Extended-spectrum β-lactamases (ESBLs) are increasing in prevalence due to selective pressure created by the use of extended-spectrum cephalosporins. Reference Wood1 In 2010, the Clinical and Laboratory Standards Institute (CLSI) issued minimum inhibitory concentration (MIC) revised breakpoints for many β-lactams such that confirmatory testing for ESBL production is no longer necessary. 2 These new breakpoints are intended to more clearly identify resistant organisms without the need for an ESBL confirmatory test, alleviating this workload in the microbiology laboratory and expediting antimicrobial susceptibility test (AST) result reporting. Reference Harris, Tambyah and Lye8 In May 2017, our microbiology laboratory changed our microbroth dilution AST panel (Beckman Coulter Microscan, Brea, CA) to comply with the CLSI MIC revised breakpoints, and we eliminated the standard reporting of an ESBL within the electronic medical record (EMR) (Supplementary Fig. 1 online). The primary objective of this study was to characterize the impact of removal of the ESBL designation from microbiology reports in the EMR on inpatient antibiotic prescribing.
Methods
This retrospective, noninterventional chart analysis study was conducted at Penn State Health Milton S. Hershey Medical Center, a 550-bed academic medical center that includes a 125-bed children’s hospital. Our antimicrobial stewardship program (ASP) is described in a separate publication. Reference Katzman, Kim and Lesher3 The Microscan LabPro Command Center database (Microscan WalkAway 96 Plus, Beckman Coulter) was used to identify adult and pediatric patients with positive cultures for ESBL-producing Enterobacteriaceae during the 6-month periods before revised breakpoints were implemented (January 1, 2017, to April 30, 2017) and after the revised breakpoints were implemented (May 1, 2017, to October 31, 2017). Patients and isolates were excluded from the current analysis if any of the following conditions applied: treatment initiated in the outpatient setting, no antibiotics prescribed, antibiotics were prescribed for an indication unrelated to the clinical culture, central nervous system (CNS) infection, organism not reported in the EMR (eg, final report of “mixed flora”), polymicrobial cultures with varying susceptibilities impacting antibiotic selection, or isolates demonstrating resistance to all β-lactam–β-lactamase inhibitor combinations (BL/BLIs) and/or carbapenems. For patients with multiple encounters, only the first instance was included.
Bacteria were identified using the MALDI-TOF (Bruker Microflex, Billerica, MA) database using scores >2.0. During evaluations before the revised breakpoints were implemented, antimicrobial susceptibilities were determined using the MicroScan Neg Combo Panel Neg/Urine Combo 51 or NUC51 (Beckman Coulter, Brea, CA), which included an ESBL screening test. During the period after the breakpoints were revised, the MicroScan MN46 panel was verified according to CLSI standards and was used for initial AST. This panel does not include ESBL screening but includes increased range of concentrations of cephalosporins (Supplementary Table 1 online). All AST was performed according to the manufacturer’s instructions using Microscan WalkAway 40SI (Beckman Coulter). Isolates with high MICs for third- and fourth-generation cephalosporins were retested using the NUC51 panel to confirm the ESBL phenotype.
Charts of patients meeting inclusion criteria were reviewed to obtain the following information: patient demographics, organism isolated, source of isolate/infection, and the definitive antibiotic treatment selected after the date and time microbiology results were posted in the EMR. Only the first antibiotic chosen on the basis of the culture and susceptibility result was recorded for the primary outcome. Antibiotics were grouped by therapeutic class for comparisons.
Statistical analysis
Statistical analysis was performed using SAS software (SAS Institute, Cary, NC). Categorical variables were compared using the Fisher exact test. Continuous variables were evaluated using the Wilcoxon rank-sum test. All statistical tests were 2-tailed, and a P < .05 was considered statistically significant.
Results
In total, 285 ESBL-producing bacteria were identified across 263 patient encounters: 115 isolates before the revised breakpoints were implemented and 170 isolates after revised breakpoints were implemented (Supplementary Fig. 2 online). Among these isolates, 57 were cultured from inpatients (50 patients) before and 79 were cultured from inpatients (66 patients) after the removal of the “ESBL” label. Patient and clinical characteristics (after applying exclusion criteria) are presented in Table 1. The most common organism isolated was Escherichia coli. The numbers of patients with ESBL bacteremia were similar between groups: 35.5% before breakpoints were revised and 32.3% after breakpoints were revised. Overall, 17 (54.8%) patients in the period before revised breakpoints were implemented and 21 patients (67.7%) in the period after revised breakpoints were implemented had infections of a urinary source, defined as patients with urine cultures growing the organism as well as those who had concomitant blood and urine cultures with the organism.
Table 1. Patient and Microbiology Laboratory Characteristics
Note. IQR, interquartile range; BL/BLI, β-lactam–β-lactamase inhibitor combination.
Definitive prescribing of carbapenems decreased from 48.4% before the breakpoint revision to 16.1% after the revision (P = .01), whereas BL/BLI use increased from 19.4% to 61.3%, respectively (P = .002) (Fig. 1). The most commonly prescribed BL/BLIs were piperacillin/tazobactam (before breakpoints were revised, n = 4, and after breakpoints were revised, n = 8) and amoxicillin/clavulanate (before breakpoints were revised, n = 2, and after breakpoints were revised, n = 9). Cefepime prescribing also increased from 3.2% to 9.7%, respectively, but this change was not statistically significant. In the subset of patients with ESBL bacteremia (of any source), a similar trend was noted, with 72.7% of patients being prescribed a carbapenem in the period before breakpoint revision compared to only 30.0% after break-point revision. BL/BLI prescribing also increased in these bacteremia patients from 18.2% to 70.0%, respectively. Considering only infections of a urinary source, the trend was similar, with 41.2% of patients receiving a carbapenem before breakpoints were revised compared to 9.5% afterward. Our ASP daily prospective audit-and-feedback approach did not change over the study period, and no additional interventions were conducted that were expected to impact these results.
Fig. 1. Antibiotic prescribing for definitive treatment of infections due to ESBL-producing organisms before implementation of revised breakpoints and removal of ESBL designation from culture reports (pre-RB) and after (post-RB).
Discussion
Close collaboration with the clinical microbiology laboratory is paramount for a successful ASP, and an important aspect of this relationship is how microbiology results are reported and interpreted by clinicians. Strategies to improve the clarity of microbiology results require few additional resources and can strongly influence prescriber behavior, for example, adding interpretive comments or implementing selective reporting of susceptibilities. Reference Morency-Potvin, Schwartz and Weinstein5,Reference Cunney and Smyth6 Musgrove et al Reference Musgrove, Kenney, Kendall, Peters, Tibbetts, Samuel and Davis7 recently described how the modification of a comment to respiratory cultures growing commensal flora favorably resulted in increased antibiotic de-escalation or discontinuation. Reference Musgrove, Kenney, Kendall, Peters, Tibbetts, Samuel and Davis7 Our study is unique in that we assessed the impact of removing potentially informative comments (ie, the term “ESBL”) rather than the addition of comments, and we observed a significant change in prescribing.
Although a reduction in carbapenem use is favorable from ASP, resistance, and cost perspectives, the clinical impact of these metrics is unclear. Carbapenems are often selected over BL/BLI combinations for nonurinary infections due to ESBL-producing organisms regardless of BL/BLI susceptibility, but this practice continues to be questioned. Reference Tamma, Han and Rock4,Reference Harris, Tambyah and Lye8 Most available data suggest that these 2 treatment options are equivalent, particularly when the BL/BLI MIC is low and for infections of a urinary source; however, most of these studies are observational in nature. Reference Vardakas, Tansarli, Rafalidis and Falagas9 In 2018, a randomized noninferiority trial of 378 patients was published that compared meropenem to piperacillin-tazobactam for bloodstream infections due to ceftriaxone-resistant E. coli and Klebsiella pneumoniae. Definitive treatment with piperacillin-tazobactam did not result in a noninferior 30-day mortality compared with meropenem, with 12.3% of patients in the piperacillin-tazobactam arm meeting the primary outcome of mortality compared to 3.7% in the meropenem arm (P = .90 for noninferiority). Reference Harris, Tambyah and Lye8 This study reaffirmed the notion that carbapenems should be preferred for bloodstream infections due to ESBL-producing organisms; however, BL/BLIs may still be appropriate for urinary tract infections. Because the removal of the ESBL label from the culture reports was associated with reduced carbapenem use for definitive treatment of these infections, there is potential for patients with severe infections, including bacteremia, to be treated with a suboptimal antibiotic. We also observed an increase in the number of patients who received definitive cefepime treatment after the “ESBL” label was removed, though this was not statistically significant due to small numbers. This factor compounds the possibility that a suboptimal antibiotic may be chosen, and studies have consistently shown that cefepime is not a good option for ESBL infections regardless of the site of infection. Reference Vardakas, Tansarli, Rafalidis and Falagas9,Reference Wang, Cosgrove and Tschudin-Sutter10
Importantly, therefore, clinicians should be informed of the possibility that an organism with ESBL production is present based on the susceptibility pattern. One way to accomplish this is to include an interpretive comment in the presence of third- and fourth-generation cephalosporin resistance (with aztreonam resistance if tested). Reference Cunney and Smyth6 Based on the results of this study, our ASP team developed the following interpretive comment that has been applied to nonurine cultures meeting the aforementioned criteria: “Possible ESBL producer. A carbapenem is considered the drug of choice for severe infections due to ESBL-producing organisms. Please contact Infectious Diseases or the Antimicrobial Stewardship Program if you have questions.” Future evaluation of the impact of this comment on antibiotic prescribing would provide additional insight into clinician practice, especially given that positive culture reports can influence therapeutic management responses, even when the organism is not considered a pathogen. Although it was not a primary aim of our study, we observed that 90.2% of inpatients with an ESBL-labeled isolate in the period before breakpoints were revised were treated compared with 73.8% in the period after they were revised (Supplementary Fig. 2 online). Nearly all of these patients had urine cultures (data not shown), which suggests that clinicians may be more inclined to treat non–clinically relevant cultures based on the label of “ESBL.” The treatment of asymptomatic bacteriuria is discouraged in national guidelines, and our findings highlight the continued importance of ASPs to aid in the clinical interpretation of these cultures.
Our study has several limitations. First, this study has a retrospective design, and only first encounters were collected, whereas patient outcomes and treatment appropriateness were not assessed. We also did not assess the antibiotic treatment decisions in outpatients in this study, which comprised more than half of the patient encounters assessed for inclusion: before breakpoints were revised, n = 58, and after breakpoints were revised, n = 89. With the growing emphasis on outpatient antimicrobial stewardship, this finding necessitates further focus on how and what antibiotics are being prescribed in the outpatient setting. Second, our study used relatively small sample size of inpatients with a high percentage of urine cultures, although we did detect a statistically significant change in antimicrobial prescribing. Finally, patients with a reported penicillin allergy were not excluded; however, this likely did not affect our results. Only 2 of the 5 patients with a reported penicillin allergy in the period before the breakpoints were revised received a carbapenem, and both had nonsevere allergies in the medical record (diarrhea and unknown). In the period after the revised breakpoints were implemented, all penicillin allergies were unknown.
The way in which results are reported by the microbiology laboratory plays an important role in ASP and antibiotic decision making in the clinical management of the patient. Implementation of CLSI revised breakpoints and removal of ESBL labeling in the EMR resulted in a significant change in inpatient antibiotic prescribing, specifically a reduction in carbapenem and increase in BL/BLI use for infections due to ESBL producers. Our data suggest that as microbiology laboratories continue to collaborate with their institutional ASPs to inform result reporting, appropriate communication of updated AST reports will influence optimal therapeutic management. In this data set, informing clinicians of the presence of a possible or confirmed ESBL producer should aid in optimal antibiotic selection, particularly for bloodstream infections.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2020.46
Acknowledgments
We would like to thank Jill Rebuck (Director of Clinical Pharmacy Services at Hershey Medical Center) for her support throughout this manuscript.
Financial support
No financial support was provided relevant to this article.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
