Bacteremia and Antimicrobial Resistance Pattern of Uropathogens Causing Febrile Urinary Tract Infection in a Pediatric University Hospital

Abstract

Introduction: Febrile urinary tract infections (UTIs) in children are among the most serious bacterial infections. Inadequate treatment can lead to kidney scarring and permanent kidney damage. Eight to ten percent of children with UTIs could have concomitant bacteremia. The study aimed to estimate the prevalence of UTI-associated bacteremia and identify common organisms causing UTIs and their antimicrobial susceptibility patterns to help guide empiric antimicrobial therapy. Methods: The current study was conducted over a 6-month period on children admitted with febrile UTIs at Alexandria University Children’s Hospital. Blood and urine samples were collected for culture and antimicrobial susceptibility. Results: A total of 103 children with a median age of 12 months (IQR 6.0-24.0) were included in the study. Concomitant bacteremia was present in 63.1% (n=65). The median temperature of 38.40 °C (IQR 38.15-38.60) and the median creatinine level of 0.18 mg/dL (IQR 0.14-0.25) were significantly higher in the bacteremic group compared to the non-bacteremic group (p=0.005, p=0.034, respectively). E. coli (n=51; 49.5%) and Klebsiella pneumoniae (n=30; 29.1%) were the most common isolated organisms. Most (n=68; 66%) of the isolated organisms were multidrug-resistant (MDR), followed by extensively drug-resistant (XDR) (n=16; 15.5%), and pan-drug-resistant (PDR) organisms (n=1; 1%). E. coli showed lower resistance to gentamicin and ceftriaxone (9.8 % and 13.7%, respectively). Conclusions: E. coli remains the most important UTI pathogen. Ceftriaxone and gentamicin are good empiric options for febrile UTIs in our hospital.

Introduction

Concurrent bacteremia is estimated to affect 8-10% of infants with febrile urinary tract infections (UTIs) [1]. It is challenging to clinically differentiate between bacteremic and non-bacteremic UTIs due to conflicting data regarding symptoms and signs between both groups [2,3]. Young ages have consistently been identified as a risk factor for bacteremia and UTI-associated morbidity, including kidney damage [4,5]. The majority of pediatric UTIs are caused by Gram-negative bacteria that enter the urinary tract via the ascending route [6]. E. coli is the most common uropathogen, accounting for approximately 80% to 90% of pediatric UTIs. Klebsiella, Proteus, Enterobacter, and Enterococcus are common uropathogens [4].

Early and appropriate treatment of febrile UTI, especially with concomitant bacteremia, will lead to decreased morbidity and better outcomes. Unfortunately, rising antibiotic resistance makes empiric prescriptions for febrile UTIs challenging. This study aimed to estimate the prevalence of concomitant bacteremia among children with febrile UTI and to identify common bacterial pathogens and their antibiotic susceptibility patterns at Alexandria University Children’s Hospital to guide antimicrobial therapy.

Methods

Study design

The current case-control study was conducted over six months at Alexandria University Children’s Hospital, a tertiary care university pediatric hospital in Alexandria, Egypt. Children ≥1 month with a temperature of ≥38 °C and symptoms suggestive of UTI were included in the current study. Neonates, children with immunodeficiency, children who were already previously diagnosed with urinary tract abnormalities, and those without withdrawn blood cultures were excluded from the study. A total of 103 children with microbiologically confirmed UTIs were enrolled for further analysis. The diagnosis of concomitant bacteremia is based on the isolation of the same organism in the patient’s urine and blood culture within 48-hour intervals [7].

Collection of data

All relevant demographic, clinical, and laboratory data were collected.

Microbiological investigations

Sample collection

All samples were collected before the start of antibiotic therapy. Two to five mL of blood for blood culture and urine samples (by clean catch midstream urine or straight catheterization) for urine culture were collected using standard collection procedures after following a strict aseptic technique. The collected samples were immediately transferred to the microbiology lab of Alexandria Children’s Hospital for processing [8].

Processing of samples

All collected samples were processed according to standard microbiological procedures [8].

1. Culture procedure

1.1. Inoculated blood culture bottles were incubated for 5 days in an automated BACT/ALERT blood culture system (bioMérieux, USA). Positive signal bottle contents were inoculated onto the surface of chocolate blood agar, blood agar, and MacConkey’s agar and streaked for isolation.

For blood culture, any recognized blood pathogen isolated from a single blood culture, or an original contaminant pathogen isolated from two blood cultures confirmed bloodstream infection.

1.2. Urine samples were cultured semi-quantitatively using a 1-µL calibrated loop. Cultures were performed using blood agar and MacConkey’s agar. Cultures were examined for colony counts and morphological types of organisms. The bacterial count in colony-forming units/mL was calculated according to the following formula: number of colonies × 1000. Definitive identification of isolates and antimicrobial susceptibility testing was done for the significant growth of the following:

-

Single organism with colony count ≥10⁵ colony forming units (CFU)/mL

-

Two organisms ≥10⁵ CFU/mL for each with significant pyuria (≥10 WBC/high power field).

-

The growth of two isolates with ≥10⁵ CFU/mL for one isolate and <10⁴ CFU/mL for the other one with significant pyuria; identification was done only for the first isolate.

The growth of two isolates with colony count <10⁵ CFU/mL for each or the growth of ≥three isolates was not processed.

2. Identification of isolated organisms

Full microbiological identification was based on Gram’s-stained film, morphological colonial characters (hemolytic or non-hemolytic colonies on blood agar, lactose, or non-lactose colonies on MacConkey’s agar), and standard biochemical tests (catalase, coagulase, oxidase, bile esculin, triple sugar iron, urease, citrate, motility, and indole tests) [8].

3. Antimicrobial susceptibility testing

Antimicrobial susceptibility was determined using the Kirby-Bauer disc diffusion method according to Clinical Laboratory Standards Institute (CLSI) guidelines (2021) [9]. E. coli (ATCC 25923) and Staphylococcus aureus (ATCC 25922) were used as controls for the test. Phenotypic detection of extended-spectrum beta-lactamases (ESBL) in Gram-negative bacilli was based on using cefoxitin (30 μg disc diffusion) and confirmed by a combined disc diffusion test [9]. Multidrug-resistant (MDR), extensively drug-resistant (XDR), and pan-drug-resistant (PDR) organisms were defined according to standard international terminology presented by ECDC (European Centre for Disease Prevention and Control) and CDC (Centre for Disease Prevention and Control) [10].

Statistical analysis

The data were analyzed using IBM SPSS software version 20.0 (IBM Corp., USA) Qualitative data were described using numbers and percentages. The Kolmogorov-Smirnov test was used to verify the normality of the distribution. Quantitative data were described using range (minimum and maximum), mean, standard deviation, median, and interquartile range (IQR). The significance of the obtained results was judged at the 5% level. A Chi-square test was performed to study the significant association between different categorical variables and the occurrence of bacteremia. Fischer-Exact and Monte-Carlo significance were employed if more than 20% of the total expected cell counts <5 at the 0.05 level of significance. The Mann-Whitney test detected a significant difference in the median quantitative variables between negative and positive bacteremia based on the variables’ distribution by the Kolmogorov-Smirnov test.

Results

During the study period, 265/1536 (17.3%) febrile children with culture-confirmed UTI were hospitalized, of whom 103 fulfilled the inclusion criteria – Figure 1. The age of the included children ranged from 1 to 72 months with a median of 12.0 (IQR 6.0-24.0) months. Most of the studied children (91.3%) (n=94) were aged less than 3 years, with an equal percentage of infants and children over the age of one year. Females were slightly higher than males (54.4% versus 45.6%).

The clinical manifestations differed significantly by age (

Table 1. Comparison between different age groups regarding clinical and laboratory characteristics.

χ²: Chi-square test. MC: Monte Carlo. H: H for Kruskal Wallis test, *: Statistically significant (p<0.05). p: p-value for comparing the age groups. CRP – C reactive protein; WBC – white blood cell count; BUN – blood urea nitrogen.

), as classical urinary symptoms were more common in children over the age of 36 months (n=9), with burning micturition accounting for 100% (n=9), malodorous urine accounting for 77.8% (n=7), and both changes in urine color and urine retention accounting for 66.7% (n=6). Non-classical symptoms including vomiting (74.5% (n=35)), irritability (14.9% (n=7)), and poor feeding (42.6% (n=20)) were common in children younger than one year. Children older than 36 months significantly (p=0.012) had longer median days of fever before admission [3 (2-15) days]. Concomitant bacteremia among children with febrile UTI was 63.1% (n=65). Most (n=61; 93.9%) of bacteremia occurred in children younger than 36 months. The median temperature (38.4 °C IQR (38.15-38.60)) and median creatinine level (0.18 mg/dL IQR (0.14-0.25)) were significantly higher in the bacteremic group compared to the non-bacteremic group, with p values of 0.005 and 0.034, respectively. In both bacteremic and non-bacteremic groups, E. coli and Klebsiella pneumoniae were the most often isolated species, with 49.5% (n=51) and 29.1% (n=30), respectively –

Table 2. Comparison between bacteremic and non-bacteremic urinary tract infections.

*: Statistically significant (p<0.05). MC: Monte Carlo correction for a p-value of Pearson Chi-square test. ^a,b Superscripts denote significant pairwise comparison by the Bonferroni test with adjusted significance. MD (IQR) – median (interquartile range); CRP – C reactive protein; WBC – white blood cell count; BUN – blood urea nitrogen; MDR – multidrug resistant; XDR – extensively drug resistant; PDR – pan drug resistant.

. ESBL-producing isolates represented 15.8% (n=6) of the non-bacteremic group. Enterococcus faecalis was the only isolated Gram-positive organism, representing 9.2% (n=6) and 10.5% (n=4) in the bacteremic and non-bacteremic groups, respectively.

Overall, there was an alarming increase in the resistance patterns of isolated organisms, 82.5% (n=85) were resistant. The MDR organisms were the most prevalent at 66.0% (n=68), followed by XDR at 15.6% (n=16) and PDR organisms at 1.5% (n=1). Although both MDR (64.6%) and XDR (18.5%) were more frequently isolated in the bacteremic group, this was not statistically significant (p=0.638). The MDR organisms were most detected among Enterococcus faecalis (80%), E. coli (74.5%), and Klebsiella pneumoniae (60%) – Figure 2.

E. coli showed moderate to high resistance to most of the tested antibiotics (47.1-75.5%) – Figure 3A. The lowest resistance was exhibited towards the following antibiotics: fosfomycin (2%); colistin (3.9%); meropenem (5.9%); levofloxacin (5.9%); gentamicin (9.8%); amoxicillin/clavulanic acid (11.8%); ceftriaxone (13.7%); and nitrofurantoin (15.7%). Klebsiella pneumoniae exhibited a very high rate of resistance (43.3%-80%) to most tested antibiotics (Figure 3B). The lowest rates of resistance were recorded for colistin (3.3%), levofloxacin (3.3), ceftriaxone (13.3%), and meropenem (10.0%). Pseudomonas aeruginosa was susceptible to most of the tested antibiotics, with moderate resistance (14.3%-28.6%) to the following antibiotics: imipenem, ciprofloxacin, ceftazidime, aminoglycosides, and cefepime.

Discussion

The current study aimed to estimate the prevalence of bacteremia and common organisms causing febrile urinary tract infections and their resistance patterns to help guide empiric antibiotic treatment. Concomitant bacteremia in UTIs is associated with a longer hospital stay and higher morbidity and mortality rates in adults and children [13]. As a result, many studies have been designed to identify early predictors of concomitant bacteremia. One of the most concerning findings in the current study is the high prevalence of bacteremia (63.0% (n=65)) among children with febrile UTIs. Other studies detected a lower rate of bacteremia (5.7%-17%) [13,15,17]. The current study’s high prevalence of bacteremic UTIs could be explained by the high prevalence of resistant organisms causing bacteremic UTIs; among the bacteremic group, 42.6% (n=42) of organisms were MDR and 18.2% (n=12) were XDR. Another explanation is that the younger the age, the higher the prevalence of bacteremia, as reported by other studies [14,15]. In the current study, 47.7% (n=31) and 46.2% (n=30) of children with bacteremia were aged 12-36 months and younger than 12 months, respectively. The young age implies a hematogenous spread of infection due to the immaturity of the immune system in this age group. Lastly, as we are in a tertiary care referral hospital and most of the referred cases are complicated, after the trial of outpatient treatment.

The median temperature of 38.40 °C (IQR 38.15-38.60) and median creatinine level of 0.18 mg/dL (IQR 0.14-0.25) were significantly higher in bacteremic children, with p values of 0.005 and 0.034, respectively. Although the median WBC count of 13.00 (×10³)/µL (IQR 10.00-16.00) was higher among the bacteremic group, this was not statistically significant (p=0.098). Comparable results were reported in several studies. Megged et al. [9] reported significantly higher creatinine in children with bacteremic UTI, and although the mean WBC (16.2±7.2) was higher, this was not statistically significant (p=0.17). Similar findings have been reported in other studies [16,17]. Elevated creatinine levels among children with bacteremia could be explained by renal hypoperfusion and dehydration due to poor oral feeding and vomiting.

In the current study, Gram-negative bacteria were the most isolated uropathogens, in 90.3% (n=93) of the cases. E. coli accounted for 49.5% of UTIs (n=51), Klebsiella pneumoniae 29.1% (n=30), Enterococcus faecalis 9.7% (n=10), Pseudomonas aeruginosa 6.8% (n=7), and Citrobacter 1.9% (n=2). This was consistent with the findings of numerous other studies [18,19,20,21]. The only Gram-positive organism isolated was Enterococcus faecalis (9.7%, n=10). Unlike in other studies, Staphylococcus aureus was identified as the causative agent of UTI in Ethiopia (27.3%) [19] and Nepal (7%) [21]. Uropathogens that cause UTIs vary depending on age, geographic location, and concomitant co-morbidities. Accordingly, conducting an epidemiological study to guide antibiotic therapy based on local antibiograms and prevalent causative agents is crucial.

Antibiotic resistance is an important problem that is acknowledged globally. It is essential to understand how antimicrobial susceptibility patterns change over time in a particular region to control the use of antibiotics, identify and manage antimicrobial resistance, and effectively treat bacterial infections. One of the most striking results in the current study is the alarming increase in the resistant organisms, as 66.0% were MDR, followed by XDR in 15.6%, and PDR organisms in 1% of isolates. Overall, 91.3% of the investigated children were under three years old, which may help to explain the high incidence of resistant organisms. The immaturity of the immune system in young infants makes them vulnerable to multiple febrile viral episodes with subsequent exposure to multiple unnecessary antibiotic courses in primary care. Therefore, the establishment of an antimicrobial stewardship program is crucial. Concordant results were found in Ethiopia [19]. The estimated MDR isolates among Gram-negative bacteria were 61%, while 22.4% and 4% of the Gram-negative organisms were XDR and PDR, respectively. Parajuli et al. [22] and Merga Duffa et al. [23] found MDR organisms in 64.9% and 73.7% of their samples, respectively.

In contrast, with a lower prevalence rate in Nepal [21], MDR and XDR organisms were detected in 32% and 5% respectively. ESBL-producing organisms have emerged as a challenge in treating community-acquired UTIs [24]. In our series, ESBL represented 5.8% (n=6) of isolates in E. coli (7.8% (n=4)) and Klebsiella pneumoniae (6.7% (n=2)). In contrast to the current results, Shrestha et al. (40%) [21], and Parajuliuli et al. (38%) [22], both found a higher rate of ESBL. Differences in the resistance pattern of organisms emphasize the importance of routine surveillance to help establish local guidelines for treating such cases.

Focusing on E. coli and Klebsiella pneumoniae antibiograms, both showed low resistance to fosfomycin (2% and 20%, respectively) and amoxicillin-clavulanate (11.8% and 20%, respectively). Therefore, these antibiotics can be used as an initial treatment for uncomplicated UTIs. Furthermore, ceftriaxone and gentamicin demonstrated lower resistance among E. coli (13.7% and 9.8%) and Klebsiella pneumoniae (13.3% and 20%) isolates, supporting their use as empiric treatment for febrile UTIs in immunocompetent children. Both exhibited low resistance to meropenem (5.9% and 10%, respectively). On the other hand, resistance to nitrofurantoin was higher in K. pneumoniae versus E. coli (73.3% versus 15.7%), discouraging its use as an empiric oral therapy for uncomplicated UTI. Similar outcomes were seen in Tanzania [25] and Turkey [26]. Both exhibited a high rate of resistance to trimethoprim-sulfamethoxazole, so it couldn’t be used as empiric therapy for UTI.

The current study has some limitations: first, it is a single-center experience; additional studies on different cohorts are required to establish a guide for antimicrobial prescription in children with febrile UTI. Second, there is a lack of radiological evaluation of children with concurrent bacteremia to identify any undiagnosed underlying urological problems.

Conclusions

There is an alarming increase in the resistance pattern of organisms causing febrile UTIs therefore regular surveillance should be carried out to guide the management of such cases and establish an antimicrobial stewardship program. Low resistance to ceftriaxone and gentamicin encourages their use as empiric therapy in children with febrile UTIs who have no known co-morbidities. Concomitant bacteremia among children with febrile UTIs was high. Children with concomitant bacteremia had significantly higher creatinine levels and a high degree of fever.