this requirement, confirming adequate statistical power (>80%) for the primary prevalence estimate. Subgroup analyses (eg, by sex, age, or specific underlying conditions) were secondary objectives; the study was adequately powered for larger subgroups (eg, sex, age categories) but had limited power for smaller subgroups (eg, neurogenic bladder, n=7), which we acknowledge as a limitation.

Common signs of suspected UTI include fever without a clear source, painful or frequent urination, abdominal or back pain, vomiting, bedwetting, irritability, poor feeding, and foul-smelling urine.^2,21 Patients were enrolled following a predefined inclusion and exclusion criterion. Patients of both sexes aged 1–12 years, with at least one underlying medical condition (Table 1), and clinically indicated urine culture ordered by a physician were included in the study. Constipation was included as a documented comorbid condition based on physician diagnosis recorded in the patient’s medical history. The presence of constipation was noted as a potential contributory factor to UTI risk, consistent with published literature demonstrating an association between constipation and increased UTI susceptibility in children.^27,28 Pediatric outpatients aged < 1 year and > 12 years, those who used antibiotics within 72 hours, and those with indwelling devices were excluded from the study. The 72-hour exclusion criterion was applied to avoid the confounding effects of recent antibiotic exposure on culture results and antimicrobial susceptibility testing. Antibiotics present in the urine of recently treated patients can suppress bacterial growth, leading to false-negative cultures or reduced zone diameters in susceptibility testing, thereby potentially overestimating resistance.^29–31 Furthermore, excluding these patients ensures that our resistance data reflect the true susceptibility profiles of community-acquired E. coli isolates in this population, providing clinically relevant information for empirical therapy selection in outpatient settings. Prior to enrolment, informed written consent was obtained from the parents of the children, and only those who consented to participate were included in the study. The study was approved by the Bioethical Committee of Quaid-I-Azam University Islamabad, Pakistan, under Ethical number BEC-FBS-QAU2018-109. All the experiments were performed in accordance with the guidelines.

Table 1.

Details of the Underlying Medical Conditions Selected for the Current Study

Type of Condition	Underlying Medical Condition	Specific Criteria
Urologic	Vesicoureteral reflux	Grades III–V
	Posterior urethral valves
	Hydronephrosis
	Duplex collecting system
Neurologic	Neurogenic bladder	Due to spina bifida, spinal cord injury, or cerebral palsy with bladder involvement
Immunologic	Primary immunodeficiency
	Chronic steroid use	≥ 2 weeks in the past 3 months
	Chemotherapy	Within the past 6 months
Metabolic/Endocrine	Diabetes mellitus type 1	HbA1c > 7.5%
Gastrointestinal/Hepatic	Constipation Chronic liver disease	Diagnosed by physician during checkup, or documented in medical history
	Short gut syndrome
Cardiac	Congenital heart disease	Only if on chronic diuretics or with prior endocarditis
Multiple	Two or more of the above conditions	From any category

For analytical purposes, patients were stratified into three age groups: 1 year (infants), 2–5 years (preschool), and 6–12 years (school age).

1 year old: This period is characterized by the highest cumulative incidence of febrile UTI in early childhood, a greater risk of renal parenchymal injury, and a higher prevalence of congenital anomalies of the CAKUT, including VUR. Infants also present with non-specific symptoms (eg, fever without a source, poor feeding, irritability), which complicates diagnosis and often leads to empirical antibiotic exposure before culture results are available.³² Younger age has been identified as an independent risk factor for infection with ESBL-producing uropathogens, alongside prior antibiotic use and underlying urologic abnormalities.³³

2–5 years (Preschool age): By preschool age, toilet training is typically completed, and bladder–bowel dysfunction (BBD) including voiding postponement, incomplete emptying, and constipation emerges as a dominant, modifiable risk factor for recurrent UTI, particularly in children with VUR.³⁴ The prevalence of high-grade congenital anomalies generally declines with age, while functional factors become more prominent in driving recurrence.²⁸ Global and regional burden analyses indicate that UTI incidence remains substantial in the 2–4-year age band, supporting the clinical relevance of this stratum for epidemiologic and resistance-pattern analyses.³⁵

6–12 years (school age): In school-aged children, UTIs are less frequent overall but are more likely to occur in the context of recognized anatomic or functional abnormalities, recurrent infection, or persistent BBD.³⁶ By this age, most significant congenital urologic conditions have already been diagnosed, and recurrent UTIs are frequently linked to voiding dysfunction and constipation rather than newly discovered structural disease.³⁷ In some populations, early adolescence within this bracket may introduce additional behavioral risk factors (eg, sexual activity), although this is more prominent beyond age 12.³⁸

Data and Sample Collection

Demographic data (age, sex, weight, and height), clinical history (history of previous UTIs, prior hospitalization, and antibiotic use), medical history, and urine analysis results were recorded from the patients.

Anthropometric measurements were performed by trained nursing staff using standardized protocols. Weight was measured to the nearest 0.1 kg using a calibrated digital scale (LifeCare Weight Scale Analog Body Weight Machine BR2017, Pakistan) with children wearing light clothing and no shoes. Height/length was measured to the nearest 0.1 cm using a stadiometer (Seca 213) for children ≥2 years who could stand unassisted, and an infantometer (Seca 416) for children <2 years. All equipment was calibrated daily using standardized weights and rods.

Urine samples were collected from the children through two approaches such as mid-stream urine (MSU) and clean-catch (CCU) according to the age and ability of the child to void on command as per the recommendation of the World Health Organization.³⁹ For children who were able to void on command, MSU was collected. They were asked to do so in a sterile container, discarding the first few milliliters of urine from the distal urethra. Parents or guardians were instructed to clean the external genitalia and collect samples. Children unable to void on command and CCU were collected, with slight modifications. Briefly, sterile pediatric urine collection bags were applied after cleaning the periureteral area. The child was held, and the back was gently stroked. The bags were checked every 30 min and were removed within one hour. If no sample was obtained, a new sterile bag was used.

The collected samples were immediately transported to the laboratory at room temperature.

Bacterial Isolation and Identification

Bacterial Isolation

A volume of 10 µL from the urine samples was incubated on cystine-lactose-electrolyte-deficient (CLED) agar plates and incubated at 37°C for 24 hours.⁴⁰ The positive samples were then sub-cultured onto chocolate, blood, and MacConkey agar. Samples with no growth after 24 h were re-incubated for an additional 24 h before being reported as negative.

Gram Staining and Biochemical Confirmation

Isolated pure colonies were subjected to Gram staining and gram-positive isolates were omitted from the study. For biochemical identification of gram-negative isolates, a series of confirmatory tests were performed to differentiate E. coli from other Enterobacteriaceae.

Fresh, pure, single colonies of the gram-negative isolates were inoculated onto triple sugar iron (TSI) agar (Oxoid, Basingstoke, Hampshire, United Kingdom) and incubated at 37°C for 24 h. TSI agar differentiates bacteria based on their ability to ferment glucose, lactose, and/or sucrose, and to produce hydrogen sulfide (H₂S). E. coli characteristically produces an acid slant and acid butt (yellow/yellow) with gas production, indicating fermentation of both glucose and lactose, and does not produce H₂S. This pattern distinguishes E. coli from Salmonella (alkaline slant with H₂S production) and Shigella (alkaline slant no gas, no H₂S).⁴¹

The isolates were inoculated onto Simmons’ citrate agar (Oxoid, Basingstoke, Hampshire, UK) and incubated at 37°C for 24 h. Citrate utilization is indicated by a color change from green to blue, reflecting alkalinity from ammonium salt utilization. E. coli is typically citrate-negative (green), distinguishing it from Klebsiella and Enterobacter species, which are citrate-positive.^19,42

Additionally, motility was assessed using sulfide indole motility (SIM) medium inoculated at 37°C for 24 hours. E. coli is motile, demonstrated by diffuse growth radiating from the inoculation line, whereas non-motile organisms such as Klebsiella and Shigella show growth confined to the stab line. The combination of motility, indole positivity (detected by Kovac’s reagent), and lack of H₂S production on SIM medium further confirms the identification of E. coli.⁴³

Molecular Identification of the Isolates

The ON-Rep-Seq technique, as previously described, was used to confirm molecular identification of the isolates.⁴⁴ DNA was extracted from freshly cultured colonies using the InstaGene™ Matrix (Bio-Rad Laboratories, California, USA), visualized using 2% gel electrophoresis, and quantified with the Qubit™ 4 fluorometer (Thermo Fisher Scientific, Massachusetts, USA). The samples were prepared for the ON-Rep PCR cycles. Briefly, a 1µL sample of DNA (~20 ng/µL) was combined with 5 µL of PCRBIO-HiFi buffer, 0.25 µL of PCRBIO-HiFi Polymerase (PCR Biosystems Ltd, London, UK), 4 µL of (GTG)5 primer, and MiliQ water to achieve a total volume of 25 µL. For PCR-1, the thermocycler settings (SureCycler 8800, Agilent, CA, USA) were configured to 95°C for 5 minutes; followed by 30 cycles of 95°C for 30 seconds, 45°C for 1 minute, and 62°C for 4 minutes, concluding with a final elongation step at 72°C for 5 minutes.

Upon finishing the initial cycle, 1 µL of the Rep-PCR-1 product was combined with 12 µL of PCRBIO-UltraMix (PCR Biosystems Ltd, London, UK), 2 µL of repBC primers (10 µM), and 10 µL of nuclease-free water. The thermocycler was calibrated to execute 3 cycles at 95°C for 30 seconds, 45°C for 1 minute, and 62°C for 4 minutes, succeeded by 10 cycles at 95°C for 30 seconds, 65°C for 1 minute, and 72°C for 4 minutes, concluding with an elongation phase at 72°C for 5 minutes.

Following Rep-PCR-2, the quantification of DNA was conducted using the Qubit^® dsDNA BR Assay Kit (Life Technologies, CA, USA). The measurement was documented using the Varioskan Flash Multimode Reader (Thermo Fischer Scientific, MA, USA), with fluorescence captured at 485/530 nm, followed by the amalgamation of 10 µL from each sample into a library.

The sequencing of the generated library was conducted using Oxford Nanopore. The sequenced (ON – rep – seq) data, following the preliminary base calling, underwent de-multiplexing of fastq files to produce read length count profiles (LCps) derived from the distribution of sequence lengths. Subsequently, readings corresponding to each peak were consolidated using VSEARCH, refined with Canu, and subsequently categorized taxonomically with adjusted quality reads. Ultimately, the clues were analysed to infer species and strains based on LCps.⁴⁴ Following PCR, 10 µL of each sample was used to create a sequencing library, which was subsequently sequenced using MinION (Oxford Nanopore Technologies, Oxford, UK). The sequenced ON-Rep-Seq data were demultiplexed into fastq format to generate read length count profiles (LCps). Reads associated with each peak were clustered using VSEARCH, refined using Canu, and taxonomically categorized.

Antimicrobial Sensitivity Testing (AST)

AST was performed using the Kirby-Bauer disk diffusion method, following the Clinical and Laboratory Standard Institute (CLSI) guidelines.⁴⁵ An antibiotic panel was selected to reflect agents commonly used in both oral and parenteral pediatric therapy. The panel included ampicillin (10 µg), amoxicillin-clavulanate (20/10 µg), cephalexin (30 µg), ceftriaxone (30 µg), cefotaxime (30 µg), meropenem (10 µg), gentamicin (10 µg), amikacin (30 µg), ciprofloxacin (5 µg), trimethoprim-sulfamethoxazole (1.25/23.75 µg), nitrofurantoin (300 µg), and colistin (10 µg). For colistin susceptibility testing, reference broth microdilution was performed in cation-adjusted Mueller–Hinton broth (CAMHB; Becton Dickinson, USA), using untreated polystyrene microdilution plates and following the CLSI–EUCAST reference method.⁴⁶ Colistin sulfate powder (Sigma-Aldrich, USA) was dissolved in sterile distilled water and serially diluted twofold in CAMHB to obtain final concentrations of 0.0625–64 µg/mL. A bacterial suspension prepared from a fresh overnight culture was adjusted to a 0.5 McFarland standard and diluted to achieve a final inoculum of approximately 5×10⁵ CFU/mL per well. Plates were incubated aerobically at 35 ± 2°C for 16–20 hours. The MIC was defined as the lowest colistin concentration showing complete inhibition of visible growth. E. coli ATCC 25922 was included as a quality-control strain. Results were interpreted using the applicable CLSI 2024 criterion for Enterobacterales: susceptible at an MIC of ≤2 µg/mL and resistant at an MIC of ≥4 µg/mL; no intermediate category was assigned. Breakpoints for susceptibility (S), intermediate (I), and resistance (R) were interpreted according to CLSI 2024 and EUCAST guidelines for cephalexin.^45,47 Ciprofloxacin was included in the susceptibility panel despite its restricted use in pediatric populations for two main reasons: (i) it serves as a sentinel marker for fluoroquinolone resistance among Enterobacteriaceae, supporting epidemiological surveillance of resistance trends; and (ii) it may be considered in specific clinical scenarios where other treatment options are exhausted, such as complicated UTIs caused by MDR organisms with limited therapeutic alternatives.^48,49 These results should not be interpreted as endorsing routine ciprofloxacin use in children, but rather as providing reference data for resistance surveillance and stewardship.

Fresh overnight cultures were used to prepare inocula, which were adjusted to a 0.5 McFarland standard and evenly spread onto Mueller–Hinton agar (MHA) plates using sterile cotton swabs. The antibiotic discs were then placed on the agar surface. After incubation at 37°C for 24 h, the zone of inhibition (ZOI) was measured and the results were interpreted as resistant, susceptible, or intermediate. E. coli ATCC 25922 was used as a quality control strain, tested with each batch of antimicrobial susceptibility testing. In addition, isolates were classified as MDR if they were not susceptible to at least one agent in three or more tested antibiotic classes.⁵⁰

In addition to the standard AST, we conducted screening and confirmatory tests for ESBL production. Screening criteria were zone diameter of ≤ 22 mm for cefotaxime or ≤ 27 for ceftazidime.⁵¹ The combined disc test, cefotaxime (30 µg) vs cefotaxime with clavulanate (30/10 µg), was used for confirmation. ESBL production was confirmed by an increase in zone diameter ≥ 5 mm in the presence of clavulanate.⁵² For greater accuracy, minimum inhibitory concentration (MIC) determination was performed on a subset of isolates, namely all ESBL-positive isolates and 20% (n = 39) of randomly selected non-ESBL isolates, using the broth microdilution method (Sensititre, Thermo Fisher).⁵³ Simple random sampling without replacement was used for randomization. Each non-ESBL isolate was assigned a unique number from 1 to 197. Using R software (version 4.6.0), 39 unique random numbers were generated (set.seed = 2023). The isolates corresponding to these numbers were selected for MIC testing.

Data and Statistics Analysis

Data were collected using Microsoft Excel vr 2019. All demographic, clinical, and laboratory data were complete for all enrolled patients, with no missing values for any variable included in the analysis. Data collection forms were reviewed for completeness at the time of enrollment, and any incomplete forms were resolved through direct communication with the families of the children and the medical team. Therefore, no missing data imputation methods were required. Data cleaning was performed in Excel, while all analyses and visualizations were performed using R v. 4.6.0. The weight and height of the children were used to calculate BMI and weight status classification. For children aged ≥ 24 months, Z-scores of BMI-for-age were calculated using the Centers for Disease Control and Prevention reference charts for 2–20 years.⁵⁴ Gender-specific LMS parameters (L = Box-Cox power transformation, M = median, S = coefficient of variation) were obtained from the CDC website (https://www.cdc.gov/growthcharts/), and Z-scores were calculated using the following formula.^54,55

where BMI is the child’s measured body mass index, and L, M, and S are the age- and sex-specific parameters from the CDC reference. Children were then classified into weight categories based on their BMI-for-age Z-scores according to CDC criteria: underweight (Z-score < −2 standard deviations [SD]), normal weight (−2 SD ≤ Z-score ≤ 1 SD), overweight (1 SD < Z-score ≤ 2 SD), and obese (Z-score > 2 SD). These cutoffs correspond to less than the 5^th percentile, 5^th to 85^th percentile, 85^th to 95^th percentile, and greater than the 95^th percentile, respectively.

Weight status classification was not performed for children under 24 months of age, as validated BMI-for-age Z-score reference standards are not available for this age group in the CDC growth charts. These children were included in all other analyses but excluded from weight status subgroup analyses. Categorical variables were compared using the Chi-square or Fisher’s exact test, when appropriate. The Mann–Whitney U-test was used for continuous variables that were not normally distributed. Post hoc test was used for multiple comparison.

Regression models were fitted using the glm() function from the base R stats package. Results were visualized using the ggplot2 package. Univariate regression was performed to identify factors associated with ESBL production and MDR status. Candidate predictor variables including sex, age (both as continuous and categorical), BMI categories, and underlying medical conditions were assessed individually in univariate models. For each univariate model, odds ratios (ORs), 95% confidence intervals (CIs), and p values were calculated.

Given the collinearity between continuous age and categorical age groups (VIF = 4.23), separate multivariate models were constructed for each age parameter to avoid variance inflation and biased coefficient estimates. The same predictor candidate variables were included in the multivariate models as were included in the univariate model, with the only difference being that in one model age was included as a continuous variable and in the second model age was included as a categorical variable. A two-sided p value < 0.05 was considered statistically significant.

Results

Baseline Characteristics

In the current study, a total of 941 pediatric outpatients were included, of which 283 (30.08%) were found to be E. coli positive and 658 (69.92%) were negative (Table 2). A significant difference was observed in terms of sex (p < 0.001). Females accounted for 72% (204/283) of the positive cases compared to 50% (331/658) of the negative cases. The patients were divided into three groups: 1 year, 2–5 years, and 6–12 years old. The majority of children were in the 6–12 years group (49%), followed by 2–5 years (38%), and 1 year old (14%). There was no significant difference in the age group distribution between patients with positive and negative results. Constipation was found to be the most common underlying medical condition overall (93%), with no significant difference in the distribution between the groups (Table 2). Furthermore, among the weight statuses assessed in children aged ≥ 2 years (n = 812), normal weight was the most common category (59%), followed by underweight (23%). No significant differences in the weight status distribution were observed (Table 2).

Table 2.

Baseline Characteristics of the Pediatric Outpatients Included in the Current Study

Characteristic	Overall n = 941 (100%)	Negative n = 658 (69.92%)	Positive n = 283 (30.08%)	p-value
Gender				< 0.001
Male	406 (43%)	327 (50%)	79 (28%)
Female	535 (57%)	331 (50%)	204 (72%)
Age group				0.6
1 Year old	129 (14%)	91 (14%)	38 (13%)
2–5 Years old	354 (38%)	241 (37%)	113 (40%)
6–12 Years old	458 (49%)	326 (50%)	132 (47%)
Underlying medical condition				0.3
Constipation	876 (93%)	619 (94%)	257 (91%)
Neurogenic Bladder	7 (0.7%)	4 (0.6%)	3 (1.1%)
UTA	37 (3.9%)	22 (3.3%)	15 (5.3%)
VUR	21 (2.2%)	13 (2.0%)	8 (2.8%)
Weight status (n = 812 for ≥2 years)				0.13
Underweight	183 (23%)	136 (24%)	47 (19%)
Normal weight	483 (59%)	326 (57%)	157 (64%)
Overweight	116 (14%)	80 (14%)	36 (15%)
Obese	30 (3.7%)	25 (4.4%)	5 (2.0%)

Notes: Significant values (p < 0.050 are written in bold text.

Abbreviations: UTA, Urinary Tract Abnormalities; VUR, Viscouretheral Reflux.

Prevalence of E. coli in Pediatric Patients

Among the 941 collected urine samples, 283 isolates were identified as E. coli through Gram staining and biochemical characterization, and their identity was confirmed through Rep Seq analysis. From the 941 samples, 283 (30.08%) showed significant bacterial growth (>10⁵ CFU/mL) on CLED agar after 24 to 48 h of incubation. All 283 isolates were subjected to Gram staining, which confirmed them as gram-negative rods. Biochemical confirmation using TSI agar, Simmons’ citrate agar, and SIM medium was performed on all isolates, which exhibited the characteristic E. coli profile. These include acid/acid reaction on TSI with gas production, citrate-negative, and motile with indole production. Molecular identification using ON-Rep-Seq was subsequently performed on all biochemically confirmed isolates, and sequence analysis confirmed as E. coli. The sequence data were submitted to the NCBI database under accession number PRJNA1483328.

A 30.08% E. coli prevenance was observed among the total patients with 95% CI = 27.2–33.1% (Figure 1A). Among the sexes, E. coli infection was significantly prevalent in 38.1% (p < 0.001) of females as compared to male patients (19.5%) (Figure 1B). Among the different age groups, E. coli was highly prevalent (31.9%) in 2–5 years old children followed by toddlers (29.5%) and 6–12 year old (28.8%), with no significant difference within the groups (Figure 1C). Among patients with documented underlying conditions, the prevalence was highest in those with neurogenic bladder (42.9%, 95% CI = 9.9–81.6%), followed by urinary tract abnormalities (40.5%, 95% CI = 24.8–57.9%) and vesicoureteral reflux (VUR) (38.1%, 95% CI = 18.1–61.6%). Notably, among the large subgroup of patients with constipation (n = 876), the prevalence was 29.3% (95% CI = 26.3–32.5%), closely mirroring the overall cohort estimate (Figure 1D). Regarding the weight status, the prevalence of E. coli varied among the groups, but the differences were not statistically significant (Figure 1E).

Figure 1.

Prevalence of E. coli in pediatric outpatients in the current study. (A) Overall prevalence of E. coli in all the collected samples. (B) Prevalence of E. coli among male and female patients. (C) Prevalence of E. coli among different age groups. (D) Prevalence of E. coli in among different underlying medical conditions. (E) Prevalence of E. coli among patients with different body weight statuses. The level of significance is presented above each group.

Antibiotic Resistance Patterns Among E. coli Isolates

Among the total E. coli isolates, 264 (93.28%) exhibited resistance to at least one antibiotic based on primary susceptibility testing. ESBL production was detected in 86 isolates (30.4%), whereas MDR was observed in 53 isolates (18.7%) (Figure 2A). Among the ESBL-producing isolates, 61.6% (53/86) were classified as MDR, compared to 38.4% (33/86) of non-MDR isolates (p < 0.001) (Figure 2B). Among the MDR isolates, resistance was predominantly observed against cephalosporins (ceftriaxone and cefotaxime 100%; cefazolin 92.5%), penicillin (ampicillin 88.7%), and trimethoprim-sulfamethoxazole (79.2%), followed by ciprofloxacin (69.8%) and tetracycline (67.9%). All isolates remained susceptible to meropenem and colistin, with amikacin (98.1%) and nitrofurantoin (96.2%) demonstrating excellent preservation (Supplementary Table S1). MIC confirmation was performed for 125 isolates, 86 ESBL-producing isolates, and 39 randomly selected non-ESBL isolates. The MIC results were consistent with the disc diffusion results. For cefotaxime, the MIC for ESBL-producing isolates were 54.70 ± 28.23 µg/mL compared to 1.83 ± 0.98 µg/mL for non-ESBL isolates. The ESBL-positive isolates had MIC values of 53.02 ± 42.77 µg/mL for cefotaxime, confirming the ESBL phenotype (Supplementary Table S2).

Figure 2.

Antibiotic resistance pattern among E. coli isolates of the current study. (A) Overall resistance pattern including resistance (resistant to at least one antibiotic), susceptible isolates, MDR and ESBL producing isolates. (B) Co-occurrence of MDR and ESBL isolates. (C) Proportion of E. coli isolates to different antibiotics. (D) Proportion of ESBL and MDR isolates in male and females. (E) Proportion of ESBL and MDR isolates in different age groups. (F) Proportion of ESBL and MDR isolates in patients with different underlying conditions. Caution should be exercised when interpreting the estimates for neurogenic bladder, VUR, and urinary tract abnormalities due to small sample sizes, which result in wide confidence intervals and statistical fragility. The resistance patterns observed in these subgroups should be considered preliminary and hypothesis-generating rather than definitive. (G) Proportion of ESBL and MDR isolates in patients with different weight statuses. The obese subgroup estimate should be interpreted with particular caution due to the very small sample size.

For individual antibiotics, resistance rates against ampicillin were 70.3% (n = 199), followed by cephalexin at 60.1% (n = 170), and trimethoprim-sulfamethoxazole at 59.4% (n = 168). Notably, no resistance was detected to meropenem or colistin. Detailed results are presented in Figure 2C.

Furthermore, the resistance pattern was assessed among different subgroups, including gender, age groups, underlying medical conditions, and body weight status. In the sex group, the number of ESBL and MDR isolates was comparatively high in females (31.86% and 20.10%, respectively), with no significant difference between the sexes. However, ESBL ESBL-producing isolates were significantly (p < 0.05) higher than MDR isolates in both male and female patients (Figure 2D).

Resistance patterns varied across age categories but did not reach statistical significance among the groups. ESBL production was significantly (p = 0.002) high in the 1-year old group (36.2%) compared to in the 2–5 years old (31.1%) and 6–12 years old groups (29.3%) (p = 0.588). Similarly, MDR rates were also significantly (p = 0.003) elevated in the 1-year old group (34.0%) relative to the 2–5 years old (29.7%) and 6–12 years old groups (28.3%) (p = 0.672) (Figure 2E). However, ESBL-producing isolates were significantly (p < 0.001) high in the 2–5 years old and 6–12 years old groups (Figure 2E).

Children with underlying medical conditions showed higher resistance proportions across all metrics, with no significant differences between the groups. ESBL production was the most prevalent in 36.2% of E. coli isolates from children with urinary tract abnormalities, followed by constipation (30.3%). ESBL-producing isolates were also high in these two groups, but the difference was not statistically significant (Figure 2F).

Analysis by weight status showed notable variations in the resistance patterns. ESBL-producing isolates were more frequent in normal-weight children (29.3%) and underweight children (27.4%) than in obese and overweight children. MDR rates were high in normal-weight children (17.2%), followed by underweight children (17.7%); however, for both ESBL and MDR, the difference was not significant. Furthermore, the ESBL-producing isolates were significantly higher (p < 0.05) only in underweight and normal-weight children than in MDR isolates (Figure 2G).

Factors Associated with Uropathogenic E. coli, ESBL-Production, and Multidrug Resistance

In univariate analysis, age was found to be significantly associated with ESBL and MDR status. Age as a continuous variable was significantly associated with ESBL production (OR = 0.908, 95% CI = 0.841–0.979; p = 0.012) and MDR status (OR = 0.898, 95% CI = 0.819–0.984; p = 0.021), indicating that the odds of resistance decreased with increasing age. When age was examined categorically, the 1‑year‑old group had the highest risk, with 2.1‑fold increased odds of ESBL production (OR = 2.138, 95% CI = 1.010–4.551; p = 0.046) and 3.6‑fold increased odds of MDR (OR = 3.598, 95% CI = 1.596–8.154; p = 0.002) compared to the combined older age groups. Children aged 2–5 years showed significantly lower odds of ESBL (OR = 0.467, 95% CI = 0.311–0.688; p = 0.001) and MDR (OR = 0.202, 95% CI = 0.119–0.323; p < 0.001). The 6–12 years age group did not show statistically significant differences in ESBL (OR = 0.656, 95% CI = 0.371–1.153; p = 0.143) or MDR (OR = 0.781, 95% CI = 0.385–1.577; p = 0.489) compared to the 1‑year‑old reference (Table 3 and Figure 3).

Table 3.

Univariate Analysis of Factors Associated with E. coli-Associated Urinary Tract Infection, ESBL-Production, and Multidrug Resistance in Pediatric Outpatients

Factors	ESBL Production		p value	MDR Status		p value
	OR	95% CI		OR	95% CI
Male	0.744	0.427–1.367	0.386	0.712	0.339–1.402	0.343
Female	1.291	0.731–2.341	0.386	1.404	0.713–2.943	0.343
Age	0.908	0.841–0.979	0.012	0.898	0.819–0.984	0.021
1 Year old	2.138	1.010–4.551	0.046	3.598	1.596–8.154	0.002
2–5 Years old	0.467	0.311–0.688	0.001	0.202	0.119–0.323	0.001
6–12 Years old	0.656	0.371–1.153	0.143	0.781	0.385–1.577	0.489
Normal weight	1.083	0.533–2.300	0.827	1.419	0579–4.013	0.470
Obese/Overweight	0.548	0.236–1.308	0.213	0.520	0.147–1.437	0.249
Underweight	0.922	0.434–1.873	0.827	0.704	0.249–1.724	0.470
Constipation	0.808	0.352–1.970	0.623	0.591	0.244–1.588	0.265
Other conditions	0.529	0.225–1.161	0.122	1.689	0.628–4.097	0.265

Notes: Significant values (p < 0.05) are written in bold text.

Figure 3.

Probability of ESBL and MDR in E. coli among pediatric outpatients, by patient age.

To address collinearity between age as a continuous variable and categorical age groups (VIF = 4.23), we constructed two separate models. In Model 1, which included continuous age, increasing age was independently associated with lower odds of both ESBL production (OR = 0.91, 95% CI: 0.84–0.98; p = 0.017) and MDR (OR = 0.91, 95% CI: 0.83–1.00; p = 0.047), suggesting a protective effect of older age. In Model 2, using categorical age groups, children aged 2–5 years exhibited significantly reduced odds of ESBL production (OR = 0.31, 95% CI: 0.014–0.70; p = 0.005) and MDR (OR = 0.22, 95% CI: 0.09–0.54; p = 0.007) compared to the reference group. Of particular clinical importance, the 1-year-old age group consistently demonstrated the highest risk across both univariate and multivariate analyses, with a significant association for ESBL (OR = 0.46, 95% CI: 0.20–1.07; p = 0.072) and a strong, statistically significant association for MDR (OR = 1.82, 95% CI: 0.63–2.69; p = 0.002), underscoring infancy as a critical risk window. Notably, sex, BMI categories, and underlying gastrointestinal conditions did not emerge as independent predictors of either ESBL production or MDR status in our adjusted models, suggesting that age particularly early childhood is the dominant demographic factor driving resistance in this outpatient pediatric cohort (Table 4).

Table 4.

Multivariate Analysis of Factors Associated with E. coli-Associated Urinary Tract Infection, ESBL-Production, and Multidrug Resistance in Pediatric Outpatients

Factors	ESBL Production		p value	MDR Status		p value
	OR	95% CI		OR	95% CI
Male	0.840	0.460–1.520	0.574	0.580	0.240–1.270	0.196
Female	1.200	0.720–2.300	0.310	1.360	0.607–2.910	0.411
Age	0.910	0.840–0.980	0.017	0.910	0.830–1.000	0.047
1 Year old	0.460	0.200–1.070	0.072	1.820	0.630–2.690	0.002
2–5 Years old	0.310	0.014–0.700	0.005	0.220	0.090–0.540	0.007
6–12 Years old	0.656	0.371–1.153	0.143	0.781	0.385–1.577	0.489
Normal weight	1.060	0.511–2.310	0.622	1.200	0560–4.000	0.380
Obese/Overweight	0.489	0.190–1.120	0.109	0.500	0.140–1.380	0.219
Underweight	0.860	0.410–1.740	0.683	0.570	0.200–1.380	0.246
Constipation	0.801	0.370–1.850	0.515	0.591	0.244–1.588	0.265
Other conditions	1.110	0.440–1.640	0.811	1.560	0.560–3.950	0.362

Notes: Age was analysed in two separate models, as continuous variable in Model 1 and as categorical variable in Model 2. Significant values (p < 0.05) are written in bold text.

Sensitivity Analysis Restricted to Constipation Patients

Given the high prevalence of constipation in our cohort, which likely reflects tertiary care referral bias, we performed a sensitivity analysis restricted to patients with constipation to assess whether the observed associations remained robust within this dominant subgroup. In this analysis, the associations between age (as a continuous variable) and both ESBL production (OR = 0.910, 95% CI = 0.840–0.990; p = 0.023) and MDR status (OR = 0.910, 95% CI = 0.820–1.00; p = 0.050) remained consistent with the overall cohort findings. Similarly, the 1-year-old and 2–5-year-old age groups showed significantly higher odds of both ESBL and MDR status compared to older children: for the 1-year-old group, ESBL (OR = 0.310, 95% CI = 0.014–0.700, p = 0.005) and MDR (OR = 0.220, 95% CI = 0.090–0.540, p = 0.001); and for the 2–5-year-old group, ESBL (OR = 0.460, 95% CI = 0.200–1.070, p = 0.072) and MDR (OR = 0.280, 95% CI = 0.110–0.700, p = 0.007) (Supplementary Table S3). These findings suggest that the age-related resistance associations are not substantially altered by the overrepresentation of constipation and remain valid within this patient subgroup.

Discussion

UTI has been reported as one of the most common E. coli-associated infections in children, the frequency of which is affected by different factors, including comorbidities.^1,3,4 The resistance of pathogenic bacteria to antibiotics and underlying medical conditions in the host makes the infection difficult to treat.^9,18 In the current study, the prevalence of E. coli and their resistance profile among pediatric outpatients with underlying medical conditions was investigated. We observed a 30% prevalence of E. coli-linked UTIs in pediatric patients, with 30% ESBL-producing and 19% MDR isolates. Our findings show a substantially higher prevalence of E. coli-associated UTI compared to 9.2% recently reported from Faisalabad, Pakistan.⁵⁶ This difference is likely reflecting the higher-risk nature of our study population, which included children with underlying medical conditions. However, our observed resistance rates are consistent with the 38–100% resistance range reported in the literature.^16,56 Similarly, the observed ESBL production rate in our study falls within the lower to middle range of the 35–100% ESBL prevalence documented in previous Pakistani studies.^15,17 This confirms that the high resistance burden previously documented in the region extends to the vulnerable pediatric population with comorbidities. Similarly, a study from Jeddah, Saudi Arabia, reported a high prevalence of E. coli-associated UTI in children (49%) but a lower ESBL production rate (20%) compared to our findings of 30% prevalence and 30% ESBL production.⁴ This difference may reflect regional differences in antimicrobial resistance patterns, with Pakistan experiencing higher rates of ESBL-producing organisms, as well as differences in patient populations and healthcare practices. However, no study has reported on E. coli in children with underlying medical conditions in Pakistan.

The high prevalence of E. coli-associated UTIs observed in our study is consistent with the well-established increased risk of UTIs in children with underlying medical conditions, including urological abnormalities, neurogenic bladder, and constipation, which are known to impair bladder emptying, cause urinary stasis, and predispose to bacterial colonization and ascending infections.^18,57 While subgroup analysis did not show statistically significant associations between specific underlying conditions and ESBL or MDR rates likely due to limited sample sizes in certain subgroups including neurogenic bladder and VUR. However, the overall high resistance burden in our cohort is clinically concerning. The lack of statistical significance in subgroup analyses should not be interpreted as absence of clinical relevance, and future multi-center studies with larger sample sizes are warranted to explore these associations more definitively.

In the current study, we found that females predominated in both the overall population and the prevalence of infection. This pattern is consistent with the well-established pattern of UTIs in children.⁵⁸ A recent study from the UK also reported that 87% of female children were infected with E. coli-associated UTI.⁵⁹ Anatomical factors, particularly the shorter urethral length and its proximity to the perianal area, have been reported as reasons for the increased susceptibility of females to UTIs.⁶⁰ We found that this gender difference is particularly noticeable in school-age children. This finding is consistent with previous studies, in which girls were reported to have 1.6-fold higher rates of UTI than male children.⁶¹

Beyond resistance patterns, our findings highlight a 30% prevalence of E. coli‑associated UTI in this high‑risk pediatric population. This rate is substantially higher than the 16% prevalence of UTI reported in children presenting with urinary symptoms in primary care.²⁸ This rate also far exceeds from 15% reported in a recent meta-analysis that has reported on the prevalence of UTI from 2000 to 2021.⁶¹ The elevated burden in our cohort reflects the increased susceptibility conferred by underlying medical conditions such as neurogenic bladder, urinary tract anomalies, and constipation, all of which are established risk factors for paediatric UTI and recurrence.^28,62

Notably, females accounted for 72% of positive cases, consistent with the well‑documented female predominance of paediatric UTI beyond infancy, attributed to anatomical factors including a shorter urethra and periurethral colonization dynamics.^63,64 The age distribution of E. coli positivity was relatively uniform across groups (29–32%), suggesting that underlying conditions, rather than age alone, drive infection risk in this population.^62,65 However, the significantly elevated odds of ESBL and MDR in 1‑year‑old children highlight that age remains a critical factor once infection is established; infants <12 months have consistently been identified as an independent risk group for ESBL‑producing UTI, likely due to higher rates of antibiotic exposure, healthcare contact, and prophylaxis in early life.^66,67

The unusually high prevalence of constipation in our study cohort warrants careful interpretation and is acknowledged as a significant limitation. While constipation is recognized as a modifiable risk factor for UTI in children, the prevalence observed in this study substantially exceeds contemporary population-based estimates. A recent systematic review and meta-analysis reported pooled functional-constipation prevalences of 8.17% in children aged 0–4 years and 11.39% in those aged 4–18 years.⁶⁸ This is likely attributable to referral bias. To address this concern, we performed a sensitivity analysis restricted exclusively to patients with constipation. Importantly, the associations between age and resistance outcomes remained consistent in this restricted analysis indicating that the age-related resistance patterns are not artifacts of the high constipation prevalence. However, we acknowledge that the extreme skew toward constipation limits our ability to draw robust conclusions about resistance patterns specific to other underlying conditions (eg, neurogenic bladder, VUR) due to small sample sizes in these subgroups. Future multicenter studies across primary, secondary, and tertiary care settings with broader representation of underlying conditions are needed to validate these findings and characterize the true distribution of underlying conditions in pediatric UTI populations.

We observed that the prevalence of E. coli infection was the highest in patients with neurogenic bladder, followed by those with urinary tract abnormalities and vesicoureteral reflux. Similar findings have been reported recently, where a multicenter study identified CAKUT as one of the major risk factors for UTIs in children.¹⁸ The pathophysiological understanding also supports our findings, as these conditions are known to impair bladder emptying, increase intravesical pressure, and cause urinary stasis, all of which predispose patients to bacterial colonization and ascending infections.^18,57 Interestingly, constipation was the most common underlying condition in our study population. Although the prevalence of E. coli and its resistance profile did not reach statistical significance, it still warrants further clinical attention. Chronic functional constipation has been claimed to increase the risk of UTIs.⁶⁹ A recent study reported a 51% prevalence of E. coli-associated UTI in patients with chronic constipation.⁷⁰ Constipation may create a reservoir for bacterial growth, thereby causing UTI.⁷¹ The high proportion of children with constipation in our study population may reflect regional healthcare-seeking patterns or dietary factors, which requires further investigation.

We observed that 93.28% of the isolates in the current study were resistant to at least one antibiotic, 30% were ESBL-producers, and 18% were MDR. Our findings are comparable with other studies in Pakistan, which reported ESBL-producing isolates in 35–100% of cases, along with a high prevalence and high MDR status.^15–17,56 However, these rates exceed those reported in many high‑income settings. For example, a recent US study documented ESBL production in only 2.6% of pediatric UTIs in 2022.⁷² A European study reported ESBL‑producing isolates in 20% of pediatric UTI cases.⁷³ This disparity likely reflects multifactorial drivers beyond antibiotic overuse alone. First, diagnostic practices differ substantially: in Pakistan, empirical antibiotic prescribing frequently precedes culture confirmation, whereas high‑income countries more often implement structured antimicrobial stewardship programs that promote targeted therapy based on susceptibility data.¹⁵ Second, healthcare access patterns shape resistance acquisition; in low‑resource settings, patients commonly experience delayed diagnosis, incomplete treatment courses, and recurrent infections, all of which intensify selective pressure for resistant organisms.⁷⁴ Third, infection control infrastructure including robust surveillance systems and enforced antibiotic restriction policies is generally more developed in high‑income countries.⁷⁵ Finally, our cohort consists exclusively of children with underlying medical conditions, a high‑risk group that may not be directly comparable to general pediatric populations studied elsewhere, and this case mix may further contribute to the elevated resistance rates observed.¹⁵ The burden of AMR in Pakistan is particularly concerning within the Eastern Mediterranean Region, where the median proportion of carbapenem-resistant Acinetobacter spp. reaches 70.3%, and regional assessments indicate that few countries have adequate infection prevention and control capacities to prevent AMR emergence and spread.¹² Recent WHO reports highlight that while 22 Eastern Mediterranean Region countries are enrolled in the Global Antimicrobial Resistance and Use Surveillance System (GLASS), concerns persist regarding data quality and representativeness, and only a few countries use AMR data to inform policy decisions.⁷⁶ In November 2025, Pakistan’s National Institute of Health, in collaboration with WHO, launched the region’s first national priority pathogen list—a critical milestone for strengthening AMR surveillance, guiding antimicrobial stewardship, and prioritizing research and development tailored to Pakistan’s epidemiological context.^13,14 These initiatives, while promising, require sustained investment in laboratory capacity and diagnostic stewardship to translate surveillance data into actionable clinical guidance.⁷⁷

Our findings showed no significant association between weight status and the presence of ESBL or MDR E. coli. This contrasts with literature, in which obesity has been linked to higher risks of infection and antimicrobial resistance, potentially mediated by altered antibiotic pharmacokinetics, chronic low‑grade inflammation, and greater healthcare exposure.⁷⁸ Several factors may explain this discrepancy. Our cohort was predominantly normal weight (59%) or underweight (23%), with only 3.7% classified as obese, limiting statistical power to detect modest effects of adiposity on resistance outcomes. Second, the very high burden of pre‑existing urological and gastrointestinal comorbidity likely overshadowed any independent contribution of weight status. Constipation and bladder–bowel dysfunction are well‑established, strong predictors of paediatric UTI and recurrence, and in neurogenic bladder populations, constipation independently doubles UTI risk.^79,80 The obesity–resistance relationship is better characterized in adults while paediatric data are more heterogeneous, with some studies reporting increased UTI risk with overweight/obesity and others showing no association, and evidence specifically linking paediatric obesity to ESBL/MDR uropathogens remains limited.^81,82 Finally, although obesity can alter antibiotic distribution and clearance, paediatric pharmacokinetic studies to date have not demonstrated consistent, clinically meaningful dosing differences for commonly used antibiotics, which may further attenuate any obesity‑related selection pressure for resistance in children.⁸¹

While looking at resistance against specific antibiotics, we observed highest resistance against ampicillin followed by cephalexin and trimethoprim-sulfamethoxazole, respectively. Similar findings have also been reported in other countries. For example, a recent study from India reported 67% of E. coli isolates from children resistant to ampicillin.⁸³ A study from Bangladesh reported that 66% of E. coli strains isolated from children are resistant to ampicillin.⁸⁴ These patterns may reflect consistent overuse of these antimicrobial agents, rendering them largely ineffective for empirical therapy. We observed low resistance to nitrofurantoin (2.8%) and amikacin (1.8%) in our study, which suggests that these antibiotics can be used as viable options for treating UTI in children. The moderate resistance observed against ciprofloxacin warrants cautious interpretation. While fluoroquinolones are not recommended as first-line agents in pediatric UTIs due to safety concerns,⁸⁵ this finding provides valuable surveillance data on the dissemination of fluoroquinolone resistance genes, which may co-transfer with ESBL determinants. The high ciprofloxacin resistance among MDR isolates further underscores the limitation of therapeutic options in this vulnerable population and highlights the necessity of resistance surveillance even for drugs with restricted pediatric use.

We used regression analysis and found that age was the most significant predictor for both ESBL and MDR E. coli prevalence in children, with 1 year old and 2–5 years old-age children being at a significantly higher risk of infection. This finding is concerning, given the vulnerability of this age group to UTIs and the limited treatment options available. Many factors may contribute to this risk, including higher rates of antibiotic exposure in early childhood, immature immune systems, and the prevalence of congenital urologic abnormalities often diagnosed in infancy.³² Similarly, studies have reported higher ESBL colonization rates in infants and 1 year olds, possibly due to recurrent hospitalizations and antibiotic courses during this developmental period.^86,87 The inverse relationship between age and resistance risk for ESBL and MDR in our study suggests that cumulative antimicrobial pressure may have a significant impact on early life. This finding underscores the critical importance of cautious antibiotic prescription in pediatric populations, especially in the first two years of life, when healthcare encounters are frequent and empirical therapy is common.⁸⁸

Although these findings provide valuable insight into local resistance patterns, the cross-sectional, single-center design limits generalizability and does not support definitive practice-changing recommendations. Nevertheless, several pragmatic institutional measures may be considered. First, the antimicrobial stewardship team should review the current empirical guideline for pediatric UTIs, particularly in view of the poor activity of ampicillin, cephalexin, and trimethoprim–sulfamethoxazole in this cohort. Nitrofurantoin may be considered for uncomplicated cystitis, whereas amikacin may be considered for severe presentations when clinically appropriate; however, these options should be guided by local susceptibility data, renal function, infection severity, and relevant safety considerations. Any revision should initially be piloted with close monitoring of clinical outcomes, adverse events, and resistance trends because single-center susceptibility patterns may change over time.⁷⁵

Second, the existing policy of obtaining a urine culture before antibiotic initiation in children with underlying conditions should be reinforced, since empirical treatment was inappropriate in nearly one-third of ESBL-positive cases in this study. This approach requires consistent adherence to current protocols rather than additional resources.²¹

More broadly, stewardship in this population should rely on continuous local surveillance rather than a single cross-sectional assessment. We therefore recommend that the institution establish quarterly, age-stratified antibiogram reporting to support timely adjustment of empirical therapy and facilitate early detection of emerging resistance. This feasible, low-cost strategy is consistent with established antimicrobial-resistance surveillance principles. Broader policy changes should await confirmation from multicenter perspective studies.²¹

The current study has some limitations. For example, this study was based in a single center, which may limit generalizability. The observed prevalence and resistance patterns are regional and may differ from those of other regions. We did not collect complete prior data from the patients, which may have caused some important confounders of resistance. Although MIC determination of a subset of isolates addressed this concern, the use of urine culture and phenotypic susceptibility testing may underestimate resistance due to the limitations of disc diffusion methods. Finally, we did not assess clinical outcomes or treatment responses, which would be useful in the interpretation of resistance data.

Conclusion

This study shows that E. coli UTIs in Pakistani children with underlying medical conditions including constipation (93% of the cohort), neurogenic bladder, urinary tract abnormalities, and vesicoureteral reflux were associated with substantial antimicrobial resistance. ESBL-producing isolates accounted for 30.4% of isolates, while 18.7% met the definition of MDR. Although resistance prevalence differed across underlying-condition categories, no individual condition was significantly associated with ESBL production or MDR. However, the small numbers of children with neurogenic bladder (n=7), urinary tract abnormalities (n=37), and vesicoureteral reflux (n=21) limited subgroup comparisons; consequently, the observed resistance pattern largely reflects that of children with constipation. Age was significantly associated with both ESBL production and MDR, with 1-year-old children having higher odds of MDR (OR=3.6). This finding supports age-informed clinical assessment and a low threshold for urine culture and susceptibility testing in younger children, rather than establishing age-specific empiric regimens from this single-centre dataset. For lower UTI in clinically stable children, nitrofurantoin may be considered when susceptibility results, renal function, and clinical presentation support its use; it is not appropriate for febrile UTI or suspected pyelonephritis. Amikacin may be considered when parenteral treatment is required, with dosing and renal monitoring guided by local protocols. These interpretations are consistent with WHO guidance that distinguishes lower from severe upper UTI and emphasizes antibiotic selection according to clinical syndrome and local resistance patterns. Given resistance exceeding 59% to ampicillin, cephalexin, and trimethoprim-sulfamethoxazole, these agents should not be selected empirically for this cohort without supportive local susceptibility data. Ciprofloxacin susceptibility results are reported for surveillance only and should not be interpreted as a recommendation for routine pediatric treatment. Recent Pakistani pediatric studies similarly identify high resistance to commonly used antibiotics and emphasize culture-guided prescribing and antimicrobial stewardship. The cross-sectional, single-centre design, predominance of constipation, and limited clinical data restrict causal inference and generalizability. Prospective multicentre studies including children with a broader range of underlying conditions, detailed information on previous antibiotic exposure and UTI history, and longitudinal follow-up are needed to validate these findings and inform locally appropriate antimicrobial-stewardship policies.

Funding Statement

This work has received no external funding.

Data Sharing Statement

The data used in this study were collected by co-author Atif Haroon from the Pakistan Institute of Medical Sciences (PIMS), Islamabad, during his research at Quaid-I-Azam University. The study was approved by the Bioethical Committee of Quaid-I-Azam University (BEC-FBS-QAU2018-109), which covers the data collection and use for research purposes. Co-author Muhammad Nadeem Khan, although currently affiliated with a Chinese institution, is a graduate of Quaid-I-Azam University and contributed to data analysis and manuscript preparation under the same approved ethics protocol. Dr. Hazrat Bilal, also affiliated with a Chinese institution, collaborated on this study through his ongoing research partnership with Muhammad Nadeem Khan. No separate data transfer agreement was required as the data were anonymized and used within the scope of the original ethics approval. The de-identified dataset is available from the first author (M. Nadeem Khan, Email: khanmuhammadnadeem@sdxiehe.edu.cn) upon reasonable request.

Ethics Approval and Informed Consent

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Bioethical Committee of Quaid-I-Azam University, Islamabad, Pakistan (BEC-FBS-QAU2018-109). Informed consent was obtained from the parents of the children, and only those who consented to participate were included in this study. All patient data were fully anonymized and de-identified prior to analysis. No identifiable personal information was included in the database.

Consent for Publication

Not applicable as all figures and tables presented in this manuscript are original.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis, and interpretation, or in all these areas, took part in drafting, revising, or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.