Risk Factors for Infection or Colonization with CTX-M Extended-
Spectrum-?-Lactamase-Positive Escherichia coli
Jennifer H. Han,aKei Kasahara,bPaul H. Edelstein,cWarren B. Bilker,d,eand Ebbing Lautenbacha,d,e
Division of Infectious Diseases, Department of Medicine,aDepartment of Pathology and Laboratory Medicine,cDepartment of Biostatistics and Epidemiology,dand
Center for Clinical Epidemiology and Biostatistics,eUniversity of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA, and Center for Infectious Diseases,
Nara Medical University, Kashihara, Japanb
Therehasbeenasignificantincreaseintheprevalenceof Enterobacteriaceae thatproduceCTX-M-typeextended-spectrum
richia coli. A case-control study was conducted within a university system from 1 January 2007 to 31 December 2008. All patients
with clinical cultures with E. coli demonstrating resistance to extended-spectrum cephalosporins were included. Case patients
were designated as those with cultures positive for CTX-M-positive E. coli, and control patients were designated as those with
non-CTX-M-producing E. coli. Multivariable logistic regression analyses were performed to evaluate risk factors for CTX-M-
positive isolates. A total of 83 (56.8%) of a total of 146 patients had cultures with CTX-M-positive E. coli. On multivariable anal-
yses, there was a significant association between infection or colonization with CTX-M-type ?-lactamase-positive E. coli and
an increased risk of colonization or infection with CTX-M-positive E. coli. Future studies will need to focus on outcomes associ-
ated with infections due to CTX-M-positive E. coli, as well as infection control strategies to limit the spread of these increasingly
ESBL-producing Enterobacteriaceae are associated with increased
morbidity, mortality, and health care costs (19, 42). In the past
decade, there has been a significant increase in the prevalence of
Given the increased mortality associated with delay in appro-
priate treatment for ESBL-associated infections (41), early recog-
ing E. coli is critical for selection of empirical antibiotic therapy
Risk factors such as severity of illness, instrumentation, and prior
teriaceae in general (4, 5, 19). However, despite emerging data
suggesting that the epidemiology of CTX-M-producing isolates is
are few published studies specifically evaluating risk factors for
CTX-M-producing Enterobacteriaceae (7, 20, 26, 37, 39, 46). Fur-
thermore, these have been limited by small sample sizes (20, 26,
37, 46) and restricted to select patient populations or types of
studies evaluating risk factors for CTX-M-producing E. coli in
the United States, where the epidemiologies of infections asso-
ciated with CTX-M-type ?-lactamases may be different due to
variation in antibiotic prescription and infection control prac-
tices. Finally, to our knowledge, our study is the first to inves-
tigate risk factors for CTX-M production in E. coli using a
?-lactamase (ESBL)-producing Gram-negative organisms
control group selected from non-CTX-M-producing Entero-
bacteriaceae demonstrating resistance, as opposed to suscepti-
bility, to extended-spectrum cephalosporins. Elucidating risk
other ESBLs) is critical, since prior work suggests that the epi-
demiologies of various resistance mechanisms among Entero-
bacteriaceae, including risk factors for isolation, may be differ-
ent (18). Therefore, we conducted the present study to evaluate
risk factors for infection or colonization with CTX-M-positive
E. coli, with the hypothesis that prior antibiotic use is a signif-
icant risk factor for isolation of CTX-M-positive E. coli.
MATERIALS AND METHODS
Study design and setting. This case-control study was conducted at two
hospitals in the University of Pennsylvania Health System (UPHS) in
Philadelphia: the Hospital of the University of Pennsylvania (HUP), a
725-bed academic tertiary care medical center, and Penn Presbyterian
Study population. All adult inpatients and outpatients with clinical
Received 30 May 2012 Returned for modification 17 July 2012
Accepted 7 August 2012
Published ahead of print 13 August 2012
Address correspondence to Jennifer H. Han, firstname.lastname@example.org.
Copyright © 2012, American Society for Microbiology. All Rights Reserved.
November 2012 Volume 56 Number 11 Antimicrobial Agents and Chemotherapyp. 5575–5580 aac.asm.org
laboratory, which processes all specimens obtained from patients at HUP
tant to ceftriaxone or ceftazidime were eligible for inclusion in the study.
coli meeting the above criteria identified during the study period.
Microbiological identification and susceptibility testing. Standard
standard methods (8, 9) using the Vitek2 semiautomated system or disk
diffusion testing. Confirmatory ESBL testing was performed using the
double-disk synergy test for nonurinary isolates, as well as for urinary
isolates that exhibited a ceftazidime or ceftriaxone MIC that was ?1
?g/ml but ?64 ?g/ml. Urinary isolates with a ceftazidime or ceftriaxone
MIC of ?64 ?g/ml, as well as those with carbapenem resistance as deter-
mined by a positive modified Hodge test (9), did not have double-disk
testing performed as these were assumed to be ESBL-producers. Finally,
the presence of the blaCTX-Mgene was detected by PCR as previously
reported (31). Therefore, cases and controls were defined solely on the
basis of CTX-M production, with case patients designated as those with
cultures positive for CTX-M-positive E. coli, and control patients desig-
nated as those who had non-CTX-M-producing, extended-spectrum
cephalosporin-resistant E. coli isolated. A previous study of the CTX-M-
positive E. coli isolated in 2007 showed that the isolates were not clonal
based on pulsed-field gel electrophoresis analysis and that multiple
CTX-M types were represented (25).
Data collection. Data were abstracted from the Pennsylvania Inte-
grated Clinical and Administrative Research Database (PICARD), which
includes demographic, laboratory, pharmacy, and billing information
resistance (2, 12, 21). The following data were collected for all subjects:
baseline demographics, inpatient or outpatient status in relation to the
culture date, origin at the time of hospital admission for inpatients (i.e.,
physician referral, transfer from another facility, or admission through
the Emergency Department), hospital location at the time of infection
30 days prior to the culture date, time of onset of nosocomial infection
date of admission or admission as a transfer from another institution),
inine of ?2.0 mg/dl or the requirement of dialysis), chronic liver disease
(i.e., cirrhosis or chronic hepatitis), chronic pulmonary disease (i.e.,
chronic obstructive pulmonary disease, asthma, or interstitial lung dis-
ease), congestive heart failure, solid organ or hematopoietic stem cell
transplantation, HIV infection, neutropenia (absolute neutrophil count
ticosteroids, in the prior 30 days. In addition, the Charlson comorbidity
tion prior to the culture date.
clinical culture date was documented. Antimicrobial therapy was catego-
rized by agent or class, including vancomycin, aminoglycosides, extended-
spectrum penicillins (e.g., piperacillin-tazobactam), antistaphylococcal
penicillins (e.g., nafcillin), other penicillins (e.g., ampicillin and am-
one, ceftazidime, and cefepime), other cephalosporins (e.g., cefazo-
lin), trimethoprim-sulfamethoxazole, fluoroquinolones, tetracyclines
(e.g., doxycycline), metronidazole, aztreonam, tigecycline, and dapto-
mycin (11, 23).
Statistical analysis. Cases and controls were characterized by poten-
tial risk factors, including demographic variables, comorbid conditions,
and prior antibiotic use. Continuous variables were compared using the
Student t test or Wilcoxon rank-sum test, and categorical variables were
compared using the ?2or Fisher exact test. Bivariable analyses were then
primary risk factor of interest. The odds ratio (OR) and 95% confidence
interval (CI) were calculated to evaluate the strength of any association.
Stratified analyses were conducted to elucidate where confounding
and interaction were likely to exist in multivariable analyses, using the
Mantel-Haenszel test for summary statistics (24). In particular, inpatient
versus outpatient status at the time of the culture date, as well as culture
site (e.g., bacteremia or urinary source), were a priori designated as po-
tential effect modifiers of the association of interest. Interaction was de-
termined to be present when the test for heterogeneity between the ORs
subsequently performed using multiple logistic regression (15), with cal-
culation of adjusted ORs with 95% CIs. A stepwise (forward-backward)
selection procedure was used, with variables with P values ?0.20 in
bivariable analyses or noted to be confounders in stratified analyses con-
if their inclusion resulted in a ?15% change in the effect measure for the
primary association of interest or if they were statistically significant on
likelihood ratio testing (27). For all calculations, a two-tailed P value of
?0.05 was considered significant.
All statistical calculations were performed using commercially avail-
able software (STATA version 11.0; StataCorp LP, College Station, TX).
Study population. A total of 146 unique patients with clinical
cultures with E. coli resistant to extended-spectrum cephalospo-
rins were identified during the 2 year study period. The mean age
of patients was 63 years (standard deviation [SD], 17.3), and 61
(41.8%) of them were male. Of the 146 patients, 81 (55.5%) were
(3.4%) were Hispanic, and the remainder were self-identified as
“other.” Furthermore, among all patients, 108 (74.0%) were hos-
pitalized at the time of the clinical culture, while 38 (26.0%) were
outpatients. Finally, the distribution of culture sources was as fol-
lows: 88 (60.3%) of the cultures were from urine, 29 (19.9%)
blood, 18 (12.3%) wound, and 11 (7.5%) respiratory tract.
Microbiological results. Of the 146 unique isolates, 83
which there were 63 (43.2%), were negative for CTX-M-type
?-lactamase. The distribution of CTX-M groups among the iso-
lates was as follows: 52 (62.6%) belonged to group I, 30 (36.1%)
belonged to group IV, and 1 (1.2%) belonged to group II. Two
unique isolates among the control group were Klebsiella pneu-
moniae carbapenemase-producing E. coli.
trol patients are shown in Table 1. Coresistance to antibiotics was
notable for a significant association between CTX-M positivity
and fluoroquinolone (P ? 0.001) and tobramycin (P ? 0.004)
Risk factors for CTX-M isolation. In bivariable analyses, sev-
eral variables were noted to be significantly associated with CTX-
M-type ?-lactamase positivity (Table 2), including hospitaliza-
tion at the time of the clinical culture (OR, 2.25; 95% CI, 1.06 to
duction, whereby bacteremia demonstrated an OR of 2.89 (95%
Han et al.
aac.asm.orgAntimicrobial Agents and Chemotherapy
an extended-spectrum penicillin (i.e., piperacillin-tazobactam)
was significantly associated with CTX-M positivity (OR, 9.05;
95% CI, 2.02 to 40.5; P ? 0.001).
On multivariable analyses of risk factors for infection or colo-
nization with CTX-M-type ?-lactamase-positive E. coli (Table 3),
there was no significant effect modification by inpatient status
(P ? 0.43), health care association (P ? 0.13), bacteremia (P ?
0.99), or urinary source (P ? 0.98). The unadjusted OR between
CTX-M-type ?-lactamase production was 1.62 (95% CI, 0.83 to
3.16; P ? 0.16). In multivariable analyses, receipt of an extended-
spectrum penicillin (i.e., piperacillin-tazobactam) in the 30 days
prior to the culture date was an independent risk factor for CTX-
M-type ?-lactamase production (OR, 7.36; 95% CI, 1.61 to 33.8;
P ? 0.01). After controlling for confounders, a urinary clinical
or colonization with CTX-M-positive E. coli (OR, 0.36; 95% CI,
0.17 to 0.77; P ? 0.008).
In this cohort study of 146 patients with both nosocomial and
community-acquired cultures with E. coli demonstrating resis-
tance to extended-spectrum cephalosporins, we found that 83
(56.8%) isolates were positive for CTX-M-type ESBLs. Further-
more, isolation of CTX-M positive E. coli was significantly associ-
ated with the recent use of an extended-spectrum penicillin and a
nonurinary source of infection or colonization.
ing previous antibiotic use (19, 30, 44). However, despite emerg-
ing data suggesting that the epidemiology of CTX-M-positive
isolates is different from those producing other types of ESBLs (4,
38), only a few studies to our knowledge have evaluated risk fac-
sample sizes (20, 26, 37, 46), lack of multivariable analyses (26),
and evaluation restricted to select populations such as hospital-
ized patients (7, 20, 26, 37) or specific types of infections such as
trol groups for comparison from non-CTX-M-positive Entero-
bacteriaceae with susceptibility to extended-spectrum cephalo-
to overestimation of the ORs for prior exposure to antibiotics,
particularly the antibiotic to which the organism associated with
infections in control but not case patients is susceptible (14). To
our knowledge, the present study is the first in the literature to
evaluate risk factors for CTX-M production in E. coli associated
with both the health care and community settings in the United
States and to utilize a control group comprised of patients with
cultures with E. coli demonstrating resistance to extended-spec-
A novel finding of our study is that isolation of CTX-M
positive E. coli from clinical specimens was associated with
prior use of an extended-spectrum penicillin in the 30 days
prior to the culture date. Prior antibiotic use has been well
described as a risk factor for infections due to ESBL-producing
Enterobacteriaceae (4). Among studies evaluating risk factors
specifically for CTX-M-positive isolates, recent receipt of ex-
tended-spectrum cephalosporins, fluoroquinolones, or a com-
bination of the above has been implicated (7, 37). In addition,
a study of 45 patients with cultures positive for CTX-M-pro-
ducing E. coli reported a significant association between the
prior use of antibiotics from the ?-lactam/?-lactamase inhib-
itor class and CTX-M production (26), but that study was lim-
ited to bivariable analyses and lacked a control group.
Why piperacillin-tazobactam use was a risk factor for CTX-M
identical piperacillin-tazobactam susceptibility frequencies be-
finding will be important in the implementation of antibiotic
stewardship measures, particularly in institutions with high rates
of extended-spectrum penicillin use. Ultimately, further work is
needed to elucidate outcomes associated with infections due to E.
coli with CTX-M-type ?-lactamases compared to other ESBLs to
help guide development of effective antimicrobial stewardship
and restriction policies.
One potential explanation for CTX-M infection or coloniza-
tion not involving antibiotic selection is that the use of extended-
spectrum penicillins, specifically piperacillin-tazobactam, may be
CTX-M-positive E. coli. Along these lines, on bivariable analyses,
case patients were more likely to be hospitalized (P ? 0.03) and
had a higher mean Charlson comorbidity score (P ? 0.01) at the
time of the culture date.
CTX-M production with nonurinary sources of colonization or
infection, and to our knowledge, this study is the first to demon-
urinary source, 0.37; 95% CI, 0.16 to 0.87; P ? 0.02). Although
CTX-M-producing Enterobacteriaceae have typically been associ-
ated with community onset infections (20, 32, 36), and most no-
tably urinary tract infections, the results of our study suggest that
pital setting and in particular in nonurinary infections such as
bacteremia. Differences in control group selection may have, in
part, explained the novel association with a nonurinary source
demonstrated in our study. Furthermore, studies suggest that
CTX-M-type ?-lactamases may be associated with virulence fac-
TABLE 1 Antibiotic susceptibility rates of E. coli resistant to extended-
spectrum cephalosporins among case and control patients
(n ? 83)
(n ? 63)
aIncludes intermediate and resistant isolates.
bSusceptibility testing was performed only for nonurinary isolates.
cSusceptibility information was unavailable for 21 patients (12 control patients, 9 case
dSusceptibility testing was performed only for urinary isolates.
Risk Factors for CTX-M
November 2012 Volume 56 Number 11 aac.asm.org 5577
tors different from those of other ESBL types (16, 33), and it is
opposed to low-inoculum infections such as urinary tract infec-
tions. However, further work is needed to elucidate the epidemi-
ology of CTX-M-positive Enterobacteriaceae as a cause of non-
community setting. Nevertheless, this finding has important im-
plications for empirical treatment of associated infections in the
hospital setting, as the antibiotic susceptibility patterns of CTX-
M-positive E. coli strains appear to differ significantly from those
of E. coli strains with production of other ESBL types (7, 20, 38).
As in previous studies (7, 20), we observed high rates of core-
sistance to fluoroquinolones in isolates from case as opposed to
control patients, with 90.4% of CTX-M-positive E. coli isolates
mycin, were relatively high and may reflect geographic variation.
active against E. coli with CTX-M-type ?-lactamases, and these
antibiotics should be considered first-line treatment for serious
infections due to these organisms.
TABLE 2 Characteristics of patients with clinical cultures with E. coli resistant to extended-spectrum cephalosporins
P OR (95% CI)b
With CTX-M (n ? 83) Without CTX-M (n ? 63)
Mean age in yrs (SD)
Emergency department admission
Physician referral on admission
Duration of hospitalization prior to the culture date, mean
Health care-associated infection
Prior admission to UPHS ? 30 days before the culture date
Mean white blood cell count ? 109/liter (SD)c
Central venous catheter
Transplant (solid organ or hematopoietic stem cell)
Chronic liver disease
Chronic pulmonary disease
Congestive heart failure
Mean Charlson comorbidity score (SD)
Receipt of any immunosuppression ? 30 days prior to the
Receipt of corticosteroids ?30 days prior to the culture
ICU location on culture date
Receipt of antibiotics ?30 days prior to the culture dated
25 (30.1)19 (30.2) 0.99 1.00 (0.49–2.04)
22 (26.5) 9 (14.3)0.10 2.16 (0.92–5.10)
aData are presented as numbers (percentages) except as noted otherwise in column 1. NA, not applicable.
bORs were unavailable for continuous variables or categorical variables with no events in one or more cells.
cData were unavailable for 24 patients (16 control patients, 8 case patients).
dAll other agents and classes of antimicrobials not shown due to P ? 0.20 on bivariable analyses.
TABLE 3 Final multivariable model of risk factors associated with CTX-
M-type ?-lactamase positivity in patients with clinical cultures with
Variable OR (95% CI)P
Charlson comorbidity score
Receipt of an extended-spectrum penicillin
?30 days prior to the culture date
Han et al.
aac.asm.org Antimicrobial Agents and Chemotherapy
were unable to differentiate colonization from infection in regard
to clinical cultures, although a substantial proportion of patients
in our study had bacteremia as the culture source. The selection
ical microbiology laboratory which processed and cultured all
specimens obtained at HUP and PPMC during the study period.
Similarly, misclassification bias is a concern in case-control stud-
ies, but case and control patients were identified based solely on
antibiotic susceptibility testing. Of note, our control group in-
cluded two patients with isolates that were carbapenem resistant
sion of these patients from the control group did not affect the
results of final multivariable analyses. Finally, the present study
istics or to other geographical locations.
In conclusion, we found that nonurinary sources of clinical
cultures, as well as the recent use of extended-spectrum penicil-
lins, conferred an increased risk of colonization or infection with
other ESBL types. Future studies will need to focus on outcomes
associated with infections due to CTX-M-producing E. coli, in-
cluding mortality, as well as elucidation of optimal infection con-
trol strategies designed to limit the spread of these increasingly
This study was supported by the National Institutes of Health
(K24AI080942 to E.L.), the Centers for Disease Control and Prevention
Department of Health (to E.L.).
The funding agencies had no role in the design and conduct of the
study, in the collection, management, analysis, and interpretation of the
data, or in the preparation, review, or approval of the manuscript.
E.L. has received research grant support from Merck, AstraZeneca,
3M, and Cubist.
siella spp. collected from intensive care units in Southern and Western
Europe in 1997–1998. J. Antimicrob. Chemother. 45:183–189.
2. Barton TD, Fishman NO, Weiner MG, LaRosa LA, Lautenbach E. 2005.
High rate of coadministration of di- or tri-valent cation-containing com-
pounds with oral fluoroquinolones: risk factors and potential implica-
tions. Infect. Control Hosp. Epidemiol. 26:93–99.
3. Bell JM, Turnidge JD, Gales AC, Pfaller MA, Jones RN. 2002. Prevalence
of extended spectrum beta-lactamase (ESBL)-producing clinical isolates
nonhospitalized patients. Clin. Infect. Dis. 49:682–690.
5. Bhavnani SM, Ambrose PG, Craig WA, Dudley MN, Jones RN. 2006.
Escherichia coli and Klebsiella species as defined by CLSI reference
methods: report from the SENTRY Antimicrobial Surveillance Pro-
gram. Diagn. Microbiol. Infect. Dis. 54:231–236.
6. Bonnet R. 2004. Growing group of extended-spectrum beta-lactamases:
the CTX-M enzymes. Antimicrob. Agents Chemother. 48:1–14.
7. Cassier P, et al. 2010. Cephalosporin and fluoroquinolone combinations
are highly associated with CTX-M beta-lactamase-producing Escherichia
coli: a case-control study in a French teaching hospital. Clin. Microbiol.
8. Clinical and Laboratory Standards Institute. 2006. Performance stan-
dards for antimicrobial disk susceptibility tests; approved standard 9th
edition: M2-A9. Clinical and Laboratory Standards Institute, Wayne, PA.
9. Clinical and Laboratory Standards Institute. 2008. Performance
standards for antimicrobial susceptibility testing; 18th information
supplement. M100-S18. Clinical and Laboratory Standards Institute,
10. Dutour C, et al. 2002. CTX-M-1, CTX-M-3, and CTX-M-14 beta-
11. Gasink LB, Zaoutis TE, Bilker WB, Lautenbach E. 2007. The categori-
zation of prior antibiotic use: impact on the identification of risk factors
for drug resistance in case control studies. Am. J. Infect. Control 35:638–
12. Gross R, et al. 2001. Impact of a hospital-based antimicrobial manage-
ment program on clinical and economic outcomes. Clin. Infect. Dis. 33:
13. Hanberger H, et al. 1999. Antibiotic susceptibility among aerobic gram-
negative bacilli in intensive care units in 5 European countries. JAMA
14. Harris AD, et al. 2002. Control-group selection importance in studies of
antimicrobial resistance: examples applied to Pseudomonas aeruginosa,
Enterococci, and Escherichia coli. Clin. Infect. Dis. 34:1558–1563.
15. Hosmer D, Lemeshow SL. 1989. Applied logistic regression. Wiley and
Sons, Inc, New York, NY.
16. Karisik E, Ellington MJ, Livermore DM, Woodford N. 2008. Virulence
factors in Escherichia coli with CTX-M-15 and other extended-spectrum
beta-lactamases in the UK. J. Antimicrob. Chemother. 61:54–58.
17. Knothe H, Shah P, Krcmery V, Antal M, Mitsuhashi S. 1983. Transfer-
able resistance to cefotaxime, cefoxitin, cefamandole and cefuroxime in
18. Lautenbach E, et al. 2009. Gastrointestinal tract colonization with fluo-
roquinolone-resistant Escherichia coli in hospitalized patients: changes
over time in risk factors for resistance. Infect. Control Hosp. Epidemiol.
19. Lautenbach E, Patel JB, Bilker WB, Edelstein PH, Fishman NO. 2001.
Extended-spectrum beta-lactamase-producing Escherichia coli and Kleb-
siella pneumoniae: risk factors for infection and impact of resistance on
outcomes. Clin. Infect. Dis. 32:1162–1171.
J. Clin. Microbiol. 45:620–626.
21. Lee I, et al. 2009. Risk factors for fluconazole resistance in patients with
Candida glabrata bloodstream infection: potential impact of control
group selection on characterizing the association between previous flu-
conazole use and fluconazole resistance. Am. J. Infect. Control 38:456–
22. Lewis JS, II, Herrera M, Wickes B, Patterson JE, Jorgensen JH. 2007.
First report of the emergence of CTX-M-type extended-spectrum beta-
system. Antimicrob. Agents Chemother. 51:4015–4021.
23. MacAdam H, Zaoutis TE, Gasink LB, Bilker WB, Lautenbach E. 2006.
Investigating the association between antibiotic use and antibiotic resis-
J. Antimicrob. Agents 28:325–332.
24. Mantel N, Haenszel W. 1959. Statistical aspects of the analysis of data
from retrospective studies of disease. J. Natl. Cancer Inst. 22:719–748.
25. McGettigan SE, Hu B, Andreacchio K, Nachamkin I, Edelstein PH.
2009. Prevalence of CTX-M beta-lactamases in Philadelphia, Pennsylva-
nia. J. Clin. Microbiol. 47:2970–2974.
26. McMullan R, Loughrey AC, McCalmont M, Rooney PJ. 2007. Clinico-
epidemiological features of infections caused by CTX-M type extended
spectrum beta lactamase-producing Escherichia coli in hospitalized pa-
tients. J. Infect. 54:46–52.
27. Mickey RM, Greenland S. 1989. The impact of confounder selection
criteria on effect estimation. Am. J. Epidemiol. 129:125–137.
28. National Nosocomial Infections Surveillance System. 2002. NNIS Sys-
2002. Am. J. Infect. Control 30:458–475.
29. Neuhauser MM, et al. 2003. Antibiotic resistance among gram-negative
bacilli in US intensive care units: implications for fluoroquinolone use.
Risk Factors for CTX-M
November 2012 Volume 56 Number 11aac.asm.org 5579