Skip to nav Skip to content

IDSA 2023 Guidance on the Treatment of Antimicrobial Resistant Gram-Negative Infections

Published by IDSA on 6/7/2023. Document is current as of 12/01/22,

A Focus on Extended-spectrum β-lactamase-Producing Enterobacterales, AmpC β-Lactamase-Producing Enterobacterales, Carbapenem-Resistant Enterobacterales, Pseudomonas aeruginosa with Difficult-to-Treat Resistance, Carbapenem-Resistant Acinetobacter baumannii, and Stenotrophomonas maltophilia 

This updated document replaces previous versions of the guidance document.

Clinical Infectious Diseases, ciad428,
Published: 18 July 2023


Pranita D. Tamma*, Samuel L. Aitken, Robert A. Bonomo, Amy J. Mathers, David van Duin, Cornelius J. Clancy

*Corresponding Author

Please submit your feedback and comments on the AMR Guidance by emailing


Antimicrobial resistance (AMR) is a global crisis. Internationally, approximately 1.3 million deaths were estimated to be directly attributable to antimicrobial resistant bacterial pathogens in 2019[1]. In the United States, antimicrobial resistant pathogens caused more than 2.8 million infections and over 35,000 deaths annually from 2012 through 2017, according to the Centers for Disease Control and Prevention (CDC) Antibiotic Resistance Threats in the United States Report[2]. The Infectious Diseases Society of America (IDSA) identified the development and dissemination of clinical practice guidelines and other guidance documents as a top initiative in its 2019 Strategic Plan [3]. IDSA acknowledged that the ability to address rapidly evolving topics such as AMR was limited by prolonged timelines needed to generate new or updated clinical practice guidelines, which are based on systematic literature reviews and employ GRADE (Grading of Recommendations Assessment, Development, and Evaluation) methodology. Additionally, when clinical trial data and other robust studies are limited or not available, the development of clinical practice guidelines is challenging. As an alternative to practice guidelines, IDSA endorsed developing more narrowly focused guidance documents for the treatment of infections where data continue to rapidly evolve. Guidance documents are prepared by a small team of experts, who answer questions about treatment based on a comprehensive (but not necessarily systematic) review of the literature, clinical experience, and expert opinion. Documents do not include formal grading of evidence, and are made available online and updated annually.

In the present document, guidance is provided on the treatment of infections caused by extended-spectrum β-lactamase-producing Enterobacterales (ESBL-E), AmpC β-lactamase-producing Enterobacterales (AmpC-E), carbapenem-resistant Enterobacterales (CRE), Pseudomonas aeruginosa with difficult-to-treat resistance (DTR-P. aeruginosa), carbapenem-resistant Acinetobacter baumannii species (CRAB), and Stenotrophomonas maltophilia. Many of these pathogens have been designated urgent or serious threats by the CDC[2]. Each pathogen causes a wide range of infections that are encountered in United States hospitals of all sizes, and that carry with them significant morbidity and mortality.

Guidance is presented in the form of answers to a series of clinical questions for each pathogen. Although brief descriptions of notable clinical trials, resistance mechanisms, and antimicrobial susceptibility testing (AST) methods are included, the document does not provide a comprehensive review of these topics. GRADE methodology was not employed. Due to differences in the molecular epidemiology of resistance and availability of specific antibiotics internationally, treatment recommendations are geared toward antimicrobial resistant infections in the United States. The content of this document is current as of December 31st, 2022. The most current version of this IDSA guidance document and corresponding date of publication is available at:

Table 1.  Suggested dosing of antibiotics for the treatment of antimicrobial resistant infections in adults, assuming normal renal and hepatic function


IDSA convened a panel of six actively practicing infectious diseases specialists with clinical and research expertise in the treatment of antimicrobial resistant bacterial infections. Through a series of virtual meetings, the panel developed commonly encountered treatment questions and corresponding suggested treatment approaches for each pathogen group. Answers include a brief discussion of the rationale supporting the suggested approaches. This guidance document applies to both adult and pediatric populations. Suggested antibiotic dosing for adults with antimicrobial resistant infections, assuming normal renal and hepatic function, are provided in Table 1. Pediatric dosing is not provided. 

General Management Recommendations

Treatment recommendations in this guidance document assume that the causative organism has been identified and that in vitro activity of antibiotics is demonstrated. Assuming two antibiotics are equally effective, safety, cost, convenience, and local formulary availability are important considerations in selecting a specific agent. The panel recommends that infectious diseases specialists and physician or pharmacist members of the local antibiotic stewardship program are involved in the management of patients with infections caused by antimicrobial-resistant organisms.

In this document, the term complicated urinary tract infection (cUTI) refers to UTIs occurring in association with a structural or functional abnormality of the genitourinary tract, or any UTI in an adolescent or adult male. In general, the panel suggests cUTI be treated with similar agents and for similar treatment durations as pyelonephritis. For cUTI where the source has been controlled (e.g., removal of a Foley catheter) and ongoing concerns for urinary stasis or indwelling urinary hardware are no longer present, it is reasonable to select antibiotic agents and treatment durations similar to uncomplicated cystitis.

Empiric Therapy

Empiric treatment decisions should be guided by the most likely pathogens, severity of illness of the patient, the likely source of the infection, and any additional patient-specific factors (e.g., severe penicillin allergy, chronic kidney disease). When determining empiric treatment for a given patient, clinicians should also consider: (1) previous organisms identified from the patient and associated antibiotic susceptibility testing (AST) data in the last twelve months, (2) antibiotic exposure within the past 30 days, and (3) local AST patterns for the most likely pathogens. Empiric decisions should be refined based on the identity and AST profile of the pathogen, as well as based on the identification of any prominent β-lactamase genes.

For DTR-P. aeruginosa, CRAB, and S. maltophilia, in particular, a distinction between bacterial colonization and infection is important as unnecessary antibiotic therapy will only further the development of resistance and may cause unnecessary antibiotic-related harm to patients. Commonly selected empiric antibiotic regimens are generally not active against CRAB and S. maltophilia infections. The decision to target treatment for CRAB and/or S. maltophilia in empiric antibiotic regimens should involve a careful risk-benefit analysis after reviewing previous culture results, clinical presentation, individual host risk factors, and antibiotic-specific adverse event profiles.

Duration of Therapy and Transitioning to Oral Therapy

Recommendations on durations of therapy are not provided, but clinicians are advised that the duration of therapy should not differ for infections caused by organisms with resistant phenotypes compared to infections caused by more susceptible phenotypes. After antibiotic susceptibility results are available, it may become apparent that inactive antibiotic therapy was initiated empirically. This may impact the duration of therapy. For example, cystitis is typically a mild infection [4]. If an antibiotic not active against the causative organism was administered empirically for cystitis, but clinical improvement nonetheless occurred, it is generally not necessary to repeat a urine culture, change the antibiotic regimen, or extend the planned treatment course. However, for all other infections, if antibiotic susceptibility data indicate a potentially inactive agent was initiated empirically, a change to an active regimen for a full treatment course (dated from the start of active therapy) is recommended. Additionally, important host factors related to immune status, ability to attain source control, and general response to therapy should be considered when determining treatment durations for antimicrobial-resistant infections, as with the treatment of any bacterial infection. Finally, whenever possible, oral step-down therapy should be considered, particularly if the following criteria are met: (1) susceptibility to an appropriate oral agent is demonstrated, (2) the patient is hemodynamically stable, (3) reasonable source control measures have occurred, and (4) concerns about insufficient intestinal absorption are not present [5].

Extended-Spectrum β-Lactamase-Producing Enterobacterales

The incidence of ESBL-E identified in bacterial cultures in the United States increased by 53% from 2012 to 2017, in large part due to a greater number of community-acquired infections [6, 7]. ESBLs are enzymes that inactivate most penicillins, cephalosporins, and aztreonam. EBSL-E generally remain susceptible to carbapenems. ESBLs do not inactivate non-β-lactam agents (e.g., ciprofloxacin, trimethoprim-sulfamethoxazole [TMP-SMX], gentamicin). However, organisms carrying ESBL genes often harbor additional genes or mutations in genes that mediate resistance to a broad range of antibiotics.

Any gram-negative organism has the potential to harbor ESBL genes; however, they are most prevalent in Escherichia coli, Klebsiella pneumoniae, Klebsiella oxytoca, and Proteus mirabilis[8-10]. CTX-M enzymes, particularly CTX-M-15, are the most common ESBLs in the United States[10]. ESBLs other than CTX-M with unique hydrolyzing abilities are also present, including variants of narrow-spectrum TEM and SHV β-lactamases with amino acid substitutions, but they have undergone less rigorous clinical investigation than CTX-M enzymes [11-14]. Routine EBSL testing is not performed by most clinical microbiology laboratories [15, 16]. Rather, non-susceptibility to ceftriaxone (i.e., ceftriaxone minimum inhibitory concentrations [MICs] ≥2 µg/mL), is often used as a proxy for ESBL production, although this threshold has limitations with specificity as organisms not susceptible to ceftriaxone for reasons other than ESBL production may be falsely presumed to be ESBL-producers [17, 18]. For this guidance document, ESBL-E will refer to presumed or confirmed ESBL-producing E. coli, K. pneumoniae, K. oxytoca, or P. mirabilis. Treatment suggestions for ESBL-E infections assume that in vitro activity of preferred and alternative antibiotics has been demonstrated.


Question 1.1: What are preferred antibiotics for the treatment of uncomplicated cystitis caused by ESBL-E?

Suggested approach: Nitrofurantoin and TMP-SMX are preferred treatment options for uncomplicated cystitis caused by ESBL-E. Ciprofloxacin, levofloxacin, and carbapenems are alternative agents for uncomplicated cystitis caused by ESBL-E. Although effective, their use is discouraged when nitrofurantoin or TMP-SMX are active. Single dose aminoglycosides and oral fosfomycin (for E. coli only) are also alternative treatments for uncomplicated cystitis caused by ESBL-E. 


Nitrofurantoin and TMP-SMX have been shown to be safe and effective options for uncomplicated cystitis, including uncomplicated ESBL-E cystitis [4, 19, 20]. Although carbapenems and the fluoroquinolones ciprofloxacin or levofloxacin are effective agents against ESBL-E cystitis [21, 22], their use for uncomplicated cystitis is discouraged when other safe and effective options are available. Limiting use of these agents preserves their activity for future infections when treatment options may be more restricted. Moreover, limiting their use reduces the risk of associated toxicities, particularly with the fluoroquinolones, which have been associated with an increased risk for prolonged QTc intervals, tendinitis and tendon rupture, aortic dissections, seizures, peripheral neuropathy, and Clostridioides difficile infections[23-26].

Treatment with a single intravenous (IV) dose of an aminoglycoside is an alternative treatment option for uncomplicated ESBL-E cystitis. Aminoglycosides are nearly exclusively eliminated by the renal route. A single IV dose is generally effective for uncomplicated cystitis, with minimal toxicity, but robust clinical trial data are lacking [27]. Oral fosfomycin is an alternative treatment option exclusively for uncomplicated ESBL-E cystitis caused by E. coli. Fosfomycin is not suggested for the treatment of infections caused by K. pneumoniae and several other gram-negative organisms which frequently carry fosA hydrolase genes that may lead to clinical failure[28, 29]. A randomized open-label trial indicated that a single dose of oral fosfomycin is associated with higher clinical failure than a five-day course of nitrofurantoin for uncomplicated cystitis [19]. Although this trial was not limited to E. coli cystitis, in a subgroup analysis exclusively of E. coli infections, outcomes remained poor in the fosfomycin group with day 14 clinical failure at 50% in the fosfomycin group versus 22% in the nitrofurantoin group[19]. The additive benefit of a second dose of oral fosfomycin for uncomplicated cystitis is not known.

The panel does not suggest prescribing amoxicillin-clavulanic acid or doxycycline for the treatment of ESBL-E cystitis. A randomized clinical trial compared a three-day regimen of amoxicillin-clavulanic acid to a three-day course of ciprofloxacin for 370 women with uncomplicated E. coli cystitis [21]. Clinical cure was observed in 58% and 77% of the women randomized to the amoxicillin-clavulanic acid and ciprofloxacin arms, respectively. The higher failure rates with amoxicillin-clavulanic acid appear associated with persistent vaginal bacterial colonization, which occurred in 45% and 10% of patients in the amoxicillin-clavulanic acid and ciprofloxacin arms, respectively [21]. The proportion of women in the trial infected with ESBL-E strains is not available. Even though data indicate that clavulanic acid may be effective against ESBLs in vitro[30, 31], this may not translate to clinical efficacy[32].  Robust data indicating that oral amoxicillin-clavulanic acid is effective for uncomplicated ESBL-E UTI are lacking.

Two clinical outcomes studies, published more than 40 years ago, demonstrated that oral tetracyclines may be effective for the treatment of UTIs [33, 34]. Both of these studies, however, primarily focused on P. aeruginosa, an organism not susceptible to oral tetracyclines, questioning the impact that antibiotic therapy had on clinical cure. Doxycycline is primarily eliminated through the intestinal tract[35]. Its urinary excretion is limited. Until more convincing data demonstrating the clinical effectiveness of oral doxycycline for the treatment of ESBL-E cystitis are available, the panel suggests against use of doxycycline for this indication. The roles of piperacillin-tazobactam, cefepime, and the cephamycins for the treatment of uncomplicated cystitis are discussed in Question 1.4, Question 1.5, and Question 1.6, respectively.

Question 1.2: What are preferred antibiotics for the treatment of pyelonephritis and cUTI caused by ESBL-E?

Suggested approach: TMP-SMX, ciprofloxacin, or levofloxacin are preferred treatment options for pyelonephritis and cUTIs caused by ESBL-E. Ertapenem, meropenem, and imipenem-cilastatin are preferred agents when resistance or toxicities preclude the use of TMP-SMX or fluoroquinolones. Aminoglycosides for a full treatment course are an alternative option for the treatment of ESBL-E pyelonephritis or cUTI.


TMP-SMX, ciprofloxacin, and levofloxacin are preferred treatment options for patients with ESBL-E pyelonephritis and cUTIs based on the ability of these agents to achieve adequate and sustained concentrations in the urine, clinical trial results, and clinical experience[36-38]. Carbapenems are also preferred agents, when resistance or toxicities prevent use of TMP-SMX or fluoroquinolones, or early in the treatment course if a patient is critically ill (Question 1.3). If a carbapenem is initiated and susceptibility to TMP-SMX, ciprofloxacin, or levofloxacin is demonstrated, transitioning to oral formulations of these agents is preferred over completing a treatment course with a carbapenem. Limiting use of carbapenem exposure will preserve their activity for future antimicrobial resistant infections.

In patients in whom the potential for nephrotoxicity is deemed acceptable, aminoglycosides (dosed based on therapeutic drug monitoring results) for a full treatment course are an alternative option for the treatment of ESBL-E pyelonephritis or cUTI [39, 40] (Table 1, Supplemental Material). Once-daily plazomicin was noninferior to meropenem in a clinical trial that included patients with pyelonephritis and cUTIs caused by Enterobacterales [41]. Individual aminoglycosides are equally effective if susceptibility is demonstrated. Of note, in January 2023 the Clinical Laboratory and Standards Institute (CLSI) revised the aminoglycoside breakpoints[16] (Table 2). 

Fosfomycin is not suggested for the treatment of pyelonephritis or cUTI given its limited renal parenchymal concentrations. However, more data are needed to evaluate the role of oral fosfomycin as an oral step-down agent for patients with pyelonephritis or cUTI, particularly when administered as a multidose regimen and after several days of preferred therapy. A clinical trial of 97 women with E. coli pyelonephritis (approximately half of patients had associated bacteremia) who received up to 5 days of IV therapy and were subsequently transitioned to either once-daily 3 g doses of oral fosfomycin or twice daily 500 mg doses of oral ciprofloxacin for 10 days of total antibiotic therapy identified similar clinical cure percentages in both groups (75% versus 65%, respectively)[42]. However, only approximately 6% of isolates were ESBL-producing, limiting generalizability to pyelonephritis caused by more drug-resistant phenotypes[42]. Fosfomycin is an alternative option for the treatment of prostatitis caused by ESBL-producing E. coli when preferred options (i.e., carbapenems, TMP-SMX, or fluoroquinolones) cannot be tolerated or do not test susceptible [43-48]. In an observational study, fosfomycin, dosed at 3 g orally daily for one week, followed by 3 g orally every 48 hours for 6 to 12 weeks, was associated with clinical cure in 36 (82%) of 44 males with chronic bacterial prostatitis [43]. Fosfomycin should be avoided for prostatitis caused by gram-negative organisms other than E. coli (Question 1.1).

Nitrofurantoin does not achieve adequate concentrations in the renal parenchyma and is not advised for pyelonephritis or cUTI. Doxycycline is also not advised for the treatment of ESBL-E pyelonephritis or cUTIs due to its limited urinary excretion and limited published comparative effectiveness studies (Question 1.1) [35]. The roles of piperacillin-tazobactam, cefepime, and the cephamycins for the treatment of pyelonephritis and cUTIs are discussed in Question 1.4, Question 1.5, and Question 1.6, respectively.


Question 1.3: What are preferred antibiotics for the treatment of infections outside of the urinary tract caused by ESBL-E?

Suggested approach: Meropenem, imipenem-cilastatin, or ertapenem are preferred for the treatment of infections outside of the urinary tract caused by ESBL-E. For patients who are critically ill and/or experiencing hypoalbuminemia, meropenem or imipenem-cilastatin are the preferred carbapenems. After appropriate clinical response is achieved, transitioning to oral trimethoprim-sulfamethoxazole, ciprofloxacin, or levofloxacin should be considered, if susceptibility is demonstrated.


A carbapenem is recommended as first-line treatment of ESBL-E infections outside of the urinary tract, based primarily on data from a large clinical trial, as described below [49]. Meropenem, imipenem-cilastatin, or ertapenem are preferred agents; ertapenem offers a more convenient option for patients needing to continue carbapenem therapy in the outpatient setting when oral treatment options are not available. For patients who are critically ill and/or experiencing hypoalbuminemia, meropenem or imipenem-cilastatin are the preferred carbapenems.

Ertapenem, in contrast to meropenem and imipenem, is highly protein bound leading to a relatively prolonged serum half-life[50]. In patients with hypoalbuminemia and critical illness, the free fraction of ertapenem increases leading to a significant decrease in the serum half-life[51-53]. An observational study of 279 patients with Enterobacterales infections found that hypoalbuminemia (defined as serum albumin <2.5 g/dL) was associated with an approximately five-times higher odds of 30-day mortality for patients receiving ertapenem compared to those receiving meropenem or imipenem-cilastatin[54]. Clinical literature regarding the use of ertapenem, relative to other carbapenems, in critically ill patients is limited and conflicting[53, 55]. However, given known pharmacokinetic (PK) alterations in patients with critical illness and some limitations in the pharmacokinetic/pharmacodynamic (PK/PD) profile of ertapenem[56, 57], the panel suggests the use of meropenem or imipenem-cilastatin, rather than ertapenem, as initial therapy in critically ill patients with ESBL-E infections. 

The clinical trial which established carbapenem therapy as the treatment of choice for ESBL-E bloodstream infections randomized 391 patients with ceftriaxone non-susceptible E. coli or K. pneumoniae (87% later confirmed to have ESBL genes) bloodstream infections to piperacillin-tazobactam 4.5 g IV every six hours or meropenem 1 g IV every eight hours, both as standard infusions (i.e., over 30 minutes). The primary outcome of 30-day mortality occurred in 12% and 4% of patients receiving piperacillin-tazobactam and meropenem, respectively [49]. Trial data were subsequently reanalyzed only including patients with clinical isolates against which piperacillin-tazobactam MICs were ≤16 µg/mL by broth microdilution, the reference standard for AST [58]. Reanalyzing the data from 320 patients, 30-day mortality was observed in 11% versus 4% of those in the piperacillin-tazobactam and meropenem arms, respectively. Although the absolute risk difference was attenuated and no longer significant in the reanalysis (i.e., the 95% confidence interval ranged from −1% to 10%) [58], the panel still suggests carbapenem therapy as the preferred treatment of ESBL-producing bloodstream infections due to the notable direction of the risk difference. Comparable clinical trial data are not available for ESBL-E infections of other body sites. Nevertheless, the panel suggests extrapolating evidence for ESBL-E bloodstream infections to other common sites of infection, namely intra-abdominal infections, skin and soft tissue infections, and pneumonia. Similarly, although the trial evaluated meropenem, the panel suggests extending the findings to imipenem-cilastatin and ertapenem, with the latter limited to patients with normal serum albumin and patients who are not critically ill.

In January 2022, the CLSI lowered the piperacillin-tazobactam breakpoints and piperacillin-tazobactam MICs of ≤8/4 µg/mL are considered susceptible for the Enterobacterales (Table 2)[59]. In the clinical trial, 77% and 94% of isolates would have been considered susceptible and susceptible dose-dependent, respectively, to piperacillin-tazobactam if applying revised the piperacillin-tazobactam interpretive criteria, indicating that in the presence of ESBL production, susceptibility may not correlate with clinical success[49, 58].

Data from observational studies support the use of oral step-down therapy for Enterobacterales bloodstream infections, including those caused by antimicrobial resistant isolates, after appropriate clinical milestones are achieved [60, 61]. Based on the known bioavailability and sustained serum concentrations of oral TMP-SMX and fluoroquinolones, these agents should be treatment considerations for patients with ESBL-E infections if (1) susceptibility to one of these agents is demonstrated, (2) the patient is hemodynamically stable, (3) reasonable source control has occurred, and (4) concerns about insufficient intestinal absorption are not present [5].

Clinicians should avoid oral step-down to nitrofurantoin, fosfomycin, amoxicillin-clavulanate, doxycycline, or omadacycline for ESBL-E bloodstream infections. Nitrofurantoin and fosfomycin achieve poor serum concentrations. Amoxicillin-clavulanate and doxycycline achieve unreliable serum concentrations.

Omadacycline is a tetracycline derivative with an oral formulation that has limited in vitro activity against ESBL-producing Enterobacterales isolates and has an unfavorable PK/PD profile for the treatment of Enterobacterales infections[62, 63]. Like other tetracyclines, omadacycline efficacy is associated with the 24-hour area under the curve to MIC ratio (AUC/MIC). In in vitro models, an AUC/MIC ratio of ~38 is needed to achieve at least a one-log kill (a standard pharmacodynamic target) for E. coli[63]. Standard oral omadacycline dosing achieves a 24-hour AUC of ~13mg*hr/L[64], suggesting limited activity of omadacycline against Enterobacterales, which have an MIC50 of 0.5 µg/mL (i.e., AUC/MIC ratio of ~26)[65]. The panel does not suggest omadacycline for the treatment of ESBL-E infections.


Question 1.4: Is there a role for piperacillin-tazobactam in the treatment of infections caused by ESBL-E?

Suggested approach: If piperacillin-tazobactam was initiated as empiric therapy for uncomplicated cystitis caused by an organism later identified as an ESBL-E and clinical improvement occurs, no change or extension of antibiotic therapy is necessary. The panel suggests TMP-SMX, ciprofloxacin, levofloxacin, or carbapenems rather than piperacillin-tazobactam for the treatment of ESBL-E pyelonephritis and cUTI, with the understanding that some data suggest the risk of clinical failure with piperacillin-tazobactam may be low. Piperacillin-tazobactam is not suggested for the treatment of infections outside of the urinary tract caused by ESBL-E, even if susceptibility to piperacillin-tazobactam is demonstrated.


Piperacillin-tazobactam demonstrates in vitro activity against a number of ESBL-E[66]. There are several concerns regarding tazobactam’s ability to function as an effective β-lactamase inhibitor. First, piperacillin-tazobactam MIC testing may be inaccurate and/or poorly reproducible when ESBL enzymes are present, or in the presence of other β-lactamase enzymes such as OXA-1, making it unclear if an isolate that tests susceptible to this agent is indeed susceptible [58, 67-70]. Second, in vitro data indicate that with increased bacterial inoculum (e.g., abscesses), piperacillin-tazobactam may no longer be effective against ESBL-E when compared to meropenem; however, the clinical implications of these findings are unclear[71-73]. Additionally, the effectiveness of tazobactam may be diminished by organisms with increased expression of ESBL enzymes or by the presence of multiple ESBL or other β-lactamases[74]. Finally, there are ESBL enzymes that are inhibitor resistant (i.e., not inhibited by β-lactamase inhibitors)[75, 76].

If piperacillin-tazobactam was initiated as empiric therapy for uncomplicated cystitis caused by an organism later identified as an ESBL-E and clinical improvement occurs, no change or extension of antibiotic therapy is necessary, as uncomplicated cystitis often resolves on its own. At least three observational studies have compared the efficacy of piperacillin-tazobactam and carbapenems for the treatment of ESBL-E pyelonephritis or cUTI [77-79]. The most robust observational study included 186 hospitalized patients from five hospitals with pyelonephritis or cUTI caused by E. coli, K. pneumoniae, K. oxytoca, or P. mirabilis, with confirmation of the presence of ESBL genes in all isolates. This study identified no difference in the resolution of clinical symptoms or 30-day mortality between the groups [77]. A randomized, open-label clinical trial investigating this question was also conducted [80]. The trial included 66 patients with ESBL-producing E. coli pyelonephritis or cUTI (with confirmation of the presence of ESBL genes) randomized to either piperacillin-tazobactam 4.5 g IV every six hours or ertapenem 1 g IV every 24 hours. Clinical success was similar between both groups at 94% for piperacillin-tazobactam and 97% for ertapenem. These studies suggest non-inferiority between piperacillin-tazobactam and carbapenems for pyelonephritis or cUTIs.

In the subgroup of 231 patients with ESBL-E bloodstream infections from a urinary source in the aforementioned clinical trial comparing the outcomes of patients with E. coli or K. pneumoniae bloodstream infections treated with piperacillin-tazobactam or meropenem (Question 1.3), higher mortality was identified in the piperacillin-tazobactam group (7% vs. 3%) [49], although it did not attain statistical significance. The panel is unable to state that piperacillin-tazobactam should be avoided for pyelonephritis or cUTIs; however, given concerns with the efficacy of tazobactam as an ESBL inhibitor and the clinical trial results, the panel has concerns with the use of piperacillin-tazobactam for the treatment of ESBL-E pyelonephritis or cUTIs, and prefers carbapenem therapy (or oral trimethoprim-sulfamethoxazole, ciprofloxacin, or levofloxacin, if susceptible), particularly in the setting of urosepsis (Question 1.2).

Observational studies have had conflicting results regarding the effectiveness of piperacillin-tazobactam for the treatment of ESBL-E bloodstream infections