ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Dec. 2004, p. 4574–4581
0066-4804/04/$08.00?0 DOI: 10.1128/AAC.48.12.4574–4581.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Vol. 48, No. 12
Bloodstream Infections Due to Extended-Spectrum
?-Lactamase-Producing Escherichia coli and Klebsiella pneumoniae:
Risk Factors for Mortality and Treatment Outcome, with Special
Emphasis on Antimicrobial Therapy
Cheol-In Kang,1Sung-Han Kim,1Wan Beom Park,1Ki-Deok Lee,1Hong-Bin Kim,1
Eui-Chong Kim,2,3Myoung-Don Oh,1,3* and Kang-Won Choe1,3
Departments of Internal Medicine1and Laboratory Medicine,2Seoul National University College of Medicine, and
Clinical Research Institute, Seoul National University Hospital,3Seoul, Republic of Korea
Received 13 January 2004/Returned for modification 24 April 2004/Accepted 29 August 2004
This study was conducted to evaluate risk factors for mortality and treatment outcome of bloodstream
infections due to extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli and Klebsiella pneu-
moniae (ESBL-EK). ESBL production in stored K. pneumoniae and E. coli blood isolates from Jan 1998 to Dec
2002 was phenotypically determined according to NCCLS guidelines and/or the double-disk synergy test. A
total of 133 patients with ESBL-EK bacteremia, including 66 patients with ESBL-producing K. pneumoniae and
67 with ESBL-producing E. coli, were enrolled. The overall 30-day mortality rate was 25.6% (34 of 133).
Independent risk factors for mortality were severe sepsis, peritonitis, neutropenia, increasing Acute Physiology
and Chronic Health Evaluation II score, and administration of broad-spectrum cephalosporin as definitive
antimicrobial therapy (P < 0.05 for each of these risk factors). In 117 of the 133 patients, excluding 16 patients
who died within 3 days after blood culture sample acquisition, the 30-day mortality rates according to definitive
antibiotics were as follows: carbapenem, 12.9% (8 of 62); ciprofloxacin, 10.3% (3 of 29); and others, such as
cephalosporin or an aminoglycoside, 26.9% (7 of 26). When patients who received appropriate definitive
antibiotics, such as carbapenem or ciprofloxacin, were evaluated, mortality in patients receiving inappropriate
empirical antimicrobial therapy was found not to be significantly higher than mortality in those receiving
appropriate empirical antimicrobial therapy (18.9 versus 15.5%; P ? 0.666). Carbapenem and ciprofloxacin
were the most effective antibiotics in antimicrobial therapy for ESBL-EK bacteremia. A delay in appropriate
definitive antimicrobial therapy was not associated with higher mortality if antimicrobial therapy was adjusted
appropriately according to the susceptibility results. Our data suggest that more prudent use of carbapenem
as empirical antibiotic may be reasonable.
Escherichia coli and Klebsiella pneumoniae are major noso-
comial pathogens causing intra-abdominal infection, urinary
tract infection, and primary bacteremia (5). These organisms
had been uniformly susceptible to oxymino-?-lactam antibiot-
ics. During the past 2 decades, broad-spectrum cephalosporins,
including oxymino-?-lactam antibiotics, have been used world-
wide, and antibiotic-resistant strains that produce extended-
spectrum ?-lactamases (ESBL) have emerged among the En-
terobacteriaceae, predominantly in E. coli and K. pneumoniae
(3, 8). Since the initial description of ESBL production by K.
pneumoniae isolates in 1983 (14), strains that are resistant to
broad-spectrum cephalosporins are being increasingly recog-
nized (2, 7). The marked increase in the incidence of infections
due to ESBL-producing organisms in recent years is of great
concern (2, 7, 21, 23). Because ESBL-producing organisms are
frequently resistant to multiple antimicrobial agents, therapeu-
tic options for these infections are severely limited.
There have been many reports of outbreaks caused by these
organisms (10, 16, 19, 23, 24), and it has been demonstrated
that ESBL production by infecting organisms adversely affects
the clinical outcome (11, 12, 15, 26). At present, carbapenems
are recommended for the treatment of infections caused by
ESBL-producing organisms. This recommendation is based
primarily on the in vitro effect, the results of animal experi-
ments (25, 28), and limited clinical data (12, 22, 26, 32). How-
ever, there are few data on treatment outcomes of bloodstream
infections due to ESBL-producing organisms. We aimed to
evaluate treatment outcomes of bloodstream infections due to
ESBL-producing E. coli and K. pneumoniae (ESBL-EK) and
the factors associated with mortality. Special emphasis was
placed on determining the correlation between the antimicro-
bial therapy and the outcome.
(This study was presented in part at the 43rd Interscience
Conference on Antimicrobial Agents and Chemotherapy, Chi-
cago, Ill., September 2003 [C.I. Kang, S. H. Kim, D. M. Kim, et
al., Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother.,
abstr. K-720, 2003].)
MATERIALS AND METHODS
Patients. The database at our clinical microbiology laboratory (Seoul National
University Hospital, Seoul, Korea) was reviewed in order to identify patients with
E. coli and K. pneumoniae bacteremia. In the period from January 1998 to
December 2002, a total of 1,154 episodes of E. coli bacteremia and 578 episodes
of K. pneumoniae bacteremia, in 1,045 and 499 patients, respectively, were
identified. Only the first bacteremic episode of each patient was included in the
* Corresponding author. Mailing address: Department of Internal
Medicine, Seoul National University College of Medicine, 28 Yongon-
dong Chongno-gu, Seoul 110-744, Republic of Korea. Phone: 82-2-
760-2945. Fax: 82-2-762-9662. E-mail: address: email@example.com.
Bacterial strains. All of the blood isolates were collected by the clinical
microbiology laboratory in our hospital. Of the stored blood isolates from pa-
tients, 982 strains of E. coli and 471 strains of K. pneumoniae were successfully
recovered for inclusion in the study. Species identification was carried out with
VITEK-GNI cards (bioMe ´rieux, Hazelwood, Mo.) by standard methods (9).
Only one isolate from each bacteremic episode was included in the analysis. If
there was more than one isolate, we selected the more resistant isolate.
Microbiologic methods. The antibiotic susceptibility of each isolate was deter-
mined by the disk diffusion method, employing the criteria of the National
Committee for Clinical Laboratory Standards (NCCLS) (17). Mueller-Hinton
agar (Becton Dickinson, Sparks, Md.) was inoculated, and antibiotic disks (Bec-
ton Dickinson) were placed. The antibiotics included in the susceptibility test
were cefotaxime, ceftriaxone, ceftazidime, aztreonam, cefpodoxime, cefoxitin,
cefepime, amoxicillin-clavulanic acid, ciprofloxacin, amikacin, gentamicin, tobra-
mycin, and imipenem. For isolates from patients who had received extended-
spectrum cephalosporin, susceptibility testing for the actual agent used for the
treatment was conducted. MICs were determined by the broth microdilution
method, as described by NCCLS (18). ESBL production was screened and
determined by the disk diffusion method according to NCCLS performance
standards (17). In brief, we determined the diameters of the inhibition zones on
cefotaxime and ceftazidime disks (30 ?g each), alone and in combination with
clavulanic acid (10 ?g). An increase of ?5 mm in zone diameter when either of
the antimicrobial agents was combined with clavulanic acid was considered
evidence of ESBL production. Isolates that either were resistant to cefotaxime,
ceftazidime, or cefpodoxime; did not synergize with clavulanic acid; or gave a
?5-mm increase in zone diameter were subjected to the double-disk diffusion
test with cefotaxime, ceftazidime, and cefepime disks (31), as described by
Thomas and Sanders (29), except that the ceftazidime and amoxicillin-clavulanic
acid disks were placed 15 mm apart (15). After incubation, an enhanced zone of
inhibition between any one of the ?-lactam disks and the clavulanic acid disk was
interpreted as presumptive evidence for the presence of an ESBL. The produc-
tion of AmpC type ?-lactamase was suspected for isolates that were resistant to
either cefotaxime or ceftazidime, did not synergize with clavulanic acid, and were
resistant to both amoxicillin-clavulanic acid and cefoxitin (27). Organisms that
demonstrated AmpC type ?-lactamase production were classified as ESBL-
producing organisms. Two control organisms, E. coli ATCC 25922 and K. pneu-
moniae ATCC 700603, were inoculated in each set of tests for quality control.
Clinical analysis. We reviewed the medical records of the patients, and a
retrospective cohort study was conducted. The data collected included age, sex,
underlying disease, primary site of infection, severity of illness as calculated by
the Acute Physiology and Chronic Health Evaluation (APACHE) II score (13),
duration of hospital stay before onset of bacteremia, antimicrobial regimen, and
any antimicrobial therapy in the 30 days prior to onset of bacteremia. The
presence of the following comorbid conditions was also documented: neutrope-
nia, presentation with septic shock, care in intensive care unit, use of immuno-
suppressive agents within 30 days prior to onset of bacteremia, corticosteroid use,
postoperative state, and invasive procedures in the 72 h prior to onset of bacte-
remia. In addition, the presence of a central venous catheter, indwelling urinary
catheter, or mechanical ventilation was assessed. As this study was retrospective,
the patients’ physicians, not the researchers, had chosen the antimicrobial ther-
The main outcome measures used were initial response to treatment and
30-day mortality rate. The initial response to treatment was assessed at 72 h after
starting antimicrobial therapy and was classified as follows: complete response
for patients who had resolution of fever, leukocytosis, and all signs of infection;
partial response for patients who had abatement but not complete resolution of
these parameters; and treatment failure for patients who had either no abate-
ment or a deterioration in any of their clinical parameters or who died (15, 32).
The 30-day mortality rate was calculated as total number of deaths/total number
Definitions. Bacteremia was defined as the finding of an organism in a blood
culture specimen. Clinically significant bacteremia was defined as at least one
positive blood culture together with clinical features compatible with systemic
inflammatory response syndrome (1), and the patients with significant bactere-
mia were included in this study.
Nosocomial infection was defined as an infection that occurred ?48 h after
admission to the hospital, an infection that occurred ?48 h after admission to the
hospital in patients who had been hospitalized in the 2 weeks prior to admission,
or an infection that occurred ?48 h after admission to the hospital in patients
that had been transferred from another hospital or nursing home. Nosocomial
bloodstream infections, as well as other nosocomial infections, were defined
according to the criteria proposed by the Centers for Disease Control and
Prevention (6). Neutropenia was defined as an absolute neutrophil count of
The antimicrobial therapies were classified as empirical or definitive, with the
former being defined as the initial therapy before the results of blood culture
were available and the latter being defined as therapy after the results of anti-
biotic susceptibility tests had been received. Empirical antimicrobial therapy was
defined as treatment that included at least one antibiotic and that was started no
later than 24 h after the index positive blood sample for culture had been drawn.
Also, definitive antimicrobial therapy was defined as antimicrobial therapy that
was continued or commenced on the day that the antibiogram results were
reported to the clinicians and that was started no later than 120 h after the index
positive blood sample for culture had been drawn. The antimicrobial therapy was
considered appropriate if the treatment regimen included antibiotics active in
vitro and the dosage and route of administration were in conformity with current
medical standards. Cephalosporin monotherapy as definitive antimicrobial ther-
apy was considered inappropriate irrespective of the MIC. Cefotaxime, ceftriax-
TABLE 1. Demographic characteristics of study population
Age, mean yr ? SD (median, range).................................54 ? 15.18 (56, 16–87)
No. male:female....................................................................85:48 (63.9%:36.1%)
No. with K. pneumoniae:E. coli...........................................66:67 (49.6%:59.4%)
APACHE II score, mean ? SD (range)...........................10.26 ? 4.60 (0–24)
Hospital stay before bacteremia, mean days ? SD ........22 ? 33.87 (10, 0–230)
No. with hospital stay of ?2 weeks....................................61 (45.9%)
No. with nosocomial infection............................................105 (78.9%)
No. with neutropenia ...........................................................24 (18%)
No. with presentation with septic shock............................20 (22.6%)
No. with care in intensive care unit...................................11 (8.3%)
No. postsurgery.....................................................................10 (7.5%)
No. with indwelling urinary catheter..................................30 (22.6%)
No. with central venous catheterization............................36 (27.1%)
No. with invasive procedure within previous 72 h...........34 (25.6%)
No. with prior use of any antibiotics within......................95 (71.4%)
previous 30 days
No. of kinds of antibiotics administered ..........................2.03 ? 1.60 (2, 0–6)
within previous 30 days, mean ? SD
TABLE 2. Thirty-day mortality rates for bloodstream infections
caused by ESBL-producing K. pneumoniae and E. coli
Disease or infection site
No. of deaths/total (% mortality)
Overall 34/133 (25.6)21/66 (31.8)13/67 (19.4)
Solid organ transplantation
Bone marrow transplantation
End stage renal disease
Primary site of infection
11/33 (33.3)7/18 (38.9) 4/15 (26.7)
VOL. 48, 2004BLOODSTREAM INFECTIONS DUE TO ESBL PRODUCERS4575
one, ceftizoxime, and ceftazidime were defined as broad-spectrum cephalospo-
Statistical analysis. The Student t test was used to compare continuous vari-
ables, and the ?2or Fisher exact test was used to compare categorical variables.
In identifying the independent risk factors for mortality, a backward stepwise
logistic regression analysis was used to control for the effects of confounding
variables. Variables with a P value of ?0.05 in the univariate analysis were
candidates for multivariate analysis, as was the main variable of interest (i.e.,
cephalosporin therapy). The Kaplan-Meier method was used for survival analy-
sis. All P values were two-tailed, with a P value of ?0.05 considered statistically
significant. The SPSS (version 10.0) software package was used for these analy-
Demographic characteristics. A total of 1,154 episodes of E.
coli bacteremia and 578 episodes of K. pneumoniae bacteremia,
TABLE 3. Factors influencing mortality in bloodstream infections due to ESBL-producing K. pneumoniae and E. coli
No. of deaths/no. of episodes
OR (95% CI)P value
Definitive antimicrobial therapy
Empirical antimicrobial therapy
Origin of infection
Presentation with septic shock
Care in intensive care unit
6.16 (1.68–22.61) 0.006
Hospital stay, days
1.72 (0.78–3.77) 0.174
APACHE II score
4576KANG ET AL.ANTIMICROB. AGENTS CHEMOTHER.
in 1,045 and 499 patients, respectively, were identified. Of the
stored blood isolates from these patients, 982 strains of E. coli
and 471 strains of K. pneumoniae were successfully recovered
for inclusion in the study. Of these, 7.7% (76 of 982) of the E.
coli strains and 15.9% (75 of 471) of the K. pneumoniae strains
were ESBL-producing organisms. A total of 133 patients, in-
cluding 67 patients with E. coli bacteremia and 66 patients with
K. pneumoniae bacteremia, were analyzed, since the remaining
had unavailable medical records or had no significant bactere-
mia. Among the 67 patients with ESBL-producing E. coli bac-
teremia and 66 patients with ESBL-producing K. pneumoniae
bacteremia, three cases and four cases, respectively, were con-
sidered to be infections caused by AmpC type ?-lactamase
A total of 63.9% of the patients were males, and their me-
dian age was 54 (range, 16 to 87) years. Demographic data and
risk factors for infection are shown in Table 1. A total of 78.9%
of the episodes were nosocomial infections, and 71.4% of the
patients had received any antibiotics within previous 30 days
(Table 1). The most common underlying disease of the study
population was solid tumor (n ? 45; 33.8%), and the most
common primary site of infection was the pancreaticobiliary
tract (n ? 58; 43.6%).
Thirty-day mortality and risk factors for mortality. The
30-day mortality rate for all patients was 25.6% (34 of 133); it
was 31.8% (21 of 66) for patients with K. pneumoniae bacte-
remia and 19.4% (13 of 67) for patients with E. coli bactere-
mia. The mortality rates according to underlying diseases and
sites of infection are shown in Table 2. The mortality rate for
patients with peritonitis was 68.4% (13 of 19), and that for
patients with unknown primary site of infection was 33.3% (11
of 33) (Table 2). There were four patients with pneumonia and
five patients with liver abscess caused by K. pneumoniae,
whereas there were no patients with such infections caused by
E. coli. Of four patients with pneumonia due to ESBL-produc-
ing K. pneumoniae, two patients died. However, all patients
with liver abscess survived.
From univariate analysis, variables significantly associated
with mortality included the following: administration of a
broad-spectrum cephalosporin as definitive antimicrobial ther-
apy, neutropenia, presentation with septic shock, care in an
intensive care unit, peritonitis, immunosuppressive treatment,
prior corticosteroid use, and increasing APACHE II score
(Table 3). However, administration of a broad-spectrum ceph-
alosporin as empirical antimicrobial therapy was not associated
with a higher mortality (odds ration [OR] ? 1.10; 95% confi-
dence interval [95% CI] ? 0.42 to 2.87; P ? 0.842) (Table 3).
Multivariate analysis using a logistic regression model, which
included the variables associated with mortality by univariate
analysis (P ? 0.05), demonstrated that the independent risk
factors for 30-day mortality were as follows: administration of
a broad-spectrum cephalosporin as definitive antimicrobial
therapy, neutropenia, peritonitis, presentation with septic
shock, and increasing APACHE II score (Table 4).
Treatment outcome. To assess the initial clinical response at
72 h after empirical antimicrobial therapy, 127 patients who
had received broad-spectrum cephalosporins, fluoroquino-
lones, or carbapenem as empirical antimicrobial therapy were
included in the analysis. The remaining six patients were ex-
cluded because they had received piperacillin with or without
an aminoglycoside, and of these, five patients had experienced
treatment failure. When the clinical response at 72 h was
assessed, patients with bloodstream infections caused by
ESBL-producing K. pneumoniae had higher a treatment failure
rate than those with infections caused by E. coli (32 of 64 [50%]
versus 21 of 63 [33.3%]; P ? 0.057). A trend toward a higher
treatment failure rate was observed for cephalosporin regi-
mens compared with noncephalosporin regimens (47 of 104
[45%] versus 6 of 23 [26%]; P ? 0.093). This trend was more
prominent in patients with infection caused by E. coli than in
those with infection caused by K. pneumoniae (39 versus 14%
[P ? 0.086] and 51 versus 44% [P ? 0.719], respectively).
A further breakdown of treatment failure according to the
MICs of cephalosporins administered as empirical antimicro-
bial therapy is shown in Table 5. There was a trend toward a
significant increase in treatment failure rate as MICs rose (P ?
0.088). Eight patients were treated with an in vitro active
broad-spectrum cephalosporin, that is, one for which the MIC
was ?8 ?g/ml, and of these, two died. Six patients received a
broad-spectrum cephalosporin with an MIC of ?2 ?g/ml, and
of these, only one experienced treatment failure. The clinical
details for these patients are summarized in Table 6.
We also assessed outcomes for patients who had received a
cephalosporin to which the organism was apparently suscepti-
ble as definitive antimicrobial therapy. There were three cases
due to ESBL producers for which the MIC of the antibiotic
which was used to treat the infection was ?2 ?g/ml, and all of
these patients were cured. The MICs of cephalosporin were
?1 ?g/ml in two cases and 2 ?g/ml in one case. However, in
three cases due to ESBL producers for which the MIC of the
cephalosporin which was used to treat the infection was 16
TABLE 4. Independent risk factors for mortality in bloodstream
infections due to ESBL-producing K. pneumoniae and E. colia
Risk factor Adjusted OR (95% CI)P value
Administration of broad-spectrum
cephalosporin as definitive
Presentation with septic shock
Increasing APACHE II score (per
9.18 (1.55–54.51) 0.015
aMultivariate analysis with a logistic regression model was used.
TABLE 5. Outcome of empirical treatment with a cephalosporin in
bloodstream infections caused by ESBL-producing E. coli and
K. pneumoniae according to the MIC of the
MIC (?g/ml) Treatment failurea,b
?1 0/2 (0)
aData are number of cases/number of episodes (percent).
bTreatment failure at 72 h after empirical antimicrobial therapy.
cNC, no case identified.
VOL. 48, 2004BLOODSTREAM INFECTIONS DUE TO ESBL PRODUCERS4577
?g/ml, all of the patients experienced treatment failure and
To assess the 30-day mortality rates according to the defin-
itive antimicrobial therapy regimens, 117 patients who had
survived during more than 3 days after onset of bacteremia
were analyzed. The remaining 16 patients, who died within 3
days after blood culture sample acquisition, were excluded
from the analysis. The 30-day mortality rates were as follows:
carbapenem, 12.9% (8 of 62), ciprofloxacin, 10.3% (3 of 29);
and others, such as cephalosporin or an aminoglycoside, 26.9%
(7 of 26). Compared with others, the carbapenem and cipro-
floxacin group had lower mortality, although statistical signif-
icance was not reached (P ? 0.128 and 0.164, respectively).
The survival curve is shown in Fig. 1.
Among 91 patients who received carbapenem or ciprofloxa-
cin as definitive antimicrobial therapy, 24 (26.4%) received
combination therapy and the remaining 67 (73.6%) received
monotherapy. There was no significant difference in mortality
between the combination therapy group and monotherapy
group (6 of 24 [25%] versus 5 of 67 [7.5%]; P ? 0.061). None
of the patients had received cefepime or cephamycin.
Influence of delay in appropriate antimicrobial therapy on
mortality. The patients who had received a broad-spectrum
cephalosporin as definitive antimicrobial therapy had signifi-
cantly higher mortality than those received a noncephalosporin
(56.3 versus 15.8%; P ? 0.001), whereas administration of a
broad-spectrum cephalosporin as empirical antimicrobial ther-
apy was not associated with higher mortality (26 versus 24.1%;
P ? 0.842). To evaluate the influence of a delay in appropriate
antimicrobial therapy on the mortality, only the 95 patients
who had received appropriate definitive antibiotics, such as
carbapenem or ciprofloxacin, which are active in vitro against
the causative microorganisms, were analyzed. The remaining
38 were excluded because they had received inappropriate
definitive antibiotics, i.e., cephalosporins or aminoglycosides.
The mean duration (? standard deviation) of the delay in
appropriate antimicrobial therapy was 3.3 ? 1.8 days. Of the 95
patients that had received appropriate definitive antibiotics,
the 30-day mortality rates were 15.5% (9 of 58) for those
treated with an appropriate empirical antibiotic regimen and
18.9% (7 of 37) for those treated with an inappropriate regi-
men (P ? 0.666) (Table 7).
In the present study we found that the independent risk
factors for mortality in bloodstream infections due to
ESBL-EK were neutropenia, peritonitis, septic shock, higher
APACHE II score, and administration of a cephalosporin as
definitive antimicrobial therapy. In addition, we demonstrated
greater initial treatment failure in the cephalosporin treatment
Suboptimal clinical outcomes have been documented when
cephalosporins are used to treat serious infections due to
ESBL-producing organisms (10, 12, 22, 26, 32). Furthermore,
several investigators have concluded that initial treatment of
bloodstream infections caused by ESBL-producing strains with
noncarbapenem agents may be associated with higher mortal-
ity than treatment with a carbapenem agent (22, 26). However,
details of the treatment of most of the patients in those studies
were not provided (22, 26). On the basis of in vitro suscepti-
bility tests, coupled with a limited number of reports of clinical
outcome, carbapenem has emerged as the agent of choice for
the treatment of serious infections associated with ESBL-pro-
ducing organisms. In the present study, the outcome of ceph-
alosporin treatment of ESBL-EK bacteremia was also poor.
TABLE 6. Clinical outcomes for 8 patients with ESBL-producing K. pneumoniae and E. coli bacteremia treated empirically with broad-
spectrum cephalosporins to which the causative organisms were susceptible in vitro
Primary site of
Organism Antibiotic regimenb
17/male BMT, neutropenicUnknown K. pneumoniae ZOX, 2 g q8h8 Failure; continued fever after 7 days,
changed to imipenem and
ciprofloxacin but patient died on
22nd day of treatment
Failure; continued fever after 3 days,
changed to imipenem and
amikacin with cure
Failure; progression to renal abscess,
changed to imipenem but patient
died on 28thday of treatment
Cure; partial response to initial
antimicrobial therapy, changed to
ciprofloxacin with cure
Cure; partial response to initial
antimicrobial therapy, changed to
ciprofloxacin with cure
Cure; complete response to initial
Cure; complete response to initial
Cure; antimicrobial therapy with
UnknownE. coli ZOX, 2 g q8h; AMK,
7.5 mg/kg q12h
Urinary tractK. pneumoniaeCAZ, 1 g q8h; AMK,
7.5 mg/kg q12h
79/male Unknown E. coli CTX, 1 g q6h;
AMK, 7.5 mg/kg q12h
75/male CholangiocarcinomaCholangitisE. coli CTX, 1 g q6h2
69/male CBD stone CholangitisE. coliCTX, 1 g q6h2
65/female CBD stone CholangitisE. coliCTX, 1 g q6h; AMK,
7.5 mg/kg q12h
CTX, 1 g q6h; AMK,
7.5 mg/kg q12h
48/maleDM Liver abscessK. pneumoniae
aBMT, bone-marrow transplantation; ESRD, end stage renal disease; DM, diabetes mellitus; CBD, common bile duct.
bZOX, ceftizoxime; CTX, cefotaxime; CAZ, ceftazidime; AMK, amikacin.
4578 KANG ET AL.ANTIMICROB. AGENTS CHEMOTHER.
However, initial treatment with a cephalosporin as empirical
antimicrobial therapy was not associated with higher mortality.
In addition, a delay in appropriate treatment of ESBL-EK did
not have an adverse outcome, even in bloodstream infections.
Similarly, in a report by Lautenbach et al., a delay in effective
treatment for infection due to ESBL-producing organisms did
not result in poorer clinical outcomes (15). Thus, more prudent
use of a carbapenem as an initial empirical antibiotic may be
justified. However, the empirical use of antibiotics should be
based on the resistance background of the microorganisms and
the underlying diseases of the patients.
There was low mortality in patients with ESBL-EK bactere-
mia associated with pancreaticobiliary tract infection. This
could result from early intervention for biliary decompression
in the majority of the patients and might indicate that non-
medical interventions, such as decompression, are also impor-
tant in the treatment of biliary infection. However, patients
with peritonitis had significantly higher mortality, and perito-
nitis was one of the independent risk factors for death. Most of
our patients with peritonitis had advanced liver cirrhosis, with
spontaneous bacterial peritonitis, and might not have tolerated
ineffective initial antimicrobial therapy, due to their impaired
Schiappa et al. suggested that treatment with noncarbapenem
agents is associated with higher mortality than treatment with
a carbapenem agent (26). Wong-Beringer et al. also demon-
strated that ceftazidime treatment was associated with treat-
ment failure in all patients with bloodstream infections due to
ESBL-producing organisms (32). Currently, the NCCLS rec-
ommends that ESBL-producing strains should be considered
by microbiology laboratories to be resistant to all penicillins,
cephalosporins, and aztreonam (17). These criteria and other
expert recommendations may promote the use of carbapenems
for the treatment of serious infections caused by ESBL-producing
organisms, limiting clinical experience with the use of other
agents. Furthermore, increased empirical use of carbapenems in
response to outbreaks of ESBL-EK infections has been accom-
panied by the rapid emergence of carbapenem resistance in other
FIG. 1. Survival curve obtained by the Kaplan-Meier method for bloodstream infections due to ESBL-producing E. coli and K. pneumoniae
according to definitive antimicrobial therapy regimens. The 16 patients who died within 3 days after blood culture sample acquisition were excluded
from the analysis. The 30-day mortality rates were as follows: carbapenem, 12.9% (8 of 62), ciprofloxacin, 10.3% (3 of 29); others, 26.9% (7 of 26).
TABLE 7. Thirty-day mortality rates in patients with
ESBL-producing K. pneumoniae and E. coli bacteremia according to
the appropriateness of empirical antimicrobial therapya
aOf 133 patients with bloodstream infection due to ESBL-EK, 95 patients who
had received appropriate antimicrobial agents, such as a carbapenem or cipro-
floxacin, as definitive therapy were examined. The remaining 38 patients were
excluded because they had received inappropriate definitive antimicrobial ther-
bData are number of deaths/number of episodes (percent).
cThis group included patients who had received antimicrobial agents to which
the ESBL-EK isolate was susceptible in vitro as initial empirical antibiotics.
dThis group included patients who had not received any empirical antimicro-
bial agents or who had received antimicrobial agents to which the ESBL-EK
isolate was not susceptible in vitro as initial empirical antibiotics.
VOL. 48, 2004 BLOODSTREAM INFECTIONS DUE TO ESBL PRODUCERS 4579
nosocomial pathogens (4, 30). Therefore, therapeutic options
other than carbapenems would be attractive. When ESBL-EK
isolates are susceptible to the fluoroquinolones, these agents have
been effective and may constitute an alternative antimicrobial
therapy (10, 20). However, data on treatment outcomes of infec-
tion due to ESBL-EK, especially on ciprofloxacin treatment, are
limited to case reports or descriptions of nosocomial outbreaks.
In the present study, the outcome of ciprofloxacin treatment was
similar to that of carbapenem treatment and was more favorable
than that of treatment with broad-spectrum cephalosporins.
In this study, outcomes for patients receiving cephalosporin
therapy were more favorable when the MICs for the infecting
organisms were ?2 ?g/ml than when they were 8 ?g/ml or
greater. In this study, there were six cases of bloodstream
infections due to ESBL producers for which the MIC of the
antibiotic which was used to treat the infection was ?2 ?g/ml.
Death occurred in only one of these patients, and two patients
required a change in therapy to effect a cure. The remaining
three patients experienced a complete response and cure.
However, the number of patients in this category is too small
to accurately state whether cephalosporins can be reliably used
in this setting.
It is unlikely that randomized controlled trials of therapy for
infections due to ESBL-EK will be performed in the near
future. Therefore, further clinical experience with antibiotics
other than carbapenem is warranted to fully evaluate the po-
tential usefulness of these antibiotics against serious infections
due to ESBL-producing organisms.
Although retrospective studies may be limited by the avail-
ability of medical records, we found that 90% of the charts
were complete and available for review. Another potential
limitation is that molecular epidemiologic analysis and charac-
terization of ESBL types was not carried out in our study.
However, there was no evidence of clonal spread, based on the
antimicrobial susceptibility patterns of the isolates (data not
shown). Also, we did not evaluate the relationship between the
type of ESBLs and treatment outcomes. This study included
three cases of infection by E. coli isolates and four cases of
infection by K. pneumoniae isolates with different mechanisms
of cephalosporin resistance, probable AmpC producers.
AmpC-producing E. coli and K. pneumoniae isolates are not
considered to be classic ESBL-producing organisms. However,
this study consisted mainly of clinical analysis regarding treat-
ment outcome, not regarding the epidemiology, laboratory de-
tection methods, or molecular characterization of ESBLs.
When data for patients infected by isolates that did not meet
the classic definition were removed, the conclusions were not
altered (data not shown). Thus, the results of this study may
not be confounded by the inclusion those isolates.
Previous studies have focused on the epidemiology, labora-
tory detection methods, and molecular characterization of
ESBL-producing organisms. However, data on treatment out-
comes are limited. In this study, we evaluated the clinical
outcome in 133 patients with bloodstream infections due to
ESBL-EK. To our knowledge, this is the largest clinical anal-
ysis so far of treatment outcome for bloodstream infections
caused by ESBL-EK.
In conclusion, independent risk factors for mortality were
severe sepsis, peritonitis, neutropenia, increasing APACHE II
score, and administration of a broad-spectrum cephalosporin
as definitive antimicrobial therapy. The outcome of cephalo-
sporin treatment for ESBL-EK bacteremia was poor. Cipro-
floxacin and carbapenems were the most effective antibiotics in
antimicrobial therapy for ESBL-EK bacteremia. However, a
delay in appropriate definitive antimicrobial therapy was not
associated with higher mortality if antimicrobial therapy was
adjusted appropriately according to the susceptibility results.
Our data suggest that ciprofloxacin may be an alternative an-
timicrobial therapy for ESBL-EK infections, even for blood-
stream infections, and that more prudent use of carbapenems
as initial empirical antibiotics may be reasonable.
1. American College of Chest Physicians/Society of Critical Care Medicine
Consensus Conference. 1992. Definitions for sepsis and organ failure and
guidelines for the use of innovative therapies in sepsis. Crit. Care Med.
2. Burwen, D. R., S. N. Banerjee, and R. P. Gaynes. 1994. The National
Nosocomial Infections Surveillance System. Ceftazidime resistance among
selected nosocomial gram-negative bacilli in the United States. J. Infect. Dis.
3. Bush, K. 2001. New ?-lactamases in gram-negative bacteria: diversity and
impact on the selection of antimicrobial therapy. Clin. Infect. Dis. 32:1085–
4. Corbella, X., A. Montero, M. Pujol, M. A. Dominguez, J. Ayats, M. J.
Argerich, F. Garrigosa, J. Ariza, and F. Gudiol. 2000. Emergence and rapid
spread of carbapenem resistance during a large and sustained hospital out-
break of multiresistant Acinetobacter baumannii. J. Clin. Microbiol. 38:4086–
5. Eisenstein, B. I., and D. F. Zaleznik. 2000. Enterobacteriaceae, p. 2294–2310.
In G. L. Mandell, J. E. Bennett, and R. Dolin (ed.), Principles and practice
of infectious diseases, 5th ed. Churchill Livingstone, Philadelphia, Pa.
6. Garner, J. S., W. R. Jarvis, T. G. Emori, T. C. Horan, and J. M. Hughes.
1988. CDC definitions for nosocomial infections. Am. J. Infect. Control
7. Itokazu, G. S., J. P. Quinn, C. Bell-Dixon, F. M. Kahan, and R. A. Weinstein.
1996. Antimicrobial resistance rates among aerobic gram-negative bacilli
recovered from patients in intensive care units: evaluation of a national
postmarketing surveillance program. Clin. Infect. Dis. 23:779–784.
8. Jacoby, G. A., and A. A. Medeiros. 1991. More extended-spectrum ?-lacta-
mases. Antimicrob. Agents Chemother. 35:1697–1704.
9. Jorgensen, J. H., J. D. Turnidge, and J. A. Washington. 1999. Antibacterial
susceptibility tests: dilution and disk diffusion methods, p. 1526–1543. In
P. R. Murray, E. J. Baron, M. A. Pfaller, F. C. Tenover, and R. H. Yolken
(ed.), Manual of clinical microbiology. 7th ed. American Society for Micro-
biology, Washington, D.C.
10. Karas, J. A., D. G. Pillay, D. Muckart, and A. W. Sturm. 1996. Treatment
failure due to extended-spectrum beta-lactamase. J. Antimicrob. Che-
11. Kim, B. N., J. H. Woo, M. N. Kim, J. Ryu, and Y. S. Kim. 2002. Clinical
implications of extended-spectrum beta-lactamase-producing Klebsiella
pneumoniae bacteremia. J. Hosp. Infect. 52:99–106.
12. Kim, Y. K., H. Pai, H. J. Lee, S. E. Park, E. H. Choi, J. Kim, J. H. Kim, and
E. C. Kim. 2002. Bloodstream infections by extended-spectrum ?-lactamase-
producing Escherichia coli and Klebsiella pneumoniae in children: epidemi-
ology and clinical outcome. Antimicrob. Agents Chemother. 46:1481–1491.
13. Knaus, W. A., E. A. Drapier, D. P. Wagner, and J. E. Zimmerman. 1985.
APACHE II: a severity of disease classification system. Crit. Care Med.
14. Knothe, H., P. Shah, V. Kremery, M. Antal, and S. Mitsuhashi. 1983. Trans-
ferable resistance to cefotaxime, cefoxitin, cefamandole and cefuroxime in
clinical isolates of Klebsiella pneumoniae and Serratia marcescens. Infection
15. Lautenbach, E., J. B. Patel, W. B. Bilker, P. H. Edelstein, and N. O. Fish-
man. 2001. Extended-spectrum ?-lactamase-producing Escherichia coli and
Klebsiella pneumoniae: risk factors for infection and impact of resistance on
outcomes. Clin. Infect. Dis. 32:1162–1171.
16. Meyer, K. S., C. Urban, J. A. Eagen, B. J. Berger, and J. J. Rahal. 1993.
Nosocomial outbreak of Klebsiella infection resistant to late-generation
cephalosporins. Ann. Intern. Med. 119:353–358.
17. National Committee for Clinical Laboratory Standards. 2000. Performance
standards for antimicrobial disk susceptibility tests. Approved standard M2-
A7, 7th ed. National Committee for Clinical Laboratory Standards, Wayne,
18. National Committee for Clinical Laboratory Standards. 2003. Methods for
dilution antimicrobial susceptibility tests for bacteria that grow aerobically;
Approved standard. NCCLS document M100-S13. National Committee for
Clinical Laboratory Standards. Wayne, Pa.
4580 KANG ET AL.ANTIMICROB. AGENTS CHEMOTHER.
19. Naumovski, L., J. P. Quinn, D. Miyashiro, M. Patel, K. Bush, S. B. Singer, Download full-text
D. Graves, T. Palzkill, and A. M. Arvin. 1992. Outbreak of ceftazidime
resistance due to a novel extended-spectrum beta-lactamase in isolates from
cancer patients. Antimicrob. Agents Chemother. 36:1991–1996.
20. Paterson, D. L. 2000. Recommendation for treatment of severe infections
caused by Enterobacteriaceae producing extended-spectrum beta-lactamases
(ESBLs). Clin. Microbiol. Infect. 6:460–463.
21. Paterson, D. L., W-C. Ko, A. V. Gottberg, S. Mohapatra, J. M. Casellas, H.
Goossens, L. Mulazimoglu, G. Trenholme, K. P. Klugman, R. A. Bonomo,
L. B. Rice, M. M. Wagener, J. G. McCormack, and V. L. Yu. 2004. Interna-
tional prospective study of Klebsiella pneumoniae bacteremia: implications of
extended-spectrum beta-lactamase production in nosocomial infections.
Ann. Intern. Med. 140:26–32.
22. Paterson, D. L., W.-C. Ko, A. V. Gottberg, J. M. Casellas, L. Mulazimoglu,
K. P. Klugman, R. A. Bonomo, L. B. Rice, J. G. McCormack, and V. L. Yu.
2001. Outcome of cephalosporin treatment for serious infections due to
apparently susceptible organisms producing extended-spectrum ?-lacta-
mases: implications for the clinical microbiology laboratory. J. Clin. Micro-
23. Pena, C., M. Pujol, C. Ardanuy, A. Ricart, R. Pallares, J. Linares, J. Ariza,
and F. Gudiol. 2001. An outbreak of hospital-acquired Klebsiella pneumoniae
bacteraemia, including strains producing extended-spectrum beta-lactamase.
J. Hosp. Infect. 47:53–59.
24. Pena, C., M. Pujol, C. Ardanuy, A. Ricart, R. Pallares, J. Linares, J. Ariza,
and F. Gudiol. 1998. Epidemiology and successful control of a large outbreak
due to Klebsiella pneumoniae producing extended-spectrum ?-lactamases.
Antimicrob. Agents Chemother. 42:53–58.
25. Rice, L. B., L. L. Carias, R. A. Bonomo, D. M. Shlaes. 1996. Molecular
genetics of resistance to both ceftazidime and beta-lactam-beta-lactamase
inhibitor combinations in Klebsiella pneumoniae and in vivo response to
beta-lactam therapy. J. Infect. Dis. 173:151–158.
26. Schiappa, D. A., M. K. Hayden, M., G. Matushek, F. N. Hashemi, J. Sullivan,
K. Y. Smith, D. Miyashiro, J. P. Quinn, R. A. Weinstein, and G. M. Tren-
holme. 1996. Ceftazidime-resistant Klebsiella pneumoniae and Escherichia
coli bloodstream infection: a case-control and molecular epidemiologic in-
vestigation. J. Infect. Dis. 174:529–536.
27. Steward, C. D., J. K. Rasheed, S. K. Hubert, J. W. Biddle, P. M. Raney, G. J.
Anderson, P. P. Williams, K. L. Brittain, A. Oliver, J. E. McGowan, and F. C.
Tenover. 2001. Characterization of clinical isolates of Klebsiella pneumoniae
from 19 laboratories using the National Committee for Clinical Laboratory
Standards extended-spectrum ?-lactamase detection methods. J. Clin. Mi-
28. Thauvin-Eliopoulos, C., M. F. Tripodi, R. C. Moellering, and G. M. Elio-
poulos. 1997. Efficacies of piperacillin-tazobactam and cefepime in rats with
experimental intra-abdominal abscesses due to an extended-spectrum beta-
lactamase-producing strain of Klebsiella pneumoniae. Antimicrob. Agents
29. Thomason, K. S., and C. C. Sanders. 1992. Detection of extended-spectrum
?-lactamases in members of the family Enterobacteriaceae: comparison of the
double-disk and three-dimensional tests. Antimicrob. Agents Chemother.
30. Troillet, N., M. H. Samore, and Y. Carmeli. 1997. Imipenem-resistant
Pseudomonas aeruginosa: risk factors and antibiotic susceptibility patterns.
Clin. Infect. Dis. 25:1094–1098.
31. Tzelepi, E., P. Giakkoupi, D. Sofianou, V. Loukova, A. Kemeroglou, and A.
Tsakris. 2000. Detection of extended-spectrum ?-lactamases in clinical iso-
lates of Enterobacter cloacae and Enterobacter aerogenes. J. Clin. Microbiol.
32. Wong-Beringer, A., J. Hindler, M. Loeloff, A. M. Queenan, N. Lee, D. A.
Pegues, J. P. Quinn, and K. Bush. 2002. Molecular correlation for the
treatment outcomes in bloodstream infections caused by Escherichia coli and
Klebsiella pneumoniae with reduced susceptibility to ceftazidime. Clin. Infect.
VOL. 48, 2004 BLOODSTREAM INFECTIONS DUE TO ESBL PRODUCERS4581