Pseudomonas aeruginosa in patients
hospitalised for COPD exacerbation: a
prospective study
C. Garcia-Vidal*, P. Almagro*, V. Romanı´*, M. Rodrı´guez-Carballeira*, E. Cuchi#,
L. Canales", D. Blasco+, J.L. Heredia1 and J. Garau*
ABSTRACT: Risk factors for Pseudomonas aeruginosa (PA) isolation in patients hospitalised for
chronic obstructive pulmonary disease (COPD) exacerbation remain controversial. The aim of our
study was to determine the incidence and risk factors for PA isolation in sputum at hospital
admission in a prospective cohort of patients with acute exacerbation of COPD.
We prospectively studied all patients with COPD exacerbation admitted to our hospital between
June 2003 and September 2004. Suspected predictors of PA isolation were studied. Spirometry
tests and 6-min walking tests were performed 1 month after the patients were discharged. High-
resolution computed tomography (HRCT) was performed in a randomised manner in one out of
every two patients to quantify the presence and extent of bronchiectasis. Patients were followed
up during the following year for hospital re-admissions.
A total of 188 patients were included, of whom 31 (16.5%) had PA in sputum at initial admission.
The BODE (body mass index, airflow obstruction, dyspnoea, exercise capacity) index (OR 2.18, CI
95% 1.26–3.78; p50.005), admissions in the previous year (OR 1.65, CI 95% 1.13–2.43; p50.005),
systemic steroid treatment (OR 14.7, CI 95% 2.28–94.8; p50.01), and previous isolation of PA (OR
23.1, CI 95% 5.7–94.3; p,0.001) were associated with PA isolation. No relationship was seen
between bronchiectasis in HRCT and antibiotic use in the previous 3 months.
PA in sputum at hospital admission is more frequent in patients with poorer scoring on the
BODE index, previous hospital admissions, oral corticosteroids and prior isolation of PA.
KEYWORDS: BODE index, bronchiectasis, chronic obstructive pulmonary disease, hospitalisation,
Pseudomonas aeruginosa infection
T
he role of bacterial infection in chronic
obstructive pulmonary disease (COPD)
exacerbation remains controversial [1, 2].
Recent studies have correlated COPD exacerba-
tion with the overgrowth of the bacterial load or
with the acquisition of a new strain of pathogenic
bacteria [3–5]. In ambulatory patients,
Haemophilus influenzae, Moraxella catarrhalis and
Streptococcus pneumoniae are the three major
pathogens isolated in COPD exacerbations, while
Pseudomonas aeruginosa (PA) is uncommon and is
usually associated with the greatest degree of
functional impairment [6–10]. Hospitalised
patients for acute exacerbation of COPD
(AECOPD) usually have more advanced disease
[11], and the infecting pathogens could be
different.
Recent European guidelines highlight the need
for specific studies on risk factors for PA in
COPD exacerbated patients [12]. This is an
important issue because empirical antibiotic regi-
mens designed to cover this pathogen are
different from those aimed to cover for the usual
microorganisms. Nevertheless, contradictory
results regarding the role of PA in sputum at
admission in COPD hospitalised patients have
been reported [13–15]. Our hypothesis was that
there would be specific risk factors that would
predict a change in flora and an increased risk for
the presence of PA in sputum in this population.
On this basis, the aim of the study was to
determine the incidence and risk factors for PA
isolation in sputum at hospital admission in a
prospective cohort of patients with AECOPD.
AFFILIATIONS
*Services of Internal Medicine,
#Microbiology,
"Radiology,
1Pneumology, Hospital Mutua de
Terrassa, University of Barcelona, and
+Hospital Clinic, Dept of Clinical
Microbiology, Hospital Mutua de
Terrassa, University of Barcelona,
Barcelona, Spain.
CORRESPONDENCE
C. Garcia-Vidal
Service of Internal Medicine
Hospital Mutua de Terrassa
Plaza Dr. Robert 5
08221 Terrassa
Barcelona
Spain
E-mail: carolgv75@hotmail.com
Received:
Jan 08 2009
Accepted after revision:
March 29 2009
First published online:
April 22 2009
European Respiratory Journal
Print ISSN 0903-1936
Online ISSN 1399-3003This article has supplementary material accessible from www.erj.ersjournals.com
1072 VOLUME 34 NUMBER 5 EUROPEAN RESPIRATORY JOURNAL
Eur Respir J 2009; 34: 1072–1078
DOI: 10.1183/09031936.00003309
Copyright�ERS Journals Ltd 2009
METHODS
Subjects
We prospectively studied all consecutive patients admitted to
our institution (Hospital Mutua de Terrassa, Barcelona, Spain)
for AECOPD between June 2003 and September 2004. All
episodes of hospital re-admissions of the cohort during the
following year, until September 2005, were also prospectively
followed up. The study was carried out in a 500-bed university
hospital. Inclusion criteria were: hospitalisation for AECOPD,
basal forced spirometry showing a forced expiratory volume in
1 s (FEV1) f70% of their reference value, and b2-agonist
reversibility of predicted FEV1 of ,15% and/or 200 mL, with
FEV1/forced vital capacity (FVC) ,70%. Exclusion criteria
included a history of asthma or bronchiectasis as a predomi-
nant illness, pneumonia or pulmonary oedema at admission,
hospitalisation for causes other than AECOPD, or patient
refusal to participate in the study. COPD exacerbation was
defined following the criteria of ANTHONISEN et al. [16].
Admission criteria were at the discretion of the emergency
room physician. For the purposes of this study, patients were
divided into two groups: those in whom PA was isolated in
sputum at hospital admission (PA group) and those in whom
PA was not isolated in sputum at hospitalisation (non-PA
group). Written informed consent was obtained from each
subject and the study was approved by the ethics committee at
the Hospital Mutua de Terrassa.
Clinical evaluation
At the initial visit, patients provided a complete clinical history
and underwent a physical examination. Information collected
included demographic characteristics, body mass index,
comorbidity (as measured by the Charlson index), previous
functional dependence (Katz score) and dyspnoea measured
by the modified Medical Research Council (mMRC).
Information on smoking history, number of hospitalisations
for COPD within the last year, time to last hospital discharge,
use of antibiotics within the last 3 months and prior admission,
use of systemic or inhaled corticosteroids, and chronic home
use of oxygen therapy prior to admission were also collected.
Chronic systemic corticosteroid use was considered when
doses equivalent to prednisone o5 mg?day-1 had been given
for at least the previous 3 months.
Microbiological studies
Spontaneous or induced sputum samples were collected at
admission and at each re-admission, before antibiotic admin-
istration (section 1 of supplementary data). Bacterial agents
were classified into potential pathogenic microorganisms
(PPMs) and non-PPMs, as previously described [3]. Only
PPMs were evaluated. In some patients with at least two
isolates of PA in sputum at different admissions, strains were
typed by pulsed-field gel electrophoresis. The presence of a
mucoid phenotype on PA isolation agar plates was recorded.
High-resolution computed tomography evaluation
The presence of bronchiectasis was assessed at the first
hospital admission. To limit radiation and costs, patients were
randomised for a high-resolution computed tomography
(HRCT) of the chest, in a 2:1 ratio. The diagnosis of
bronchiectasis was based on standard criteria. HRCT scans
were interpreted by two experienced radiologists blinded to
the patients’ clinical grouping and microbiological status.
Posterior consensus was reached in case of disagreement.
The bronchiectasis score is detailed in section 2 of the
supplementary data.
Follow-up
A follow-up visit took place ,1 month after discharge. At this
visit, forced spirometry and bronchodilatador testing were
performed according to standard techniques [17]. The 6-min
walking test was performed following the American Thoracic
Society recommendations [18]. The BODE (body mass index,
airflow obstruction, dyspnoea, exercise capacity) index was
also calculated as the sum score proposed by CELLI et al. [19]. In
case of hospital re-admission within the first month after
discharge, the patient was followed up for 1 month after
reaching clinical stability. All patients were followed up for
hospital re-admission during the year after discharge.
Statistical analysis
Sample size calculation was based on FEV1 values; post-
bronchodilator FEV1 40% of predicted for PA patients, and
50% for non-PA patients, with an a- and b-error of 0.05 and 0.1,
respectively. A 10% patient loss was assumed. Accordingly,
the calculated size was 234 patients. To assess factors
associated with PA isolation, we compared the PA and non-
PA groups. To detect significant differences between groups
we used the Chi-squared test with continuity correction for
categorical variables. Quantitative variables were analysed
using an unpaired t-test or their corresponding nonparame-
trical tests when the distribution of data so required. The
relationship between bronchiectasis score and FEV1 was
calculated with the Pearson correlation coefficient. For multi-
variate analysis a logistic regression model was constructed
with PA isolation as a dependent variable. In this model
independent variables included the most clinically relevant
variables that were found to be significant in bivariate analysis.
Data analysis was performed using the SPSS for Windows
software package (version 11; SPSS Inc., Chicago, IL, USA). In
all analyses, we considered p-values f0.05 to be statistically
significant. All reported p-values are two-tailed.
RESULTS
Patient characteristics
Over the study period, a total of 254 patients with a suspected
diagnosis of AECOPD were admitted to hospital. Of these, 66
patients were excluded for the following reasons: impossibility
to perform spirometry or lack of spirometric criteria (26
(10.2%) patients); pneumonia (25 (9.8%) patients); bronchiec-
tasis as a main manifestation of disease (six (2.3%) patients);
idiopathic fibrosis (four (1.6%) patients); and others (five
(2.0%) patients). The studied population were predominantly
male (95%), with a mean¡SD age of 72.1¡10.0 yrs and length
of stay of 11¡8.7 days. Sociodemographic and functional
characteristics are shown in table 1.
Of the 188 patients included, 157 (83.5%) were in the non-PA
group and 31 (16.5%) were in the PA-group. When comparing
groups, no significant differences in age or sex were observed.
PA isolation in sputum was more frequent in patients with the
worst values of Katz score (4.7 versus 5.6; p50.03) and the
mMRC (3.4 versus 2.7; p50.001). The PA group had a stronger
C. GARCIA-VIDAL ET AL. COPD AND SMOKING-REALTED DISORDERS
c
EUROPEAN RESPIRATORY JOURNAL VOLUME 34 NUMBER 5 1073
history of smoking than patients in the non-PA group (mean
pack-years of smoking 73.7 versus 56.8; p50.02). Previous
hospital admissions in the last month (32.3% versus 14.0%;
p50.001) and number of previous episodes of admission
within the last year (3.1 versus 0.9; p50.002) were more
frequent in the PA group. Acute and chronic steroid therapy
were more frequent in patients in the PA group (22.6% versus
6.6% (p50.012) and 12.9% versus 2.5% (p,0.01), respectively).
No relationship was found between PA isolation and the use of
inhaled steroids or antimicrobial use in the last 3 months.
Respiratory parameters significantly associated with the
presence of PA in sputum at admission were: severity of
disease as measured by post-bronchodilator FEV1 (mean of
FEV1 38.7% versus 45.9%; p50.012), the poorest values for the
6-min walking test (217.5 m versus 343.7 m; p,0.001) and
chronic home oxygen therapy (32.3 versus 15.6; p50.041). The
BODE index was also significantly associated with PA isolation
(7.3 versus 5.4; p50.0005). As a summary, data of significant
variables in bivariate analysis are shown in table 2.
A HRCT scan of the chest was performed in 88 randomised
patients. Patients with HRCT were similar with respect to age,
previous admissions, corticosteroid use, FEV1, FVC and other
measured physiological parameters, compared to those who
did not undergo HRCT scanning. 46 (52%) patients had
significant detectable bronchiectasis on HRCT (two or more
dilated bronchi; global score in percentage o5.6%). Of those in
whom bronchiectasis was detected, the median (range) score
was 25 (5–56)% (fig. 1). No statistical relationship was seen
between the total bronchiectasis score and FEV1 measurement
(r250.059; p50.058), 6-min walking test (r250.002; p50.784) or
BODE index (r250.009; p50.94). Similarly, no relationship was
found between the score of bronchiectasis and positive
bacterial culture (p50.76) or PA isolation in sputum at
admission (95% CI 0.99–1.05; p50.09).
Microbiological findings
Of the 188 patients included, 119 (63.3%) had good quality
sputum in the first hospital admission. Non-PPMs were
isolated in 55% of these patients while PPMs were found in
45% of cases. A single bacterial species was isolated in 50
patients, two species in three patients, and three species in one
patient. PA was the most frequently isolated species in patients
with valid sputum (31 (26%) out of 119 cases) followed by S.
pneumoniae and H. influenzae (11 (9.2%) cases each). The
presence of other microorganisms was infrequent.
During the initial admission and the subsequent year of
prospective follow-up, a total of 469 episodes of hospitalisation
due to COPD exacerbation were collected (134 in PA group
and 335 in non-PA group), and valid sputum was collected in
220 (47%) episodes. Patients with positive bacterial cultures of
sputum (any microorganism) had a lower FEV1 than patients
with negative sputum cultures at admission (p50.003).
As shown in figure 2, the global incidence of PA isolation in
the index of hospitalisation and re-admissions during the
subsequent year was 23.18% of all episodes. H. influenzae (11%)
and S. pneumoniae (10%) remained common aetiologies for
COPD exacerbation in patients requiring hospitalisation.
Patients in the PA-group were re-admitted more frequently
than patients in the non-PA group (p50.001), and were more
likely to present valid sputum during these re-admissions than
patients in the non-PA group (p,0.001). The relationship
between different microorganisms and FEV1 is shown in
figure 3.
Among the 31 patients with PA at the first admission, PA was
isolated in 12 patients only once, twice in 11 patients, and in
eight patients at least three times, in sputum cultures
performed during subsequent admissions. Previous isolation
of PA was associated with a higher probability of a new PA
isolation (p,0.001). Molecular typing of 41 PA strains from 10
patients obtained in different exacerbations showed that the
current strain was identical to the original strain in seven (70%)
patients and in 37 (90%) of the samples. Of note, all persisting
PA strains were non-mucoid (six patients) with a single
instance of persisting mucoid strain (one patient) (see
supplementary data). Table 3 shows the antibiotic suscept-
ibility pattern of PA isolates. Of the 31 patients in the PA
group, only four patients received empirical antibiotic treat-
ment with pseudomonal coverage (quinolones: n53, ceftazi-
dime: n51). 12 patients received antipseudomonal treatment
when the microbiological results were known.
Factors associated with PA isolation in multivariate analysis
Table 4 summarises the results of multivariate analysis of
factors potentially associated with PA isolation. Significant
variables associated independently with PA isolation were
BODE index (OR 2.18, 95% CI 1.26–3.78; p50.005), number of
TABLE 1 Characteristics of patients hospitalised with
acute chronic obstructive pulmonary disease
exacerbation during the period of study
Subjects n 188
Age yrs 72.11¡10.0
Male sex 178 (94.7)
PBD FEV1 L 1.04¡0.37
PBD FEV1 % pred 44¡14.52
Severity according to GOLD stage
Stage II, moderate 65 (34.6)
Stage III, severe 95 (50.5)
Stage IV, very severe 28 (14.9)
Walking test m 330¡105
Charlson index 2.17¡1.3
Katz score 5.46¡1.3
Dyspnoea mMRC 2.78¡1.2
Smoking pack-yrs 59.84¡35.4
Patients with a hospital admission in the previous month 32 (17)
Episodes of admission in previous year 1.27¡2.0
Days hospitalised in previous year 12.84¡25.9
Antimicrobials in last 3 months 62 (33)
Body mass index 27.8¡5.2
Inhaled steroids 140 (76.5)
Systemic steroids 17 (9.3)
Data are presented mean¡SD or n (%), unless otherwise specified. PBD: post-
bronchodilator; FEV1: forced expiratory volume in 1 s; % pred: % predicted;
GOLD: Global Initiative for Chronic Obstructive Pulmonary Disease; mMRC:
modified Medical Research Council.
COPD AND SMOKING-REALTED DISORDERS C. GARCIA-VIDAL ET AL.
1074 VOLUME 34 NUMBER 5 EUROPEAN RESPIRATORY JOURNAL
hospital admissions in the previous year (OR 1.65, 95% CI
1.13–2.43; p50.005), systemic steroid treatment (OR 14.7, 95%
CI 2.28–94.8; p50.01), and previous isolation of PA (OR 23.1,
95% CI 5.7–94.3; p,0.001).
DISCUSSION
In this prospective study we offer a comprehensive evaluation
of the incidence and risk factors for PA isolation in a large
prospective cohort of patients hospitalised for AECOPD. To
our knowledge, this study is the first to assess specifically, in a
prospective cohort, the multidimensional risk factors of PA
isolation in sputum in patients hospitalised for AECOPD. The
most important finding of our study is the strong relationship
between PA isolation at hospital admission and several
markers of respiratory functional impairment. Additionally,
our study shows that the incidence of PA in sputum in this
population is high (23% of total episodes). Specifically, this
TABLE 2 Characteristics of 188 patients hospitalised with acute chronic obstructive pulmoanry disease#
PA group" Non-PA group+ p-value OR (95% CI)
Katz index 4.74¡1.9 5.61¡1.1 0.03 0.67 (0.51–0.90)
Dyspnoea mMRC 3.43¡0.8 2.66¡1.2 0.001 2.07 (1.22–3.50)
Smoking pack-yrs 73.7¡39.7 56.76¡33.8 0.02 1.01 (1.003–1.02)
Hospital admission in last month 10 (32) 22 (14) 0.001 2.95 (1.16–7.48)
Admissions in previous year 3.06¡3.5 0.91¡1.3 0.002 1.47 (1.19–1.81)
Days hospitalised in previous year 38.3¡51.7 8¡12.7 0.004 1.04 (1.02–1.06)
Antibiotic treatment in previous 3 months 11 (35.5) 51 (32.5) 0.828 1.14 (0.51–2.57)
Systemic corticosteroids 7 (22.6) 10 (6.6) 0.01 3.57 (1.17–10.88)
Chronic systemic corticosteroids 4 (12.9) 4 (2.5) 0.01 5.66 (1.3–24)
PBD FEV1 % 38.7¡12.2 45.9¡14.7 0.01 0.96 (0.93–0.99)
Walking test m 217.5¡103 343.7¡98.5 ,0.001 0.99 (0.98–0.99)
BODE index 7.32¡1.72 5.42¡2.53 ,0.001 1.45 (1.17–1.78)
Number of Anthonisen criteria 0.735
1 6 (19) 40 (26)
2 8 (26) 41 (26)
3 17 (55) 76 (48)
Previous isolation PA 19 (61.3) 12 (7.6) ,0.001 122 (25.5–590)
Home oxygen therapy 10 (32.3) 24 (15.3) 0.04 0.48 (0.19–1.1)
Data are presented as mean¡ SD or n (%), unless otherwise stated. PA: Pseudomonas aeruginosa; mMRC: modified Medical Research Council; PBD: post-
bronchodilator; FEV1: forced expiratory volume in 1 s; BODE: body mass index, airflow obstruction, dyspnoea, exercise capacity. #: PA group compared with 157
patients in the non-PA group; ": n531; +: n5157.
50
40
30
20
10
0
P
at
ie
nt
s
%
Bronchiectasis score %
0–5.6 11.1–22 28–44 >44
FIGURE 1. Total bronchiectasis score in 88 patients. In total, 46 (52%) patients
had significant detectable bronchiectasis (two or more dilated bronchi; global score
in percentage o5.6%) on high-resolution computed tomography. The overall
bronchiectasis score was expressed as a percentage: (Brochiectasis score /
bronchiectasis maximum score) 6100. Patients with a score of 0 or 5.6 (less than
two affected segments) were considered normal.
60
50
40
30
20
10
0
P
at
ie
nt
s
%
Bacterial isolates
Non-PPM PA SP HI ETB MC
FIGURE 2. Bacterial isolates of patients with valid sputum. &: i