Acinetobacter is emerging as an important pathogen in traditional and nontraditional healthcare settings. Its ability to infected healthy hosts and its propensity to develop antimicrobial drug resistance have caused concern among the infectious diseases community. Our study assessed the clinical outcomes of patients infected with MDR Acinetobacter compared with outcomes of patients infected with susceptible Acinetobacter strains and patients without Acinetobacter infections among a large cohort. We demonstrated that patients infected with MDR strains of Acinetobacter have longer lengths of stay in both the hospital and ICU than patients infected with drug-susceptible Acinetobacter and patients without Acinetobacter infection when we controlled for severity of illness. We found a trend toward increased mortality rates among patients with MDR Acinetobacter infection. However, the difference was not statistically significant when we controlled for severity of illness.
According to NNIS, Acinetobacter
species caused 7% of ICU healthcare-associated pneumonias in 2003 compared with 4% in 1986 (p<0.001) (4
). The proportion of ICU healthcare-associated urinary tract infections and surgical site infections caused by Acinetobacter
also increased significantly from 1986 to 2003 (p<0.001) (4
). Furthermore, the proportions of Acinetobacter
isolates reported to NNIS that were resistant to ceftazidime, amikacin, and imipenem all increased significantly during that period (p<0.001). Healthcare-associated outbreaks of MDR Acinetobacter
infection have been reported in Asia, Europe, North America, and among US service members injured in the Middle East (7
). These findings have brought control of MDR Acinetobacter
infections to the forefront of discussion.
Investigating the effect of multidrug resistance on clinical outcomes presents multiple methodologic challenges that have been explicitly addressed in our study design. Confounding risk factors associated with mortality rates and antimicrobial drug resistance, such as age, severity of illness, and underlying disease (18
) must be controlled for in the study design or analysis. Our results differ from those of researchers who examined outcomes of Acinetobacter
infections without controlling for these confounders (14
), which makes their findings difficult to interpret. We assessed and controlled for severity of illness and underlying disease with 2 measurements: the APACHE score, which included age, physiologic parameters, and selected underlying diseases; and a separate Charlson index, which included a broader range of underlying diseases. Both measurements have been validated, although the APACHE score has only been studied in its original form (19
). Because patients infected with Acinetobacter
have worse clinical outcomes than those who are colonized with the organism (11
), we separated Acinetobacter
infection from colonization on the basis of standardized, validated Centers for Disease Control and Prevention (Atlanta, GA, USA) NNIS definitions for nosocomial infection (16
) and applied them uniformly to MDR Acinetobacter
and susceptible references. We compared outcomes of MDR Acinetobacter
infections with those of 2 reference groups and showed an association of MDR Acinetobacter
infection with both increased hospital and ICU lengths of stay, regardless of the reference group selected. As one would predict on the basis of results of a study by Kaye et al., the effect of multidrug resistance was greater compared with uninfected than susceptible references (15
Because of the lack of a standard definition for multidrug resistance in the literature, we defined multidrug resistance as resistance to all or all but 1 antimicrobial drug class commonly prescribed for treatment of patients with gram-negative infections, with the exclusion of polymyxins (26
). This definition has 2 advantages. First, it is a strict standard and is readily accepted by clinicians as representative of multidrug resistance. Second, it allows for a clear distinction between susceptible and MDR Acinetobacter
strains because we excluded isolates that showed intermediate resistance (strains resistant to all but 2 commonly prescribed antimicrobial drug classes).
The association of MDR Acinetobacter
infections with worse clinical outcomes could be related to discordant empiric antimicrobial drug therapy. Previous studies that examined the effects of delayed concordant antimicrobial therapy on patient outcomes have shown conflicting results (27
). We examined this issue and found that patients with MDR Acinetobacter
infections who received discordant empiric antimicrobial drug therapy were not more likely to die or have a longer hospital length of stay than patients who received concordant empiric drug therapy; however, they were more likely to have a longer ICU length of stay. On the basis of these results, to what extent discordant empiric antimicrobial drug therapy affects clinical outcomes of MDR Acinetobacter
infection is not clear.
Determining optimal infection control approaches to MDR Acinetobacter
has been complicated by the lack of agreement on the clinical significance of Acinetobacter
infections. The Hospital Infection Control Practices Advisory Committee guideline for isolation precautions in hospitals recommends targeting increased infection control efforts toward “resistant bacteria judged by the infection control program to be of special clinical and epidemiologic significance” (30
). We found that MDR Acinetobacter
infections are independently associated with increased hospital and ICU lengths of stay. This finding, combined with increased risk for in-hospital transmission of the organism (31
), supports recommendations to implement aggressive control measures to limit the transmission of MDR Acinetobacter
in healthcare settings.
Several limitations of this study merit discussion. Because of the lack of available data to calculate a standard APACHE III score for non-ICU patients, we modified the APACHE III score by excluding variables that were unavailable for non-ICU patients. However, our findings support the validity of this scoring system as a measure for severity of underlying illness. Mean APACHE scores were higher in MDR Acinetobacter
–infected patients than in both reference groups and progressed stepwise from no infection to MDR infection. These findings are expected because drug-resistant infections reportedly occur in sicker patients (24
). Univariate analysis showed that a modified APACHE score was also associated with mortality rates (p<0.001), which further supports its validity as a measure of illness severity.
The lack of available reference patients with similar exposure times to several of the case-patients (5 susceptible references and 7 uninfected references) was a second limitation because we were obligated to exclude unmatched MDR Acinetobacter–infected patients from our analysis. These excluded patients typically had prolonged exposure times and tended to have long hospital and ICU lengths of stay after infection. Exclusion of these patients decreased the power of our study and likely biased our results toward showing no difference in hospital or ICU length of stay between the groups. Finally, the lack of a difference in mortality rates, according to multivariable analysis, could mean that MDR Acinetobacter are not more virulent than nonresistant strains or that the sample size in this study lacked the power to show a difference.
Our study indicates that infections with MDR Acinetobacter are independently associated with the adverse clinical outcomes of prolonged hospital and ICU lengths of stay compared with the outcomes for uninfected patients and those infected with drug-susceptible Acinetobacter. This is the first study evaluating length of stay and mortality rates associated with MDR Acinetobacter infection while controlling for important confounders such as severity of illness and underlying disease. These data emphasize the need for aggressive infection control strategies to prevent MDR Acinetobacter infection and its adverse effects on hospitalized patients.