Introduction

Acinetobacter baumannii is a gram-negative bacilli that can survive for prolonged periods in the environment and cause urinary tract infections, skin and soft-tissue infections, pneumonia, bacteremia, meningitis, and osteomyelitis. This organism has emerged as a significant nosocomial pathogen in hospitalized patients worldwide and is associated with high mortality rates [1–3]. The SCOPE surveillance study, a large prospective study of nosocomial bloodstream infections (BSIs) conducted in the United States from 2000–2001, ranked A. baumannii as the tenth leading cause of BSIs in the country (0.6 BSIs per 10,000 admissions; 1.3% of all monomicrobial BSIs). However, A. baumannii ranked third in crude mortality (34.0%), behind only Candida sp. (39.2%) and Pseudomonas aeruginosa (38.7%) [4]. A. baumannii is also a common cause of late-onset ventilator-associated pneumonia (>5 days after admission) and is associated with a higher mortality (30–75%) than other bacteria [5–7]. The National Nosocomial Infections Surveillance System determined the epidemiology of gram-negative bacilli in intensive care units (ICUs) in the United States for the most common types of nosocomial infections during the period 1986–2003. Acinetobacter species ranked fourth (6.9%) in prevalence among gram-negative organisms causing pneumonia in ICUs. However, it was the only gram-negative organism with a significant increase in prevalence from 1986 to 2003 [8].

Multidrug-resistant (MDR) infections are a growing problem and have been reported worldwide. A. baumannii has been become increasingly difficult to treat because of the emergence of strains resistant to all drugs or all but one commonly prescribed antibiotics. These MDR strains are sometimes only susceptible to polymyxins (colistin and polymyxin B), a class of antibiotics that has not been used commonly for decades and is more toxic than the commonly prescribed antibiotics [1–2, 9, 10].

Our clinical microbiology laboratory suspected that our institution was experiencing an increase in A. baumannii infections and was concerned about resistance. A cross-sectional study was conducted at our institution with two objectives: first, to determine the A. baumannii antibiogram for 2006, and second, to assess the appropriateness of antibiotic therapy for A. baumannii infections.

Methods

A cross-sectional study was conducted using electronic pharmacy and microbiology laboratory patient data from January 1 2006 through December 31 2006 from a tertiary care hospital in Augusta, Georgia. A computer-generated list was created from PathNet (Cerner, Kansas City, MO, USA) to identify all patients by medical record number (MRN) with an A. baumannii isolate. From the list of MRNs, a database was created from the microbiology laboratory data, which included the MRN, date of admission, site of isolate, date of collection, and date of final verification. Next, a computer-generated database was created from PharmNet (Cerner, Kansas City, MO, USA) to gather the antibiotic data for the same list of patients identified in PathNet.

An isolate was included in the final database if it was the first A. baumannii isolate per site of infection per admission. Duplicate isolates were counted if they were collected >14 days after the previous isolate during the same admission. Isolates were from all adult (age ≥13 years) inpatient admissions with a positive culture for A. baumannii. Isolates were collected from both sterile and nonsterile sites. In addition, patient demographic data, including age, gender, and hospital unit, were collected. Infections were considered to be nosocomial if the isolate was collected >48 h after hospital admission.

Antibiotics susceptibility testing

Minimum inhibitory concentrations (MICs) were determined using a commercial broth microdilution (MicroScan negative MIC type 30 panel on a WalkAway instrument, Dade Behring, West Sacramento, CA, USA). Testing was conducted for the following antibiotics: amikacin, ampicillin-sulbactam, cefepime, ceftazidime, ciprofloxacin, gentamicin, imipenem, levofloxacin, meropenem, tetracycline, ticarcillin-clavulanate, tobramycin, and trimethoprim-sulfamethoxazole.

In this study, we reported MDR as nonsusceptibility, which was defined as the sum of the intermediate and resistant isolates. Antibiotic selection was considered to be inappropriate if the infection was not being effectively treated at the time when both the A. baumannii isolate was identified and its antibiotic susceptibility pattern was reported.

Data analysis

The two databases were linked electronically by the patient’s MRN in Access (Microsoft, Redmond, WA, USA). The data was analyzed using SAS version 9.1 (SAS Institute Inc., Cary, NC, USA).

Results

A total of 129 isolates were identified using the inclusion criteria from January 1 2006 through December 31 2006. Table 1 lists the sites of infection. Eighty-one (62.8%) isolates were from respiratory cultures. A respiratory culture was defined as an isolate collected via bronchoalveolar lavage, nasopharyngeal aspirate, sputum, throat swab, and tracheostomy aspirate. The mean time to final culture verification was 6.1 days.

Table 1 Site of infection for 129 Acinetobacter baumannii isolates in 2006
Full size table

The isolates came from 80 patients, who had a combined total of 86 admissions. The mean length of stay for the 86 admissions was 28.6 days. Sixty of the 86 (69.8%) admissions included one day or more in the ICU. Sixty-two (72.1%) of the 86 admissions contained an isolate that was classified as a nosocomial infection.

Table 2 shows the institution’s antibiogram for A. baumannii in 2006. A. baumannii was highly susceptible to imipenem (97.7%) and meropenem (95.3%). Among the aminoglycosides, A. baumannii was highly susceptible to amikacin (93.6%) and tobramycin (92.3%), but had reduced susceptibility to gentamicin (40.5%). The isolates were found to have reduced susceptibility to ampicillin-sulbactam (46.5%) and tetracycline (41.4%).

Table 2 Antibiogram for 129 A. baumannii isolates in 20061
Full size table

Multidrug nonsusceptibility

Only 35 (27.1%%) isolates were susceptible to all thirteen of the tested antibiotics. Six (4.7%) isolates were nonsusceptible to only one antibiotic. Eighty-eight (68.2%) isolates were nonsusceptible to three or more antibiotics. No isolates were nonsusceptible to all thirteen antibiotics (Table 3).

Table 3 Multidrug nonsusceptibility of 129 A. baumannii isolates, 2006
Full size table

Appropriate antibiotic selection

One hundred twenty-nine antibiotic regimen were identified from the pharmacy database. Out of the 129 regimens, 76 (58.9%) regimens were susceptible against the isolates based on their susceptibility patterns. Of 53 regimens that were nonsusceptible, the most common antibiotics ordered were levofloxacin [34 (64.2%)], cefepime [6 (11.3%)], ampicillin-sulbactam [4 (7.6%)], and gentamicin [4 (7.6%)].

A total of 67 regimens were identified when controlling for empiric use. Empiric regimens were defined as regimens that were administered for 6 days or less. This definition was based on the mean time to final culture verification (6.1 days) identified in the study. Out of the 67 regimens, 41 (61.2%) regimens were susceptible against the isolates based on their in vitro susceptibility patterns. Of 26 regimens that were nonsusceptible, the most common antibiotics administered were levofloxacin [20 (76.9%)] and cefepime [3 (11.5%)].

Discussion

Overall, the A. baumannii isolates at our institution were highly susceptible to carbapenems, with 97.7% of the isolates susceptible to imipenem and 95.3% susceptible to meropenem. The potential for the emergence of carbapenem resistance is a concern because A. baumannii isolates are frequently resistant to aminoglycosides, fluoroquinolones, and third-generation cephalosporins. Recent epidemiologic studies from around the world have reported A. baumannii isolates that are resistant to carbapenems [11–13].

We also found the isolates to be susceptible in vitro to two aminoglycosides, amikacin (93.6%) and tobramycin (92.3%). Gentamicin had reduced susceptibility (40.5%) for unknown reasons. Despite the high susceptibility to carbapenems and aminoglycosides, we found 88 (68.2%) isolates that were nonsusceptible to three or more of the antibiotics tested.

Antibiotic selection was only appropriate in 58.9% of the regimens and 61.2% of the regimens when adjusting for empiric use (regimens of duration >6 days). Using inappropriate antibiotics can result in treatment failures. These results emphasize the importance of monitoring cultures and their susceptibilities. Ibrahim et al. reported a reduction in fatality in patients receiving appropriate antibiotics regimens (61.9% vs. 28.4%, RR=2.18) [14]. Zaragoza et al. conducted a study on inadequate empirical antibiotics treatment in ICU patients with BSIs. The study reported that 23.5% of patients with a BSI received inappropriate empiric antibiotics treatment. The study failed to demonstrate a relationship between mortality and the administration of inappropriate empiric antibiotics in ICU patients with bacteremia [15]. However, inappropriate therapy can lead to increased morbidity, medication cost, length of stay, and, potentially, mortality.

The limitations of the present study include its study design, possible concurrent infections, and not obtaining the data by diagnosis-related groups (DRG). A cross-sectional study design only provides descriptive statistics. However, cross-sectional studies do provide information about the topic for the specified period of time. We obtained all data electronically. By conducting a retrospective chart review, we could have gathered data on mortality, severity of disease, colonization, and concurrent infections. Our percentage of appropriate antibiotic selection may have improved if we could have proved that some antibiotics were being used for a separate infection. Lastly, we do not know if these patients were primarily in the hospital for an A. baumannii infection. Obtaining the data by DRG would have helped clarify this. Our mean length of stay was 28.6 days. Part of the length stay could be due to other diseases, which could be identified by a DRG.

Our study found a high prevalence of MDR A. baumannii at our institution. However, A. baumannii remains highly susceptible to the carbapenems and aminoglycosides. We encourage our practitioners to analyze the susceptibility pattern of each isolate when ordering an antibiotic regimen, which will help increase our rate of appropriate antibiotic selection.