High frequency of antibiotic resistance among Gram-negative isolates in intensive care units at 10 Swedish hospitals
Abstract
Objective: To investigate the resistance rates among Gram-negative isolates in Swedish intensive care units (ICUs).
Method: During 1994–95, members of the Swedish Study Group collected, on clinical indication, 502 consecutive initial isolates of Gram-negative bacteria from patients admitted to ICUs at 10 Swedish hospitals and performed minimal inhibitory concentration (MIC) determinations with the Etest. Breakpoints were defined according to the criteria of the Swedish Reference Group for Antibiotics (SRGA).
Results: The distribution of bacterial species was: Escherichia coli > Klebsiella spp. > Enterobacter spp. > Pseudomonas aeruginosa > Haemophilus spp. > Proteus spp. > Stenotrophomonas maltophilia > Citrobacter spp. > Acinetobacter > Pseudomonas. spp. > Morganella morganii > Serratia spp. Together these constituted 97% of all isolates. The frequencies of resistance for all the initial Gram-negative isolates were: ceftazidime 6.8%, cefotaxime 14.9%, ceftriaxone 18.5%, cefuroxime 44.1%, ciprofloxacin 4.2%, co-trimoxazole 17.8%, gentamicin 5.8%, imipenem 8.6%, piperacillin 20.2%, piperacillin/tazobactam 12.9% and tobramycin 5.8%.
Conclusions: Among Gram-negative isolates in Swedish ICUs, a very high frequency of resistance was seen to cefuroxime, and rather high frequencies of resistance to cefotaxime, ceftriaxone, piperacillin and piperacillin/tazobactam. These drugs cannot be recommended for further use as empirical monotherapy for severe ICU-acquired Gram-negative infections in ICUs in Sweden.
INTRODUCTION
In a recent European prevalence study it was shown that 20.6% of intensive care unit (ICU) patients had an ICU-acquired infection [1]. The nosocomial infection rates among ICU patients are 5–10 times higher than among general ward patients [2,3] and most epidemics originate in ICUs [3–6], ICU-acquired infections are often associated with microbiologica isolates of resistant organisms [4,7]. Selective pressure exerted by heavy antibiotic use, as in ICUs, is the driving force behind the emergence of antibiotic resistance; eventually, bacteria develop resistance to practically all antibiotics [8]. Surveillance of antibiotic resistance is especially important in ICUs, since infection rates are much higher there than in other hospital wards [2,3]. The best current strategy for preventing ICU infections and the spread of antibiotic-resistant bacteria involves emphasizing handwashing between patient contacts, keeping the use of invasive devices to a minimum, and the restriction of antibiotic use [4,9,10]. A lower incidence of antibiotic resistance among Gram-negative blood isolates has been reported in Sweden as compared to the incidence in southern Europe [11–15]. However, in a recent survey of bacterial resistance of all isolates collected from 1993 to 1994 in four ICUs at a Swedish university hospital, β-lactam resistance among Enterobacter spp. and Enterococcus spp. showed an increase [16]. A simultaneous survey of all blood isolates from the four hospitals in the same county revealed the same emergence of resistance, but a lower rate compared to those in the ICUs [16]. The aim of this study was to investigate the prevalence of antibiotic resistance among 100 consecutively collected Gram-negative bacteria in 10 different ICUs in Sweden.
MATERIALS AND METHODS
Study design and culture collection
A Swedish multicenter susceptibility testing study was performed on isolates collected from August 1994 to June 1995 from patients admitted to ICUs in 10 different hospitals. The cultures were performed consecutively on clinical indications. The mean collection time was 7 months (range 3.5–8.5 months). Four hospitals were tertiary university hospitals and six were primary county hospitals. Seven hundred and fifty-nine consecutive Gram-negative bacterial isolates were collected from 347 patients admitted to the ICUs. Of these isolates, 502 were initial isolates (Table 1) and 257 were repeat isolates of the same species from the same patient. The repeat isolates were not included in the analysis of data except for a comparison between the 502 (initial) isolates and the 759 (initial and repeat) isolates. The sources of the isolates are shown in Table 2.
| Isolates (%) | Patients | |
|---|---|---|
| Escherichia coli | 128 (25.5) | 128 |
| Klebsiella spp. | 70 (13.8) | 68 |
| Enterobacter spp. | 68 (13.4) | 62 |
| Pseudomonas aeruginosa | 59 (11.8) | 59 |
| Haemophilus spp. | 39 (7.7) | 38 |
| Proteus spp. | 31 (6.2) | 31 |
| Stenotrophomonas maltophilia | 28 (5.5) | 28 |
| Citrobacter spp. | 21 (4.2) | 20 |
| Acinetobacter spp. | 16 (3.2) | 16 |
| Pseudomonas spp. | 12 (2.4) | 12 |
| Serratia spp. | 9 (1.8) | 9 |
| Morganella morganii | 7 (1.4) | 7 |
| Others | 14 (2.8) | - |
| Total | 502 (100) |
| Isolates | (%) | |
|---|---|---|
| Respiratory aspirate/drain/lavage/pleural/sputum | 149 | 29.7 |
| Urine | 81 | 16.1 |
| Tracheal | 66 | 13.1 |
| Abdominal wound/drain/bile/peritoneum | 63 | 12.5 |
| Ear/nose | 45 | 9.0 |
| Blood | 33 | 6.6 |
| Skin and soft tissue | 21 | 4.2 |
| Others | 44 | 8.8 |
| Total | 502 | 100 |
Susceptibility testing
Minimal inhibitory concentration (MIC) was determined using Etest (AB BIODISK, Solna, Sweden) [17–19]. Resistant strains were defined according to the MIC breakpoints of the Swedish Reference Group of Antibiotics (SRGA) [20]. These SRGA breakpoints for susceptible/resistant isolates are shown in Table 3, together with corresponding breakpoints from the National Committee for Clinical Laboratory Standards (NCCLS) [21].
| SRGA | NCCLS | |||
|---|---|---|---|---|
| Susceptible | Resistant | Susceptible | Resistant | |
| Imipenem | ≤4 | ≥16 | ≤4 | ≥16 |
| Ceftazidime | ≤4 | ≥16 | ≤8 | ≥32 |
| Cefotaxime | ≤4 | ≥16 | ≤8 | ≥64 |
| Ceftriaxone | ≤4 | ≥16 | ≤8 | ≥64 |
| Cefuroxime | ≤4 | ≥16 | ≤8 | ≥32 |
| Piperacillin | ≤16 | ≥32 | ≤16 | ≥128 |
| Piperacillin + P. aeruginosa | ≤16 | ≥32 | ≤64 | ≥128 |
| Piperacillin/tazobactam | ≤16 | ≥32 | ≤16/4 | ≥128/4 |
| Gentamicin | ≤4 | ≥8 | ≤4 | ≥16 |
| Tobramycin | ≤4 | ≥8 | ≤4 | ≥16 |
| Co-trimoxazole | ≤32 | ≥64 | ≤2/38 | ≥4/76 |
| Ciprofloxacin | ≤1 | ≥8 | ≤1 | ≥4 |
An attempt to detect extended-spectrum β-lactamases (ESβL) of Escherichia coli and Klebsiella spp. was performed by using two Etest strips: one with ceftazidime alone and one with ceftazidime in combination with clavulanic acid (AB BIODISK, Solna, Sweden). Isolates with a reduction of ceftazidime MIC by ≥ 3 two-fold dilutions in the presence of clavulanic acid were considered ESβL-positive [22].
RESULTS
Distribution of bacterial species
The distribution of bacterial species is shown in Table 1.
Sources of the isolates
The sources of the isolates are shown in Table 2.
Susceptibility
The resistance rates among all initial Gram-negative isolates and all initial + repeat isolates and the resistance rates for Escherichia coli, Klebsiella spp., Enterobacter spp., Pseudomonas aeruginosa and Stenotrophomonas maltophilia to the antibiotics are shown in Figure 1.
The resistance rates among all initial Gram-negative isolates and all initial + repeat isolates, and the resitance rates for initial isolates of Escherichia coli, Klebsiella spp., Enterobacter spp., Pseudomonas aeruginosa and Stenotrophomonas maltophilia.
MIC-distributions
The MIC distributions for E. coli, Klebsiella spp., Enterobacter spp, P. aeruginosa and S. maltophilia to selected antibiotics are shown in Figure 2.
MIC distributions of initial isolates for ceftazidime, cefuroxime, ciprofloxacin, imipenem and tobramycin in Escherichia coli, Klebsiella spp. and Enterobacter spp., for ceftazidime, ciprofloxacin, imipenem and tobramycin in P. aeruginosa, and for ceftazidime, ciprofloxacin, co-trimoxazole and tobramycin in S. maltophilia.
University hospital ICUs versus county hospital ICUs
Four of the participating hospitals were tertiary university hospitals (248 initial isolates) and six were primary county hospitals (254 initial isolates). The percentage resistance rates in the university hospital ICUs/county hospital ICUs were: ceftazidime 7.7%/5.9% cefotaxime 14.1%/15.7%, ceftriaxone 21.8%/15.0%, cefuroxime 45.3%/42.9% ciprofloxacin 4.4%/3.9%, co-trimoxazole 13.8%/21.7%, gentamicin 8.1%/14.6%, piperacillin/tazobactam 15.3%/10.6% and tobramycin 8.5%/3.1%.
Extended-spectrum β-lactamases (ESβLs)
One of a total of 128 E. coli isolates and four of a total of 70 Klebsiella spp. isolates had an Etest MIC ratio for ceftazidime and ceftazidime/clavulanic acid indicating the presence of EβBLs.
DISCUSSION
This study showed that of all the consecutively collected initial Gram-negative isolates in Swedish ICUs, 44.1% were resistant to cefuroxime, 14.9% to cefotaxime, 18.5% to ceftriaxone, 17.8% to co-trimoxazole, 20.2% to piperacillin and 12.9% to piperacillin/tazobactam. Such a high antibiotic resistance level among Gramnegative isolates has not, to our knowledge, been previously reported from Scandinavia. Resistance surveillance studies in Sweden have earlier focused mainly on blood isolates, and only low-level antibiotic resistance has been reported in these studies [11–15]. However, a trend towards increasing resistance among Enterobacter spp. and Enterococcus spp. to β-lactam antibiotics could be seen in an investigation of antibiotic resistance during 1993–94 in the ICUs of one Swedish university hospital [16].
The breakpoints for susceptibility and resistance to cefuroxime divide the main MIC populations of E. coli, Klebsiella spp. and E. cloacae into two parts when SRGA breakpoints are used, but not when NCCLS breakpoints are used (Figure 2, Table 3). To circumvent this problem, species-related breakpoints may have to be defined. In contrast to cefuroxime the SRGA and NCCLS breakpoints for susceptibility and resistance to cefotaxime, ceftazidime and ceftriaxone of E. coli and Klebsiela spp. are several dilution steps from the main MIC population, which makes it hard to detect bacteria with acquired low-level resistance. This also supports a change to species-related breakpoints.
The resistance rates for all Gram-negative bacteria to ceftazidime, ciprofloxacin, gentamicin, imipenem and tobramycin were 6.8%, 4.2%, 5.8% 8.6% and 5.8%, respectively (Figure 1). These figures indicate that, of the tested antibiotics, ceftazidime, ciprofloxacin, gentamicin, impipenem and tobramycin are the best empirical treatments for Gram-negative infections in ICUs in Sweden. However, these drugs may select multiresistant Enterococcus faecium, which is an increasing problem in Swedish ICUs [16]. This is in agreement with unpublished results from seven of 10 Swedish ICUs participating in this study shwing that 24% of Enterococcus spp. had decreased susceptibility to ampicillin.
Enterobacter was the third most frequent Gramnegative genus in this study (Table 1). The resistance rates among Enterobacter spp. to ceftazidime, cefotaxime, cefuroxime, piperacillin and piperacillin/tazobactam were 26.5%, 26.5%, 26.5%, 61.8%, 26.5% and 26.5%, respectively, whereas no Enterobacter spp. were resistant to imipenem, gentamicin or ciprofloxacin (1, 2). This resistance is probably due to selection of Enterobacter spp. producing high levels of chormosomal class I β-lactamases, which cause resistance to all second- and third-generation cephalosporins [23–26]. Previous administration of thirdgeneration cephalosporins is more likely to be associated with multiresistant blood isolates of Enterobacter spp., causing a higher mortality rate, than administration of other antibiotics, as shown in a study by Chow et al [27]. The resistance rates for P. aeruginosa were 1.7% for ceftazidime, 6.8% for ciprofloxacin, 6.8% for gentamicin, 8.5% for piperacillin, 8.5% for piperacillin/tazobactam, 13.8% for imipenem and 5.1% for tobramycin (Figure 1). The β-lactamase inhibitor tazobactam in piperacillin/tazobactam is not stable to the class 1 β-lactamases produced by P. aeruginosa or Enterobacter spp., which is confirmed by the same level of resistance for the combination with tazobactam compared to piperacillin alone, as shown in this study (Figure 1).
The frequencies of resistance among all Gramnegative bacteria were higher for most antibiotics in the university hospital ICUs compared with the county hospital ICUs. This may be a result of more extensive antibiotic use in the university hospital ICUs due to a higher prevalence of severe cases causing longer ICU stays.
No analysis was performed in the present study to determine whether the isolates collected caused infection or only reflected colonization of the critically ill patient. However, only a small number of all isolates could be considered opportunistic, and colonization with multiresistant pathogens is very often a prerequisite for infection [28].
The multiple antibiotic-resistant S. maltophilia is of major concern primarily in immunocompromised cancer patients and transplant recipients [29]. Risk analysis has shown that mechanically ventilated ICU patients receiving antibiotics, especially carbapenems are at increased risk of colonization/infection with S. maltophilia [29]. A more liberal indication for isolation of patients colonized with S. maltophilia is now appropriate, and a more judicious use of carbapenems will decrease the selective pressure on multiresistant S. maltophilia.
In conclusion, this study showed a very high frequency of resistance among Gram-negative isolates in Swedish ICUs to cefuroxime, and a rather high frequency of resistance to cefotaxime, ceftriaxone, piperacillin and piperacillin/tazobactam (Figure 1). These drugs cannot be recommended for further use as empirical monotherapy for severe ICU-acquired Gramnegative infections in ICUs in Sweden. Furthermore, these high frequencies of resistance among Gramnegative isolates, not previously reported in Sweden, pose a threat to patients in Swedish ICUs. To reduce the spread of resistant bacterial isolates there is a need for identification of reservoirs and isolation of patients with resistant isolates and, if possible, the eradication of these isolates by means of judicious antibiotic treatment. Antibiotic resistance surveillance programs associated with registration of antibiotic consumption are necessary to promote optimal use of antibiotics, especially in ICUs.
Acknowledgment
This study has previously, in part, been presented at the First European Congress of Chemotherapy, Glasgow, 14–17 May 1996 (abstract T 109). For financial support we would like to thak the Swedish Society for Medical Research, County of Östergötland, Tore Nilsons Fund for Medical Research and MSD Sweden.
We thank Associate Professor Barbro Olsson-Liljequist at the Swedish Institute for Infectious Disease Control for her constructive criticism of this paper.
