Skip Navigation

Extended-Spectrum β-Lactamase-Producing Escherichia coli and Klebsiella pneumoniae: Risk Factors for Infection and Impact of Resistance on Outcomes

  1. Ebbing Lautenbach1,2,
  2. Jean Baldus Patel3,
  3. Warren B. Bilker2,
  4. Paul H. Edelstein3, and
  5. Neil O. Fishman1
  1. 1 Division of Infectious Diseases, Department of Medicine, University of Pennsylvania Medical Center, Philadelphia
  2. 2 Department of Biostatistics and Clinical Epidemiology, University of Pennsylvania Medical Center, Philadelphia
  3. 3 Department of Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, Philadelphia
  1. Reprints or correspondence: Dr. Ebbing Lautenbach, Presbyterian Medical Center, University of Pennsylvania, Wright-Saunders Building, Suite W-250, 39th and Market Streets, Philadelphia, PA 19104-6021 (ebbing{at}mail.med.upenn.edu).

  1. Presented in part: 36th annual meeting of the Infectious Diseases Society of America, Denver, 12–15 November 1998.

 
Next Section

Abstract

The prevalence of antibiotic resistance among extended-spectrum β-lactamase (ESBL)-producing Escherichia coli and Klebsiella pneumoniae has increased markedly in recent years. Thirty-three patients with infection due to ESBL-producing E. coli or K. pneumoniae (case patients) were compared with 66 matched controls. Total prior antibiotic use was the only independent risk factor for ESBL-producing E. coli or K. pneumoniae infection (odds ratio, 1.10; 95% confidence interval, 1.03–1.18; P = .006). Case patients were treated with an effective antibiotic a median of 72 hours after infection was suspected, compared with a median of 11.5 hours after infection was suspected for controls (P < .001). ESBL-producing E. coli or K. pneumoniae infection was associated with a significantly longer duration of hospital stay and greater hospital charges (P = .01 and P < .001, respectively). Finally, many ESBL-producing E. coli and K. pneumoniae isolates were closely related. ESBL-producing E. coli and K. pneumoniae infections have a significant impact on several important clinical outcomes, and efforts to control outbreaks of infection with ESBL-producing E. coli and K. pneumoniae should emphasize judicious use of all antibiotics as well as barrier precautions to reduce spread.

Initial enthusiasm surrounding the introduction of oxyimino β-lactam agents in 1981 was quickly tempered by the emergence of pathogens that are resistant to these agents. Extended-spectrum β-lactamase (ESBL)-producing organisms were first isolated in Germany in 1983 [1], and outbreaks of infection due to these organisms soon occurred in several European centers [2, 3]. The first ESBL-producing isolates in the United States were reported in 1989 [4, 5], and shortly thereafter, a number of outbreaks of infection due to such organisms were reported [6, 7].

The incidence of infections due to organisms resistant to β-lactam agents has increased sharply in recent years. At 144 hospitals in the United States, the prevalence of ceftazidime-resistant Klebsiella pneumoniae increased from 1.5% in 1989 to 3.6% in 1991 [8]. The prevalence of such organisms in intensive care units (ICUs) in the United States increased from 3.6% in 1990 to 14.4% in 1993, and it was as high as 21.8% in large teaching hospitals [9].

It is imperative that risk factors for infections due to ESBL-producing organisms be clearly identified so that effective strategies to limit outbreaks of these infections may be developed. Several studies have attempted to elucidate risk factors for infections due to ESBL-producing organisms, but the results have been widely disparate. Although this disparity may be due in part to true differences in the epidemiology of different outbreaks, other potential reasons for this lack of consensus include failure to distinguish colonization from true infection [6, 10–12], lack of a comparison group [6, 7, 10, 13], small numbers of patients [12–14], inclusion of only specific patient populations [7, 13], and the limiting of the investigation to patients in ICUs [11, 12]. In addition, little is known regarding what role, if any, that resistance plays in the prediction of negative outcomes.

Beginning in June 1997, we noted a marked increase in the number of ESBL-producing isolates at our institution. We conducted the current study to identify risk factors for infection with ESBL-producing Escherichia coli or K. pneumoniae and to determine whether clinical outcomes differed between patients with infections caused by resistant organisms and those with infections caused by susceptible organisms.

Previous SectionNext Section

Methods

Risk factors for infection due to ESBL-producing organisms. This investigation was conducted at the University of Pennsylvania Medical Center, a 725-bed academic tertiary care hospital located in Philadelphia. To assess risk factors for infection due to ESBL-producing E. coli or K. pneumoniae, a matched case-control study was conducted. All case patients and control patients were identified through records of the Clinical Microbiology Laboratory, which processes and cultures all specimens obtained at this institution. All patients for whom culture results were positive for E. coli or K. pneumoniae from 1 June 1997 through 31 May 1998 were eligible for inclusion in the study. Designation as a case patient or a control patient was based solely on whether the infecting organism was found to demonstrate ESBL resistance.

Each patient was included as a case patient only once. If ESBL-producing E. coli or K. pneumoniae was isolated on multiple occasions, only the first episode of infection was reviewed. Potential controls were identified among hospitalized patients who were infected with non-ESBL-producing E. coli or K. pneumoniae during the same period. Controls were matched in a 2 : 1 ratio to case patients according to the following 3 variables: species of infecting organism, anatomic site of infection, and date of isolation. Only patients who met the Centers for Disease Control and Prevention's criteria for infection were included [15]. Nosocomial acquisition of infection was defined as follows: infection that occurred >48 h after admission to the hospital; infection that occurred <48 h after admission to the hospital, for patients who had been hospitalized within the 2 weeks prior to admission; and infection that occurred <48 h after admission to the hospital, for patients who had transferred from an outside hospital or nursing home.

Potential risk factors for ESBL-producing E. coli or K. pneumoniae infection were ascertained by means of a review of medical records. Data obtained included age, sex, race, hospital location, number of hospital days prior to infection, and severity of illness, as calculated by means of the Acute Physiological and Chronic Health Evaluation (APACHE) II score [16]. The presence of a central venous catheter, urinary catheter, or mechanical ventilation was also assessed. Finally, all antimicrobial therapy that was administered in the 30 days prior to admission was documented.

The presence of the following comorbid conditions was documented: hepatic dysfunction, malignancy, diabetes mellitus, renal insufficiency (indicated by a creatinine level of >2.0 mg/dL or the requirement of dialysis [17]), HIV infection, neutropenia, corticosteroid use, prior organ transplantation, use of an immunosuppressive agent in the 30 days prior to admission to the hospital, and surgical procedure or trauma in the 30 days prior to admission.

Role of ESBL-resistance in outcomes. To evaluate the effect of ESBL-producing E. coli or K. pneumoniae infection on clinical outcome, a retrospective cohort study including the same case- and control patients was conducted. The following outcomes were assessed: clinical outcome, microbiological outcome, mortality attributable to infection, duration of hospital stay after infection, and hospital charges accrued after infection.

Clinical outcome 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 of abnormalities in the above parameters without complete resolution; “failure,” for patients who had absence of abatement or deterioration in any clinical parameters; and “uncertain,” for patients who had intermittent or recurrent signs and symptoms that were not clearly attributable to infection [18]. Microbiological outcome was classified as follows: “definite response,” for patients whose cultures were sterile after a course of antimicrobial therapy; “probable response,” for patients whose cultures were sterile during a course of antimicrobial therapy; “failure,” for patients who had persistent isolation of the organism after at least 3 days of antimicrobial therapy; and “uncertain,” for patients who had intermittent isolation of the pathogen with no clear temporal association with antimicrobial therapy or absence of subsequent cultures to assess microbiological response [18].

“Mortality directly attributable to bacteremia” was defined as death in the setting of clinical evidence of active infection and a positive culture result. “Mortality indirectly attributable to bacteremia” was defined as infection that caused failure or further compromise of an organ system and death that occurred as a result of organ failure. The proportion of deaths directly and indirectly attributable to infection defined the attributable mortality rate [18, 19].

Microbiological methods. Susceptibilities to all antimicrobial agents were determined according to criteria of the National Committee for Clinical Laboratory Standards [20] by means of either a semiautomated system (MicroScan WalkAway System, NC16 panel; Dade Behring) or disk diffusion susceptibility testing. K. pneumoniae and E. coli isolates for which the MIC of ceftazidime was >2 µg/mL were suspected of producing an ESBL or AmpC-type β-lactamase. Such isolates were subjected to the double-disk diffusion test, as described by Thomson and Sanders [21], with the exception that the ceftazidime and amoxicillin/clavulanic acid disks were placed 15 mm apart. AmpC-type β-lactamase production was suspected in isolates for which the MIC of ceftazidime was elevated (>2 µg/mL) that were resistant to both amoxicillin/clavulanic acid and cefoxitin. Isolates that demonstrated AmpC-type β-lactamase production were also classified as ESBL-producing organisms.

Selected isolates were evaluated for relatedness by means of genomic fingerprinting by using the typing procedure described by Maslow et al. [22]. Fingerprinting data were interpreted according to the established guidelines [23]. To identify possible plasmid-mediated resistance, bacterial conjugations were performed according to the broth mating procedure described by Rice et al. [7]. For K. pneumoniae donor cells, the recipient strain was E. coli MB4800 (strA derivative of E. coli SK2267 [24], which was a gift of L. Silver, Merck Research Laboratories, Rahway, NJ). Transconjugants were selected by plating the conjugation mixture (donor and recipient cells) on laked blood media that contained ceftazidime (3 µg/mL) and streptomycin A (1000 µg/mL). For E. coli donor cells, the recipient strain was E. coli SM10λpir [25], and the transconjugants were selected by plating the conjugation mixture on laked blood media that contained ceftazidime (3 µg/mL) and kanamycin (200 µg/mL).

Statistical analysis. Categorical and continuous variables were compared by use of the Mantel-Haenzel test and conditional logistic regression, respectively [26]. Multivariable analysis was performed by means of conditional logistic regression [26]. All variables for which a P value of ⩽.15 was determined by means of bivariable analysis were considered for inclusion in an explanatory multivariable model. Building of the model began with inclusion of certain key variables based on a priori hypotheses (i.e., duration of antibiotic therapy and total number antibiotics administered) as well as variables (e.g., age and sex) believed likely to influence the association between the key variables and the outcome of interest. The effect of inclusion of the matching variables in the final multivariable model was also evaluated.

Conditional logistic regression was used to evaluate possible associations between ESBL-producing E. coli or K. pneumoniae infection and categorical outcomes. The association between ESBL-producing E. coli or K. pneumoniae infection and continuous outcomes was evaluated by use of generalized estimating equation regression for clustered data [27], following log transformation of the continuous outcomes (e.g., length of stay) to approximate more closely a normal distribution. In evaluating the association between ESBL-producing E. coli or K. pneumoniae infection and various outcomes, we controlled for certain variables (i.e., APACHE II score and duration of hospitalization prior to infection), which, on the basis of a priori hypotheses, were believed likely to influence the association between ESBL-producing E. coli or K. pneumoniae infection and outcomes of interest. A 2-tailed P value of <.05 was considered significant. All statistical calculations were performed by use of standard programs in STATA, version 5.0 (Stata).

Previous SectionNext Section

Results

During the study period, ESBL-producing E. coli or K. pneumoniae isolates were identified in 44 case patients, of whom 38 met criteria for infection [15]. Of these 38 case patients, 33 (86.8%) had medical records available for review.

Of the 33 patients with ESBL-producing E. coli or K. pneumoniae infection, 25 (75.8%) of them had infections due to K. pneumoniae and 8 (24.2%) had infections due to E. coli. The sites of infection were as follows: urinary, in 17 patients (51.5%); wound, in 5 (15.2%); central venous catheter, in 4 (12.1%); blood, in 3 (9.1%); respiratory, in 3 (9.1%); and abdominal, in 1 (3.0%).

Case patients were significantly younger than were control patients; also, they were more frequently male and had higher APACHE II scores (table 1). In addition, case patients were more likely to have nosocomial infection, to have had longer hospitalizations prior to infection, and to have a central venous catheter or urinary catheter in place than were controls (table 1). No significant differences were noted when hospital locations of case patients and control patients were compared. Nine case patients (27.3%) and 13 controls (19.7%) were located on a medical floor (OR, 1.62; 95% CI, 0.57–4.69; P = .37), whereas 6 case patients (18.2%) and 6 control patients (9.1%) were located on a surgical floor (OR, 0.57; 95% CI, 0.19–1.69; P = .31). Thirteen case patients (39.4%) and 17 control patients (25.8%) were located in an ICU (OR, 1.87; 95% CI, 0.78–4.52; P = .16).

Table 1
View larger version:
Table 1

General characteristics of patients with infection due to extended spectrum β-lactamase (ESBL)-producing Escherichia coli or Klebsiella pneumoniae (case patients) and patients with infection due to non-ESBL-producing E. coli or K. pneumoniae (control patients) in a study of the risk factors for infection and the role of resistance in negative therapeutic outcomes in a setting of an outbreak of infection.

When the comorbid conditions of the 2 groups were compared, case patients were significantly less likely than were control patients to have malignant disease (table 2). Renal insufficiency and diabetes mellitus were more common among case patients than they were among control patients, although the differences were not statistically significant. Finally, case patients had significantly greater prior cumulative antibiotic exposure (in terms of both total number of antibiotics and total duration of antibiotic treatment) as well as greater exposure to extended-spectrum cephalosporins, fluoroquinolones, aminoglycosides, cotrimoxazole, vancomycin, and metronidazole than did control patients (table 3).

Table 2
View larger version:
Table 2

Comorbid conditions in patients with infection due to extended-spectrum β-lactamase (ESBL)-producing Escherichia coli or Klebsiella pneumoniae (case patients) and patients with infection due to non-ESBL-producing E. coli or K. pneumoniae (control patients).

The only variable that remained an independent risk factor for ESBL-producing E. coli or K. pneumoniae infection after multivariable analysis was duration of antibiotic therapy (OR for each additional day of antibiotic therapy, 1.10; 95% CI, 1.03–1.18; P = .006). In addition, there was a borderline significant association between ESBL-producing E. coli or K. pneumoniae infection and presence of a central venous catheter (OR, 9.85; 95% CI, 0.87–111.34; P = .06) and diabetes (OR, 5.10; 95% CI, 0.87–30.00; P = .07). The variables “age” and “sex” were also included in the final model, but they were not associated with ESBL-producing E. coli or K. pneumoniae infection. Inclusion of the variables “duration of hospital stay prior to infection” and “nosocomial versus community acquisition of infection” in the model did not significantly alter the effect sizes of the primary model variables and, therefore, they were not included in the final explanatory model.

Results of antimicrobial susceptibility testing for the 33 ESBL-producing E. coli or K. pneumoniae isolates are shown in figure 1. Imipenem was the only agent to which these isolates did not demonstrate resistance. As treatment (determined on the basis of final results of antimicrobial susceptibility testing), 14 case patients (42.4%) received levofloxacin, 7 (21.2%) received an aminoglycoside, 5 (15.2%) received cotrimoxazole, 5 (15.2%) received imipenem, and 2 (6.1%) received doxycycline.

Figure 1
View larger version:
Figure 1

Results of antimicrobial susceptibility testing for 33 extended-spectrum β-lactamase-producing Escherichia coli and Klebsiella pneumoniae isolates recovered in an outbreak. n, Number of isolates tested.

Although patients with infections due to susceptible organisms were treated with an effective antibiotic (i.e., an agent to which the infecting organism was susceptible) a median of 11.5 h after a specimen was sent for culture, patients with infections due to resistant organisms were treated with an effective antibiotic a median of 72 h after the specimen was sent for culture (OR for each additional hour of delay in effective therapy, 1.05; 95% CI, 1.02–1.07; P < .001).

Of the case patients, 25 (75.8%) had complete or partial clinical response to therapy compared with 55 (83.3%) of control patients (OR, 1.43; 95% CI, 0.95–2.14; P = .08; table 4). When microbiological outcomes were compared, 19 case patients (57.6%) and 22 control patients (33.3%) had complete or partial response to therapy (OR, 0.61; 95% CI, 0.39–0.93; P = .02; table 4). Of note, microbiological outcomes for case patients for whom repeated cultures were not performed were considered to be nonresponses, and control patients were much less likely than case patients to have repeated cultures performed (42.4% vs. 63.6%, respectively).

Infection with ESBL-producing E. coli or K. pneumoniae was significantly associated with a greater median hospital charge accrued subsequent to infection than was non-ESBL-producing E. coli or K. pneumoniae infection ($66,590 vs. $22,231, respectively; charges were 2.90 times higher for case patients than for control patients; 95% CI, 1.76–4.78; P < .001). This difference remained significant after multivariable analysis was performed that controlled for APACHE II score and duration of hospitalization prior to infection (charges were 1.71 times higher for case patients than for control patients; 95% CI, 1.01–2.88; P = .04).

Infection with ESBL-producing E. coli or K. pneumoniae was also associated with a longer median duration of hospital stay subsequent to infection than was non-ESBL-producing E. coli or K. pneumoniae infection (11 vs. 7 days, respectively; median duration of hospital stay for case patients was 1.76 times greater than that for control patients; 95% CI, 1.17–2.64; P = .01). This association remained significant after multivariable analysis was performed that controlled for APACHE II score at the time of infection (median duration of hospital stay for case patients was 1.73 times greater than that for control patients; 95% CI, 1.14–2.65; P = .01) but not after controlling for both APACHE II score and duration of hospitalization prior to infection (median duration of hospital stay for case patients was 1.23 times greater than that for control patients; 95% CI, 0.81–1.87; P = .34).

Finally, 5 case patients (15.2%) had mortality attributable to infection, compared with 6 control patients (9.1%; OR, 1.91; 95% CI, 0.49–7.42; P = .35). Of the 5 case patients whose mortality was attributable to ESBL-producing E. coli or K. pneumoniae infection, 1 (20%) received appropriate therapy within 72 h of admission to the hospital. In comparison, 15 (53.6%) of 28 case patients who survived received appropriate therapy within 72 h of the time that the specimen was sent for culture. Of 8 case patients whose infections involved the bloodstream, 3 died. Only 1 (33.3%) of these 3 patients received appropriate therapy within 72 h of admission to the hospital, while 4 (80%) of 5 patients who survived received appropriate therapy during this period.

Molecular analysis of resistant E. coli and K. pneumoniae isolates provided evidence of nosocomial transmission. Restriction patterns of chromosomal DNA from 11 (KP4-KP9, KP11-KP13, KP15, and KP16) of the 13 K. pneumoniae isolates analyzed were identical or differed by ⩽4 bands, which indicated either a clonal relationship or a close relationship, respectively (figure 2). Likewise, the restriction patterns of chromosomal DNA from E. coli isolates EC2 and EC6 indicated that these isolates were closely related, and the restriction patterns of isolates EC3 and EC4 demonstrated a clonal relationship with each other. These data suggest that many of the resistant isolates were acquired from an external source rather than from the patient's endogenous flora.

Table 3
View larger version:
Table 3

Prior antibiotic exposure of patients with infection due to extended spectrum β-lactamase (ESBL)-producing Escherichia coli or Klebsiella pneumoniae (case patients) and patients with infection due to non-ESBL-producing E.coli or K. pneumoniae (control patients).

Figure 2
View larger version:
Figure 2

Pulsed-field gel electrophoresis patterns for chromosomal DNA from extended-spectrum β-lactamase-producing isolates recovered in an outbreak. Top, Klebsiella pneumoniae isolates; bottom, Escherichia coli isolates. The number of each lane corresponds to the number of the isolate (e.g., the isolates in lane 1 of top and lane 1 of bottom are KP1 and EC1, respectively). The lanes labeled λ correspond to a pulsed-field λ ladder size standard (Bio-Rad). kb, Kilobase.

A conjugative plasmid conferring increased resistance to ceftazidime was detected in 7 of 16 K. pneumoniae isolates and 1 of 7 E. coli isolates (table 5). The inability to detect a conjugative plasmid in all isolates was not unexpected, because other researchers have reported similar results [28]. Besides conferring ceftazidime resistance, all of the plasmids conferred resistance to gentamicin and tobramycin and, in the case of 2 plasmids from isolates KP3 and KP16, amikacin resistance. Finally, 5 of the plasmids transmitted resistance to cotrimoxazole (data not shown). Colocalization of aminoglycoside and cotrimoxazole resistance with ceftazidime resistance on a conjugative plasmid is consistent with the increased OR associated with prior aminoglycoside and cotrimoxazole therapy.

Table 4
View larger version:
Table 4

Clinical and microbiological outcomes for patients with infection due to extended-spectrum β-lactamase (ESBL)-producing Escherichia coli or Klebsiella pneumoniae (case patients) and patients with infection due to non-ESBL-producing E. coli or K. pneumoniae (control patients).

Table 5
View larger version:
Table 5

Summary of characterized isolates from patients with infection due to extended-spectrum β-lactamase-producing Escherichia coli or Klebsiella pneumoniae.

Previous SectionNext Section

Discussion

Past attempts to identify risk factors for infection due to ESBL-producing organisms have come to very different conclusions. This fact may be partly because most previous studies failed to distinguish colonization with such pathogens from true infection [6, 10–12]. It has been suggested that risk factors for infection and colonization may indeed be different [29]. For example, although patients with severe underlying diseases may have altered host defenses (and, therefore, greater risk of infection), host defenses are of little importance in the determination of whether a patient is colonized by an organism that is a natural inhabitant of the gastrointestinal tract [29]. To identify specific patients with an increased likelihood of developing a clinical infection due to ESBL-producing E. coli or K. pneumoniae, we have emphasized the elucidation of those factors that predict true infection.

Although early correlational studies suggested an association between antimicrobial use and the emergence of ESBL-producing E. coli or K. pneumoniae infections [6, 10, 13], these studies failed to control for potential confounding factors and lacked comparison groups. Subsequent case-control studies either noted no association between antimicrobial use and infection due to ESBL-producing organisms [11] or an association that did not remain after the researchers controlled for other variables [14, 29, 30]. In fact, even 5 years after ESBLs were first described in the United States, it was noted that the true nature of the selective effect of antibiotics in fostering these epidemics was not clear [31]. We found total exposure antimicrobial agents to be the only independent predictor of infection with ESBL-producing E. coli or K. pneumoniae; to our knowledge, this is the first time that such an independent association has been reported.

We found that the use of certain antibiotics was associated with infection with ESBL-producing E. coli or K. pneumoniae. In addition to greater exposure to extended-spectrum cephalosporins, we also noted that a significantly greater number of case patients than control patients had been exposed to cotrimoxazole and aminoglycosides. Conjugation experiments demonstrated that resistance to these drugs is associated with the presence of a conjugative plasmid that confers resistance to aminoglycosides and, in some cases, cotrimoxazole in addition to the ESBL phenotype. Colocalization of multiple resistance genes to a plasmid carrying ESBL-mediating genes has been reported elsewhere [32]. These data suggest that the spread of ESBLs may be partly due to the selective effect of other antibiotics.

Although we found no association between hospital location and infection with ESBL-producing E. coli or K. pneumoniae, molecular epidemiological studies of these isolates clearly indicated the nosocomial spread of ESBL-producing E. coli or K. pneumoniae isolates. On the basis of these results, as well as the strong association between infection with ESBL-producing E. coli or K. pneumoniae and antibiotic use, we hypothesize that multiple events may be required for infection with ESBL-producing E. coli or K. pneumoniae to develop. First, the patient must acquire ESBL-producing E. coli or K. pneumoniae through contact with a colonized health care worker or contaminated fomite. Second, the isolate must emerge as a result of the selective effect of antibiotic use.

These findings have important implications with regard to possible interventions aimed at curbing such outbreaks. Efforts should emphasize limiting contact transmission of resistant isolates as well as controlling antibiotic use. Most interventions aimed at limiting antibiotic use have focused on restricting the use of extended-spectrum cephalosporins and have met with only modest success [6, 7, 10]. As demonstrated in our study, infection with ESBL-producing E. coli or K. pneumoniae may be related to the use of several different antibiotics, but it is most closely associated with total antibiotic use. Elimination of one class of antibiotics, or the substitution of one class for another, without fully addressing the widespread problem of inappropriate antibiotic use will likely result in continued inadequate control of such outbreaks of infection. Emphasis must be placed on the rational and judicious use of all antimicrobial agents.

Despite the fact that patients with infections due to ESBL-producing organisms did not receive appropriate antimicrobial therapy until a median of 72 h after infection was first suspected, these delays in treatment did not result in worse clinical outcomes. One explanation for this lack of association may be that treatment efficacy is affected by the site of infection. Empirical antimicrobial therapy for urinary tract infection with agents to which the organism is resistant may be successful because of the high antibiotic levels achieved in the urine and/or local defense mechanisms. Of note, several studies have found that the conditions of patients with urinary tract infections due to ESBL-producing organisms improve despite the patients having received treatment with agents to which the organisms are resistant [3, 6, 13]. The potential impact of inappropriate antimicrobial therapy might become clearer if one focuses on more serious infections (e.g., bacteremia). Indeed, a 1997 study of patients with bloodstream infection due to ESBL-producing K. pneumoniae found a 75% crude mortality rate among patients who received ineffective initial empirical therapy, compared with 28% among patients whose initial therapy was active against the organism [33].

There are additional outcome measures that are of importance in assessing the impact of infections due to resistant organisms. We found that infection with ESBL-producing E. coli or K. pneumoniae was associated with significantly longer durations of hospital stay and increased hospital charges than were infections due to susceptible organisms. These results suggest that, even if infections due to ESBL-producing organisms have little impact on overall mortality, their impact on cost may still be of great significance.

There are several potential limitations to our study. Although the possibility of selection bias is normally of concern in a case-control study, all case patients and control patients were identified through the same microbiology laboratory. Because control patients were selected from the same hospital-based population that provided the case patients and were matched according to clearly defined criteria, the potential of selection bias should be small, with the exception of what was introduced through lost charts. Although misclassification bias is likewise often of concern, both case- and control patients were identified solely on the basis of data from antimicrobial susceptibility testing. Because these tests were conducted without prior knowledge of the status of the patients with regard to possible exposures or outcomes of interest, there is not likely to be any differential misclassification bias.

Although any retrospective chart review study may be limited by the availability and completeness of medical records, we found that >85% of charts were available for review. Even though information concerning in-hospital antibiotic use was available from the medical records, the possibility exists for inaccuracies in data concerning antibiotic use at an outside medical facility or as an outpatient. However, data regarding recent antibiotic use would have been assessed at the time of the patient's admission, before the doctor knew whether the patient had developed infection due to a resistant or susceptible organism. Therefore, the nature of the bias due to missing information, if present, is likely not differential, which would bias the results toward the null hypothesis. Despite this potential bias, we found a highly significant difference in antibiotic use in these 2 groups.

In conclusion, we found that total antimicrobial exposure was the only independent predictor of ESBL-producing E. coli or K. pneumoniae infection. Furthermore, molecular epidemiological analysis revealed that many of the ESBL-producing E. coli or K. pneumoniae isolates were closely related. These results suggest that if attempts to control outbreaks of infections due to these organisms are to be successful, such efforts must emphasize the rational and judicious use of all antimicrobial agents. Careful attention to barrier precautions to prevent the nosocomial spread of ESBL-producing E. coli or K. pneumoniae infections must also be stressed. Finally, ESBL-producing E. coli or K. pneumoniae infection was associated with a significantly longer duration of hospital stay and greater hospital charges, thereby demonstrating that these infections have an important impact on clinical outcomes.

Previous SectionNext Section

Footnotes

  • This study was reviewed and approved by the Committee on Studies Involving Human Beings of the University of Pennsylvania School of Medicine.

  • Received June 26, 2000.
  • Revision received August 21, 2000.
Previous Section
 

References

  1. ↵
    1. Knothe H,
    2. Shah P,
    3. Kremery V,
    4. Antal M,
    5. Mitsuhashi S
    . Transferable resistance to cefotaxime, cefoxitin, cefamandole and cefuroxime in clinical isolates of Klebsiella pneumoniae and Serratia marcescens. Infection 1983;11:315-7.
    CrossRefMedlineWeb of Science
  2. ↵
    1. Sirot D,
    2. Sirot J,
    3. Labia R,
    4. et al
    . Transferable resistance to third-generation cephalosporins in clinical isolates of Klebsiella pneumoniae: identification of CTX-1, a novel β-lactamase. J Antimicrob Chemother 1987;20:323-34.
    Abstract/FREE Full Text
  3. ↵
    1. Brun-Buisson C,
    2. Legrand P,
    3. Phillippon A,
    4. Montravers F,
    5. Ansquer M,
    6. Duval J
    . Transferable enzymatic resistance to third-generation cephalosporins during nosocomial outbreak of multiresistant Klebsiella pneumoniae. Lancet 1987;2:302-6.
    CrossRefMedline
  4. ↵
    1. Quinn JP,
    2. Miyashiro D,
    3. Sahm D,
    4. Flamm R,
    5. Bush K
    . Novel plasmid-mediated β-lactamase (TEM-10) conferring selective resistance to ceftazidime and aztreonam in clinical isolates of Klebsiella pneumoniae. Antimicrob Agents Chemother 1989;33:1451-6.
    Abstract/FREE Full Text
  5. ↵
    1. Jacoby GA,
    2. Medeiros AA,
    3. O'Brien TF,
    4. Pinto ME,
    5. Jiang J
    . Broad-spectrum transmissible β-lactamases. N Engl J Med 1989;319:723-4.
    CrossRefWeb of Science
  6. ↵
    1. Meyer KS,
    2. Urban C,
    3. Eagen JA,
    4. Berger BJ,
    5. Rahal JJ
    . Nosocomial outbreak of Klebsiella infection resistant to late-generation cephalosporins. Ann Intern Med 1993;119:353-8.
    Abstract/FREE Full Text
  7. ↵
    1. Rice LB,
    2. Willey SH,
    3. Papanicolaou GA,
    4. et al
    . Outbreak of ceftazidime resistance caused by extended-spectrum β-lactamases at a Massachusetts chronic-care facility. Antimicrob Agents Chemother 1990;34:2200-9.
    Abstract/FREE Full Text
  8. ↵
    1. Burwen DR,
    2. Banerjee SN,
    3. Gaynes RP
    ; The National Nosocomial Infections Surveillance System. Ceftazidime resistance among selected nosocomial gram-negative bacilli in the United States. J Infect Dis 1994;170:1622-5.
    Abstract/FREE Full Text
  9. ↵
    1. Itokazu GS,
    2. Quinn JP,
    3. Bell-Dixon C,
    4. Kahan FM,
    5. Weinstein RA
    . 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 1996;23:779-84.
    Abstract/FREE Full Text
  10. ↵
    1. Rice LB,
    2. Eckstein EC,
    3. DeVente J,
    4. Shlaes DM
    . Ceftazidime-resistant Klebsiella pneumoniae isolates recovered at the Cleveland Department of Veterans Affairs Medical Center. Clin Infect Dis 1996;23:118-24.
    Abstract/FREE Full Text
  11. ↵
    1. Piroth L,
    2. Aube H,
    3. Doise JM,
    4. Vincent-Martin M
    . Spread of extended-spectrum β-lactamase-producing Klebsiella pneumoniae: are β-lactamase inhibitors of therapeutic value? Clin Infect Dis 1998;27:76-80.
    Abstract/FREE Full Text
  12. ↵
    1. Pena C,
    2. Pujol M,
    3. Ardanuy C,
    4. et al
    . Epidemiology and successful control of a large outbreak due to Klebsiella pneumoniae producing extended-spectrum β-lactamases. Antimicrob Agents Chemother 1998;42:53-8.
    Abstract/FREE Full Text
  13. ↵
    1. Naumovski L,
    2. Quinn JP,
    3. Miyashiro D,
    4. et al
    . Outbreak of ceftazidime resistance due to a novel extended spectrum β-lactamase in isolates from cancer patients. Antimicrob Agents Chemother 1992;36:1991-6.
    Abstract/FREE Full Text
  14. ↵
    1. Schiappa DA,
    2. Hayden MK,
    3. Matushek MG,
    4. et al
    . Ceftazidime-resistant Klebsiella pneumoniae and Escherichia coli bloodstream infection: a case-control and molecular epidemiologic investigation. J Infect Dis 1996;174:529-36.
    Abstract/FREE Full Text
  15. ↵
    1. Garner JS,
    2. Jarvis WR,
    3. Emori TG,
    4. Horau TC,
    5. Hughes JM
    . CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988;16:128-40.
    CrossRefMedlineWeb of Science
  16. ↵
    1. Knaus WA,
    2. Drapier EA,
    3. Wagner DP,
    4. Zimmerman JE
    . APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818-29.
    MedlineWeb of Science
  17. ↵
    1. Linden PK,
    2. Pasculle AW,
    3. Manez R,
    4. et al
    . Differences in outcomes for patients with bacteremia due to vancomycin-resistant Enterococcus faecium or vancomycin-susceptible E. faecium. Clin Infect Dis 1996;22:663-70.
    Abstract/FREE Full Text
  18. ↵
    1. Lautenbach E,
    2. Schuster MG,
    3. Bilker WB,
    4. Brennan PJ
    . The role of chloramphenicol in the treatment of bloodstream infection due to vancomycin-resistant Enterococcus. Clin Infect Dis 1998;27:1259-65.
    Abstract/FREE Full Text
  19. ↵
    1. Noskin GA,
    2. Peterson LR,
    3. Warren JR
    . Enterococcus faecium and Enterococcus faecalis bacteremia: acquisition and outcome. Clin Infect Dis 1995;20:296-301.
    Abstract/FREE Full Text
  20. ↵
    National Committee for Clinical Laboratory Standards. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically: approved standard. 4th ed. Wayne, PA: National Committee for Clinical Laboratory Standards; 1997. NCCLS document M7-A4.
  21. ↵
    1. Thomason KS,
    2. Sanders CC
    . Detection of extended-spectrum beta-lactamases in members of the family Enterobacteriaceae: comparison of the double-disk and three-dimensional tests. Antimicrob Agents Chemother 1992;36:1877-82.
    Abstract/FREE Full Text
  22. ↵
    1. Persing DH,
    2. Smith TF,
    3. Tenover FC,
    4. White TJ
    1. Maslow JN,
    2. Slutsky AM,
    3. Arbeit RD
    . Application of pulsed-field gel electrophoresis to molecular epidemiology. In: Persing DH, Smith TF, Tenover FC, White TJ, editors. Diagnostic molecular microbiology (principles and applications). Washington, DC: American Society for Microbiology; 1993.
  23. ↵
    1. Tenover FC,
    2. Arbeit RD,
    3. Goering RV,
    4. et al
    . Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing. J Clin Microbiol 1995;33:2233-9.
    FREE Full Text
  24. ↵
    1. Ostroff GR,
    2. Pene JJ
    . Molecular cloning with bifunctional plasmid vectors in Bacillus subtilis. I. Construction and analysis of B. subtilis clone banks in Escherichia coli. Mol Gen Genet 1984;193:299-305.
    CrossRefMedlineWeb of Science
  25. ↵
    1. Miller VL,
    2. Mekalanos JJ
    . A novel suicide vector and its use in construction of insertion mutation: osmoregulation of outer membrane proteins and virulence determinants in Vibrio cholerae requires toxR. J Bacteriol 1988;170:2575-83.
    Abstract/FREE Full Text
  26. ↵
    1. Kleinbaum DG,
    2. Kupper LL,
    3. Morgenstern H
    . Epidemiologic research: principles and quantitative methods. New York: Van Nostrand-Reinhold; 1982.
  27. ↵
    1. Liang K-Y,
    2. Zeger SL
    . Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13-22.
    Abstract/FREE Full Text
  28. ↵
    1. Piddock LJ,
    2. Walters RN,
    3. Jin YF,
    4. Turner HL,
    5. Gascoyne-Binzi DM,
    6. Hawkey PM
    . Prevalence and mechanism of resistance to “third-generation” cephalosporins in clinically relevant isolates of Enterobacteriaceae from 43 hospitals in the UK, 1990–1991. J Antimicrob Chemother 1997;39:177-87.
    Abstract/FREE Full Text
  29. ↵
    1. Lucet JC,
    2. Chevret S,
    3. Decre D,
    4. et al
    . Outbreak of multiply resistant Enterobacteriaceae in an intensive care unit: epidemiology and risk factors for acquisition. Clin Infect Dis 1996;22:430-6.
    Abstract/FREE Full Text
  30. ↵
    1. Pena C,
    2. Pujol M,
    3. Ricart A,
    4. et al
    . Risk factors for faecal carriage of Klebsiella pneumoniae producing extended spectrum β-lactamase (ESBL-KP) in the intensive care unit. J Hosp Infect 1997;35:9-16.
    CrossRefMedlineWeb of Science
  31. ↵
    1. Quinn JP
    . Clinical significance of extended-spectrum beta-lactamases. Eur J Clin Microbiol Infect Dis 1994;13(Suppl 1):S39-42.
  32. ↵
    1. Sirot J,
    2. Chanal C,
    3. Petit A,
    4. Sirot D,
    5. Labia R,
    6. Gerbaud G
    . Klebsiella pneumoniae and other Enterobacteriaceae producing novel plasmid-mediated β-lactamases markedly active against third-generation cephalosporins: epidemiological studies. Rev Infect Dis 1988;10:850-9.
    MedlineWeb of Science
  33. ↵
    1. Paterson DL,
    2. Ko WC,
    3. Mohapatra S,
    4. et al
    . Program and abstracts of the 37th Interscience Conference on Antimicrobial Agents and Chemotherapy (Toronto). Washington, DC: American Society for Microbiology; 1997. Klebsiella pneumoniae bacteremia: impact of extended-spectrum β-lactamase (ESBL) production in a global study of 216 patients.
| Table of Contents

This Article

  1. Clin Infect Dis. (2001) 32 (8): 1162-1171. doi: 10.1086/319757
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 54 (5)
  1. Clinical Infectious Diseases
  1. Alert me to new issues

Most