Antibiotic Susceptibility Among Aerobic Gram-negative Bacilli in Intensive Care Units in 5 European Countries FREE

More than 20% of patients admitted to European intensive care units (ICUs) develop an ICU-acquired infection.¹ A high prevalence of decreased antibiotic susceptibility among gram-negative bacilli has been reported from ICU patients in France,^2- 3 Belgium,⁴ Germany,⁵ and the Netherlands during 1990 and 1991,⁶ the United States between 1990 and 1993,⁷ and Belgium⁸ and Sweden⁹ during 1994 and 1995.

The National Nosocomial Infections Surveillance System of the Centers for Disease Control and Prevention monitors secular trends of antibiotic resistance rates of nosocomial pathogens in US hospitals.^10- 11 No corresponding international surveillance system is operational in Europe. The aim of this study was to conduct a cross-country comparison of the incidence of antibiotic resistance among aerobic gram-negative bacilli from patients in ICUs in Belgium, France, Portugal, Spain, and Sweden using determination of minimum inhibitory concentration.

Multicenter susceptibility testing was performed using aerobic gram-negative bacilli isolated from patients admitted to ICUs between June 1994 and June 1995. The number of hospitals included was 18 in Belgium, 40 in France, 20 in Portugal, 30 in Spain, and 10 in Sweden, and these hospitals were chosen to represent different geographic areas of each country. Intensive care unit–acquired infections and ICU mortality have been the topic of a large, international, European epidemiological study.¹ Consecutive specimens collected on the basis of clinical indications were cultured and tested. A total of 11,249 gram-negative bacterial strains were isolated from 7308 ICU patients. Of these, 9166 were initial isolates (first recovered during ICU admission) and 2083 were repeat isolates of the same species from the same patients. Repeat isolates were not included in this analysis. The number of initial isolates and patients in the various countries were as follows: Belgium, 1435 isolates from 1029 patients; France, 2988 isolates from 2205 patients; Portugal, 2067 isolates from 1882 patients; Spain, 2174 isolates from 1845 patients; and Sweden, 502 isolates from 347 patients; for a total of 9166 isolates from 7308 patients in all countries. Decreased antibiotic susceptibility was analyzed for pathogens collected from all sites and for blood isolates alone.

Minimum inhibitory concentrations for amikacin (except in Sweden), ceftazidime, ceftriaxone, ciprofloxacin, gentamicin, imipenem, piperacillin, and piperacillin-tazobactam were determined using the E test (AB BIODISK, Solna, Sweden).¹² All laboratories performed quality control using reference strains of Escherichia coli ATCC 35218 and Pseudomonas aeruginosa ATCC 27853 to confirm consistency of materials, methods, and results. Decreased susceptibility (ie, resistant and intermediate categories) was defined using the minimum inhibitory concentration break points recommended by the National Committee for Clinical Laboratory Standards.¹³ The break points for susceptible and resistant categories were as follows: imipenem, 4 mg/L or less and 16 mg/L or more; ceftazidime, 8 mg/L or less and 32 mg/L or more; ceftriaxone, 8 mg/L or less and 64 mg/L or more; piperacillin, 16 mg/L or less and 128 mg/L or more (except P aeruginosa, ≤64 mg/L and ≥128 mg/L); piperacillin-tazobactam, 16/4 mg/L or less and 128/4 mg/L or more (except P aeruginosa, ≤64/4 mg/L and ≥128/4 mg/L); gentamicin, 4 mg/L or less and 16 mg/L or more; amikacin, 16 mg/L or less and 64 mg/L or more; and ciprofloxacin, 1 mg/L or less and 4 mg/L or more.¹³

The main sources of isolates were respiratory tract (42%), urine (26%), blood (14%), abdomen (11%), and skin and soft tissue (7%).

The distribution of bacterial species is shown in Table 1. The most frequently isolated organisms in all countries were Enterobacteriaceae (59%) followed byP aeruginosa (24%). The most frequent species of Enterobacteriaceae were E coli, Enterobacter species, Klebsiella species, Serratia species, Proteus species, Morganella morganii, and Citrobacter species. The species included in Table 1 constituted 97% of the total strains in Belgium, 96% in France, 98% in Portugal, 98% in Spain, and 95% in Sweden.

The incidence of decreased antibiotic susceptibility by species and country is shown in Table 2, indicating significant differences for some drug-pathogen combinations between countries.

In summary, we found that the percentage of decreased antibiotic susceptibility across all species and drugs was highest in Portuguese ICUs, followed by French, Spanish, Belgian, and Swedish ICUs. The highest incidence of resistance was seen in all countries among P aeruginosa, Enterobacter species, Acinetobacter species, and Stenotrophomonas maltophilia. In Portugal and France, there also was a high (20%-34%) level of decreased susceptibility to ceftazidime among Klebsiella species. The most active agents were amikacin and imipenem (Table 2). This is similar to findings from a large North American ICU study conducted between 1990 and 1993 of antibiotic resistance among gram-negative isolates.⁷

The high incidence of decreased susceptibility among gram-negative bacteria seen in France (Table 2) was similar to the findings of a French study performed during 1991.³ The overall resistance to most antibiotics found in Belgium⁸ (Table 2) was similar to that found in 1990 in studies performed in Belgium,⁴ Germany,⁵ and the Netherlands.⁶ However, because the methods of this and previous studies were different, a direct comparison of the findings is difficult. The high bacterial inoculum (10⁸ colony-forming units [CFUs] per milliliter) provided by E test that was used in this study is a sensitive tool for detecting resistant subpopulations that may cause development of resistance during therapy. In previous studies, a microdilution method with a lower inoculum (10⁴ CFUs/mL)^3- 6 was used that may not detect resistant subpopulations as easily as the E test.

The lower resistance levels found in Swedish ICUs compared with Belgium, France, Portugal, and Spain (Table 2) may be due to several different factors. These include prudent use of antibiotics, lower total antibiotic administration, and better compliance with infection control measures, such as hand disinfection and barrier precautions by health care workers.

Because our data represent all consecutive gram-negative isolates initially recovered from patients in the participating ICUs, they include pathogens isolated from patients with community- and hospital-acquired infections. No analysis was done to determine whether the isolates tested caused infection or only colonized these critically ill patients. The isolates from specimens other than blood may reflect colonization; however, colonization is often a prerequisite for infection. We found similarly high levels of decreased susceptibility among blood isolates compared with isolates from all sites (Table 2).

Bacterial resistance to ceftazidime is a serious emerging problem. Approximately 40% of Enterobacter cloacae were resistant to ceftazidime, according to studies of ICU isolates in the United States between 1987 and 1991¹⁰ and between 1994 and 1995,¹⁴ which corresponds to the findings in our study (Table 2). This is probably related to production of stably derepressed chromosomal class-1 β-lactamase, which hydrolyzes β-lactam antibiotics other than carbapenems.¹⁵ Previous use of third-generation cephalosporins is more likely to cause the selection of resistance to β-lactams in blood isolates of Enterobacter species, which is associated with higher mortality.¹⁶ An increased ciprofloxacin resistance among Enterobacter species was seen in Belgium and France (Table 2) compared with previous studies performed in 1990⁴ and 1991.³

In Portugal and France, 36% and 26%, respectively, of Klebsiella pneumoniae showed decreased susceptibility to ceftazidime (data not shown). This was probably caused by the production of extended-spectrum β-lactamases^15,17 and is concordant with a study by Livermore and Yuan.¹⁸ According to Jarlier et al,³ decreased susceptibility to ceftazidime among K pneumoniae in French ICUs was 36% in 1991. In the United States, resistance to ceftazidime among K pneumoniae in ICU isolates increased from 3.6% in 1990 to 14.4% in 1993.⁷ Markedly lower resistance rates of Klebsiella species to all β-lactam antibiotics was observed in this study in Belgium compared with a previous study performed in 1990 (Table 2).

The lower susceptibility of P aeruginosa to imipenem compared with ceftazidime was consistent across all 5 countries in this study. The mechanism of resistance to imipenem among P aeruginosa (mainly decreased penetration through outer membrane) is well investigated and may occur more readily than development of resistance to ceftazidime (mainly related to hyperproduction of class 1 β-lactamases).¹⁵

Acinetobacter species (Table 2) and S maltophilia (data not shown) were the most resistant pathogens to all investigated drugs. Although not the most virulent gram-negative pathogen, Acinetobacter is an increasingly infectious threat, especially for patients receiving broad-spectrum antimicrobial therapy and requiring life support.^19- 21 A Spanish study²⁰ has shown that Acinetobacter isolates, usually acquired in the ICU, are multiresistant and may cause severe infections associated with a high mortality rate. Riley et al²¹ recently described the failure to stop the spread of gentamicin-resistant Acinetobacter baumannii in an Australian ICU despite infection control measures.

The high levels of decreased susceptibility among most species of gram-negative bacilli to ceftazidime, ceftriaxone, piperacillin, piperacillin-tazobactam, gentamicin, and ciprofloxacin in Belgium, Portugal, France, and Spain (Table 2) further limits use of these drugs for empirical monotherapy. Spread of multiresistant bacteria in the ICU can be minimized with infection control measures such as rigorous hand disinfection between patient contacts, barrier precautions, and isolation of patients infected with resistant isolates. We suggest that hospitals should have an active program for online antibiotic resistance surveillance of common drugs, using quantitatively accurate minimum inhibitory concentration methods, to constantly evaluate antibiotic administration and pharmacy formulary options. These measures have the potential to minimize morbidity and mortality associated with infections caused by multiresistant bacteria in critically ill patients.^18,22- 25 However, studies to develop and evaluate the most effective strategies to minimize selection and spread of resistant organisms in the ICU are sorely needed, as are investigations to better quantify their clinical and economic consequences.