Escherichia coli and Klebsiella pneumoniae are the major pathogens of community-onset infections, such as urinary tract and intra-abdominal infections [1]. Guidelines recommend the use of fluoroquinolones, cephalosporins, and β lactam/β lactamase inhibitor combinations as treatment options [2,3]. Third-generation cephalosporins (3GCs) are now used widely, due to the global prevalence of fluoroquinolone resistance [4-6]. Recently, 3GC resistant E. coli and K. pneumoniae producing extended-spectrum β-lactamases (ESBL), or imported AmpC β-lactamases have been emerging in community-onset infections [7-9].

A nationwide survey reported that community-onset infection caused by ESBL-producers is increasing in South Korea [10]. However, the data on 3GC-resistant E. coli and K. pneumoniae in community-onset infections at secondary and primary care hospitals are limited, as previous studies have focused on tertiary care hospitals. To this end, we performed a risk factor analysis for 3GC resistance in E. coli and K. pneumoniae bacteremia in the emergency department (ED) of a secondary care hospital.

One thousand two hundred and eight episodes of community-onset bacteremia were identified, of which 551 (45.6%) and 206 (17.1%) were caused by E. coli and K. pneumoniae, respectively, at the Daegu Fatima Hospital ED during the study period. E. coli and K. pneumoniae were present in 85% (318/373), 71% (304/429), and 35% (53/153) of urinary tract infections, intra-abdominal infections, and pneumonia, respectively (Fig. 1). Five hundred and forty-eight patients with E. coli bacteremia, and 183 patients with K. pneumoniae bacteremia, whose clinical data were available, were included in the analysis. The mean age was 65.93 ± 13.8 years and 62.4% (458/734) were female. The most common infections were urinary tract (43.1%) infections, followed by intra-abdominal (39.0%), lower respiratory tract (7.2%), and primary bacteremia (6.7%) infections.

Among 731 clinical isolates of E. coli and K. pneumoniae, 93.8% were susceptible and 6.2% were resistant to 3GC. Resistance rates of ampicillin/sulbactam, trimethoprim/sulfamethoxazole, fluoroquinolone, tobramycin, cefazolin, 3GC, and amikacin were 45.8% (245/535), 22.8% (165/725), 19.6% (143/731), 11.4% (83/728), 9.8% (68/695), 6.2% (45/731), and 1.3% (9/715), respectively. Rate of resistance rates to the antibiotics, with the exception of imipenem, were significantly higher in the 3GC-resistant group (p < 0.01 for all tested antibiotics except imipenem). Rate of resistance rates to the antibiotics according to organism and infection site are shown in Table 1. Thirty-two (4.4%) were resistant to a 3GC and f luoroquinolone combination, and six (0.8%) were resistant to a 3GC and amikacin combination (Table 1).

The rate of 3GC resistance increased significantly in 2009 (10.6%), compared to the previous annual average rate from 2003 to 2008 (6.1%; p = 0.03) (Fig. 2). Of 45 3GC-resistant isolates, resistance rates to fluoroquinolone, trimethoprim/sulfamethoxazole, piperacillin/tazobactam, amikacin, and imipenem were 71.1%, 55.6%, 20.0%, 13.3%, and 2.2%, respectively and 41 (91.1%) were ESBL-producers. From 2003 to 2008, 53.1% (17/32) of 3GC-resistant isolates were susceptible to cefepime, but in 2009, 84.6% (11/13) were susceptible to cefepime (p = 0.03) (Fig. 2).

Comparisons of clinical characteristics between the 3GC-susceptible and resistant groups are shown in Table 2. There were no significant differences in the median Pitt bacteremia score and the CCI between the two groups (Table 2).

Bivariate analyses showed that prior admission to hospital (p < 0.01), prior antibiotic use (p < 0.01), metastatic cancer (p < 0.01), and neutropenia (p = 0.04) were significantly associated with 3GC resistance in E. coli and K. pneumoniae bacteremia (Table 3). In a multivariate logistic regression analysis, previous exposure to antibiotics was an independent risk factor for 3GC resistance (odds ratio, 5.02; 95% confidence interval, 1.54 to 16.39; p = 0.01) (Table 3).

Among 73 isolates from patients with known prior antibiotics use, 13 (17.8%) were resistant to 3GC, 10 (13.7%) were resistant to a 3GC-fluoroquinolone combination and two (2.7%) were resistant to a 3GC-amikacin combination. Excluding the 104 follow-up loss cases, 627 cases were included in the final analysis. There was no significant difference between the 3GC-susceptible and -resistant groups in treatment success rates (p = 0.1) or 30-day mortality (p = 0.9). Further analysis according to species and sites of infection showed similar results (Table 4).

3GC is frequently used as an initial empirical antibiotic in primary and secondary hospitals. However, recent studies have warned of an increasing prevalence of infections in community-onset infections due to 3GC-resistant organisms, which produce ESBL or AmpC β-lactamase [9,13,14]. A recent nationwide multicenter study of bacteremia reported that 9.5% of community-onset E. coli bacteremia was caused by ESBL-producers [10]. However, there is no certainty that the findings from the previous studies are applicable to secondary and primary hospitals, because the majority of the data came from tertiary care hospitals. The Daegu Fatima Hospital, however, is a secondary care hospital which is more community-oriented, and most of the ED patients present without a referral letter. Therefore, the results of this study may better reflect the conditions in primary and secondary hospitals.

We found that the proportion of cefepime-susceptible organisms among the 3GC-resistant organisms increased in accordance with an increasing prevalence of 3GC resistance. ESBL-producing strains of Enterobacteriaceae showed different resistance patterns to various cephalosporins, which may ref lect the different abilities of ESBL types [15]. A nationwide study reported that 3CG-resistant organisms, such as ESBL-producers, are disseminated throughout the South Korean community, which may be associated with nationwide dissemination of the CTX-M ESBLs [16]. The increased proportions of cefepime-susceptible organisms among 3GC-resistant organisms might reflect the increase in a particular type of CTX-M ESBL.

Secondly, 3GC resistance had less impact on treatment outcomes (except for pneumonia patients) even though 83% of patients with a 3GC-resistant infection received inappropriate antibiotics. There is an ongoing debate about the association between ESBL-producing organism and clinical outcomes [17-21]. We suggest the main reason for the reduced impact of 3GC-resistance in our hospital was due to the lower severity of the condition of our patients: 86% of patients had a Pitt bacteremia score ≤ 3. These findings are likely similar to those in other primary or secondary care hospitals in Korea.

We found that only 0.8% of community-onset E. coli and K. pneumoniae bacteremia was simultaneously resistant to 3GC and amikacin. ESBL-producers are frequently resistant to fluoroquinolone, trimethoprim-sulfamethoxazole, and aminoglycoside [12,22]. Rodriguez-Bano et al. [9] reported that 64%, 63%, and 15% of ESBL-producers in community-onset E. coli bacteremia were resistant to ciprof loxacin, trimethoprim-sulfamethoxazole and amikacin, respectively. In this study, 71.1% and 13.3% of 3GC-resistant E. coli and K. pneumoniae were also resistant to f luoroquinolone and amikacin, respectively. However, among 731 isolates from community-onset E. coli and K. pneumoniae bacteremia, less than 1% were simultaneously resistant to 3GC and amikacin. Therefore, instead of carbapenem, a 3GC-amikacin combination could be an option for treating patients with community-onset gram-negative bacteremia with a high risk of resistance, unless those patients have an impaired renal function. The piperacillin/tazobactam resistance rate was low (2.1%) in this study. Recent retrospective studies demonstrated that piperacillin/tazobactam had comparable outcomes to carbapenem [23,24]. Therefore, piperacillin/tazobactam could also be an option for treating patients with a high risk of resistance in primary and secondary care hospitals.

Because the resistance rate to ampicillin/sulbactam was ~45%, its empirical use for treating infections caused by gram-negative organisms could be inappropriate.

This study had some limitations. Firstly, it was conducted at a single center, and therefore our data should be interpreted and applied with caution. Secondly, we were unable to collect detailed demographic information on the patients as the study was performed retrospectively. This prevented calculation of risk factors for 3GC-resistance other than previous antibiotics use. Thirdly, our data on ESBL production were generated from an automated data collection system; thus there are likely some discrepancies between the susceptibility data in this study and standard methods. We did not perform ESBL-conf irmation tests or genotype analysis. We identified an increasing trend in 3GC resistance but could not define the association between the increasing 3GC resistance and the dissemination of CTX-M ESBLs in the community.

To conclude, the rate of 3GC resistance in community-onset infections in primary and secondary care hospitals in Korea is increasing, and prior antibiotic use is an independent risk factor. Carbapenem should be considered an empirical antibiotic for patients with community-onset bacteremia and a high risk of resistance. However, in spite of the increasing rate of 3GC resistance in community-onset bacteremia, simultaneous resistance to 3GC and amikacin is rare, and most patients had a less severe infection. We suggest a combination of 3GC and aminoglycoside as an empirical antimicrobial option for treating patients with community-onset infections with a high risk of resistance in primary or secondary care hospitals in Korea.

1. Resistant rate of third generation cephalosporin (3GC) is increasing but, resistance to aminoglycoside is rare in primary and secondary care hospital in Korea.

2. Combination of 3GC and aminoglycoside could be an empirical option for community-onset infections with a high risk of resistance in primary or secondary care hospitals.