FormalPara Key Summary Points

Why carry out this study

The prevalence of antimicrobial resistance among outpatients with uncomplicated urinary tract infection (uUTI) in the United States (US) is poorly characterized.

Co-resistance rates among E. coli and K. pneumoniae urinary isolates from US female outpatients were investigated in this study.

What was learned from this study

We found high rates of co-resistance among E. coli and K. pneumoniae urinary isolates to antimicrobial agents widely used to treat outpatients with uUTIs.

Our results highlight that there are limited effective oral treatment options for resistant E. coli and K. pneumoniae urine isolates.

Characterizing patterns of co-resistance among uropathogens causing community-acquired UTIs is critical for understanding the unmet need for effective empiric treatment options for outpatient urinary tract infections (UTIs).

Introduction

Uncomplicated urinary tract infections (uUTIs) occur in women with no anatomical or functional urinary tract abnormalities, or complicating comorbidities [1]. Most urinary tract infections (UTIs) are uncomplicated, and affect 10–12% of female patients in the United States (US) annually [2]. uUTIs caused by antimicrobial-resistant Enterobacterales, including extended-spectrum β-lactamase-producing (ESBL+)/third-generation cephalosporin (3GC) Escherichia coli (E. coli) and Klebsiella pneumoniae (K. pneumoniae), are a serious global health threat [3,4,5].

E. coli is the causative uropathogen in 80–90% of uUTIs in female patients [6], but other Enterobacterales [7,8,9] and Gram-positive isolates like Staphylococcus saprophyticus and Enterococcus spp. can also cause uUTIs. While the clinical epidemiology of uUTIs has remained relatively stable over the past decade, ESBL+/3GC and multidrug resistance (MDR) phenotypes among female outpatient urine isolates have increased [7,8,9].

Urine cultures are generally not ordered for community-acquired uUTIs, and recommended treatment is mostly empiric. Therefore, assessing the current prevalence of antimicrobial resistance (AMR) from urine culture is challenging, and limited surveillance data characterize AMR prevalence among outpatient urinary isolates in the US. As increasing AMR will likely reduce treatment effectiveness [10], it is important to raise awareness among treating physicians.

Kaye et al. evaluated the prevalence and geographic distribution of AMR in the US using the largest urine E. coli isolate database to date (N = 1,513,882) [11]. The AMR prevalence was high and varied significantly between US regions; ESBL + and MDR phenotypes also increased during the study period (2011–2019). Building on this work regarding MDR in E. coli, we assessed the prevalence of co-resistance to ≥ 1 drug class, and characterized all resistance combinations observed, among urinary E. coli and K. pneumoniae isolates from US female outpatients to further understand treatment options for patients with MDR uropathogens [11].

Methods

Study Design and Patients

This was a retrospective cohort study of female outpatients aged ≥ 12 years with ≥ 1 positive urine culture containing 30-day non-duplicate E. coli or K. pneumoniae urine isolates (i.e., presumed uUTI). The minimal amount of data per eligible patient was ≥ 3 months (between 2011 and 2019) with antimicrobial susceptibility testing (AST) results recorded at one of the 304 US facilities included in the BD Insights Research Database (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) [11]. Eligible non-duplicate isolates were the first urine E. coli or K. pneumoniae isolate per patient collected within 30 days and were classified as not susceptible (NS) if AST results were intermediate or resistant to each antibiotic tested [12].

Isolates were classified into four resistance phenotypes: NS to fluoroquinolones (FQ NS), NS to trimethoprim/sulfamethoxazole (SXT NS), NS to nitrofurantoin (NTF NS), and ESBL+/3GC NS (ESBL production determined via commercial panel and/or NS to ceftriaxone, cefotaxime, ceftazidime, or cefepime). Co-resistance phenotypes were characterized as all possible combinations of the four pre-defined resistance phenotypes described above, and therefore included co-resistance to two, three, or four of the individual drug classes.

Statistical Analysis

Descriptive statistics were used to calculate the number and percentage of isolates and co-resistance phenotype combinations observed.

Compliance with Ethics Guidelines

The study was performed using a de-identified limited retrospective dataset that was deemed exempt from patient informed consent by the New England Institutional Review Board/Human Subjects Research Committee (Wellesley, MA, USA). The study was conducted in compliance with Health Insurance Portability and Accountability Act requirements.

Results

Escherichia coli

Across 1,513,882 non-duplicate (30-day) E. coli urine culture isolates, 856,918 isolates with ≥ 1 resistance phenotype were analyzed (of which 44.8, 37.3, 11.2, and 6.6% were SXT NS, FQ NS, ESBL+/3GC NS, and NTF NS, respectively; Fig. 1).

Fig. 1
figure 1

Proportion of urine culture isolates with ≥ 1 resistance phenotype demonstrating specific individual resistance. 856,918 non-duplicate (30-day) E. coli isolates (with ≥ 1 resistance phenotype) and 187,459 non-duplicate (30-day) K. pneumoniae isolates (with ≥ 1 resistance phenotype) were evaluated for co-resistance. AMR antimicrobial resistance, E. coli Escherichia coli, ESBL+/3GC extended-spectrum β-lactamase-producing/third-generation cephalosporin, FQ fluoroquinolone, K. pneumoniae Klebsiella pneumoniae, NS not susceptible, NTF nitrofurantoin, SXT trimethoprim/sulfamethoxazole

Full size image

The following co-resistance phenotype combinations were observed among E. coli isolates (Fig. 2a): of 96,306 ESBL+/3GC NS isolates, 72.4% were co-resistant to FQs, 56.7% to SXT, and 11.9% to NTF; 46.6% of ESBL+/3GC NS isolates were co-resistant to FQ + SXT, 9.4% to FQ + NTF, and 8.0% to SXT + NTF, while 6.8% had all four resistance phenotypes. Of 319,354 FQ NS isolates, 21.8% were also ESBL+/3GC NS, 51.6% were co-resistant to SXT, and 8.0% to NTF; 4.9% of FQ NS isolates were co-resistant to SXT + NTF, and 2.0% had all four resistance phenotypes. Of 384,304 SXT NS isolates, 14.2% were also ESBL+/3GC NS, 42.9% were co-resistant to FQs, and 6.8% to NTF; 4.1% of SXT NS isolates were co-resistant to FQ + NTF and 1.7% had all four resistance phenotypes. Of NTF NS isolates (n = 56,954), 20.1% were also ESBL+/3GC NS, 44.7% were co-resistant to FQs, and 46.0% to SXT; 27.4% of NTF NS isolates were co-resistant to FQ + SXT, and 11.5% had all four resistance phenotypes.

Fig. 2
figure 2

Co-resistance phenotype combinations observed among E. coli (a) and K. pneumoniae (b) urine culture isolates. 856,918 non-duplicate (30-day) E. coli isolates (with ≥ 1 resistance phenotype) and 187,459 non-duplicate (30-day) K. pneumoniae isolates (with ≥ 1 resistance phenotype) were evaluated for co-resistance from female outpatients with UTI (2011–2019). Some isolates had overlapping susceptibilities/antimicrobial resistance types, hence individual phenotype totals do not reflect the total number of isolates evaluated. The categories are not mutually exclusive as the isolates are grouped based on the initial phenotype observed. E. coli Escherichia coli, ESBL+/3GC extended-spectrum β-lactamase-producing/third-generation cephalosporin, FQ fluoroquinolone, K. pneumoniae Klebsiella pneumoniae, NS not susceptible, NTF nitrofurantoin, SXT trimethoprim/sulfamethoxazole, UTI urinary tract infection

Full size image

Klebsiella pneumoniae

Across 250,719 non-duplicate (30-day) K. pneumoniae urine culture isolates, 187,459 isolates with ≥ 1 resistance phenotype were analyzed (of which 12.7, 5.8, 5.9, and 75.5% were SXT NS, FQ NS, ESBL+/3GC NS, and NTF NS, respectively; Fig. 1).

The following co-resistance phenotype combinations were observed among K. pneumoniae isolates (Fig. 2b): of 11,065 ESBL+/3GC NS isolates, 54.9% were co-resistant to FQs, 65.7% to SXT, and 75.5% to NTF; 45.7% of ESBL+/3GC NS isolates were co-resistant to FQ + SXT, 45.1% to FQ + NTF, 52.2% to SXT + NTF, and 38.2% had all four resistance phenotypes. Of 10,962 FQ NS isolates, 55.4% were ESBL+/3GC NS, 65.7% were co-resistant to SXT, and 79.6% to NTF; 54.0% of FQ NS isolates were co-resistant to SXT + NTF and 38.6% had all four resistance phenotypes. Of 23,887 SXT NS isolates, 30.4% were ESBL+/3GC NS, 30.1% were co-resistant to FQs, and 69.7% to NTF; 24.8% of SXT NS isolates were co-resistant to FQ + NTF and 17.7% had all four resistance phenotypes. Of 141,545 NTF NS isolates, 5.9% were ESBL+/3GC NS, 6.2% were co-resistant to FQs, and 11.8% to SXT; 4.2% of NTF NS isolates were co-resistant to FQ + SXT and 3.0% had all four resistance phenotypes.

Discussion

This study describes high rates of co-resistance among E. coli and K. pneumoniae urinary isolates to antimicrobial agents widely used to treat outpatient uUTIs. Among isolates with ≥ 1 resistance phenotype (E. coli: n = 856,918; K. pneumoniae: n = 187,459), individual resistance to SXT or FQs among E. coli isolates was 44.8 and 37.3%, respectively, while individual resistance among K. pneumoniae isolates was 12.7 and 5.8%, respectively. The ESBL+/3GC NS phenotype was observed in 11.2% of E. coli isolates and 5.9% of K. pneumoniae isolates. Among ESBL+/3GC NS E. coli isolates, 72.4, 56.7, and 46.6% were co-resistant to FQs, SXT, and FQ + SXT, respectively. Among K. pneumoniae isolates with ≥ 1 resistance phenotype, 69–80% were co-resistant to NTF. Additionally, 65.7% of ESBL+/3GC NS K. pneumoniae isolates were co-resistant to SXT, 54.0% of FQ NS isolates were co-resistant to SXT + NTF, and 45.7% of ESBL+/3GC NS isolates were co-resistant to FQ + SXT. Our findings regarding co-resistance phenotypes containing ESBL+/3GC NS indicate that effective oral treatment options are limited for resistant E. coli strains and more limited for resistant K. pneumoniae strains. NTF resistance among K. pneumoniae is problematic given that NTF is one of the most prescribed empiric therapies for outpatient UTI. Future guidelines should consider evaluation of co-resistance and related thresholds to guide clinician prescribing of effective antimicrobial therapy that addresses unmet need in uUTI.

This study adds to our previous findings regarding AMR prevalence among E. coli and Klebsiella species (Klebsiella spp.) isolates in the US [11, 13]. The current study also provides much-needed multicenter AMR surveillance data and highlights the importance of assessing patterns of co-resistance among uropathogens causing community-acquired UTIs. Indeed, we previously reported that 14.4 and 3.8% of E. coli isolates have ≥ 2 and ≥ 3 drug resistance phenotypes, respectively [11], while more than 10% of Klebsiella spp. and K. pneumoniae isolates have ≥ 2 drug resistance phenotypes [13], but these new analyses of co-resistance among resistant isolates highlight the limitations of the currently available uUTI antibiotics. Raising awareness of antimicrobial co-resistance phenotypes should help inform empiric treatment decisions. Periodic reviews of uropathogen susceptibility patterns can also highlight the need for prescribing behavior changes and facilitate guideline updates to improve appropriate antimicrobial use, though incentives may also be needed [14].

A recent study across nine US centers from 2015–2019 highlighted a 19.4% NS rate to initial antimicrobial treatment among 2366 uUTI episodes with higher 28-day antibiotic dispensing rates [15], but few studies inform on co-resistance among urine isolates from patients with community-acquired uUTI. Critchley et al. [16] also assessed co-resistance among 1831 E. coli UTI isolates (nosocomial and community-acquired) from 30 participating US centers during 2017 and found similar patterns of co-resistance to FQs and SXT. In that study, among FQ (levofloxacin)-resistant E. coli, 56% of isolates were co-resistant to SXT. Similarly, 43% of SXT-resistant E. coli isolates were co-resistant to FQs. High rates of resistance to SXT (56%) were also reported among 287 ESBL + E. coli isolates, consistent with the 57% observed in our study.

Our study has limitations. AST and results were based on local laboratory practice, and microbiology laboratory data could not definitively be linked to a uUTI due to lack of symptom data, International Classification of Diseases diagnostic codes, and pharmacy claims. More than one isolate from the same patient could have been included if collected > 30 days after the previous isolate; the study population could therefore include patients with recurrent UTIs, which may result in overestimating the prevalence of AMR.

Given the empiric nature of current uUTI prescribing, urine culture ordering practices likely vary with disease severity; therefore, AMR results could be biased towards higher resistance since not all uUTI patients have a culture. The prevalence of AMR to fosfomycin was not assessed due to its limited use and methodological challenges in testing.

Conclusions

In summary, we found high rates of co-resistance in E. coli and K. pneumoniae outpatient urine isolates, emphasizing the importance of surveillance studies to inform appropriate antibiotic prescribing practices and updates to treatment guidelines, in order to optimize outpatient UTI treatment. The high frequency of co-resistance phenotype combinations highlights the need for novel diagnostics and oral antibiotics for treatment of outpatient UTI.