Introduction

Acinetobacter baumannii, an aerobic, non-fermentative, Gram-negative coccobacillus, has emerged as an increasingly important nosocomial pathogen. Previously considered as an organism of low virulence, most commonly capable of colonizing than infecting [1, 2], it has become invasive, causing life-threatening infections in hospital patients especially among the critically ill, resulting in a significant morbidity and mortality. It is capable of prolonged environmental survival, and has both inherent and acquired mechanisms of resistance to many available antimicrobial agents [1, 3–6]. Carbapenems are usually the antimicrobial agents of choice for treatment of serious infections caused by multidrug-resistant A. baumannii [5, 7]. However, carbapenem-resistant A. baumannii (CRAB) isolates have been increasingly reported worldwide in recent years [3–6, 8, 9] and are of great importance because they limit the treatment options and may contribute to an increased mortality [10, 11].

Among the wide range of infections caused by A. baumannii, bacteremia is one of the most significant. Along with ventilator associated pneumonia (VAP), it is associated with a mortality of about 40–60% in most studies, and is strongly related to an intensive care unit (ICU) stay [12–16]. Although numerous clinical studies regarding A. baumannii infections have been published, including those from Greek ICUs [17, 18], only a few have focused on CRAB bacteremia [14–22] and none of them exclusively in ICU patients. Thus, information concerning the incidence, the risk factors for and the effect of CRAB bacteremia, compared to carbapenem-susceptible A. baumannii (CSAB), on clinical outcome in ICU patients is still limited.

Following an outbreak of multidrug-resistant A. baumannii in our unit in 2003 [23], an increased incidence of CRAB infections was observed [24]. The purpose of this study was to identify the incidence of and potential risk factors for bacteremia caused by CRAB versus CSAB in our critically ill patients and to determine their effect on patients’ outcome.

Patients and methods

This prospective, observational study was carried out in the ICU of Evangelismos Hospital in Athens, Greece, from September 2004 through January 2006. This is a 25-bed university ICU in a 1000-bed, tertiary-care, teaching hospital for adults. It serves critically ill medical, surgical and trauma patients with the exception of those with acute coronary syndromes and transplantation, managed in special units.

All ICU patients with ICU-acquired bacteremia due to A. baumannii, during the study period were included in the study. The data were prospectively collected and included demographics, diagnostic category classified as medical, surgical, and trauma non surgical, co-morbidities, illness severity, development of acute respiratory distress syndrome (ARDS) and organ dysfunction, use of mechanical ventilation, length of ICU stay, laboratory examinations and antibiotic therapy regimen. The severity of illness was evaluated at ICU admission and at the first day of onset of A. baumannii bacteremia by use of the previously validated APACHE II and SOFA scoring systems [25, 26]. Delta (Δ) APACHE II and Δ SOFA score were also calculated as previously described [27], as the difference in score values between the day of bacteremia and the day of ICU admission. All patients had at least one central venous catheter, a peripheral arterial catheter, and a urinary catheter. Additional exposure to intravascular devices (i.e. pulmonary artery catheter, continuous veno-venous hemofiltration catheter, or intraaortic balloon catheter), was recorded. Other data included previous receipt of immunosuppressive agents, corticosteroid use, and development of bacteremia due to bacteria other than Acinetobacter and/or fungemia during the ICU stay. For patients who had more than one episode of A. baumannii bacteremia, only data for the first episode were analyzed.

Definitions

ICU-acquired A. baumannii bacteremia was defined as bacteremia due to A. baumannii that occurred more than 48 h after ICU admission. Blood culture specimens were ordered by the attending physicians in the presence of signs and symptoms of systemic inflammatory response syndrome (SIRS) [28], or when infection was suspected on clinical rounds. Sources of bacteremia were defined according to the Centers for Disease Control and Prevention criteria [29]. Documentation of more than one source was defined as multiple-source bacteremia.

Carbapenem resistance in A. baumannii was defined as in vitro resistance to imipenem and/or meropenem. The term “recent CRAB ventilator associated pneumonia (VAP)” was defined as the presence of lower respiratory tract infection with CRAB, prior to A. baumannii bacteremia detection. VAP was diagnosed in patients who presented a new or progressive infiltrate on chest radiograph, after being on mechanical ventilation longer than 48 h, purulent bronchial secretions and presence of signs and symptoms of SIRS. In the absence of new infiltrate the diagnosis of ventilator associated tracheobronchitis was made. Samples of bronchial secretions were collected either by bronchoscopy or after insertion of a sterile catheter and quantitative cultures were performed.

Exposure to antimicrobial drugs was defined as antimicrobial therapy given during the ICU stay prior to the first blood sample collection that subsequently revealed A. baumannii. To examine the impact of CRAB on mortality, the empiric antibiotic treatment given by the attending physicians was also recorded. “Appropriate” initial empiric therapy was defined on the basis of in vitro susceptibility data. If a patient received at least one antimicrobial agent to which the A. baumannii strain was susceptible, within 48 h of blood culture collection, the initial antimicrobial therapy was considered appropriate. If none of the antibiotics to which A. baumannii was susceptible were included in the treatment within the 48 h, antimicrobial therapy was considered inappropriate. Mortality was assessed at the time of discharge from the ICU.

Microbiological methods

Blood cultures were performed using the BACTEC 9240 system (Becton–Dickinson Sparks, MD, USA). Identification of the blood isolates was performed by standard methodology and by using the commercially available ID panels of the semi-automated WIDER I system (Fransisco Sorie Melguiso SA) and VITEK2 Gram-negative and -positive ID cards (bioMERIEUX, Marcy l’Etoile, France). Susceptibility of the isolates by determination of the minimum inhibitory concentrations (MICs) to different antimicrobial agents was performed by the VITEK 2 system. Resistance of A. baumannii isolates to imipenem and meropenem was verified by determination of minimum inhibitory concentrations (MICs) using Etest (AB Biodisk, Solna, Sweden) strips. Interpretation breakpoints of ≤4 mg/L as susceptible, 8 mg/L as intermediately susceptible, and ≥16 mg/L as resistant were used according to the Clinical and Laboratory Standards Institute recommendations [30]. Intermediate susceptibility was considered as resistance.

Statistical analysis

Continuous variables were expressed as mean value ± standard deviation or as median and inter-quartile range when they were not normally distributed. Comparisons between patients with CSAB and CRAB bacteremia were made by the Student’s t test or Mann–Whitney U test for continuous variables as indicated, and the chi-squared test or Fisher’s exact test for categorical variables, when appropriate. In identifying the independent risk factors for development of carbapenem resistance and also for mortality, backward stepwise logistic regression analyses were performed to control for the effects of confounding factors. The variables initially entered into the models were those that were statistically significant in the univariate analyses. A p value less than 0.05 was considered statistically significant. All statistical analyses were performed using the SPSS version 11.5 for Windows (SPSS Inc. Chicago, IL, USA).

Results

During the study period, among 842 consecutively admitted ICU patients with an ICU stay of more than 48 h, 96 (11.4%) developed bacteremia with A. baumannii: 66 (68.5%) due to CSAB and 30 (31.3%) due to CRAB. The mean (±SD) patient age was 57.1 ± 18.9 years. The median length of ICU stay was 30.5 days (IQR 16.0–51.0 days) and the time from ICU admission to A. baumannii bacteremia was 12.0 days (median, IQR 6.0–19.8 days).

Respiratory tract infection was the most common source of A. baumannii bacteremia either resistant or susceptible to carbapenems, observed in 49 (51.0%) patients, followed by central venous catheter in 15 (15.6%) patients, whereas multiple sources including respiratory tract, urine, surgical wounds, or cerebrospinal fluid infections were recognized in 18 patients. A focus of infection was not identified in 14 patients (Table 1).

Table 1 Comparison by univariate analysis of the clinical characteristics of patients with Acinetobacter baumannii bacteremia with and without carbapenem resistance
Full size table

Risk factors for development of bacteremia due to CRAB among patients with A. baumannii bacteremia

Results of the univariate analysis of potential risk factors for CRAB among the patients with A. baumannii bacteremia are shown in Tables 1 and 2. No differences in terms of gender, age, co-morbidities, days of mechanical ventilation and length of ICU stay prior to A. baumannii bacteremia were found between patients with CSAB and those with CRAB bacteremia. Compared to those with CSAB, patients with CRAB bacteremia were significantly more likely to have lower initial severity of illness at admission, as indicated from a lower value of APACHE II score (16.3 ± 7.2 vs. 20.0 ± 6.8, p = 0.018). An increasing illness severity during the ICU stay, as indicated by an increased Δ APACHE II score, between admission and the day of the first positive blood culture for A. baumannii bacteremia was associated with CRAB (1.6 ± 5.8 vs. −1.5 ± 6.0, p = 0.018). Patients with CRAB bacteremia were more likely to have additional intravascular devices compared to those with CSAB (70.0 vs. 42.4%, p = 0.012). In addition, patients with CRAB bacteremia had the respiratory system most frequently as the source of bacteremia compared to those with CSAB (66.7 vs. 43.9%, p = 0.039), followed by multiple-sources (24.2 vs. 6.7%, p = 0.041), Table 1. Specifically, recent VAP due to CRAB was strongly associated with development of CRAB bacteremia (3.0 vs. 43.0%, p < 0.001). As expected, administration of appropriate empiric antimicrobial therapy was most commonly given to patients with CSAB bacteremia than to those with a CRAB one (75.8 vs. 56.7%), although this difference was marginally significant (p = 0.059).

Table 2 Antimicrobial agents given from ICU admission until A. baumannii bacteremia day (Mann–Whitney U test)
Full size table

Exposure to antimicrobial agents before the first positive blood culture for A. baumannii is shown in Table 2. The duration of prior use of carbapenems (p = 0.046), linezolid (p = 0.015) and colistin (p < 0.001) was associated with CRAB bacteremia acquisition.

By multivariate regression analysis, the best independent risk factors for CRAB bacteremia acquisition were recent VAP with CRAB (OR 16.74, 95% CI 3.16–88.79, p = 0.001), presence of additional intravascular devices (OR 3.93, 95% CI 1.19–13.0, p = 0.025), and APACHE II score on admission (OR 0.89, 95% CI 0.80–0.97, p = 0.009), Table 3. Prior exposure to carbapenems, linezolid, and colistin was not significantly associated with CRAB acquisition in the multivariate analysis.

Table 3 MICs of imipenem and meropenem for carbapenem susceptible and resistant A. baumannii isolates
Full size table

Resistance of A. baumannii isolates to carbapenems

MICs of imipenem and meropenem for CRAB and CSAB isolates are presented in Table 3.

No isolate presented resistance to colistin.

Clinical outcome

The all-cause mortality rate for patients with A. baumannii bacteremia was 45.8%. By univariate analysis APACHE II score and SOFA score on the day of bacteremia were significantly higher in non-survivors compared to survivors (20.6 ± 6.6 vs. 15.8 ± 5.1, and 9.4 ± 3.3 vs. 6.3 ± 2.6, p < 0.001, respectively). Also patients who died were older (62 ± 18 vs. 53 ± 19, p = 0.027) and had increased WBC count and lower values of serum albumin on the day of A. baumannii bacteremia compared to those who survived [14.1 × 103 ± 9 × 103 vs. 11.1 × 103 ± 4.6 × 103, p = 0.047, and 2.9 ± 0.5 vs. 3.1 ± 0.5, p = 0.019, respectively (mean ± SD)]. In addition, patients with A. baumannii bacteremia who died were more likely to have additional intravascular devices (61.2 vs. 38.8%, p = 0.002) and ARDS (77.8 vs. 22.2%, p = 0.03) than patients who survived. There was no statistically significant difference in mortality between patients with CRAB and CSAB bacteremia (43.3 vs. 46.9% respectively, p = 0.740), Table 1.

Multivariate analysis using a logistic regression model including the variables associated significantly with mortality in the univariate analysis, showed that the best factors independently associated with mortality were the severity of organ failure, as estimated by the SOFA score (OR 1.42, 95% CI 1.20–1.67, p = 0.001) and increased WBC count (OR 1.09, 95% CI 1.01–1.19, p = 0.036), both at the onset of A. baumannii bacteremia (Table 4).

Table 4 Independent risk factors for carbapenem-resistant Acinetobacter baumannii acquisition and for mortality, among patients with A. baumannii bacteremia (n = 96); multivariate analyses
Full size table

Discussion

The main findings of the present study are that among ICU patients with A. baumannii bacteremia, recent VAP due to CRAB and presence of excess intravascular devices were the best independent risk factors for CRAB bacteremia development. No significant difference in mortality between patients with CRAB and CSAB bacteremia was demonstrated. The severity of organ failure and increased WBC count, both at A. baumannii bacteremia onset, were independently associated with mortality.

Acinetobacter baumannii frequently causes respiratory infections in mechanically ventilated patients [14, 15, 21]. Indeed, in the present study, the most frequent source of A. baumannii bacteremia was the respiratory tract. Interestingly, as shown in Table 1, patients with CRAB bacteremia were more likely to have a respiratory tract infection as the source of bacteremia, than those with a CSAB one (66.7 vs. 44%, p = 0.039). More importantly, multivariate analysis demonstrated that recent VAP due to CRAB was the best independent risk factor for CRAB bacteremia, a finding that represents an addition to the literature in terms of risk assessment for CRAB bacteremia in ICU patients.

Besides deficiencies in the implementation of infection control guidelines resulting in patient to patient transmission, CRAB occurrence may be facilitated by the selection pressure of previous antimicrobial use. A literature review [31] revealed that multidrug-resistant A. baumannii acquisition is related to carbapenems and third-generation cephalosporins prior use. However, in three recent studies, no association between CRAB acquisition and specific antibiotic classes was demonstrated, probably indicating that any previous antibiotic agent use could be implicated by ablating patients’ pre-existing microflora [10, 21, 32]. The present study has shown, by univariate analysis, that longer exposure to carbapenems, linezolid and colistin was associated with CRAB bacteremia development. These antibiotics were highly used within our ICU during the study period because of the predominance of multidrug-resistant pathogens [24, 33]. Specifically, colistin had been given almost exclusively to patients who subsequently developed CRAB bacteremia and in five of them for more than 4 weeks because of persistence of infections due to Gram-negative bacteria susceptible only to colistin. However, multivariate analysis did not reveal an independent association of these antimicrobial agents with the development of CRAB bacteremia and this is in accordance with the above-mentioned studies [10, 21, 32]. A possible explanation of the different results of the multivariate analysis is that these antimicrobial agents had been given to patients by the attending physicians because of VAP and probably have a confounding role rather than causative. Notably, a positive correlation between recent VAP due to CRAB and colistin administration before CRAB bacteremia onset was found (p < 0.001) (data not shown).

Surprisingly, the risk for CRAB bacteremia acquisition was greater in patients with lower APACHE II score at ICU admission compared to those with CSAB bacteremia (Tables 1, 3). Interestingly, this is in agreement with a previous study showing that ICU patients infected or colonized with CRAB isolates belonged to group 1 of the McCabe classification (chronic or curable disease) in a significantly greater proportion [34]. It should be noted that the illness severity at admission although it is a predictor of mortality, it may not necessarily be a predictor of infection with antibiotic-resistant organisms [35]. However, these “less sick” patients with CRAB bacteremia, as compared to those with CSAB, had a deteriorating course during the ICU stay, as shown by a positive Δ APACHE II score, probably reflecting increased number of medical and nursing interventions and more invasive management. In fact, in accordance with previous studies [21, 34], the presence of excess intravascular devices was independently associated with the risk of CRAB bacteremia acquisition (Tables 1, 3).

Data on the impact of CRAB bacteremia, compared to that due to CSAB, on clinical outcome in ICU patients are limited. Mortality rate for patients with multidrug-resistant Acinetobacter infections was not significantly higher from susceptible references in two studies in hospital [36] and in ICU patients [37]. Largely consistent with those studies, a relation between CRAB bacteremia and increased risk of death, among patients with A. baumannii bacteremia, was not found in the present study. On the contrary, multidrug-resistant A. baumannii acquisition was associated with adverse outcome in the study by Abbo et al. [11], indicating the need for further studies.

The reported differences on patients’ outcome could be related, at least in part, to the antimicrobial treatment given. However, antibiotic regimens for extensively resistant A. baumannii bacteremia did not have a significant influence on patient outcome in a recent study [15] and this accords with the findings of the present study. Although not plausible at first, some explanations might be suggested. First, according to the definition of the “appropriate initial therapy”, the use of a carbapenem was considered “inappropriate” in patients with CRAB bacteremia. However, there is evidence that carbapenem use along with an aminoglycoside or ampicillin/sulbactam or colistin seemed to be effective in the treatment of A. baumannii infections, including those caused by CRAB strains [38]. Second, the majority of our patients with CRAB bacteremia, as shown in Table 1, received appropriate initial antimicrobial therapy and additionally, immediately after the blood culture results became available, all patients received colistin. Therefore, the ability of this study to elucidate this issue is rather limited.

Remarkably, among the patients who developed CRAB bacteremia, ten had already been on colistin on the day of bacteremia. This probably manifests the difficulty of this antibiotic to control effectively the primary site of infection, despite the fact that the pathogen was susceptible in vitro.

A limitation of this study was that a Pulse Field Gel Electrophoresis was not performed to show the similarity between CRAB strains isolated from blood and from the respiratory specimens. However, the antibiotic susceptibility data presented almost identical A. baumannii phenotypes indicating that the VAP caused the blood stream infection in these cases. Importantly, according to data provided from our Microbiology Department, blaOXA-58 carrying A. baumannii isolates were detected from all CRAB isolates, largely consistent with findings from other Greek hospitals [39, 40].

In conclusion, among our ICU patients with A. baumannii bacteremia, CRAB isolates frequently are implicated. Recent VAP due to CRAB and excess use of intravascular devices were the most important risk factors for CRAB bacteremia development. Patients with CRAB, although had a lower illness severity on admission, compared to those with CSAB, had a deteriorating ICU course. Mortality was not different between patients with CSAB and CRAB bacteremia. Finally, severity of organ failure and increased WBC count at A. baumannii bacteremia onset were independently associated with ICU mortality.