Since the late 1980s, infections caused by extended-spectrum β-lactamase (ESBL)-producing organisms have been reported worldwide.Reference Jacoby, Medeiros, O’Brien, Pinto and Jiang 1 – Reference Bell, Turnidge, Gales, Pfaller and Jones 8 Patients infected with ESBL-producing E. coli or Klebsiella can have poor outcomes due to delays in receiving appropriate antimicrobial therapy and limited treatment options. Strains harboring ESBLs are resistant to third-generation cephalosporins, extended-spectrum penicillins, and monobactams.Reference Einhorn, Neuhauser, Bearden, Quinn and Pendland 9 , Reference Rice 10 Additionally, many ESBL-producing strains of E. coli and Klebsiella demonstrate coresistance to other antimicrobial agents such as fluoroquinolones, aminoglycosides, and tetracyclines.Reference Pitout 11
Trends among surveillance studies suggest an increase in the proportion of ESBL-producing strains among isolates collected from E. coli or Klebsiella infections.Reference Jones, Biedenbach and Gales 12 – Reference Karlowsky, Jones, Mayfield, Thornsberry and Sahm 15 However, the incidence of E. coli or Klebsiella infections caused by ESBL-producing strains in the United States is unclear. Knowledge of the incidence and trends of ESBL infections will aid in directing appropriate empiric therapy for patients suspected of having ESBL-E. coli or ESBL-Klebsiella infections and will promote prevention efforts in places with high incidence of these resistant organisms. We performed a systematic literature review to identify the incidence and trends of ESBL-E. coli or ESBL-Klebsiella infections in the United States from 2000 to 2015.
METHODS
Search Strategy
MEDLINE via Ovid, Cochrane Library Databases via Wiley, Cumulative Index to Nursing and Allied Health Literature via the Elton Bryson Stephens Company (EBSCO), Scopus, National Health Service Economic Evaluation Database and Web of Science were searched by a health sciences librarian to identify papers published from January 2000 to July 2014 for articles pertaining to the incidence of ESBL-E. coli or ESBL-Klebsiella infections. The searches were rerun in August of 2015 to identify studies published in the interim. The librarian used terms associated with ESBL; created a filter for incidence, prevalence, and mortality using a pre-established incidence/prevalence search filter (http://www.systematicreviewsjournal.com/content/2/1/68); and used a pre-established filter to identify economic studies (http://www.cadth.ca/media/pdf/H0490_Search_Filters_for_Economic_Evaluations_mg_e.pdf). A list of all search terms is provided in Online Supplemental Appendix 1. Additionally, reference lists from retrieved articles and reviews were examined for relevant studies that were not identified in the initial literature searches.
Inclusion and Exclusion Criteria
Included studies provided information on the incidence of ESBL-E. coli or ESBL-Klebsiella infections published between January 2000 and August 2015 and included the year 2000 or later within their study period. Only multicenter studies (≥2 centers) were included to accurately determine the incidence of ESBL infections within the US population because single-center studies were more likely to represent outbreaks. Studies that provided the number of patients with an ESBL infection and a denominator that included both infected and uninfected patients such as patient days, device days, number of admissions, or number of discharges were included. Studies that provided denominators that only included patients who were infected with non-ESBL organisms were excluded. Additionally, articles were excluded that were not written in English, contained duplicated data, or pertained to ESBL organisms from food, animals, or the environment. Institutional review board approval was not required for this systematic literature review.
Data Extraction and Quality Assessment
Titles, abstracts, and manuscripts were screened by 1 investigator (J.M. or V.C.) for inclusion in the systematic literature review. Moreover, 2 investigators independently abstracted data for included studies. Disagreements were reexamined by both investigators and were resolved by consensus. Reviewers abstracted data on study design, study population, year, number of patients, description of denominator, definition of ESBL, organism examined, inclusion and exclusion criteria, and an assessment of the potential risk of bias for cohort studies using the Newcastle Ottawa score (Online Supplemental Appendix 2).Reference Stang 16
RESULTS
Article Review
Among the 51,419 studies identified for possible inclusion in the systematic review, 605 were further assessed through manuscript review. Most studies did not meet the inclusion criteria based on the title or abstract of the study. Finally, 9 studies were included, and 596 studies were excluded (Figure 1).
FIGURE 1 Flow diagram of searched studies.
Table 1 provides a summary of the 9 included studies.Reference Freeman, Sexton and Anderson 17 – Reference Swami, Liesinger, Shah, Baddour and Banerjee 25 Among the 9 studies, 7 were observational studies and 2 were interventional studies. The interventions included screening urine cultures with E. coli or Klebsiella for ESBLReference Han, Bilker and Nachamkin 18 and restricting the use of ceftazidime and ceftriaxone.Reference Lipworth, Hyle and Fishman 22 Furthermore, 6 studies evaluated the incidence of ESBL-producing E. coli and Klebsiella among hospital-onset infections, and 2 studies evaluated the incidence among community-onset infections.
TABLE 1 Studies That Reported the Incidence of Extended-Spectrum β-Lactamase (ESBL) Infections
NOTE. KLEB, Klebsiella species; CLABSI, central line-associated bloodstream infections; CAUTI, catheter-associated urinary tract infection; VAP, ventilator-associated pneumonia.
Incidence of ESBL-Producing E. coli and Klebsiella Infections Among All Patients
Incidence rates varied by year, population, and denominator (ie, patient days, device days, discharges, admissions, infants, and person years). Overall, from 1997 to 2011, incidence rates ranged from 0 to 51 cases per 10,000 admissions. During 1999–2006, incidence rates increased from 8.55 to 16.64 cases for ESBL-Klebsiella infections per 10,000 discharges (P<0.05) among patients admitted to 15 hospitals in Brooklyn, New York.Reference Landman, Bratu and Kochar 20 Throughout 2005–2009, the incidence rate of ESBL-E. coli infections increased from 5.82 to 21.53 cases per 100,000 person years within Olmsted County, Minnesota.Reference Swami, Liesinger, Shah, Baddour and Banerjee 25 From 2005 to 2008, incidence rates significantly increased among patients with ESBL-E. coli infections (range, 0–1.203 cases per 10,000 patient days; P<.001) but not for ESBL-Klebsiella infections (range, 0.837–1.011 per 10,000 patient days; P=0.785) among patients admitted to 16 community hospitals in North Carolina.Reference Freeman, Sexton and Anderson 17 Among 3 hospitals within the University of Pennsylvania Health System, a statistically significant increase in incidence rates were observed among patients with ESBL-E. coli or ESBL-Klebsiella infections after implementing the screening of urine cultures with E. coli or Klebsiella for ESBL (range during 2005–2009, 1.42–2.16 cases per 10,000 patient days preintervention and postintervention periods; P=.006).Reference Han, Bilker and Nachamkin 18
Conversely, from 1997 to 2002, Lipworth et alReference Lipworth, Hyle and Fishman 22 reported a decrease in the incidence rate of ESBL-E. coli and ESBL-Klebsiella infections in 2 university-affiliated hospitals (Hospital A, 4.2–2.3 per 10,000 patient days; Hospital B, 4.5–3.5 per 10,000 patient days).Reference Lipworth, Hyle and Fishman 22 However, the studies by Han et alReference Han, Bilker and Nachamkin 18 and Lipworth et alReference Lipworth, Hyle and Fishman 22 focused on interventions targeting ESBL infections as the outcome. Lipworth et al implemented an antimicrobial formulary, whereas Han et al implemented a hospitalwide urine culture screening protocol.
Incidence of ESBL-Producing E. coli and Klebsiella Infections by Age
Of the 9 studies reviewed, 2 focused on infections in neonates,Reference Larson, Cimiotti and Haas 21 , Reference Stoll, Hansen and Higgins 24 whereas the other 7 studies did not target a specific age group. These studies found that the incidence of ESBL-E. coli infections ranged from 0 to 1.67 per 10,000 infants (Table 1).Reference Larson, Cimiotti and Haas 21 , Reference Stoll, Hansen and Higgins 24 Larson et alReference Larson, Cimiotti and Haas 21 reported zero ESBL-K. oxytoca infections and an incidence of ESBL-K. pneumoniae infections of 51.11 per 10,000 admissions.
Swami et alReference Swami, Liesinger, Shah, Baddour and Banerjee 25 stratified the incidence rates by age for 2005–2009 and found statistically significant trends in incidence over time for patients with ESBL-E. coli infections in the 18–64-year group (range, 7–15 cases per 100,000 person years; P=.03) and the 65–79-year group (range, 28–73 cases per 100,000 person years; P=.02), but they did not find a significant increase in the 0–17-year group (range, 3–8 cases per 100,000 person years; P=.43) or the >80-year group (range, 84–91 cases per 100,000 person years; P=.97).Reference Swami, Liesinger, Shah, Baddour and Banerjee 25
Community-Onset Versus Hospital-Onset ESBL Infections
Definitions for community-onset infections and hospital-onset infections varied by study (Table 2). Of the 6 studies that defined community-onset and hospital-onset infections, 1 study defined hospital-onset infection as an infection occurring >72 hours after admission, while 3 other studies defined this as infections occurring >48 hours after admission. In addition, 2 studies used the National Healthcare Safety Network definitions for device-associated infections (≥3 days after admission). Furthermore, 2 studies differentiated between healthcare-associated infections and hospital-onset infections, while the other 4 studies did not.
TABLE 2 Definitions of Community-Onset Infections and Hospital-Onset Infections
NOTE. N/A, not available; LTCF, long-term-care facility.
Among device-related hospital-onset infections reported to the National Healthcare Safety Network in 2006–2007, the incidence rates of extended-spectrum cephalosporin-resistant infections were 0.064 cases per 1,000 device days for ventilator-associated pneumonia, 0.061 cases per 1,000 device days for catheter-associated urinary tract infections, and 0.051 cases per 1,000 device days for catheter-associated bloodstream infections.Reference Hidron, Edwards and Patel 19 Soe et alReference Soe, Edwards, Sievert, Ricks, Magill and Fridkin 23 used National Healthcare Safety Network data from 2011 to report incidence rates of central-line–associated bloodstream infections from intensive care units within 25 US states. The incidence by state ranged from 0 to 0.37 infections per 1,000 central-line days.Reference Soe, Edwards, Sievert, Ricks, Magill and Fridkin 23 Among patients from community hospitals in North Carolina, most of both the ESBL-E. coli (82.7%) and ESBL-Klebsiella (71.1%) infections were community-onset infections, and approximately half of the patients in each group were nursing home residents (ESBL-E. coli, 51.9%; ESBL-Klebsiella, 50.0%).Reference Freeman, Sexton and Anderson 17 From 2005 to 2009, Han et alReference Han, Bilker and Nachamkin 18 did not observe a significant difference in hospital-onset ESBL-E. coli and ESBL-Klebsiella infections (1.08–1.23 cases per 10,000 patient days; P=.51), even though this study focused on an intervention to reduce hospital-onset infection rates, perhaps the increased rate was due to the intense screening of the urine samples.Reference Han, Bilker and Nachamkin 18 Similarly, during that same period, Swami et alReference Swami, Liesinger, Shah, Baddour and Banerjee 25 did not observe a significant difference in either hospital-onset (range, 0–1 per 100,00 person-years; P=.98) or hospital-associated (range, 4–7 per 100,00 person years; P=.14) ESBL-E. coli bacteriuria rates. However, a significant difference was observed for community-onset ESBL infections (range, 6–14 per 100,000 person years; P=.01).Reference Swami, Liesinger, Shah, Baddour and Banerjee 25
DISCUSSION
This systematic literature review examined the incidence of ESBL-E. coli and ESBL-Klebsiella infections among the US population. Even with an increasing trend, incidence rates varied among the included studies. The variation is potentially due to differences in defining ESBL-producing organisms. Identifying true ESBL incidence rates can be challenging because laboratories need to specifically test for ESBLs, which may not occur on all E. coli and Klebsiella isolates and may vary throughout the study period. Second, inconsistencies existed among the patient populations included in the studies. Some of the studies examined neonates, while others included older adults. Furthermore, the populations included were from various regions throughout the United States. Castanheira et alReference Castanheira, Farrell, Krause, Jones and Sader 26 also found differences in rates between US census regions when examining the proportion of ESBL-E. coli or ESBL-Klebsiella pneumonia over all E. coli or Klebsiella pneumonia isolates.Reference Castanheira, Farrell, Krause, Jones and Sader 26 For E. coli, ESBL rates ranged from 5.2% to 20.1%, and for Klebsiella pneumonia, rates ranged from 3.2% to 34.7%, with the lowest rates reported in the West North Central region and the highest rates in the Mid-Atlantic region.Reference Castanheira, Farrell, Krause, Jones and Sader 26
Many articles were excluded from our systematic literature review because they did not report an appropriate denominator to calculate incidence rates. Instead, these studies reported the proportion of isolates that were identified as ESBL-producing among the total number of isolates for that organism (eg, resistance rates). Thus, without uninfected population denominators, increases in ESBL might be due to increases in ESBL producers, all strains of that organism or both. Therefore, only estimates with population-level denominators that include uninfected patients accurately report incidence rates.
In general, ESBL infections have increased from 1997 to 2011, with a slightly higher incidence rate for ESBL-Klebsiella infections compared with ESBL-E. coli infections. Other studies not included in this systematic literature review also reported increased ESBL infections when examining the proportion E. coli and Klebsiella that are ESBL producing. The SENTRY Antimicrobial Surveillance Program reported the ESBL phenotype in 7% of the 2,768 Klebsiella isolates tested in 1997–2000 from 30 US hospitals.Reference Jones, Biedenbach and Gales 12 In 2011–2013, the proportion of Klebsiella isolates with an ESBL phenotype increased to 15% (840 of 5,580) among isolates collected from 79 US hospitals.Reference Castanheira, Mills, Costello, Jones and Sader 13 Among E. coli isolates, 2 national surveillance programs (SENTRY Antimicrobial Surveillance Program and The Surveillance Network) reported a proportion of ESBL phenotype isolates around 1%–8% in 1997–2000, while in 2011–2013, the International Network for Optimal Resistance Monitoring program reported an increased proportion of 12% among isolates collected from US hospitals.Reference Castanheira, Mills, Costello, Jones and Sader 13 – Reference Karlowsky, Jones, Mayfield, Thornsberry and Sahm 15
Studies included in our analysis reported statistically significant increasing incidence rates for community-onset ESBL infections, while this was not frequently observed among hospital-onset ESBL infections.Reference Han, Bilker and Nachamkin 18 , Reference Swami, Liesinger, Shah, Baddour and Banerjee 25 A recent systematic review of the prevalence of ESBL fecal colonization among healthy individuals reported 14% colonization worldwide, with a 2% rate in the Americas.Reference Karanika, Karantanos, Arvanitis, Grigoras and Mylonakis 27 Therefore, ESBL infections may be increasing more rapidly within the community setting. However, the definitions of community-onset infections and hospital-onset infections differed throughout the studies, making direct comparisons difficult.
This study had several limitations. First, the quality of any systematic literature review is reflected by the conditions of the studies included. Second, publication bias may be present because studies that reported an increase in ESBL infection rates may be more likely to be published compared with studies that found no change. In addition, we only included studies from the United States; therefore, our results may not be generalizable to other countries.
In conclusion, the incidence rates of ESBL-E. coli and ESBL-Klebsiella infections are increasing throughout the United States. These findings may inform empiric antimicrobial treatment, and tracking the emergence of these infections is critically important for guiding antimicrobial stewardship and infection control programs. ESBL-containing bacteria are important targets for emerging state and national surveillance systems and multicenter or nationwide studies are needed to estimate current incidence rates. Without proper estimates for the burden of ESBL infections and other multidrug-resistant pathogens, it remains difficult to appropriately target research and prevention efforts.
ACKNOWLEDGMENTS
Financial support. The study was funded by Centers for Disease Control and Prevention, Safe Healthcare, Epidemiology, and Prevention Research Development (SHEPheRD) Program (contract 200-2011-42039 to M.S. and E.N.P.), by the VA Health Services Research and Development (HSR&D) Career Development Awards (award 11-215 to M.L.S. and award CDA 11-210 to R.N.), and by the CADRE Centers of Innovation (COIN) (grant nos. I50HX001240 and I50HX001242).
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2017.156
