Journal of Hospital Infection (2009) 73, 143e150

Available online at www.sciencedirect.com

www.elsevierhealth.com/journals/jhin

Risk factors and impact of nosocomial

Acinetobacter baumannii bloodstream infections
in the adult intensive care unit: a caseecontrol
study
T.-N. Jang a,b,c,*, S.-H. Lee b, C.-H. Huang a,b,c, C.-L. Lee b, W.-Y. Chen a
a
Section of Infectious Diseases, Department of Medicine, Shin Kong Wu Ho-Su Memorial Hospital, Taipei,
Taiwan
b
Infection Control Committee, Shin Kong Wu Ho-Su Memorial Hospital, Taipei, Taiwan
c
School of Medicine, Catholic Fu-Jen University, Taipei, Taiwan

Received 16 January 2009; accepted 10 June 2009

Available online 27 August 2009

KEYWORDS Summary During a nine-year study period, 96 episodes of nosocomial

Acinetobacter bloodstream infection (BSI) due to Acinetobacter baumannii were
baumannii; identified in the adult intensive care units (ICUs) of Shin Kong Wu Ho-Su
Bloodstream infection;
Memorial Hospital. Seventy-seven (80.2%) of these were available for
Intensive care unit;
Risk factors
matching in terms of age, sex, primary diagnosis of ICU admission, ICU
ward, and disease severity. Univariate analysis showed that central venous
catheter use, ventilator use, prior A. baumannii colonisation, and respira-
tory and cardiovascular organ failure were significantly associated with
acquiring A. baumannii BSI in the ICU. By multivariate analysis, only prior
A. baumannii colonisation [odds ratio (OR): 3.81; P < 0.001] and cardiovas-
cular failure (OR: 2.24; P ¼ 0.04) were identified as independent risk
factors. The lower respiratory tract (32/77; 41.6%) was the most frequent
source of infection, followed by intravascular catheters (13/77; 16.9%).
Cumulative survival curves for patients with A. baumannii BSI and control
patients showed no significant difference (30 day crude mortality: 29.9%
and 27.3%, respectively; P ¼ 0.916). However, the mean length of ICU
and hospital stay and mean hospital cost of patients with A. baumannii
BSI significantly increased, with an estimated 8.7 days excess length of
ICU stay, 19.1 days excess hospital stay, and US $8480 extra hospital costs.
Imipenem and meropenem remained the most active antimicrobial agents,

* Corresponding author. Address: Section of Infectious Diseases, Department of Medicine, Shin Kong Wu Ho-Su Memorial Hospital,
95 Wen Chang Road, Shih Lin District, Taipei 111, Taiwan. Tel: þ886 228 33 2211; fax: þ886 228 33 1111.
E-mail address: m002137@ms.skh.org.tw

0195-6701/$ - see front matter ª 2009 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.jhin.2009.06.007
144
both with 95.5% susceptibility (MIC50 ¼ 0.25 and 0.5, respectively). Improv-
ing hand hygiene of healthcare workers and aseptic care of vascular
catheters and endotracheal tubes are important measures to prevent
A. baumannii colonisation and decrease the incidence of BSI.
ª 2009 The Hospital Infection Society. Published by Elsevier Ltd. All rights
reserved.
Introduction
Acinetobacter baumannii, previously known as
Acinetobacter calcoaceticus variant anitratus, is a
non-fermenting Gram-negative aerobic coccobacil-
lus found extensively in the natural environment. It
can survive environmental desiccation for weeks,
a characteristic that promotes transmission through
fomites in hospitals.1 Recently, it has emerged as an
important nosocomial pathogen mainly affecting
patients with impaired host defences in the inten-
sive care unit (ICU) setting.2 Moreover, of particular
concern is the increasing prevalence of A. baumannii
isolates with resistance to many of the commercially
available antibiotics, including cephalosporins,
aminoglycosides, quinolones, and carbapenems.3
Clinical infections caused by A. baumannii in-
clude pneumonia, meningitis, endocarditis, peri-
tonitis, skin and soft-tissue infections, bloodstream
infections (BSIs), and catheter-related and urinary
tract infections.4 Clinical manifestations of
A. baumannii BSI may range from benign transient
bacteraemia to fulminant disease and septic shock
associated with an overall mortality (a crude mor-
tality rate) as high as 46%.5,6
Although various predisposing factors have
been analysed in different series, long-term
caseecontrol studies in a large cohort of critically
ill patients to determine the risk factors for
ICU-acquired A. baumannii BSI are rare. To improve
the understanding of risk factors and clinical and
economic impact associated with A. baumannii
BSI, this nine-year retrospective matched casee
control study was conducted in the adult ICU of
Shin Kong Wu Ho-Su Memorial Hospital.
Methods
Setting
Shin Kong Wu Ho-Su Memorial Hospital is a 921-bed
tertiary care centre in the city of Taipei, Taiwan.
Its ICUs (medical, surgical, neurological and
neurosurgical, respiratory and advanced ICUs)
have a total of 58 beds for critically ill adult
patients with an annual occupancy of about 2700
T.-N. Jang et al.
patients. During the study period (October 1997
through September 2006), patients in the ICUs
with one or more positive A. baumannii blood cul-
tures were evaluated. Only adults (aged 18
years) were included in the study.
Case
A case of clinically significant A. baumannii BSI was
defined as any patient with one or more blood cul-
tures positive for A. baumannii, where clinical ev-
idence of infection was present, such as fever,
leucocytosis, chills, and rigors, and where these
symptoms were present for more than 8 h. Nosoco-
mial ICU-acquired BSI was defined as the isolation
of one or more positive blood cultures in a patient
occurring 48 h after admission to the ICU.
Control
Each significant A. baumannii BSI case was
matched with a control patient without A. bau-
mannii BSI after controlling for age (within one de-
cade), sex, primary diagnosis of ICU admission, ICU
ward, and disease severity [Acute Physiological As-
sessment and Chronic Health Evaluation (APACHE
II) score 5 points]. Control patients were re-
quired to have had an ICU stay at least as long as
the matched case patient’s stay before the onset
of A. baumannii BSI. To ensure the same duration
of risk exposure, this time was called T0 and
defined as follows: for case patients, T0 was the
time the first A. baumannii positive blood sample
was obtained; and for control patients, T0 was
the time from ICU admission to the time at which
the matched case patient had his or her first blood
culture positive for A. baumannii.7
Clinical parameters
Recent surgery was defined as a surgical procedure
performed within 14 days of the onset of
A. baumannii BSI. Prior antibiotic therapy was de-
fined as the use of a systemic antimicrobial agent
for at least 48 h within the preceding 10 days. Col-
onisation and infection were defined according to
previously published criteria for the study of
Risk factors for A. baumannii bloodstream infections
nosocomial infections.8 The diagnosis of pneumo-
nia due to A. baumannii required new or progres-
sive infiltrates on chest radiographic examination
and the isolation of A. baumannii from a blood cul-
ture. If pulmonary infiltrates were absent and the
same A. baumannii strain was cultured from blood
and sputum or bronchial secretions, the patient
was categorised as tracheobronchitis and bacter-
aemia. Catheter-related bacteraemia was diag-
nosed when the same organism was isolated in
blood cultures and in semiquantitative catheter
tip cultures (15 or more colonies), and if no
primary site other than the intravascular catheter
was identified. The presence of organ system
failure was evaluated by using the criteria
described by Fagon et al.9 Antimicrobial therapy
was considered to be appropriate if one or more
antimicrobial agents were active in vitro against
A. baumannii when the agent(s) was given at an
adequate dosage via an appropriate route for at
least two days within three days of obtaining
a blood culture. Mortality was defined as death
within 30 days of the first positive blood culture.
Antimicrobial susceptibility testing
Blood specimens were obtained by percutaneous
venepuncture and inoculated in a two-bottle blood
culture system (Organon Teknika, Durham, NC,
USA). A. baumannii was identified with the auto-
matic API 20NE system (API-BioMérieux, LaBalme
Les Grotles, France). Isolates identified as mem-
bers of the A. baumannii/calcoaceticus complex
were grouped with A. baumannii. Minimum inhibi-
tory concentrations (MICs) of antimicrobial agents
for the isolates were determined by the agar dilu-
tion method using the guidelines of the Clinical
and Laboratory Standards Institute.10
Statistical analysis
Proportions were compared by the c2-test or Fish-
er’s exact test, as appropriate, whereas continu-
ous variables were compared using independent
samples t-test. The length of hospital stay and
costs of hospitalisation were compared between
case patients and control patients. Hospital costs
were obtained from each patient’s invoice. The
excess length of stay and extra costs attributed
to A. baumannii infection were defined as the dif-
ferences between cases and controls. To deter-
mine independent risk factors, a multivariate
logistic regression model was used. All variables
with P < 0.05 on univariate analysis were included
in the multivariate model. All P-values were two-
145
tailed and P < 0.05 was considered statistically sig-
nificant. Survival curves were prepared using the
KaplaneMeier method and univariate survival dis-
tributions were compared using the log rank test.
The SPSS software package, version 13.0 (SPSS
Inc., Chicago, IL, USA) was used for all analyses.
Results
During the nine-year study period, 106 episodes of
nosocomial BSI due to A. baumannii in the adult
ICUs were found. Of these, 10 episodes did not ful-
fil the criteria and were excluded from the study. A
total of 96 episodes of significant nosocomial BSI
were identified, with an infection rate of 0.56 ep-
isodes per 1000 patient-days. The incidence was
relatively stable with 10.7  5 cases per year.
The mean interval between ICU admission and
identification of A. baumannii BSI was
16.05  14.85 days. Of these, 77 cases were
matched according to the specified criteria and
were enrolled in the study. The characteristics of
cases and their matched controls are shown in
Table I. The average age, sex ratios and primary
ICU admission diagnosis were almost the same.
The mean APACHE II scores of cases and controls
were 22.48  5.88 and 22.18  6.27, respectively,
which indicated no difference in disease severity
between the two groups.
Risk factors for acquiring A. baumannii BSI in
the ICU are shown in Table II. By univariate analy-
ses, the following variables were significantly asso-
ciated with A. baumannii BSI in the ICU: central
venous catheter use, ventilator use, prior A. bau-
mannii colonisation, and respiratory and cardio-
vascular organ failure. Variables identified as
significant risk factors were included in the step-
wise logistic regression analyses. After adjustment
for confounding factors, only prior A. baumannii
colonisation [odds ratio (OR): 3.81; P < 0.001]
and cardiovascular failure (OR: 2.24; P ¼ 0.04)
were independent risk factors associated with
A. baumannii BSI.
Sources of infection were identified in 58.4% (45/
77) of the cases. The respiratory tract (32; 41.6%)
was the most frequent source, followed by intra-
vascular catheter (13; 16.9%). For the rest (32;
41.6%), the source of infection could not be identi-
fied. Of the 32 cases originating from the respiratory
tract, 27 cases were diagnosed as having ventilator-
associated pneumonia. The remaining five cases
were classified as tracheobronchitis due to no
obvious new or progressive pulmonary infiltration
on chest radiographic examination.
146 T.-N. Jang et al.

Table I Demographic data and primary diagnosis of cases and matched controls
Parameter Cases (%) Controls (%) P-value
N ¼ 77 N ¼ 77
Agea 69.24  14.63 69.75  14.14 0.827
Male sex 33 (42.9) 33 (42.9) NS
Primary ICU admission diagnosis
Pneumonia 21 (27.3) 21 (27.3) NS
Intracranial haemorrhage 13 (16.9) 14 (18.2) NS
Sepsis 8 (10.4) 8 (10.4) NS
Respiratory failure 7 (9.1) 7 (9.1) NS
Acute myocardial infarction 5 (6.5) 4 (5.2) NS
Gastrointestinal bleeding 4 (5.2) 4 (5.2) NS
Coronary artery disease 2 (2.6) 3 (3.9) NS
Craniotomy 3 (3.9) 3 (3.9) NS
Acute cholecystitis 2 (2.6) 2 (2.6) NS
Aortic aneurysm 2 (2.6) 2 (2.6) NS
Intra-abdominal surgery 2 (2.6) 2 (2.6) NS
Shock 2 (2.6) 2 (2.6) NS
Cerebral infarct 2 (2.6) 1 (1.3) NS
Othersb 4 (5.2) 4 (5.2) NS
APACHE II scorea 22.48  5.88 22.18  6.27 NS
Duration of ICU stay (days)a 35.49  29.66 26.78  19.87 0.034
Duration of hospital stay (days)a,c 77.08  70.56 58.01  42.31 0.045
Costs of hospitalisation (US $)a,c 30,651  23,495 22,171  14,756 0.009
ICU, intensive care unit; APACHE, Acute Physiological Assessment and Chronic Health Evaluation.
a
Mean  SD.
b
Others, including chronic obstructive pulmonary disease with acute exacerbation, hyperosmolar hyperglycaemic state, heart
failure, and multiple traumatic injury.
c
Seventy-six matched caseecontrol pairs were included in the analysis of hospital stay and cost.

The mean time (SD) from infection to death was
10.26  9.33 days. The crude 30 day mortality rates
for the case and control patients were 29.9% (23/77)
and 27.3% (21/77), respectively. Thus, the esti-
mated attributable mortality rate due to A. bau-
mannii BSI was only 2.6%. Cumulative survival
curves for patients with A. baumannii BSI and con-
trol patients are compared in Figure 1. The differ-
ence of 30 day mortality between the two groups
was not significant (P ¼ 0.916; log rank test). The
mean length of ICU stay (SD) for patients with
A. baumannii BSI was 35.49  29.66 days, and was
26.78  19.87 days for the matched controls
(P ¼ 0.034). The mean length of total hospital stay
(SD) was 77.08  70.56 days for the cases and
58.01  42.31 days for the controls (P ¼ 0.045),
and the mean costs of hospitalisation (SD) were
US $30,651  23,495 and $22,171  14,756 for the
cases and controls, respectively (P ¼ 0.009). Thus,
A. baumannii BSI patients had 8.7 and 19.1 days ex-
cess length of ICU and hospital stay, respectively,
and $8480 extra hospital costs, compared with
matched controls (Table I).
Among the 77 enrolled A. baumannii BSI cases, 67
isolates were retrieved for MIC, MIC50, MIC90, and
susceptibility studies (Table III). Among the 11
tested antimicrobials, imipenem and meropenem
were the most active against A. baumannii,
followed by ampicillin-sulbactam, levofloxacin,
cefepime, ciprofloxacin, ceftazidime and piperacil-
lin-tazobactam. Forty-nine out of 77 cases had
received early appropriate antibiotic treatment.
Of these, piperacillin-tazobactam was used in 10
(20.4%) patients, ceftazidime in 10 (20.4%), fluoro-
quinolones in 10 (20.4%), piperacillin in six (12.2%)
and carbapenems in five patients (10.2%). In 10
cases, aminoglycosides were used in combination.
Patients who received early appropriate antibiotic
treatment had a lower mortality rate than those
who received inappropriate antibiotic treatment
(24.5% vs 39.3%). However, the difference was not
significant.
Discussion
A. baumannii was previously considered an organ-
ism of low virulence.6 It is now increasingly re-
ported as an important nosocomial pathogen
implicated in outbreaks of respiratory, blood, cen-
tral nervous system, and wound infections.6,11e14
In one report from an Israeli teaching hospital,
Risk factors for A. baumannii bloodstream infections 147

Table II Risk factors in acquiring A. baumannii bloodstream infection in the intensive care unit by univariate
analysis
Risk factors Cases (%) Controls (%) P-value Odds ratio
N ¼ 77 N ¼ 77 (95% CI)
Invasive procedures
Central venous catheter 74 (96.1) 64 (83.1) 0.017 5.01 (1.37e18.37)
Arterial catheter 44 (57.1) 32 (41.6) 0.076 1.88 (0.99e3.56)
Ventilator 73 (94.8) 64 (83.1) 0.004 3.71 (1.15e11.94)
Chest tube 4 (5.2) 5 (6.5) 1.000 0.79 (0.20e3.06)
Nasogastric tube 74 (96.1) 72 (93.5) 0.719 1.71 (0.40e7.43)
Abdominal drain 7 (9.1) 11 (14.3) 0.452 0.60 (0.22e1.64)
Urinary catheter 70 (90.9) 67 (87.0) 0.607 1.49 (0.54e4.15)
Total parenteral nutrition 4 (5.2) 8 (10.4) 0.367 0.47 (0.14e1.64)
Recent major surgery 23 (29.9) 22 (28.6) 1.000 1.07 (0.53e2.13)
Prior antibiotic treatment 75 (97.4) 76 (98.7) 1.000 0.49 (0.04e5.56)
Third-generation cephalosporins 25 (32.5) 19 (24.7) 0.372 1.47 (0.73e2.97)
Fluoroquinolones 14 (18.2) 17 (22.1) 0.688 0.79 (0.36e1.73)
Aminoglycosides 35 (45.5) 45 (58.4) 0.147 0.59 (0.31e1.12)
Immunosuppression 17 (22.1) 10 (13.0) 0.204 1.90 (0.81e4.47)
Neutropenia 1 (1.3) 0 (0.0) 1.000 NA
Prior A. baumannii colonisation 40 (52.0) 16 (20.8) <0.001 4.12 (2.03e8.38)
Organ failures
Respiratory 33 (42.9) 18 (23.4) 0.017 2.46 (1.23e4.92)
Cardiovascular 36 (46.8) 18 (23.4) 0.004 2.88 (1.44e5.75)
Renal 29 (37.7) 25 (32.5) 0.612 1.26 (0.65e2.44)
Hepatic 4 (5.2) 6 (7.8) 0.744 0.65 (0.18e2.39)
CI, confidence interval; NA, not available.

acinetobacter had advanced from the fourth most

common bacterium resulting in bloodstream infec-
1.0 tions in 1997 to the most frequent cause in 2002.15
In addition to an ICU stay, risk factors for
A. baumannii colonisation and infection are recent
0.8 surgery, central vascular catheterisation, tracheos-
tomy, mechanical ventilation, enteral feedings,
APACHE II score, underlying malignancy, prior
Cumulative survial

infection, organ system failure, prolonged hospital-

0.6
isation, and prior antibiotic use (in particular,
third-generation cephalosporins and aminoglyco-
sides).5,11,16e19 Most of these potential risk factors
0.4 have been identified by univariate analysis but are
not part of a caseecontrol study, and thus may
simply reflect underlying disease severity or
a need for critical care. A prospective cohort study
0.2
in the ICU has demonstrated that immunosuppres-
sion, unscheduled hospital admission, respiratory
failure on ICU admission, previous antimicrobial
0.0 therapy, previous sepsis in the ICU, and invasive
procedures index are independently associated
0 5 10 15 20 25 30
with A. baumannii BSI.20 Grupper et al.’s three-
Day after T0
year prospective caseecontrol study found that
Figure 1 Cumulative survival curves of patients with mechanical ventilator use, central venous catheter
Acinetobacter baumannii bloodstream infection (solid in place, and previous antibiotic treatment were
line) and their matched control patients (dashed line) significantly associated with acinetobacter BSI in
(P ¼ 0.916; log rank test). univariate analysis.7
148 T.-N. Jang et al.

Table III Minimum inhibitory concentration (MIC), MIC50, MIC90 values and susceptibilities of 67 bloodstream
infection isolates to 11 antibiotics
Antimicrobial agents MIC (mg/mL) CLSI Susceptibility
breakpoints rate (%)
(mg/mL)
Range 50% 90% Sensitive
Ampicillin-sulbactam 0.5 to >256 4 32 8/4 77.6
Piperacillin-tazobactam 2 to >128 16 128 16/4 59.7
Ceftazidime 1 to >256 8 256 8 59.7
Cefepime 0.5 to 256 4 64 8 62.7
Aztreonam 2 to 256 32 128 8 a 7.5
Imipenem 0.03 to 32 0.25 2 4 95.5
Meropenem 0.125 to 64 0.5 1 4 95.5
Amikacin 1 to >256 8 >256 16 58.2
Levofloxacin 0.06 to 32 0.25 16 2 71.6
Ciprofloxacin 0.25 to >256 0.5 128 1 59.7
Colistin 1 to 128 2 8 2 53.7
a
No definite Clinical and Laboratory Standards Institute (CLSI) breakpoint was available but we used 8 as the breakpoint
according to those of third-generation cephalosporins.

In the present study, prior antibiotic therapy
and immunosuppression were similar in both
groups. None of the invasive procedures, including
recent major surgery, was identified as a risk
factor by multivariate analysis. Only prior A. bau-
mannii colonisation and cardiovascular organ fail-
ure were the independent risk factors associated
with A. baumannii BSI. Prior A. baumannii coloni-
sation had the strongest association with the
acquisition of A. baumannii BSI (OR: 3.81;
P < 0.001). This finding was consistent with a previ-
ous caseecontrol study in which prior colonisation
with A. baumannii was identified as the most sig-
nificant independent risk factor of A. baumannii
BSI (OR: 26.23; P ¼ 0.0002) in a burns ICU.21
An intravascular catheter and the lower re-
spiratory tract are two major portals of entry for
A. baumannii, with the respiratory tract as the
leading infection source in this (41.6%) and other
studies whereas the intravascular catheter was
the leading infection source in other investiga-
tions.6,7,12,22e24 The source of A. baumannii BSI
cannot be identified in 41.6% of cases in the pres-
ent study. This is comparable to previous reports in
which the source of A. baumannii BSI was undeter-
mined in 36.9e38% of cases.7,22
Mortality in the current series is 29.9% and
comparable to the range of 22e46% reported in
other published series.5,12,22e24 Nevertheless,
crude mortality rates reflect deaths due to pa-
tients’ underlying diseases as well as deaths due
to infections. Our previous study of nosocomial
Gram-negative BSI in the critically ill revealed un-
derlying illnesses as the independent risk factors
for death.25 An earlier study of acinetobacter bac-
teraemia also reported that outcome of the infec-
tion correlated more closely with the type of
underlying illness than with other factors.12
In this study design, adequate matching be-
tween case patients and control patients elimi-
nated the influence of comorbid conditions.
Mortality attributable to A. baumannii BSI was
only 2.6%, while the cumulative survival curves
for patients with A. baumannii BSI and control pa-
tients showed no significant difference. This result
is similar to a previous caseecontrol study in ICU
which also showed that A. baumannii bacteraemia
is not associated with a significantly increased
mortality rate in critically ill patients.26 Another
caseecontrol study in a burns ICU also revealed
that only 7% death was directly related to A. bau-
mannii bacteraemia.21 These investigators con-
cluded that A. baumannii was an opportunistic
pathogen and that underlying illnesses seemed to
play a more important role than the infection itself
as the cause of death.
Although the mortality rate of A. baumannii BSI
did not show significant difference compared with
the controls, the mean length of ICU and hospital
stay and mean hospital cost significantly in-
creased, with an estimated 8.7 days excess length
of ICU stay, 19.1 days excess hospital stay, and
US $8480 extra hospital cost. These results were
similar to those of a previous caseecontrol study
of nosocomial BSI in critically ill patients that re-
vealed extra ICU stay of 8 days, hospital stay of
24 days, and hospital costs of $40,890 among
survivors.27
Risk factors for A. baumannii bloodstream infections
In recent years, the emergence of carbapenem-
resistant or even pandrug-resistant A. baumannii
has been an increasing problem for both the con-
trol and treatment of nosocomial infections.3 A re-
port of 399 episodes of acinetobacter BSI from a UK
tertiary care centre showed carbapenem resis-
tance rising from 0% in 1998 to 55% in 2006.28 In
Taiwan, carbapenem-resistant A. baumannii was
also prevalent. A previous study in a university hos-
pital showed that the incidence of carbapenem-re-
sistant A. baumannii as causes of nosocomial
infection increased from 5.88% in 1993 to 21.5%
in 2000. The study also mentioned that this trend
correlated with the increasing use of carbapenem
and ciprofloxacin, but not with the use of
extended spectrum cephalosporins and aminogly-
cosides.3 A caseecontrol study of multidrug-resis-
tant A. baumannii BSI from southern Taiwan
showed that the group with resistant strains had
a 21.8% attributable mortality.29
However, in the current study, even in the ICU
setting, imipenem and meropenem remained rel-
atively sensitive with 95.5% susceptibility. The high
susceptibility of the isolates against carbapenem
may be ascribed to the hospital’s strict antibiotic
control policy, because all use of carbapenem
requires the approval of an infection specialist.
Furthermore, the low prevalence of carbapenem-
resistant A. baumannii in the isolates may explain
the low attributable mortality in this study.
Some limitations, however, are inherent in the
present study. The most important of these in-
volves the retrospective design. Since the possibil-
ity of selection bias is a concern in retrospective
caseecontrol studies, we tried to avoid the bias by
using strict inclusion criteria. Each enrolled pair
should be matched with sex, age, the same ICU
ward, major admission diagnosis, disease severity
and possible close ICU admission date. Additional
matching on length of ICU stay prior to the onset of
A. baumannii BSI includes that control subjects
also had an equivalent number of ‘days at risk’
for the development of an A. baumannii BSI.
Thus the potential for selection bias is expected
to be small. In addition, retrospective matched
studies are subject to selection bias if many poten-
tial cases are excluded. We failed to match 19 pa-
tients who developed A. baumannii BSI with
control subjects. Four of these patients died
(21%); thus, inclusion of these cases into the co-
hort would not have changed the study findings.
Because of the difficulty in matching cases with
appropriate controls, we only used one case to
one control.
In summary, this study highlights the impor-
tance of A. baumannii colonisation as a major
149
risk factor in acquiring A. baumannii BSI in criti-
cally ill patients. The lower respiratory tract and
intravascular catheters are the most frequently
identified sources of this infection. Improving
hand hygiene of healthcare workers and aseptic
care of vascular catheters and endotracheal tubes
are important measures to prevent A. baumannii
colonisation and reduce the incidence of BSI. Strict
antibiotic control is the most important measure in
the prevention of drug resistance and in reducing
the impact of growing A. baumannii BSI.
Conflict of interest statement
None declared.
Funding sources
None.
References
1. Getchell-White SI, Donowitz LG, Groschel DH. The inani-
mate environment of an intensive care unit as a potential
source of nosocomial bacteria: evidence for long survival
of Acinetobacter calcoaceticus. Infect Control Hosp Epi-
demiol 1989;10:402e407.
2. Valles J, Leon C, Alvarez-Lerma F, et al. Nosocomial bacter-
emia in critically ill patients: a multi-center study evaluat-
ing epidemiology and prognosis. Clin Infect Dis 1997;24:
387e395.
3. Hsueh PR, Teng LJ, Chen CY, et al. Pan-drug-resistant Aci-
netobacter baumannii causing nosocomial infections in
a university hospital, Taiwan. Emerg Infect Dis 2002;8:
827e832.
4. Allen DM, Hartman BJ. Acinetobacter species. In:
Mandell GL, Bennett JE, Dolin R, editors. Principles and
practice of infectious diseases. 6th edn. Philadelphia:
Churchill Livingstone; 2005. p. 2632e2636.
5. Chen YC, Chang SC, Hsieh WC, Luh KT. Acinetobacter cal-
coaceticus bacteremia: analysis of 48 cases. J Formos Med
Assoc 1991;90:958e963.
6. Smego RA. Endemic nosocomial Acinetobacter calcoaceti-
cus bacteremia: clinical significance, treatment, and
prognosis. Arch Intern Med 1985;145:2174e2179.
7. Grupper M, Sprecher H, Mashiach T, Finkelstein R.
Attributable mortality of nosocomial Acinetobacter
bacteremia. Infect Control Hosp Epidemiol 2007;28:
293e298.
8. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC
definitions for nosocomial infections. Am J Infect Control
1988;16:128e140.
9. Fagon JY, Chastre J, Novara A, Medioni P, Gibert C. Charac-
terization of intensive care unit patients using a model
based on the presence or absence of organ dysfunctions
and/or infection: the ODIN model. Intensive Care Med
1993;19:137e144.
10. Clinical and Laboratory Standards Institute. Performance
standards for antimicrobial disk susceptibility testing:
16th informational supplement. M100eS17. Wayne, PA:
CLSI; 2007.
11. Cefai C, Richards J, Gould FK, McPeake P. An outbreak of
Acinetobacter respiratory tract infection resulting from
150
incomplete disinfection of ventilatory equipment. J Hosp
Infect 1990;15:177e182.
12. Tilley PAG, Roberts FJ. Bacteremia with Acinetobacter spe-
cies: risk factors and prognosis in different clinical settings.
Clin Infect Dis 1994;18:896e900.
13. Siegman-Igra Y, Bar-Yosef S, Gorea A, Avram J. Nosocomial
Acinetobacter meningitis secondary to invasive procedures:
report of 25 cases and review. Clin Infect Dis 1993;17:
843e849.
14. Sherertz RJ, Sullivan ML. An outbreak of infections with Aci-
netobacter calcoaceticus in burn patients: contamination of
patients’ mattresses. J Infect Dis 1985;151:252e259.
15. Jerassy Z, Yinnon AM, Mazouz-Cohen S, et al. Prospective
hospital-wide studies of 505 patients with nosocomial bac-
teraemia in 1997 and 2002. J Hosp Infect 2006;62:230e236.
16. Garnacho-Montero J, Ortiz-Leyba C, Fernandez-Hinojosa E,
et al. Acinetobacter baumannii ventilator-associated pneu-
monia: epidemiological and clinical findings. Intensive Care
Med 2005;31:649e655.
17. Lortholary O, Fagon JY, Hoi AB, et al. Nosocomial acquisi-
tion of multiresistant Acinetobacter baumannii: risk factors
and prognosis. Clin Infect Dis 1995;20:790e796.
18. Mulin B, Talon D, Viel JF, et al. Risk factors for nosocomial
colonization with multiresistant Acinetobacter baumannii.
Eur J Clin Microbiol Infect Dis 1995;14:569e576.
19. Peacock JE, Sorrell L, Sottile FD, Price LE, Rutala WA. Nos-
ocomial respiratory tract colonization and infection with
aminoglycoside-resistant Acinetobacter calcoaceticus var.
anitratus: epidemiologic characteristics and clinical signifi-
cance. Infect Control Hosp Epidemiol 1988;9:302e308.
20. Garmendia JLG, Leyba OC, Montero JG, et al. Risk factors
for Acinetobacter baumannii nosocomial bacteremia in crit-
ically ill patients: a cohort study. Clin Infect Dis 2001;33:
939e946.
T.-N. Jang et al.
21. Wisplinghoff H, Perbix W, Siefert H. Risk factors for nosoco-
mial bloodstream infections due to Acinetobacter bauman-
nii: a caseecontrol study of adult burn patients. Clin Infect
Dis 1999;28:59e66.
22. Chen HP, Chen TL, Lai CH, et al. Predictors of mortality in
Acinetobacter baumannii bacteremia. J Microbiol Immunol
Infect 2005;38:127e136.
23. Seifert H, Strate A, Pulverer G. Nosocomial bacteremia due
to Acinetobacter baumannii. Clinical features, epidemiol-
ogy, and predictors of mortality. Medicine 1995;74:
340e349.
24. Rolston K, Guan Z, Bodey GP, Elting L. Acinetobacter cal-
coaceticus septicemia in patients with cancer. South Med
J 1985;78:647e651.
25. Jang TN, Kuo BIT, Shen SH, et al. Nosocomial gram-negative
bacteremia in critically ill patients: epidemiologic charac-
teristics and prognostic factors in 147 episodes. J Formos
Med Assoc 1999;98:465e473.
26. Blot S, Vandewoude K, Colardyn F. Nosocomial bacteremia
involving Acinetobacter baumannii in critically ill patients:
a matched cohort study. Intensive Care Med 2003;29:
471e475.
27. Pittet D, Tarara D, Wenzel RP. Nosocomial bloodstream in-
fection in critically ill patients: excess length of stay, extra
costs, and attributable mortality. J Am Med Assoc 1994;
271:1598e1601.
28. Wareham DW, Bean DC, Khanna P, et al. Bloodstream infec-
tion due to Acinetobacter spp: epidemiology, risk factors
and impact of multi-drug resistance. Eur J Clin Microbiol In-
fect Dis 2008;27:607e612.
29. Lee NY, Lee HC, Ko NY, et al. Clinical and economic impact
of multidrug resistance in nosocomial Acinetobacter bau-
mannii bacteremia. Infect Control Hosp Epidemiol 2007;
28:713e719.