Académique Documents
Professionnel Documents
Culture Documents
Study objectives: To describe the causative organisms and factors associated with bacterial
pneumonia and to assess its impact on the outcome of hospitalized patients with HIV.
Design: Prospective, observational.
Setting: A university-affiliated medical center.
Methods: We included 1,225 consecutive hospital admissions, from April 1995 through March
1998, of 599 adults with HIV. We collected data on APACHE II (acute physiology and chronic
health evaluation II) score, leukocyte and CD4ⴙ lymphocyte counts, length of hospital stay, ICU
admission rate, and case-fatality rate. Chest radiographs and laboratory results were reviewed.
The presence of bacterial pneumonia was noted.
Results: Bacterial pneumonia was diagnosed in 111 hospitalizations (9%): 80 (72%) were
community-acquired infections. The CD4ⴙ lymphocyte count was lower (median, 38 vs 66/L,
p ⴝ 0.0027), APACHE II score higher (17 vs 13, p < 0.0001), length of hospital stay longer
(median, 6 vs 4), and ICU admission (28% vs 9%) and case-fatality rates (21% vs 4%) higher in
patients with bacterial pneumonia compared with those without bacterial pneumonia. The most
common pathogen was Pseudomonas aeruginosa (32 admissions), followed by Streptococcus
pneumoniae (22 admissions), Staphylococcus aureus (16 admissions), and Haemophilus influenzae
(11 admissions). Thirty-three (30%) of the pneumonias were bacteremic. Bacteremia was more
common in pneumococcal than in pseudomonal pneumonia (95% vs 9%, p < 0.0001). Compared
with patients with pneumococcal pneumonia, patients with pseudomonal pneumonia had lower
leukocyte and CD4ⴙ lymphocyte counts, longer hospital stay, and similar case-fatality rate.
Conclusions: P aeruginosa is becoming a common cause of both community-acquired and
nosocomial bacterial pneumonia in hospitalized patients with HIV, especially in those with low
leukocyte and CD4ⴙ lymphocyte counts. (CHEST 2000; 117:1017–1022)
Key words: AIDS; bacterial pneumonia; HIV; ICU admission; mortality; Pseudomonas aeruginosa infection; Strepto-
coccus pneumoniae infection
Table 4 —Differences in Length of Hospital Stay, ICU penicillin-resistant compared with 44% in Spain.17 In
Admission Rate, and In-Hospital Mortality Between a study from France of HIV-infected inpatients with
Patients With or Without Bacterial Pneumonia bacterial pneumonia, 75% of the pneumococci had
Bacterial Pneumonia decreased susceptibility to penicillin.11 In our study,
only 9% of the pneumococci had decreased suscep-
Yes No
Variable (N ⫽ 111) (N ⫽ 1,1114) p Value tibility to penicillin.
Bacteremic and nonbacteremic P aeruginosa
Length of hospital stay, 9.1 ⫾ 9.4 (6) 5.9 ⫾ 6.2 (4) ⬍0.0001
infections are becoming more common in patients
days (median)
ICU admission, No. (%) 31 (28) 102 (9) ⬍0.0001 with HIV.18 An effective defense against P aerugi-
In-hospital mortality, No. 23 (21) 44 (4) ⬍0.0001 nosa depends on functions of the epithelial cell
(%) barrier, antibodies, complement system, phago-
cytic cells, and lymphocytes.19 Traditionally, P
aeruginosa infection has been associated with
neutropenia, hospitalization, central venous cath-
11 of the 36 pneumococcal pneumonias (31%) eters, burn wounds, bronchiectasis, and cystic
were bacteremic.5 In our study, 95% of the pneu- fibrosis.19 The development of bronchiectasis and
mococcal pneumonias were bacteremic, suggest- defects in granulocytes, chemotaxis, phagocytosis,
ing underdiagnosis of nonbacteremic pneumococ- and bacterial killing have been described in HIV-
cal pneumonia. Because our study was purely infected patients and have predisposed them to
observational, we did not have any control on the infection with P aeruginosa.2,3,20,21 In a prospec-
management plan of the individual patients. The tive study of HIV-infected inpatients from France,
common medical practice in our emergency de- 6 of the 33 identified pathogens causing bacterial
partment included giving at least one IV dose each pneumonia (18%) were P aeruginosa.11 In the
of trimethoprim/sulfamethoxazole and a third-gen- multicenter study by the Pulmonary Complica-
eration cephalosporin to ill-looking HIV-infected tions of HIV Infection Study Group, P aeruginosa
patients with suspected pneumonia before obtain- accounted for 6 of the 89 organisms identified as
ing respiratory specimens. This practice may have causing bacterial pneumonia (7%).5 Other studies
led to a falsely decreased incidence of pneumo- have highlighted the importance of P aeruginosa
coccal pneumonia in our study. pneumonia in patients with HIV infection, includ-
Penicillin remains the drug of choice for patients ing a rise in its incidence more recently.22–26 Most
infected with a susceptible strain of S pneumoniae. of these infections are community-acquired.22–26
In the United States, 7% of S pneumoniae are The incidence of P aeruginosa infection in hospi-
P aeruginosa S pneumoniae
Variable (n ⫽ 32) (n ⫽ 22) p Value
talized patients with HIV is estimated to be 5.8% and therapeutic approaches varied according to
and on the rise.26 In our study, the incidence of each patient’s primary physician. The initiation of
pneumonia from P aeruginosa in hospitalized pa- empiric antibiotic therapy before obtaining respi-
tients with HIV infection was 2.6%. Unlike previ- ratory specimen samples may have led to the
ous studies, we found P aeruginosa, not S pneu- underdiagnosis of bacterial pneumonia, pneumo-
moniae, to be the most common pathogen causing coccal in particular. Although there were patients
bacterial pneumonia in our patients. The underdi- whose clinical presentation and response to ther-
agnosis of pneumococcal pneumonia in our study, apy were consistent with bacterial pneumonia,
caused by technical difficulties in culturing S they were not considered as having bacterial pneu-
pneumoniae and empiric antibiotic use before monia unless they had microbiologic evidence of
obtaining adequate respiratory samples, may bacterial pneumonia. This may have led to under-
partly explain this unexpected finding. The low
estimation of the incidence of bacterial pneumo-
CD4⫹ lymphocyte count in our patients with
nia. Because we were unable to document the
pseudomonal pneumonia was similar to the find-
patients’ antiretroviral treatment and trimetho-
ings in other studies.22,25
Atypical bacterial pneumonia was diagnosed in prim/sulfamethoxazole use before hospitalization,
only one patient in our study. The Pulmonary Com- we could not determine their effect on the inci-
plications of HIV Infection Study Group reported dence and type of the bacterial pneumonia.
only one case of L pneumophila among 237 episodes In summary, we have described the incidence and
of bacterial pneumonia in HIV-positive patients.5 types of bacterial pneumonia in hospitalized patients
There are case reports of pneumonia caused by with HIV infection. Pseudomonal pneumonia is be-
Bordetella bronchiseptica in patients with HIV infec- coming a common pulmonary complication, espe-
tion.27,28 B bronchiseptica was the only unusual cially in patients with low leukocyte and CD4⫹
bacterial pathogen found to cause pneumonia in our lymphocyte counts. Compared with pneumococcal
study. pneumonia, pseudomonal pneumonia is associated
Recent trends among persons dying of HIV infec- with a lower incidence of bacteremia and a longer
tion in the United States show an increase in the hospital stay. Despite the low CD4⫹ lymphocyte
percentage of deaths associated with bacterial pneu- and leukocyte counts associated with pseudomonal
monia.29 An Italian study of 350 episodes of bacterial pneumonia, the mortality rate is only 19%. We need
pneumonia reported a case-fatality rate of 27%.12 to closely monitor the changes in disease patterns,
The case-fatality rate was 21% in our study. Com- including the pathogens causing pneumonia, in dif-
pared with patients without bacterial pneumonia, ferent institutions and regions. Our observations
patients with bacterial pneumonia had longer length suggest that pseudomonal pneumonia should be
of hospital stay and higher ICU admission and included in the differential diagnosis of both com-
case-fatality rates in our study. This indicates the munity-acquired and nosocomial pneumonia in pa-
adverse impact of bacterial pneumonia on the mor- tients with HIV infection.
bidity and mortality of hospitalized patients with
HIV infection. ACKNOWLEDGMENT: We thank Dr. David L. Armbruster
Our study has several limitations. The diagnostic for editing our manuscript.