Extended-spectrum β-lactamases (ESBLs) are enzymes produced by gram-negative bacilli that have the ability to inactivate β-lactam antibiotics containing an oxyimino group (i.e., third-generation cephalosporins and aztreonam). They are known as “extended-spectrum” because they are able to hydrolyze a broader spectrum of β-lactam antibiotics than the simple parent β-lactamases from which they are derived. ESBLs are plasmid-mediated β-lactamases, most commonly found in Klebsiella pneumoniae (but increasingly in Escherichia coli, Proteus mirabilis, and other gram-negative bacilli). They are structurally quite different from the inducible chromosomal β-lactamases produced by organisms such as Enterobacter cloacae, Citrobacter species, Serratia marcescens, and Pseudomonas aeruginosa, which also result in inactivation of third-generation cephalosporins.
ESBLs are a product of the use of third-generation cephalosporins; they were unknown before the introduction of these antibiotics in the early 1980s. In parallel with the widespread use of these antibiotics throughout the world has been the detection of ESBLs in every inhabited continent. Despite their wide prevalence, there is widespread ignorance regarding ESBL-producing organisms in many countries, including the United States.
Much of this lack of awareness of clinicians stems from inadequate reporting of the presence of ESBLs by clinical microbiology laboratories. Although MICs of third-generation cephalosporins for ESBL-producing organisms are many times higher than those for non-ESBL producers of the same species, these MICs may not reach breakpoint values for resistance. However, treatment failures were observed for these cephalosporins when they were used against ESBL-producing organisms, despite apparent in vitro susceptibility [1–5]. The observation that MICs of cephalosporins may increase 4- to 100-fold as bacterial inoculum increases and that animal studies showing failure of cephalosporins in the treatment of infections due to ESBL-producing organisms for which cephalosporin MICs are modest further support the practice of reporting all ESBL-producing isolates as resistant to cephalosporins.
In a recent survey in Europe, 37% of ESBL-producing organisms were misreported as susceptible to third-generation cephalosporins . Tenover et al.  recently sent 3 ESBL-producing organisms and 1 AmpC-producing organism to 38 microbiology laboratories in Connecticut. Only 7 laboratories (18%) correctly categorized all 4 isolates as potential ESBL producers and reported all results of cephalosporin susceptibility testing as “resistant,” as suggested by guidelines of the National Committee for Clinical Laboratory Standards (NCCLS) .
In part, the lack of awareness regarding ESBL production by gram-negative bacilli, compared with that regarding vancomycin resistance in enterococci, for example, exists because there is no simple marker of its presence. Some investigators have suggested ceftazidime resistance as a suitable marker . In North America, Europe, and Australia, most (but not all) ESBL-producing organisms are resistant to ceftazidime. However, <50% of Klebsiella isolates in the United States have their susceptibility to ceftazidime tested . In South America and eastern Europe, where different ESBLs exist, ESBL-producing organisms often appear susceptible to ceftazidime but resistant to cefotaxime and ceftriaxone. In a study from Boston, 31% of ESBL-producing organisms also appeared susceptible to ceftazidime . Hence, relying on ceftazidime resistance as the marker for presence of ESBLs can be unreliable.
To increase the correct reporting of ESBL production, the NCCLS published new guidelines for screening and confirmatory testing for ESBL production by K. pneumoniae, Klebsiella oxytoca, and Escherichia coli in January 1999 . These guidelines call for screening with a variety of extended-spectrum cephalosporin antibiotics. For laboratories using the disk diffusion test as their primary method of antibiotic susceptibility testing, revised diameters of zones of inhibition for cefpo-doxime, ceftazidime, aztreonam, cefotaxime, and ceftriaxone have been reported, which can be used as a screening tool for ESBL production. For laboratories using broth dilution methods, ESBL production should be suspected when MICs of cef-podoxime, ceftazidime, aztreonam, cefotaxime, or ceftriaxone are >1 µg/mL. (Traditionally, susceptibility to the extended-spectrum cephalosporins and aztreonam has been assumed when MICs are ≤8 µg/mL.) Testing to confirm that the in vitro effect of ceftazidime or cefotaxime is enhanced by the presence of clavulanic acid is regarded as phenotypic confirmation of the presence of ESBLs.
Current NCCLS recommendations are that clinical microbiology laboratories should report all ESBL-producing strains as resistant to all penicillins, cephalosporins, and aztreonam . We would go further and suggest that clinical microbiology laboratories should explicitly report to the clinician that ESBLs are present. The presence of an ESBL has important implications for antibiotic therapy for patients with serious infections due to ESBL-producing organisms . As Lucet et al.  point out in this issue of Clinical Infectious Diseases, the presence of an ESBL also has major infection control implications. The excuse from microbiologists that ESBL production should not be reported because “clinicians do not know what ESBLs are” should be approached by education and leadership from infectious disease physicians.
Numerous outbreaks of infection with ESBL-producing organisms have been reported worldwide; in at least 30 of these outbreaks, molecular methods were used to determine whether genotypic relatedness existed among outbreak strains [3, 14–44]. In every 1 of these outbreaks, common strains were isolated from ≥2 patients. Although ESBL-producing organisms can survive in the hospital environment [31, 45], we know of only one outbreak in which a removable environmental focus was found . More importantly, several investigators isolated ESBL-producing K. pneumoniae from the hands of medical and nursing staff [31, 44, 45]. Without doubt, patient-to-patient transmission of ESBL-producing organisms primarily occurs via the hands of hospital staff.
Unfortunately, in many hospitals, outbreaks have now been replaced by endemicity of varying levels of intensity. It is in this situation that Lucet et al. attempted to reduce transmission of ESBL-producing organisms. Such organisms had been present in their hospital since 1985. The molecular epidemiology of colonization and infection with ESBL-producing organisms with established endemicity can be complex, with multiple different genotypes coexisting. However, patient-to-patient transmission of organisms still occurs.
Lucet et al. approached the control of endemic ESBL-producing organisms by a program aimed at reducing patient-to-patient transmission. A strength of their study was that they documented that merely reinforcing the importance of hand washing was not sufficient to reduce nosocomial acquisitions of the organisms. Instead they critically evaluated situations in which the likelihood of patient-to-patient transmission of ESBL-producing organisms might occur. For example, they reviewed nursing work practices to minimize the risk that “breaks in the continuity of care” (such as nurses answering telephone calls or leaving a patient's room in the middle of a wound dressing to get extra supplies) would lead to infection control breakdowns. They realized that transfer of patients from intensive care units (ICUs) represented an enormous opportunity for transmission of ESBL-producing organisms to other parts of the hospital. Finally, they understood that uninformed health care staff visiting an ICU (e.g., surgeons or radiology technicians) represented even more of an infection control problem in the ICU than did regular ICU nursing staff. With this approach, Lucet et al. were able to reduce the isolation of not only ESBL-producing organisms but also methicillin-resistant Staphylococcus aureus (MRSA) and Acinetobacter baumannii.
It is surprising that no effort was made to restrict use of antibiotics in their hospital during the study period. Indeed, use of third-generation cephalosporins increased by 60% during the study period. Given the close relationship between release of extended-spectrum cephalosporins and emergence of ESBLs and the frequency with which patients colonized with ESBL-producing organisms recently received treatment with such antibiotics, this approach would have seemed logical. Other investigators tackled the problem of widespread occurrence of ESBL-producing organisms at their institution by reducing cephalosporin use, either through formal restriction of availability  or by education and increasing availability of alternatives . A reduction in isolation of ESBL-producing organisms occurred when imipenem , piperacillin/tazobactam , ticarcillin/clavulanate , or cefepime  replaced third-generation cephalosporins as “workhorse” antibiotics for widespread empirical use.
Unfortunately, as has been widely publicized, 1 antibiotic resistance problem may be substituted for another by replacement of antibiotic classes for heavy empirical use [26, 47]. Even more importantly, effects of these practices on mortality or length of stay in the ICU have not been measured.
ESBL-producing organisms do exist in major medical centers worldwide, regardless of the fact that microbiologists may not be reporting their presence. We suggest a strategy for the control of ESBL-producing organisms that was adapted from approaches used to control other nosocomial pathogens (table 1).
ESBL-producing organisms may become the gram-negative equivalent of vancomycin-resistant enterococci and MRSA. The advent of MRSA made obsolete the valuable workhorse antibiotics for treatment of suspected staphylococcal infections, nafcillin and cefazolin. ESBLs threaten to make obsolete the entire group of intravenously administered third-generation cephalosporins. Coinciding with the emergence of glycopeptide-intermediate S. aureus, pan-resistant ESBL-producing K. pneumoniae has now been reported [48, 49]. There seems to be no simple solution to these developments: the problem is that overprescription of antibiotics remains rampant [50, 51]. In addition to infection control measures, new and innovative approaches to limiting empirical usage of antibiotics may be the answer .