Intensive Care Med (2002) 28:1718–1723 DOI 10.1007/s00134-002-1521-1
Bin Du Yun Long Hongzhong Liu Dechang Chen Dawei Liu Yingchun Xu Xiuli Xie
Received: 29 January 2002 Accepted: 11 September 2002 Published online: 17 October 2002 © Springer-Verlag 2002
B. Du (✉) · Y. Long · H. Liu · D. Chen D. Liu Department of Critical Care Medicine, Peking Union Medical College Hospital, 1 Shuai Fu Yuan, Beijing 100730, P. R. China e-mail:
[email protected] Tel.: +86-10-65296091 Fax: +86-10-65124875 Y. Xu · X. Xie Department of Clinical Microbiology, Peking Union Medical College Hospital, Beijing, P.R. China
ORIGINAL
Extended-spectrum beta-lactamase-producing Escherichia coli and Klebsiella pneumoniae bloodstream infection: risk factors and clinical outcome
Abstract Objectives: To study the risk factor for nosocomial bacteremia caused by Escherichia coli or Klebsiella pneumoniae producing extended-spectrum beta-lactamase (ESBL) and the influence on patient outcome. Design: Retrospective, single-center study of consecutive bacteremic patients. Settings: A university-affiliated teaching hospital. Patients: A total of 85 patients with nosocomial bacteremia due to E. coli or K. pneumoniae were enrolled. Intervention: None. Measurements and main results: The demographic characteristics and clinical information including treatment were recorded upon review of patients’ records. The primary end point was hospital mortality. Twenty-seven percent of isolates produced ESBLs. Previous treatment with 3rd-generation cephalosporins was the only independent risk factor for bacteremia due to ESBL-producing pathogens [odds ratio (OR) 4.146, P=0.008]. Antibiotic treatment was considered appropriate in 71 cases (83%), and failed in 23 patients (27%). Twentyone patients (25%) died in the hospi-
Introduction Antimicrobial resistance among Gram-negative bacilli represents a major problem in nosocomial infection [1]. These organisms elaborate beta-lactamases, which can hydrolyze the amide bond in the beta-lactam ring of antibiotics. Since 1983, extended-spectrum beta-lactamase (ESBL)-producing organisms have been isolated
tal. Antibiotic treatment failure was the only independent risk factor for hospital mortality (OR 15.376, P=0.001). Inappropriate antibiotic treatment might lead to significantly higher mortality rate (7/14 vs 14/71, P=0.016). Patients treated with imipenem were more likely to survive while those receiving cephalosporin treatment tended to have a poorer outcome (1/19 vs 14/40, P=0.023). Conclusions: More judicious use of cephalosporins, especially 3rd-generation cephalosporins, may decrease ESBLproducing E. coli or K. pneumoniae bacteremia, and also improve patient outcome. Keywords Escherichia coli · Klebsiella pneumoniae · Bacteremia · Beta-lactamases · Risk factors · Hospital mortality Abbreviations ESBL extendedspectrum beta-lactamase · APACHE acute physiology and chronic health evaluation · SD standard deviation OR odds ratio · CI confidence interval · ICU intensive care unit
around the world [2, 3, 4]. In the United States, the prevalence of Klebsiella pneumoniae and Escherichia coli resistant to 3rd-generation cephalosporins in ICU was 8.9% and 3.2%, respectively, and was even higher in large teaching hospitals [5]. ESBLs are especially dangerous because they are plasmid-associated, and the plasmids may be exchanged among a variety of bacterial species.
1719
Partly because of the difficulty to differentiate infection from colonization, a number of studies have elucidated inconsistent results concerning the risk factors and clinical significance of infections due to ESBL-producing organisms [6, 7]. We therefore conducted a retrospective study of nosocomial bacteremia caused by ESBL-producing E. coli and K. pneumoniae, in order to identify the possible risk factor for infection and hospital mortality.
Patients and methods
Microbiological methods Susceptibilities to all antimicrobial agents were determined according to the criteria of the National Committee for Clinical Laboratory Standards by disk diffusion [10]. An isolate with ceftazidime inhibition zone ≤22 mm was suspicious for ESBL production. Synergy between caftazidime and clavulanate was detected by placing a disk of Augmentin (20 µg of amoxicillin plus 10 µg of clavulanic acid) and a disk of ceftazidime 30 mm apart (centerto-center). A clear-cut extension of the edge of the ceftazidime inhibition zone toward the disk containing clavulanic acid was interpreted as synergy, which suggested the production of ESBL. The double-disk synergy test was considered positive when decreased susceptibility to ceftazidime was combined with synergy between ceftazidime and Augmentin [11].
Setting and design The investigation was conducted at Peking Union Medical College Hospital, a university-affiliated teaching hospital with approximately 1,200 beds. A retrospective review was performed on the records of the Clinical Microbiology Laboratory from January 1997 to December 1999 to identify all patients who had a documented episode of nosocomial bacteremia caused by E. coli or K. pneumoniae during hospitalization. Each patient was included only once. If multiple blood cultures from the same patient were positive for the above organisms, only the first episode was reviewed and recorded. Upon review of the patient’s record, the following data were recorded: age, sex, hospital location, and severity of illness, as calculated by APACHE (acute physiology and chronic health evaluation) II score [8]. The presence of a urinary catheter or mechanical ventilation was also assessed. In addition, all antibiotic therapy and operations performed in the 2 weeks prior to the bacteremic episode were documented. The presence of the following comorbidity was documented: malignancy, diabetes mellitus, solid organ transplantation, hypertension, coronary heart disease, neutropenia (white blood cell count less than 0.5×109/l), corticosteroid use, use of an immunosuppressive agent within 30 days prior to bacteremic episode. Definitions Nosocomial bacteremia was defined as a recognized pathogen cultured from one or more blood cultures that were obtained more than 48 h after hospital admission. Blood cultures were obtained primarily via peripheral venipuncture using a standard sterile technique. Central venous catheters were used to obtain blood cultures only if they had been freshly placed and prior to breaking the sterile field that was used for catheterization. Patients with acute leukemia or blastic relapse of chronic leukemia were classified into the category of rapidly fatal underlying disease based on the criteria described by Kreger et al. [9]. A body site was considered to be a source of bacteremia on the basis of evidence of inflammation and isolation from local sites of the same organism as found in the blood. When no cultural confirmation of local sites of origin could be obtained, or when more than one site fulfilled these criteria, or when death occurred too rapidly to document the primary infection focus, patients were classified as of an “unknown” source. Appropriate antibiotic therapy was defined if the infecting organism was subsequently found to be susceptible in vitro to the drug administered and if that antibiotic was given intravenously. The success of antibiotic treatment was considered if patients had resolution or abatement of clinical manifestation related to infection, such as fever, leukocytosis, and all signs of infection. Treatment failure was defined if patients had absence of abatement or deterioration in any clinical parameters.
Statistical analysis Values are presented as the mean±SD (continuous variables) or as a percentage of the group from which they were derived (categorical variables). Continuous variables were compared with the use of Student’s t-test. The chi-square test or Fisher’s exact test was used to compare categorical variables. For determination of independent predictors for treatment failure and hospital mortality, odds ratio was estimated on the basis of multivariate logistic regression analysis. All comparisons were unpaired and all tests of significance were two-tailed. A P value less than 0.05 was considered as statistically significant.
Results A total of 85 consecutive patients with bacteremia due to E. coli or K. pneumoniae were enrolled (Table 1). Fortyseven percent of the patients were male, and 53% were female. The mean age was 48.7±19.7 years, and mean APACHE II score was 15.6±5.6. Risk factors for ESBL-production bacteremia Blood cultures yielded 60 E. coli isolates, and 25 K. pneumoniae isolates. Among these 85 pathogens, 23 (27%) were found to produce ESBL, including 16 E. coli isolates and seven K. pneumoniae isolates (P >0.05). According to univariate analysis, previous treatment with cephalosporins, especially 3rd-generation cephalosporins, was more likely to be associated with isolation of ESBL-producing organisms in the blood (Table 1). Multivariate logistic regression analysis identified previous treatment with 3rd-generation cephalosporins as the only independent risk factor for bacteremia due to ESBL-producing pathogens [odds ratio (OR) 4.146, 95% confidence interval (CI) 1.448–11.875, P=0.008]. Treatment As shown in Table 2, a majority (89%) of patients received monotherapy rather than combination therapy. Third-generation cephalosporins (n=40), imipenem (n=19), beta-lac-
1720
Table 1 Characteristics and medical conditions of patients with E. coli or K. pneumoniae bacteremia. (ESBL extended-spectrum betalactamase, 3GC 3rd-generation cephalosporins, LOS length of stay) Variable
ESBL-positive (n=23)
ESBL-negative (n=62)
Total (%) (n=85)
P value
Male sex Age APACHE II score
13 49.7±23.6 15.3±6.1
27 48.4±18.3 15.7±5.4
40 (47%) 48.7±19.7 15.6±5.6
0.287 0.782 0.754
Clinical manifestations Fever (temperature ≥37.5 °C) Abnormal mental status Shock
23 1 1
61 2 8
84 (99%) 3 (4%) 9 (11%)
1.000 1.000 0.433
Underlying disease Malignancy Solid organ transplantation Hypertension Coronary artery disease Diabetes mellitus Rapidly fatal
13 2 3 3 4 3
30 0 11 10 6 7
43 (51%) 2 (2%) 14 (17%) 13 (15%) 10 (12%) 10 (12%)
0.505 0.071 0.749 1.000 0.448 1.000
6 6 5
14 14 19
20 (24%) 20 (24%) 24 (28%)
0.735 0.735 0.418
16 8 2 3
41 17 5 7
57 (67%) 25 (29%) 7 (8%) 10 (12%)
0.765 0.508 1.000 1.000
Invasive procedure Urinary catheter Surgery Mechanical ventilation
6 6 3
18 21 10
24 (28%) 27 (32%) 13 (15%)
0.789 0.493 0.725
Primary infection focus Abdomen Urinary tract Lower respiratory tract Central venous catheter Surgical site Other Unknown
5 2 4 2 1 0 9
16 5 3 5 1 2 30
21 (25%) 7 (8%) 7 (8%) 7 (8%) 2 (2%) 2 (2%) 39 (46%)
0.699 1.000 0.082 1.000 0.470 1.000 0.447
History of medical treatment Previous antibiotics Previous beta-lactams Previous cephalosporins Previous 3GC Previous penicillins Previous fluoroquinolones Previous other antibiotics
19 14 13 12 1 4 1
42 26 17 13 8 13 3
61 (72%) 40 (47%) 30 (35%) 25 (29%) 9 (11%) 17 (20%) 4 (5%)
0.176 0.120 0.013 0.005 0.433 1.000 1.000
Others Admission to intensive care unit LOS more than 2 weeks before bacteremia
4 10
5 28
9 (11%) 38 (45%)
0.245 0.890
Immunocompromise Corticosteroids Other immunosuppression agents Chemotherapy Laboratory findings Hemoglobin ≤110 g/l Elevated liver function tests Creatinine >1.5 mg/dl Neutropenia (1.5 mg/dl Previous beta-lactams LOS more than 2 weeks before bacteremia Inappropriate antibiotic treatment Treatment with 3GC Treatment with imipenem Treatment failure
14.6±5.3 28 1 2 26 22 7 26 18 8
18.6±5.5 15 9 5 14 16 7 14 1 15
tam/beta-lactamase inhibitors (n=13), aminoglycosides (n=12), and fluoroquinolones (n=8) were among the most commonly used antibiotics, and in that order. Third-generation cephalosporins were found to be less commonly used in immunocompromised patients (5/20 vs 35/65, P=0.024), mechanically ventilated patients (2/13 vs 38/72, P=0.013), and patients with urinary catheter (7/24 vs 33/61, P=0.038). Imipenem was more often employed in patients receiving previous treatment with cephalosporins (10/30 vs 9/55, P=0.073), especially 3rd-generation cephalosporins (10/25 vs 9/65, P=0.012), and in patients admitted to ICU (5/9 vs 14/76, P=0.024). Moreover, univariate analysis showed a trend toward more use of beta-lactam/beta-lactamase inhibitors in immunocompromised hosts (6/20 vs 7/65, P=0.069). Antibiotic treatment was considered appropriate in 71 cases (83%). Among 14 patients with inappropriate antibiotic treatment, seven were treated with 3rd-generation cephalosporins, three were receiving fluoroquinolones, one patient was treated with beta-lactam/beta-lactamase inhibitor, and one patient was treated with aminoglycoside. The other two patients did not receive any antibiotics despite positive blood cultures. Combination therapy did not improve the accuracy of antibiotic treatment (9/9
OR (95%CI)
P value
3.214 (1.105, 9.350) 47.250 (5.470, 408.175) 9.688 (1.718, 54.615) 2.923 (1.038, 8.232) 6.109 (1.976, 18.891) 4.701 (1.227, 13.514) 2.923 (1.038, 8.232) 0.128 (0.016, 1.024) 17.500 (5.260, 58.226)
0.004 0.028 0.000 0.009 0.038 0.001 0.016 0.038 0.033 0.000
vs 62/76, P=0.344). Patients on steroid treatment were more likely to receive inappropriate antibiotic therapy (7/20 vs 7/65, P=0.011), while the status of ESBL production and previous treatment with 3rd-generation cephalosporins were associated with a trend toward inappropriate antibiotic therapy (6/23 vs 8/62 and 7/25 vs 7/60, respectively), although without statistical significance (P=0.189 and 0.105, respectively). Outcome Antibiotic treatment failed in 23 patients (27%), including 18 patients with monotherapy, four patients with combination therapy of 3rd-generation cephalosporins and aminoglycosides, and one patient without any antibiotic treatment. Among the 18 patients with monotherapy, nine were treated with 3rd-generation cephalosporins, four were treated with imipenem, two were treated with aminoglycosides, while beta-lactam/beta-lactamase inhibitors and fluoroquinolone were used in two and one patients, respectively. The status of ESBL production showed no association with treatment failure (6/23 vs 17/62, P=0.902); however, inappropriate antibiotic thera-
1722
Table 4 The association of ESBL, antibiotic treatment, and hospital mortality in patients with bacteremia due to E. coli or K. pneumoniae. (NA not available)
* P
