CHEST Original Research

HEALTH-CARE-ASSOCIATED PNEUMONIA

Why Mortality Is Increased in

Health-Care-Associated Pneumonia
Lessons From Pneumococcal Bacteremic Pneumonia
Jordi Rello, MD, PhD; Manel Luján, MD; Miguel Gallego, MD; Jordi Vallés, MD, PhD;
Yolanda Belmonte, MD; Dionisia Fontanals, PhD; Emili Diaz, MD, PhD;
and Thiago Lisboa, MD, PhD; for the PROCORNEU Study Group*

Background: A cohort of patients with bacteremic Streptococcus pneumoniae pneumonia was

reviewed to assess why mortality is higher in health-care-associated pneumonia (HCAP) than in
community-acquired pneumonia (CAP).
Methods: A prospective cohort of all adult patients with bacteremic pneumococcal pneumonia
attended at the ED was used.
Results: One hundred eighty-four cases were classified as CAP and 44 (19%) as HCAP. Fifty-two
(23%) were admitted to the ICU. Three (1.5%) isolates were resistant to b-lactams, and only two
patients received inappropriate therapy. The CAP cohort was significantly younger (median age
68 years, interquartile range [IQR] 42-78 vs 77 years, IQR 67-82, P , .001). The HCAP cohort
presented a higher Charlson index (2.81 6 1.9 vs 1.23 6 1.42, P , .001) and had higher severity of
illness at admission (altered mental status, respiratory rate . 30/min, Pao2/Fio2 , 250, and multi-
lobar involvement). HCAP patients had a lower rate of ICU admission (11.3% vs 25.5%, P , .05),
and a trend toward lower mechanical ventilation (9% vs 19%, P 5 .17) and vasopressor use (9% vs
18.4%, P 5 .17) were documented. More patients in the HCAP cohort presented with a pneumo-
nia severity index score . 90 (class IV-V, 95% vs 65%, P , .001), and 30-day mortality was signifi-
cantly higher (29.5% vs 7.6%, P , .001). A multivariable regression logistic analysis adjusting for
underlying conditions and variables related to severity of illness confirmed that HCAP is an inde-
pendent variable associated with increased mortality (odds ratio 5 5.56; 95% CI, 1.86-16.5).
Conclusions: Pneumococcal HCAP presents excess mortality, which is independent of bacterial sus-
ceptibility. Differences in outcomes were probably due to differences in age, comorbidities, and
criteria for ICU admission rather than to therapeutic decisions. CHEST 2010; 137(5):1138–1144

Abbreviations: aOR 5 adjusted odds ratio; CAP 5 community-acquired pneumonia; CLSI 5 Clinical and Laboratory
Standards Institute; CURB-65 5 confusion, urea nitrogen, respiratory rate, BP, 65 years of age or older; HCAP 5 health-
care-associated pneumonia; IQR 5 interquartile range; MIC 5 minimum inhibitory concentration; OR 5 odds ratio;
PSI 5 pneumonia severity index

Health-care-associated pneumonia (HCAP) is cur-

rently considered a distinct subset of pneumonia
Moreover, patients with severe HCAP or CAP may
need intensive care or mechanical ventilation and
associated with severe disease, long hospital stay, and their condition may be complicated by shock or mul-
high mortality rates.1,2 HCAP has not been exten- tiple organ failure. Unfortunately, the information
sively studied or clarified, especially in terms of its available on severe HCAP is limited. We therefore
etiology and management. It is not clear whether the decided to analyze the implications of HCAP in a
poor outcome observed in patients with HCAP is cohort of patients with bacteremic pneumococcal
related to the presence of more comorbidities or to a pneumonia.
higher incidence of antibiotic-resistant bacteria and The key question that we address in this article is
inappropriate empirical antibiotic treatment.3 There- whether classifying someone as HCAP adds any
fore, a study comparing pneumococcal HCAP and information that is not captured by existing severity
pneumococcal community-acquired pneumonia (CAP) indices. Prior studies4,5 clearly show that patients with
may be of interest. HCAP are older and have more comorbidities, and

1138 Original Research

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians
therefore have higher pneumonia severity index sion: age, sex, origin, alcohol abuse and tobacco consumption,
(PSI) scores.6 However, this information is already hospitalization and antibiotic therapies in last 3 months, CURB-65
(Confusion, urea nitrogen, respiratory rate, BP, 65 years of age or
captured by the PSI. It would be interesting to assess older) score, underlying diseases including Charlson comorbidity
predicted and actual mortality based on the actual index,7 vaccination status with the 23-valent polysaccharide
PSI score. If mortality in patients with HCAP is pre- vaccine (patients who received their last administration of the
dicted by standard CAP severity indexes, there is no vaccine in the 5 years before pneumonia were considered as vac-
real justification for categorizing them separately. cinated), symptoms in the days prior to and at admission, relevant
physical signs, blood gases, WBC and RBC counts, serum creati-
Our objectives were to compare differences in epi- nine, and chest radiograph pattern. The PSI was calculated using
demiology, use of vasopressors, need for mechanical a validated prognostic scoring system,6 with PSI . 90 (class IV-V)
ventilation, and ICU admission. We also aimed to being considered as severe pneumonia.
establish whether differences in outcomes for HCAP During hospitalization, the following variables were also
were due to differences in age, comorbidities, thera- recorded: in vitro susceptibility for strains, need for ICU admis-
sion, vasoactive drugs and mechanical ventilation, major compli-
peutic decisions, or criteria for ICU admission. cations, outcome (30-day mortality), length of stay for survivors,
and antibiotic therapy prescribed. Antibiotic therapy and decision
for ICU admission were not standardized and were left to the cri-
Materials and Methods teria of the attending physician.

Study Population Microbiologic Evaluation

Patients admitted between January 1999 and June 2007 to an Blood samples were processed using the BacT-Alert system
800-bed teaching hospital (500 beds for acute care patients and (BioMerieux; Lyon, France). Identification was carried out by
300 for long-term hospitalization) were prospectively identified colony morphology on blood agar, Gram stain, Optochin-disk sus-
by one of the authors (M. L.). All patients aged ⱖ 18 years who ceptibility, and the results of a latex agglutination test.
received a diagnosis of pneumonia (fever, productive cough, chest Minimum inhibitory concentration (MIC) values were deter-
pain, shortness of breath, and crackles on auscultation in addition mined by the microdilution method in cation-adjusted Mueller-
to a chest radiograph interpreted as pneumonia), and whose blood Hinton broth, supplemented with 5% lysed horse blood.
cultures obtained within the first 48 h of hospitalization showed Susceptibility was assessed post hoc, converting MIC values to
growth of Streptococcus pneumoniae, were included. Exclusion categories of susceptibility according to the 2008 Clinical and
criteria were bacteremia from other sources, presence of concur- Laboratory Standards Institute (CLSI) breakpoints.8 All strains
rent meningitis or endocarditis at admission, and HIV infection. were sent to the Pneumococcus Reference Laboratory of the
Cases were retrospectively considered HCAP in accordance Instituto de Salud Carlos III in Majadahonda for verification of
with the Infectious Disease Society of America/American Tho- sensitivity to the antibiotics and serotyping using the Quellung
racic Society guidelines,3 if one of the following criteria was met: reaction and/or dot blot assay, with the use of antisera provided by
hospitalization in an acute-care hospital for a minimum of 2 days the Statens Serum Institut (Copenhagen, Denmark).
within 90 days of the infection, residence in a nursing home or
long-term care facility, recent IV antibiotic therapy, chemother- Definitions
apy, wound care within the past 30 days of the current infection,
or attendance at a hospital or hemodialysis clinic. The study was Definitions were underlying conditions: alcohol consumption
approved by the institutional review board, and informed consent . 60 g/d; history of lung comorbidity including COPD and inter-
was waived. stitial lung disease; cardiac comorbidity: diagnosis of congestive
Demographic and clinical data of the CAP and HCAP cohorts heart failure, coronary artery disease, or advanced valvulopathy; renal
were compared. The following variables were recorded at admis- disease: chronic renal failure, with creatinine levels ⱖ 1.5 mg/dL;
liver disease: biopsy-proven cirrhosis or diagnosis of viral or toxic
Manuscript received September 15, 2009; revision accepted chronic liver disease; diabetes mellitus: treatment with oral antidi-
November 9, 2009. abetics or insulin; immunocompromise: corticosteroid therapy
Affiliations: From the Critical Care Department (Drs Rello, with 4 mg prednisolone/d or equivalent for . 1 month, or history
Diaz, and Lisboa), Joan XXIII University Hospital, IISPV, Rovira i of cancer chemotherapy over the past 6 months. Antibiotic treat-
Virgili University; the Pneumology Service (Drs Luján, Gallego, ment was considered concordant if the isolated strain was fully
and Belmonte), Critical Care Service (Dr Vallés), and UDIAT sensitive in vitro to at least one of the antibiotics administered.9
Centre diagnòstic (Dr Fontanals), Corporació Parc Tauli, Institut Empyema was defined as a pleural fluid with macroscopic pus or
Universitari Parc Taulí, Departament de Medicina, Universitat bacterial growth of the sample.
Autònoma de Barcelona, Sabadell, Spain.
*A complete list of participants is located in the appendix.
Funding/Support: This study was supported in part by CIBER Statistical Analysis
enfermedades respiratorias (CIBERes) 06/06/36, AGAUR 2009/
SGR/1226, FISS 08/0452, and FISS 04/1500. The Proyecto Cor- Quantitative variables for age, median values, and interquartile
porativo de Neumonía (PROCORNEU) Study Group is supported range (IQR) were reported and compared with the Mann-Whitney
by CIBERes, Instituto de Salud Carlos III. U test. For Charlson index, mean 6 SD was given and compared
Correspondence to: Jordi Rello, MD, PhD, Critical Care with the Student t test. For categorical variables, frequencies and
Department, Joan XXIII University Hospital, Carrer Mallafre percentages were reported and compared using the x2 test and
Guasch, 4, Tarragona 43007, Spain; e-mail: [email protected]
Fisher exact test when appropriate. The risk of death was assessed
© 2010 American College of Chest Physicians. Reproduction
of this article is prohibited without written permission from the by multivariable logistic regression analysis, considering variables
American College of Chest Physicians (www.chestpubs.org/ with P , .1 in the univariate analysis. We restricted the number of
site/misc/reprints.xhtml). variables included in the multivariable model following the rule of
DOI: 10.1378/chest.09-2175 at least five to seven events (deaths) per variable.10 Thus, variables

www.chestpubs.org CHEST / 137 / 5 / MAY, 2010 1139

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians
showing colinearity (multilobar involvement, respiratory rate and Table 1—Demographic Data, Underlying Conditions,
Pao2/Fio2 ratio, and shock, need for vasopressors, and ICU Clinical, Radiologic, and Microbiologic Findings in
admission) were not included in the model. Similarly, comorbidi- 228 Patients With Bacteremic Community-Acquired
ties were grouped as the Charlson index. The appropriateness of Pneumonia and Health-Care-Associated Pneumonia
the model was assessed by Hosmer-Lemeshow goodness-of-fit.11
Moreover, the survival was analyzed with the Kaplan-Meier Variables Data
method, and curves for HCAP vs CAP were compared with the
Median age (IQR) 70 (49-79)
log-rank analysis. The significance level of all analyses was HCAP 44 (19)
defined as P , .05. Data were analyzed using the SPSS-15 statis- 23-Valent pneumococcal polysaccharide 21/172 (12)
tical package (SPSS, Inc.; Chicago, IL). vaccinea
Alcoholism 30 (13)
Comorbidities
Results Cardiomyopathy 52 (23)
Lung disease 63 (27)
During the study period, all consecutive patients Hepatic disease 26 (11)
Renal 22 (9)
with bacteremic pneumococcal pneumonia were Charlson index, mean 6 SD 1.54 6 1.65
identified at the ED. Twenty were excluded because ICU admission 52 (23)
of HIV infection. Forty-four (19%) had HCAP and Pao2/Fio2 ratio , 250 74 (32)
184 had CAP. Hospitalization in the previous 90 days Vasopressor use 38 (16)
(25 patients, 56.8%) was the most common condition Multilobar involvement 101 (44)
Pleural effusion 71 (31)
in the HCAP cohort, with 16 (38.5%) patients admit- Empyema 18 (8)
ted from long-term care facilities. Only eight were Therapy
receiving cancer chemotherapy, and one patient was Discordant therapy 2 (1)
receiving hemodialysis. Six patients met more than Combination therapy 118 (51)
one condition. PSI score I–III 65 (28)
PSI score IV-IV 163 (72)
The demographic features are presented in Table 1. CURB-65 score
Two distinct populations from the community pre- 0-2 141 (61.8)
sented with pneumococcal pneumonia (Table 2). The ⱖ3 87 (38.2)
HCAP cohort presented more frequently with base- 30-d mortality 27 (12)
line comorbidities (84% vs 60%, P , .01) and had Data given as No. (%) unless otherwise specified. CAP 5 community-
higher Charlson index. Overall, 72% of the entire acquired pneumonia; CURB-65 5 confusion, urea nitrogen, respira-
cohort presented a risk of class IV or V PSI score; tory rate, BP, 65 years of age or older; HCAP 5 health-care-associated
pneumonia; IQR 5 interquartile range; PSI 5 pneumonia severity
however, the HCAP cohort presented more severe index.
illness overall and a significantly higher PSI score aThose vaccinated within the last 5 y.

(Table 2). A trend toward higher hypoxemia was

associated with HCAP but did not correlate with
greater mechanical ventilation requirement. At same also documented as resistant to cephalosporins (one
time, there was a trend toward less use of vasopres- patient in the HCAP cohort). Resistance to mac-
sors in HCAP, with the distribution of hypotensive rolides was 36% in the HCAP cohort vs 13% in the
patients not different among patients with HCAP and CAP cohort (P , .01, Table 2). However, only two
CAP. Indeed, most patients with HCAP were admit- patients (one in each group) received inappropriate
ted to medical wards, with a significantly less frequent therapy (2.3% vs 0.5%, P 5 .34). Available records
ICU hospitalization in the HCAP (11.3%) than in the documented that exposure to the 23-valent pneumo-
CAP cohort (25.5%, P , .05). Finally, medical orders coccal vaccine in the prior 5 years was uncommon.
about therapeutic limitation were explicitly stated in Indeed, 4.5% of HCAP patients were vaccinated
22 patients, 15 in the HCAP group and seven in the compared with 10% of CAP (P 5 .45). Complications
CAP group (34% vs 3.8%, P , 0,001). and outcomes are shown in Table 2. CAP and HCAP
The most frequently prescribed antibiotic thera- cohorts again presented differences. Patients with
pies in patients with HCAP were combination ther- HCAP had higher rates of complications, which were
apy, including third-generation cephalosporin and generally associated with S pneumoniae bacteremia.
macrolides (36%) or amoxicillin-clavulanate (25%). Interestingly, among HCAP subgroups, patients with
In patients with CAP, the combination therapy with prior hospitalization achieved the highest mortality
third-generation cephalosporin and macrolides was (12 out of 25, 48%), being 18% in elderly care group
administered in 46%, amoxicillin-clavulanate in 21%, (three of 18) and 12% (one of eight) in the chemotherapy
and nonpseudomonal fluoroquinolones in 11%. group. No significant differences in age, comor-
Using the 2008 CLSI breakpoints, nonsusceptibility bidities, and severity of illness were documented among
to penicillin (MIC ⱖ 4 mg/mL) was documented in subgroups. Univariate analysis (Table 3) revealed that
only three of 228 (1.5%) episodes. Three isolates were 30-day mortality was influenced by several underlying

1140 Original Research

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians
Table 2—Univariate Analysis for Conditions Associated With HCAP vs CAP in Bacteremic Pneumococcal Pneumonia

Variable HCAP, n 5 44 CAP, n 5 184 P Value

Median age (IQR) 77 (68-83) 68 (42-79) , .001
Charlson index 2.81 6 1.9 1.23 6 1.42 , .001
Comorbidities
Cardiomyopathy 14 (31.8) 39 (21.1) .16
Lung disease 21 (47.7) 42 (22.8) , .01
Liver disease 9 (20.4) 18 (9.7) .06
Renal disease 6 (13.6) 16 (8.6) .39
Pao2 /Fio2 (, 250) 18 (40.9) 56 (30.4) .21
Shock 11 (25) 53 (28.8) .71
Vasopressor use 4 (9) 34 (18.4) .17
ICU admission 5 (11.3) 47 (25.5) , .05
Multilobar involvement 18 (40.9) 83 (45) .73
Mechanical ventilation 4 (9) 35 (19) .17
Pleural effusion 17 (38.6) 54 (29.3) .27
Macrolide resistance 16 (36.3) 22 (11.9) , .01
PSI score class IV-V 42 (95.4) 121 (65.7) , .001
CURB-65 ⱖ 3 22 (50) 65 (35) .08
30-d mortality 13 (29.5) 14 (7.6) , .001
LOS . 10 d 12/29 (41.3) 84/167 (50.2) .82
Data given as No. (%) unless otherwise noted. LOS 5 length of stay. See Table 1 for expansion of other abbreviations.

conditions, reflected by higher Charlson index. Other OR [aOR] 5 3.65, 95% CI, 1.54-8.66; Hosmer-
complications (level of consciousness, vasoactive Lemeshow goodness-of-fit P 5 .942) than patients
drugs use, multilobar involvement, Pao2/Fio2 ratio) with CAP. After adjustment for CURB-65 score (0-2
were also associated with 30-day mortality. The vs ⱖ 3) patients with HCAP had a significantly higher
HCAP cohort suffered a higher mortality rate (29.5% 30-day mortality (aOR 5 4.58; 95% CI, 1.91-10.94;
vs 7.6%, P , .001). The Kaplan-Meier survival curves Hosmer-Lemeshow goodness-of-fit P 5 .550) than
(Fig 1) of HCAP vs CAP cohorts showed statistically patients with CAP.
significant differences (P 5 .001). After adjustment for When only class IV-V patients (n 5 163) were ana-
PSI class (class I-III vs class IV-V), patients with HCAP lyzed, HCAP was associated with higher mortality in
had a significantly higher 30-day mortality (adjusted the univariate analysis (OR 5 3.72; 95% CI, 1.56-8.90)
and after adjustment for age (aOR 5 3.76; 95% CI,
1.54-9.15; Hosmer-Lemeshow goodness-of-fit P 5 .830)
Table 3—Univariate Analysis of Potential Predisposing compared with CAP. Figure 2 shows the estimated
Factors for 30-d Mortality Among 228 Patients With
Bacteremic Pneumococcal Pneumonia

Predisposing Factors OR for Death (95% CI) P Value

Age . 65 y 2.09 (0.84-5.17) .14

Alcoholism 2.70 (1.03-7.10) .05
HCAP 5.09 (2.18-11.87) , .001
Comorbidity
Cardiomyopathy 1.83 (0.77-4.37) .22
Lung disease 2.35 (1.03-5.35) .05
Liver disease 2.58 (0.93-7.15) .10
Renal disease 0.72 (0.16-3.28) 1.00
Charlson index 1.41 (1.13-1.76) , .01
Severity at admission
Altered mental status 3.37 (1.44-7.91) , .01
RR . 30/min 3.30 (1.38-7.91) , .01
Pao2 /Fio2 , 250 6.30 (2.60-15.23) , .01
Acute renal failure 1.23 (0.55-2.77) .68
Multilobar involvement 6.79 (2.47-18.60) .001
Vasopressor use 4.43 (1.86-10.55) .001
ICU admission 3.22 (1.39-7.41) , .01
Antibiotic therapy
Monotherapy 1.68 (0.73-3.85) .22
Figure 1. Kaplan-Meier survival curve for HCAP vs CAP cohorts
OR 5 odds ratio; RR 5 respiratory rate. See Table 1 for expansion of (censored at 30 days). CAP 5 community-acquired pneumonia;
other abbreviation. HCAP 5 health-care-associated pneumonia.

www.chestpubs.org CHEST / 137 / 5 / MAY, 2010 1141

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians
 monia, and evaluates the implications of HCAP in
the ICU. Our findings endorse the concept that
patients with pneumococcal HCAP present signifi-
cant differences in terms of demographics, severity of
illness, and complications. These differences are not
only of academic interest, since HCAP was associated
with higher mortality. Interestingly, inappropriate
therapy was uncommon when isolates were classified
using 2008 CLSI breakpoints.8 Although PSI was pri-
mary designed to assess severity in patients with CAP,
when patients with HCAP were analyzed we found
an excess of mortality comparing with the original
studies evaluating this score (Fig 2). A lower ICU
admission and a trend toward lower rates of mech-
anical ventilation and need for vasopressors were
reported for patients having HCAP.
Figure 2. Estimated and observed mortality for patients with PSI
In a retrospective cohort study, Kollef et al12
classes IV and V. PSI 5 pneumonia severity index. See Figure 1 reviewed a large multi-institutional database of more
legend for expansion of other abbreviations. than 4,500 positive-culture patients and identified
20% of patients as having HCAP. S pneumoniae was
and observed mortality for patients with PSI classes documented in 16% of patients with CAP and 5% of
IV and V. Finally, a multivariable logistic regression patients with HCAP. In a study in Spain that used a
analysis (Table 4) identified a significantly higher similar design,4 HCAP was identified in 17.3% of the
30-day mortality in patients with HCAP (OR 5 5.56; cohort, S pneumoniae being the predominant organism.
95% CI, 1.86-16.5), patients with Pao2/Fio2 ratio In an Italian study,1 increased severity, longer hospi-
, 250 (OR 3.90; 95% CI, 1.42-10.71), and need for talization, and higher mortality were reported for the
vasopressor therapy (OR 5 4.22; 95% CI, 1.37-12.9); HCAP cohort. Unfortunately, this study reported
Hosmer-Lemeshow goodness-of-fit P 5 .360. The only limited information on organisms and ICU care.
model did not change when patients receiving discor- A recent study reported that patients with HCAP
dant therapy were excluded. Also, when patients with were more likely to receive inappropriate empirical
CURB-65 ⱖ 3 were analyzed (n 5 87), HCAP was also antibiotic coverage than those with CAP.2 The same
associated with higher mortality in the univariate anal- group found an association between inappropriate
ysis (OR 5 3.14; 95% CI, 1.04-9.43) and after adjust- empirical initial therapy and mortality in a logistic
ment for age (aOR 5 3.15; 95% CI, 1.01-9.83; regression analysis (OR 5 2.88; 95% CI, 1.46-5.67).
Hosmer-Lemeshow goodness-of-fit P 5 .605) com- Late escalation did not improve survival.13 Accord-
pared with CAP. When patients not admitted to the ing to the 2005 American Thoracic Society/Infec-
ICU were analyzed, patients with HCAP were older, tious Disease Society of America guidelines, patients
had higher Charlson index and more hypoxemia, and
were more likely to be class IV/V PSI score (Table 5).
Table 5—Characteristics of 176 Patients With
Bacteremic Pneumococcal Pneumonia Not Admitted to
ICU
Discussion
HCAP, CAP,
This article compares CAP with HCAP specifically Variable n 5 39 n 5 137 P Value
in monomicrobial bacteremic pneumococcal pneu- Median age (IQR) 77 (72-83) 70 (46-79) .001
Charlson index, mean 6 SD 2.92 6 1.99 1.13 6 1.36 , .001
Comorbidities 32 (82) 81 (59) , .01
Table 4—Results of the Multivariable Regression Cardiomyopathy 13 (33) 23 (17) , .05
Logistic Analysis for 30-d Mortality Lung disease 18 (46) 34 (25) , .05
Liver disease 8 (20.5) 11 (8) , .05
Variable OR (95% CI) P Value
Renal disease 6 (15) 9 (6.5) .10
HCAP 5.56 (1.86-16.59) , .05 Pao2/Fio2 (, 250) 13 (33) 23 (17) , .05
Need for vasopressors 4.22 (1.37-12.99) .01 Multilobar involvement 13 (33) 43 (31) .84
Pao2/Fio2 , 250 3.90 (1.42-10.71) , .01 PSI score class IV-V 37 (94) 79 (57) , .001
Alcoholism 2.04 (0.61-6.74) .24 CURB-65 ⱖ 3 18 (46) 35 (25) .01
Altered mental status 1.56 (0.57-4.24) .38 30-d Mortality 10 (26) 5 (3.6) , .001
Charlson comorbidities index 1.24 (0.94-1.63) .11
Data given as No. (%) unless otherwise specified. See Table 1 for expansion
See Tables 1 and 3 for expansion of abbreviations. of abbreviations.

1142 Original Research

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians
 Table 6—Relationship Between Death/ICU Admission in Different Health-Care Associated Infections Studies

Current Study Carratalà et al4 McDonald et al20

Variable HCAP CAP HCAP CAP HCAI CAI

Mortality, No. (%) 13 (29.5) 14 (7.6) 13 (10.3) 26 (4.3) 35 (19.7) 18 (13.6)
ICU admission, No. (%) 5 (11.3) 47 (25.5) 8 (6.3) 52 (8.7) 40 (22.5) 28 (21.2)
Death/ICU ratio 2.6 0.3 1.6 0.5 0.9 0.6
CAI 5 community-acquired infection; HCAI 5 health-care-associated infection. See Table 1 for expansion of other abbreviations.

with HCAP and CAP are at risk for infection with terms of predisposing factors and outcomes. Another
the same multidrug-resistant pathogens.3 However, strength was that it focused on one pathogen with
in the HCAP group a substantial number of patients definite diagnosis (blood cultures), but a full picture
were also infected with microorganisms that cause of pneumococcal disease requires inclusion of non-
CAP (pneumococcus, Legionella, and even virus). bacteremic cases. An additional strength is that our
Many recent publications have reviewed the role of study is a prospective analysis and the diagnosis was
gram-negative bacilli or Staphylococcus aureus,14-17 based on clinical grounds. This contrasts with many
but acceptance of the HCAP definition is still reported HCAP studies in which the diagnosis was
controversial. A recent respiratory forum18 suggested retrospective and based on administrative records.
describing different categories based on the following: One limitation, which is also inherent to prior stud-
(1) need for mechanical ventilation, (2) prior antibi- ies on HCAP,1,2,4,5,9,12-14 is the fact that the distribu-
otics for more than 3 days within the past 6 months, tion of HCAP in other geographical areas is expected
and (3) poor functional status, defined as an activi- to be different. Other limitations include lack of a
ties of daily living score of . 12.5. Wunderink19 also measure of patient mobility and that the HCAP
noted that the distinction between HCAP and CAP patient population is skewed to patients with a his-
has never been totally clear. tory of previous hospitalization and those coming
In the present study, only five patients with HCAP from long-term facilities. In addition to these prior
received mechanical ventilation. The diagnosis of reports on HCAP, our study provides additional
HCAP was significantly associated with increased insight into the demographics of S pneumoniae HCAP
30-day mortality but did not increase the probability and is relevant as it provides additional evidence in
of receiving discordant therapy when assessed using the field of health-care-associated infections, in the
2008 CLSI breakpoints. Previous reports have classi- light of recently published guidelines for manage-
fied resistant organisms as nonsusceptible, based on ment of HCAP.3
older breakpoints.8 Therefore, the number of patients In summary, HCAP accounts for a significant num-
receiving therapies classified as inappropriate was ber of cases of pneumococcal pneumonia, which,
substantially higher in the past, at least for pneumo- using traditional definitions, would be classified as
coccal HCAP. CAP. The HCAP cohort was significantly older, had
An important issue is that of nonaggressive or lim- more comorbidities, higher mean PSI scores, and a
ited treatment. The HCAP group was more severely significantly higher mortality rate. HCAP accounts
ill, but made much lower use of advanced care for a significantly lower proportion of patients admit-
modalities. How much of the excess mortality was ted to the ICU, probably because most of these
because of physician or patient preference for limit- HCAP patients were elderly and had more comor-
ing the extent of support given due to “terminal” bid conditions. Our findings suggest that, at least for
comorbidities? Although in the HCAP group mortal- pneumococcal HCAP, excess mortality was due in
ity risk was more than twice the ICU admission rate part to the decision of the physician or patient to limit
(29.5% vs 11.3%), in patients with CAP the risk for the extent of support given to patients of advanced
mortality was one-third of ICU admission rate (7.6% age and with comorbidities.
vs 25.5%). This information is concordant with find-
ings reported for HCAP4 and bacteremia,20 suggest-
ing that these patients are less likely to be admitted in Appendix
the ICU despite higher mortality (Table 6) and should The Proyecto Corporativo de Neumonía (PROCORNEU)
be considered as a separate entity. Study Group is formed by E. Bouza (Hospital Gregorio
Our study has some strengths and limitations. Marañon); J. Liñares (Hospital Bellvitge); J. Rello (IISPV);
A. Torres (Hospital Clinic i Provincial); E. Perez-Trallero
One is its sample size: although the HCAP cohort (Hospital Donosti); V. Ausina (Hospital Germans Trias I Pujol);
contained fewer than 50 patients from a single cen- J. Valles (Hospital Parc Taulí); E. Garcia (CSIC); and A. Gonzalez
ter, it was large enough to identify distinctions in de la Campa (ISCIII).

www.chestpubs.org CHEST / 137 / 5 / MAY, 2010 1143

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians
 Acknowledgments identification, management, and outcomes—proceedings
Author contributions: Dr Rello: contributed to study design, of the HCAP Summit. Clin Infect Dis. 2008;46(Suppl 4):
analysis of data, writing the first draft, and creating the final ver- S296-S334.
sion of the manuscript. 6. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to
Dr Luján: contributed to study design, enrolling patients, rec- identify low-risk patients with community-acquired pneumo-
ording variables, conducting follow-up, collecting samples, nia. N Engl J Med. 1997;336(4):243-250.
analyzing data, writing the first draft, and creating the final version 7. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new
of the manuscript. method of classifying prognostic comorbidity in longitu-
Dr Gallego: contributed to study design, analysis of data, writing dinal studies: development and validation. J Chronic Dis.
the first draft, and creating the final version of the manuscript.
Dr Vallés: contributed to enrolling patients, recording variables, 1987;40(5):373-383.
conducting follow-up, collecting samples, and creating the final ver- 8. Clinical and Laboratory Standards Institute. Performance
sion of the manuscript. Standards for Antimicrobial Susceptibility Testing. Seventeenth
Dr Belmonte: contributed to enrolling patients, recording vari- Informational Supplement. Document M100-S17. Wayne, PA:
ables, conducting follow-up, collecting samples, and creating the Clinical and Laboratory Standards Institute; 2008.
final version of the manuscript. 9. Lujan M, Gallego M, Fontanals D, Mariscal D, Rello J.
Dr Fontanals: contributed to enrolling patients, recording var- Prospective observational study of bacteremic pneumococcal
iables, conducting follow-up, collecting samples, and creating the pneumonia: effect of discordant therapy on mortality.
final version of the manuscript. Crit Care Med. 2004;32(3):625-631.
Dr Diaz: contributed to study design and creating the final version
of the manuscript. 10. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events
Dr Lisboa: contributed to analysis of data, writing the first draft, per variable in logistic and Cox regression. Am J Epidemiol.
and creating the final version of the manuscript. 2007;165(6):710-718.
Financial/nonfinancial disclosures: The authors have reported 11. Hosmer DW, Hosmer T, Le Cessie S, Lemeshow S. A com-
to CHEST that no potential conflicts of interest exist with any parison of goodness-of-fit tests for the logistic regression
companies/organizations whose products or services may be dis- model. Stat Med. 1997;16(9):965-980.
cussed in this article. 12. Kollef MH, Shorr A, Tabak YP, Gupta V, Liu LZ, Johannes
Other contributions: We are indebted to Michael Maudsley for RS. Epidemiology and outcomes of health-care-associated
help in editing the manuscript and David Suárez for statistical
pneumonia: results from a large US database of culture-positive
support.
pneumonia. Chest. 2005;128(6):3854-3862.
13. Zilberberg MD, Shorr AF, Micek ST, Mody SH, Kollef MH.
Antimicrobial therapy escalation and hospital mortality among
patients with health-care-associated pneumonia: a single cen-
References ter experience. Chest. 2008;134(5):963-968.
1. Venditti M, Falcone M, Corrao S, Licata G, Serra P; Study 14. Webster D, Chui L, Tyrrell GJ, Marrie TJ. Health care-
Group of the Italian Society of Internal Medicine. Outcomes associated Staphylococcus aureus pneumonia. Can J Infect
of patients hospitalized with community-acquired, health care- Dis Med Microbiol. 2007;18(3):181-188.
associated, and hospital-acquired pneumonia. Ann Intern Med. 15. Anand N, Kollef MH. The alphabet soup of pneumonia: CAP,
2009;150(1):19-26. HAP, HCAP, NHAP, and VAP. Semin Respir Crit Care Med.
2. Micek ST, Kollef KE, Reichley RM, Roubinian N, Kollef MH. 2009;30(1):3-9.
Health care-associated pneumonia and community-acquired 16. Restrepo MI, Anzueto A. The role of gram-negative bacteria
pneumonia: a single-center experience. Antimicrob Agents in healthcare-associated pneumonia. Semin Respir Crit Care
Chemother. 2007;51(10):3568-3573. Med. 2009;30(1):61-66.
3. American Thoracic Society; Infectious Diseases Society of 17. Lam AP, Wunderink RG. The role of MRSA in healthcare-
America. Guidelines for the management of adults with associated pneumonia. Semin Respir Crit Care Med. 2009;
hospital-acquired, ventilator-associated, and healthcare- 30(1):52-60.
associated pneumonia. Am J Respir Crit Care Med. 2005;171(4): 18. Tillotson GS, Ball P. Meeting report: Fourth Forum on
388-416. Respiratory Tract Infections, Sitges, Spain, 8-11 February
4. Carratalà J, Mykietiuk A, Fernández-Sabé N, et al. Health 2007. J Antimicrob Chemother. 2007;60(3):464-469.
care-associated pneumonia requiring hospital admission: 19. Wunderink RG. Healthcare-associated bacteremia: stirring
epidemiology, antibiotic therapy, and clinical outcomes. Arch the mud. Crit Care Med. 2006;34(10):2685-2686.
Intern Med. 2007;167(13):1393-1399. 20. McDonald JR, Friedman ND, Stout JE, Sexton DJ, Kaye KS.
5. Kollef MH, Morrow LE, Baughman RP, et al. Health care- Risk factors for ineffective therapy in patients with blood-
associated pneumonia (HCAP): a critical appraisal to improve stream infection. Arch Intern Med. 2005;165(3):308-313.

1144 Original Research

Downloaded from chestjournal.chestpubs.org by Kimberly Henricks on May 4, 2010

© 2010 American College of Chest Physicians