Urinary tract infections (UTIs) represent one of the most common outpatient infections treated with antibiotics, primarily affecting young women. Escherichia coli is the predominant bacteria in both uncomplicated cystitis and acute pyelonephritis (APN), followed by other Enterobacterales such as Klebsiella spp. and Proteus mirabilis. In a population-based study over a 5-year period, the annual rate of outpatient pyelonephritis among the female population in the United States was 12 to 13 cases per 10,000 people [1]. The management of APN in outpatient settings necessitates the use of highly bioavailable oral antibiotics obtaining high urine concentrations, such as fluoroquinolones or trimethoprim-sulfamethoxazole (TMP-SMX). While the incidence of APN remains relatively stable, the rates of antibiotic resistance to both fluoroquinolones and TMP-SMX are increasingly evident [1,2]. Additionally, the emergence of extended-spectrum beta-lactamase (ESBL)-producing Enterobacterales in the community poses a significant concern due to the common cross-resistance to fluoroquinolones and TMP-SMX among these isolates [2].

The 2010 IDSA and ESCMID guidelines recommend the use of oral ciprofloxacin, levofloxacin, or TMP-SMX for the management of APN [3]. These guidelines do not provide separate frameworks of recommendations for inpatient and outpatient settings. The empiric use of oral fluoroquinolones is recommended when the prevalence of fluoroquinolone resistance is <10%, whereas the use of TMP-SMX is only encouraged when in vitro susceptibility is known. When the prevalence of community fluoroquinolone resistance exceeds 10% and a fluoroquinolone is used empirically, an initial dose of a long-acting parenteral agent, such as ceftriaxone or an aminoglycoside, is recommended. An initial dose of a long-acting parenteral agent is also recommended if TMP-SMX is being used empirically. The use of oral beta-lactams may also be considered for a prolonged treatment duration, but they are noted to be less effective than fluoroquinolones and TMP-SMX due to reduced penetration into the site of infection in the kidneys. Therefore, an initial parenteral dose of ceftriaxone or aminoglycoside should be considered in this scenario. The duration of therapy ranges between 5 and 14 days depending on the antibiotic due to differences in bioavailability and tissue penetration. Due to the low prevalence in the community at the time when the guidelines were written, no specific antibiotic regimens were recommended for empiric therapy in the setting of APN in individuals with a high risk of infections due to ESBL-producing Enterobacterales.

Studies from the early 2000s demonstrated that the rate of E. coli resistance to fluoroquinolones in the United States was <7%, although certain regions (e.g., Southwest) had an elevated resistance rate of up to 20% [4]. In a more recent cross-sectional study of 10 emergency department (ED) sites surveyed, the prevalence of fluoroquinolone resistance increased substantially. with all sites having a rate of >10% [2]. For TMP-SMX resistance, the prevalence in the United States is at least 20% and up to 40% based on more recent surveillance data, although healthcare-associated UTIs were also considered in this study [4,5]. Similar trends in antibiotic resistance in Europe are also noted. Studies evaluating antimicrobial resistance among urinary isolates from Europe demonstrated resistance rates of 30% for both fluoroquinolones and TMP-SMX [6,7]. While ESBL-producing Enterobacterales have traditionally been of a low concern for community-acquired UTIs, data have emerged recently showing an increasing prevalence (up to 12%) owing to the globally disseminated, multidrug-resistant clone, ST131 E. coli, that accounts for 85% of these infections through the production of CTX-M-15 beta-lactamases [2].

Precision medicine applied in infectious diseases is a relatively new concept that proposes the customization of therapeutics and medical decisions based on an individual’s clinical and environmental characteristics rather than a one-size-fits-all approach [8]. When selecting an empiric antibiotic regimen, precision medicine leverages the host’s risk factors along with the local antibiotic susceptibility patterns to provide the best predicted patient-specific antibiogram. The use of local antibiograms and resistance risk factor prediction tools to deliver the most effective treatment regimen is a sentiment echoed throughout nearly all infectious syndromes. This is especially important when prescribing antibiotics in the outpatient setting, where timely culture and susceptibility reports are not always feasible. Therefore, the development of local antibiograms and the verification of risk factors for antibiotic resistance specific to the patient level are imperative to optimize the appropriate use of antibiotics.

In a study by Dunne and colleagues that included 15 institutions in the United States, based on urinary isolates of Enterobacterales from over 5000 patients, investigators found resistance rates of >20% to fluoroquinolones, TMP-SMX, and beta-lactam antibiotics [9]. This study compared outpatient prescriptions, the resistance of pathogens to those prescriptions, and the frequency of a second prescription. Of note, 22% of patients were prescribed an antibiotic that the urinary pathogen was resistant to. These patients were twice as likely to receive an additional antibiotic prescription or be hospitalized within 28 days of the initial prescription. The authors found that the risk of treatment failure, outside of having a resistant pathogen to the antibiotic prescribed, included an age greater than 60 years old, diabetes mellitus, and male sex.

While the selection of an empiric antibiotic regimen is often guided by the appropriate minimum acceptable susceptibility (MAS) based on a local antibiogram, other patient-specific markers should also be factored into the threshold of MAS. Haggard and colleagues demonstrated a proportional increase in the median MAS target among a cohort of clinically trained pharmacists with an increasing severity of Gram-negative infections and poor clinical prognoses. Specifically, the median MAS for critically ill patients with bloodstream infections and a poor or guarded prognosis was 90%, whereas the MAS for patients with APN and a good prognosis was 85% [10]. Another study focused on empiric antibiotic selection for APN with a proposed algorithm that incorporated multiple patient risk factors (e.g., risks for antibiotic resistance and severity of illness) in determining the MAS. In particular, the MAS might be 85% for a healthy patient with uncomplicated APN, as opposed to 90% in a patient with complicating factors [8]. This supports the establishment of a patient-specific MAS, which allows for flexibility in determining the most appropriate empiric antibiotic regimen for each individual patient.

Establishing an institutional antibiogram may not always be feasible, especially at smaller hospitals where there may be a lack of personnel and resources. Similarly, the outpatient setting also lacks resources to complete an antibiogram. Non-variable challenges include few pathogen isolates and therefore do not meet the CLSI standard of 30 isolates per species. Microbiologic testing may not be widely available in rural areas or may be limited to send out testing, creating a barrier to the primary source data (i.e., Vitek, etc.). Once an antibiogram is completed, if not appropriately analyzed and distributed, it may send the wrong message to prescribers and influence antibiotic ordering incorrectly. There may be a lack of knowledge on interpreting and utilizing the antibiogram effectively. Education is key for prescribers to enact change in prescribing patterns based on an antibiogram.

If a hospital chooses to complete an institutional antibiogram, ensuring correctness is of utmost importance. Upon a review of 37 hospitals in the Duke Antimicrobial Stewardship Outreach Network (DASON) group, the authors found that 25% of hospitals reported antibiograms with only >30 isolates per species. Additionally, CLSI required standards were observed correctly in only 9% of antibiograms [11]. If a hospital is concerned with the number of isolates needed to meet CLSI standards, an alternative approach for smaller community hospitals is to produce a regional antibiogram for more accurate prescriber practices [12]. Another alternative approach is to use a longer period for susceptibility analysis.

While an antibiogram is an important antimicrobial stewardship tool that can aid in the selection of empiric antibiotic regimens, there are limitations with the applicability of traditional antibiograms. Most hospitals typically distribute a general antibiogram that includes isolates from all sources and all units, which may be difficult to interpret as different units may have different resistance rates owing to the complexity of patients served (i.e., intensive care units versus medical ward units). Secondly, antibiograms do not distinguish between community-acquired versus hospital-acquired infections, whereby the latter are generally associated with higher rates of resistance. Thirdly, outpatient physician offices do not have clinic-specific antibiograms available for patients that have routine visits. Fourthly, a traditional antibiogram may include colonizing organisms, which could influence the susceptibility rates. Fifthly, the distribution of quantitative data such as the minimum inhibitory concentration (MIC) is not represented by the antibiogram, which is pertinent for the optimization of antibiotic dosing, especially in individuals with altered pharmacokinetic parameters (e.g., extreme body habitus and augmented renal clearance). Lastly, static antibiograms may be of limited use in empiric antibiotic selection in patients with recurrent or recent infections.

Pathogen susceptibility may be incorrectly interpreted when isolates from all units within an institution are combined as a cumulative antibiogram. Pathogens recovered from intensive care units (ICUs) typically have significantly higher rates of resistance compared to those from non-ICU locations due to the inherent risk for the repeated, prolonged exposure to broad-spectrum antibiotics and the severity of illness [13,14]. Therefore, the selection of empiric antibiotic regimens based on the hospital’s cumulative antibiogram could potentially underestimate the rate of resistance for patients in ICU locations, resulting in the use of suboptimal antibiotic agents.

Creating a urine-specific antibiogram may be helpful for determining empiric therapy, especially in outpatient and ED settings. In a study by Rabs and colleagues, 2284 pathogens were isolated from a urinary source. Overall, the antibiotic susceptibility increased; however, the prevalence of ESBL E. coli isolates was greater in the urinary antibiogram versus the standard (13% versus 9%, p < 0.001) [15]. In a veteran population of 2494 urine isolates that were analyzed, the antibiotic resistance for fluoroquinolones and TMP-SMX significantly differed when comparing E. coli isolates alone versus all urinary isolates (28% vs. 39%, p < 0.001) [16]. These findings demonstrate the potential utility of a urine-specific antibiogram.

Contrary to the traditional antibiogram, a patient-specific antibiogram accounts for numerous variables that are known to impact a patient’s risk of having a resistant organism. Overly and colleagues demonstrated that a patient-specific antibiogram based on the inclusion of key elements, including prior antibiotic exposure, prior microbiology, a history of resistant pathogens, diagnoses, age, and location in the hospital, provided a better prediction of antibiotic susceptibility compared to the traditional antibiogram [17]. More importantly, the antimicrobial resistance rates to all first-line oral options for APN have exceeded 20% in institutional antibiograms in most parts of the world. This makes population-level antibiograms obsolete in the selection of empiric oral antimicrobial therapy for APN.

Two single-center studies conducted in the United States examined independent risk factors for UTIs due to fluoroquinolone-resistant bacteria. Shah and colleagues identified multiple variables associated with fluoroquinolone resistance including the male sex, residence at a skilled nursing facility, an outpatient procedure within 1 month of the index infection, and prior fluoroquinolone use within 3 months and within 3 to 12 months of the index infection [18]. In addition, Rattanaumpawan and colleagues identified other variables that could contribute to fluoroquinolone resistance including a recent exposure to metronidazole or TMP-SMX, medicine service, hydronephrosis, and renal insufficiency [19].

The independent risk factors predicting TMP-SMX resistance in UTIs have been reported by several studies [20,21]. In a retrospective single health system review that included primarily females (72%) with E. coli community-acquired complicated UTIs (61%), the authors found that a prior UTI or urinary colonization with TMP-SMX resistant isolates and prior use of TMP-SMX within the past 12 months of the index infection were associated with a risk of TMP-SMX resistance [20]. Another study examined seven years of urine culture data collected from 574 female patients who presented to primary care clinics with uncomplicated UTIs due to E. coli in Michigan. The findings suggested that the recent use of TMP-SMX was associated with a higher risk of TMP-SMX resistance than the recent use of any other antibiotic (OR 16.74 [95% CI 2.90–96.95] vs. OR 2.37 [1.14–4.95], respectively) [22]. Another retrospective case–control study of patients with community-acquired UTIs due to E. coli, recruited from a Veterans Affair ambulatory clinic, also found that prior exposure to antibiotics including TMP-SMX, fluoroquinolones, and tetracyclines within the past 6 months increased the risk of TMP-SMX resistance. Furthermore, the study also demonstrated that as the number of previous antibiotic courses increased, the frequency of TMP-SMX resistance also increased. This ranged from a TMP-SMX resistance of 8% to 39% for patients with no prior antibiotics vs. patients with three or more prior antibiotics, respectively [21].

The increasing rates of ESBL-producing Enterobacterales detected in the community is concerning to all clinicians. In a large, multicenter, retrospective case–control study of adult patients admitted with community-acquired UTIs due to ESBL-producing Enterobacterales, those with a history of repeated UTIs, the presence of a urinary catheter at the time of admission, and prior exposure to outpatient antibiotics within the past 3 months had an increased risk for UTI due to ESBL-producing Enterobacterales after controlling for the acute severity of illness and comorbid conditions [23]. Another study examined the risk factors for ESBL-production among patients with a bloodstream infection (BSI) due to Enterobacterales. Notably, the majority of patients had a urinary source of infection and 79% had community-onset BSI. The risk factors for BSI due to ESBL-producing Enterobacterales included prior beta-lactam and/or fluoroquinolone use within the past 90 days, an outpatient gastrointestinal and/or genitourinary procedure within the past month, and a prior infection and/or colonization with an ESBL-producing organism within the past 12 months [24].

In a study by Goodman and colleagues, a clinical risk score with 14 variables was compared to a 5-variable decision tree model to predict patients’ risk of ESBL bacteremia. While this study focused on BSI, the authors found that the decision tree model, which included variables such as a history of infection due to ESBL-producing bacteria, local ESBL rates, indwelling hardware, prior antibiotic courses, and age, was more user-friendly but had lower discrimination when compared to the clinical risk score. This is important to consider when developing a risk score tool for the empiric selection of antibiotics based on predicted resistance [25].

A patient-specific risk factor evaluation would lead to a higher likelihood of selecting the appropriate antibiotic agent upon the initial patient presentation in the ambulatory setting. This would likely lead to faster clinical improvement in patients with APN. It may also result in a reduced risk of clinical deterioration, ultimately leading to hospitalization. This approach would also reduce the likelihood of future antibiotic switches that may be required if the bacterial isolate in the urine was resistant to the prescribed antibiotic. This will reduce the risk of exposure to multiple classes of antibiotics for the treatment of the same infection, which may subsequently reduce the risk of colonization with multi-drug resistant pathogens and Clostridioides difficile infection. Frequent switches in ambulatory antibiotic prescriptions may also lead to a reduction in patients’ compliance with therapy due to cost and other logistical barriers to the second prescription. In addition, multiple antibiotic prescriptions may lead to a higher likelihood of developing an adverse drug reaction.

As demonstrated in the study by Dunne and colleagues, patients with resistant pathogens often need an alternative prescription, which places strain on the healthcare system [9]. Another US study conducted by Jorgensen and colleagues analyzed the rate of emergency department return visits (ERVs) in relation to UTIs [26]. In this study, 15% of patients had an ERV, and 47% of those were related to a UTI. Of note, among the patients that had an ERV, 47% had a resistance to the discharge antibiotic. Once again, this study reiterates the burden of incorrectly stratifying patients in a one-size-fits-all category for antibiotic prescribing. This can lead to return primary care visits, return ED visits, or even hospitalization. It is of utmost importance to ensure that patients are receiving the correct antibiotic based on patient-specific risk factors from the beginning of their treatment course to prevent future complications.