Download Causes of Fever in Patients Infected with Human Immunodeficiency

320
Causes of Fever in Patients Infected with Human Immunodeficiency Virus Who
Were Admitted to Boston City Hospital
Lawrence M. Barat, Julia E. Gunn, Kathleen A. Steger,
Chris J. Perkins, and Donald E. Craven
From The Maxwell Finland Laboratory for Infectious Diseases, Section
of Infectious Diseases, Boston City Hospital and Boston University
School of Medicine; Adult Clinical AIDS Program, Department of
Medicine, Boston City Hospital; and Epidemiology and Health Services
Sections, Boston University School of Public Health,
Boston, Massachusetts
We prospectively studied causes of fever in patients with human immunodeficiency virus (HIV)
infection that required admission to a municipal hospital. A total of 168 HIV-infected persons were
admitted for 220 episodes of fever: 72% were male, 80% were nonwhite, 65% reported prior injection
drug use, and 74% had a baseline CD4 lymphocyte count of <2oo/mm3 • Bacterial infections,
principally pneumonia, accounted for >60% of the episodes; Streptococcus pneumoniae and Staphylococcus aureuswere most commonly isolated. Pneumocystis carinii pneumonia (PCP) and disseminated infection with Mycobacterium avium complex (MAC) comprised 53% of the remaining sources
of fever. In comparison with episodes of fever due to nonbacterial causes, those associated with
common bacterial infections were significantly more likely to involve patients with a history of
injection drug use (P = .02), higher admission leukocyte count (P < .004), shorter duration of fever
(P = .003), shorter hospital stays (P = .0001), and a CD4 count of >loo/mm3 (P = .002). We
conclude that bacterial infection, especially pneumonia, is a common cause of fever in HIV-infected
patients admitted to our hospital. Patients with bacterial infections are more likely to report a
history of injection drug use and have CD4 counts of > lOO/mm3, shorter duration of fever, decreased
length of hospitalization, and lower mortality than patients with fever due to PCP, disseminated
MAC infection, or other causes.
Fever is a frequent symptom in people infected with HIV.
Infection is a common cause of fever, especially in the symptomatic stages of HIV disease [1-16]. In the United States,
fever may be associated with opportunistic infections, such as
Pneumocystis carinii pneumonia (PCP) or disseminated Mycobacterium avium complex (MAC) infection [1-3], or with bacterial infections due to Streptococcus pneumoniae, Staphylococcus aureus, and Haemophilus injiuenzae [1-16].
Previous detailed studies of fever in HIV-infected persons
have focused on fever of unknown origin or on outpatients
[17-19]. Sepkowitz and co-workers described causes of fever
in an outpatient cohort of predominantly homosexual males
with advanced HIV disease [20]. Approximately one-half of
the cohort had fevers during the study, most of which were
caused by opportunistic infections, such as PCP or MAC infection, or by bacterial infections due to infected central venous
catheters. Our study analyzes causes of fever in HIV-infected
persons admitted for fever to the medical service at a municipal
Received 31 October 1995; revised 3 April 1996.
Presented in part at the 7th International Conference on AIDS (abstract no.
MB2413), held in Florence in June 1991, and at the 8th International Conference on AIDS (abstract no. PoB3832), held in Amsterdam in July 1992.
Reprints or correspondence: Dr. Donald E. Craven, Thorndike 311, Boston
City Hospital, 818 Harrison Avenue, Boston, Massachusetts 02118.
Clinical Infectious Diseases 1996;23:320-8
© 1996 by The Universityof Chicago. All rights reserved.
1058--4838/96/2302-0017$02.00
hospital, and it examines differences in risk factors and outcome for those with bacterial infections vs. other causes of
fever.
Methods
Study Population
This prospective study was conducted in the inpatient medical service at Boston City Hospital from November 1990
through March 1991 and again from October 1991 through
March 1992. Eligibility requirements were admission to the
inpatient medical service at Boston City Hospital and evidence
of HIV infection, either by documentation of HIV-1 seropositivity or diagnosis of AIDS on the basis of the 1987 criteria
of the Centers for Disease Control and Prevention (CDC) [21].
In order to be enrolled, a patient had to have an oral temperature of ~100.4°F on 2 successive days (or of ~101°F during
any single measurement within the first 48 hours of admission)
and had to undergo a diagnostic evaluation for the cause of
fever. Inpatients were identified through the Clinical AIDS
Program's HIV liaison service, which is notified of all HIVinfected patients at the time of admission. Most of the patients
were followed in the Boston City Hospital's Adult Clinical
AIDS Program, which provides care to >850 HIV-infected
patients, of whom >70% have a history of injection drug use,
65% are nonwhite, 30% are women, and 15% are homeless.
Demographic information, medical history, laboratory data,
description of the hospital course, results of diagnostic studies,
em 1996;23 (August)
Fever and HIV Infection
and relevant outcomes were collected from a medical record
review and recorded on standardized forms by one of the authors (L.B., C.P., or lG.). Data from autopsy reports were
included for the two subjects on whom a postmortem examination was performed; both had non-Hodgkin's lymphoma. The
most recent absolute CD4 lymphocyte count (from within 6
months of the admission) was used for analysis. Mortality rates
were based on deaths during the admission.
In our investigation, 168 HIV-infected study patients had
220 episodes of fever that required hospitalization; these 168
included all HIV-infected patients admitted for fever to the
medical inpatient service during the study period. The cause
of fever was determined on the basis of a clinical diagnosis
proven by microbiological confirmation or a probable diagnosis
without microbiological confirmation; some patients had multiple causes of fever.
Diagnostic Criteria for Causes of Fever
The diagnosis of bacterial infection as a cause of fever was
made clinically, with microbiological confirmation by culture,
if possible; infections caused by mycobacterial species were
analyzed separately. The criteria for a diagnosis of bacterial
pneumonia were a clinical history of cough with sputum production; radiographic evidence of a new, focal alveolar infiltrate
or consolidation; and resolution of symptoms with antibiotic
therapy. Endocarditis was diagnosed as either probable or possible, with use of the criteria of von Reyn et al. [22]. Central
venous catheter infection was diagnosed either by the presence
of local cellulitis or purulent drainage at the catheter site or by
growth of a pathogen, such as Staphylococcus epidermidis, in
cultures of both the blood and catheter. Pyuria and growth of
a single species in urine culture (> 105 cfu/mL) were necessary
for the diagnosis of urinary tract infection.
Cellulitis was defined as a localized area of warmth, erythema, swelling, and tenderness, with or without purulent exudate, that responded to antibiotic therapy. A soft-tissue abscess
was defined by aspiration or drainage of purulent material from
the affected area. Osteomyelitis was defined by growth of a
pathogen in cultures of bone or blood, with radiographic or
radionuclide evidence of bone inflammation or destruction.
Septic arthritis was diagnosed following growth of a pathogen
in cultured joint fluid. A diagnosis of sinusitis was based on a
history of localized headaches, the air-fluid level on a sinus
radiograph or CT scan, and a clinical response to antibiotic
therapy.
Bacteremia was defined as growth of a pathogenic species
in blood culture. For cases in which these organisms could be
skin contaminants, two positive blood cultures were required.
Only primary bacteremia (that lacking a primary focus of infection) was considered as a separate cause of fever.
Opportunistic infections were diagnosed according to criteria
established in 1987 by the CDC [21]. Tumors required histologic diagnosis. Pancreatitis was defined as hyperamylasemia
321
with sonographic or radiographic evidence of pancreatic inflammation. Diagnosis of drug fever required the exclusion of
any other obvious potential sources of fever and the resolution
of fever following withdrawal of the suspected medication.
Comparative Analyses
Demographic information, baseline clinical status, laboratory findings, and outcomes for episodes in which a cause of
fever was identified were compared to those for episodes of
unexplained fever. Episodes of fever with a single cause vs.
those with multiple causes were analyzed in a similar fashion.
An analysis was performed comparing 109 episodes caused
by a single diagnosed bacterial infection (n = 104) or two
bacterial infections (n = 5) with 69 episodes due to nonbacterial, or other, causes (65 of these had a single cause and 4 had
two such causes). "Other" causes of fever included all parasitic, mycobacterial, viral, and fungal infections, as well as
noninfectious diagnoses such as drug fever, pancreatitis, and
tumor. Thirteen episodes of fever that were due to a bacterial
infection and another cause were excluded from this analysis.
Data Analysis
Data were entered into a Foxpro (Microsoft Corp., Redmond,
WA) database. Analysis was performed with use of the Statistical Package for the Social Sciences (SPSS, Chicago) [23]. A
two-tailed X2 test was utilized to derive P values for discrete
variables [24]. When cells contained less than five expected
values, a Fisher's exact test was performed. Risk ratios were
calculated as relative risks with 95% confidence intervals. Probabilities for nonparametric continuous variables were assessed
with Wilcoxon's rank sum test.
Results
Study Population
During the l l-month study period, 168 study subjects had
220 episodes of fever (figure 1). Of these 168 subjects, 131
(78%) had one episode, 26 (15%) had two episodes, 8 had
three episodes, 2 had four episodes, and 1 had five episodes
of fever requiring admission to the hospital. Of the 220 episodes, the majority involved patients who were male and nonwhite, had a history of injection drug use, and had AIDS or
advanced HIV disease (table 1). Antiretroviral medications had
been prescribed in 123 episodes (56%), and prophylactic regimens against PCP had been administered in 96 (44%), almost
half of which were with aerosolized pentamidine.
Causes of Fever
A cause of fever was identified in 191 (87%) of the 220
episodes. A single cause of fever was diagnosed in 169 episodes
322
Barat et al.
In 29 episodes,
nodiagnosis
was made
65
other
diagnoses
10 bacterial
infections in
5 episodes
13bacterial
infections
and 13other
diagnoses in
13episodes
1996;23 (August)
monary infections with S. pneumoniae accounted for 33% of
all microbiologically confirmed bacterial infections and twothirds of all culture-positive episodes of pneumonia. S. aureus
was the second most common blood pathogen isolated (25%);
endocarditis, central venous catheter infection, and pneumonia
were common primary sources. Escherichia coli was the causative agent in two-thirds of the episodes of fever due to urinary
tract infections. Bacterial infections that were not confirmed
by culture included 33 episodes of pneumonia, 8 skin and softtissue infections, and 5 episodes of sinusitis.
A total of 43 bacterial, mycobacterial, and fungal pathogens
were isolated from blood (figure 3). Two-thirds of all blood-
168patients admitted
with fever
104
bacterial
infections
ern
8 other
diagnoses
in4
episodes
Table 1. Demographic and clinical data from 220 episodes of fever
in 168 HIV-infected inpatients at Boston City Hospital.
Variable, with regard to patients
Total of127
bacterial infections
Figure 1. Summary of data regarding 168 patients with 220 episodes of fever requiring admission to the hospital: 131 patients who
had a single episode and 37 patients who had multiple episodes
(n = 89). No cause of fever was identified in 29 episodes (13%); in
the remaining 191 episodes, either one (n = 169) or two (n = 44)
causes of fever were diagnosed. Overall, we identified 127 bacterial
infections and 86 other causes of fever (P. carinii pneumonia, 31
episodes; mycobacterial infection, 15; fungal infection, 9; viral infection, 8; parasitic infection, 3; tumor, 7; pancreatitis,S; drug fever, 2;
aca1culous cholecystitis, 3; and miscellaneous, 3.
(88%), and two causes were identified in 22 episodes. Compared with the 191 episodes of fever in which the cause was
diagnosed, the 29 episodes in which no cause could be identified occurred significantly (P < .05) more often in males (90%
vs. 70%), nonwhites (97% vs. 77%), and patients who had a
serum albumin level of <3.3 mg/dL (39% vs. 74%), a serum
aspartate aminotransferase level greater than three times the
upper limit of normal (29% vs. 10%), a fever for <2 study
days (82% vs. 50%), and a hospital stay of <7 days (72% vs.
35%).
Bacterial Infections
Of the 213 causes of fever identified in 191 episodes of
fever, 60% were infections with common bacterial pathogens
(table 2). Community-acquired bacterial pneumonia accounted
for 73 (57%) of the total number of bacterial infections (figure
2). Bacterial infections were diagnosed in 88 (54%) of the 163
episodes of fever in patients with a baseline CD4 count of
<200/mm3 ; 50 (57%) of these 88 episodes were due to pneumonia.
Of the 127 bacterial infections, 65% were proved by microbiological confirmation and 35% were presumed (table 3). Pul-
Age (y)
18-30
31-40
41-50
>50
Sex
No. (%) of
episodes
41
125
41
13
(19)
(57)
(19)
(6)
Male
159 (72)
Female
Race
61 (28)
White
Black
Hispanic
Risk factor
Injection drug user (IDU)
Homosexual or bisexual male
Homosexual or bisexual male and IDU
Heterosexual
Other*
Clinical status t
AIDS
Symptomatic (no AIDS)
Asymptomatic
No prior evaluation
CD4 lymphocyte count
<200/mm 3
200- 500/mm3
>500/mm3
Not available
Prescribed medication
Zidovudine
Didanosine
Trimethoprim-sulfamethoxazole
Dapsone
Aerosolized pentamidine
Isoniazid
Rifampin
Ciprofloxacin
Nystatin or miconazole troches
Ketoconazole or fluconazole
45 (20)
145 (66)
30 (14)
140
20
3
51
7
(64)
(9)
(1)
(23)
(3)
104
65
39
12
(47)
(30)
(18)
(5)
163
26
8
23
(74)
(12)
(4)
(10)
103
20
19
31
46
(47)
(9)
(9)
(14)
(21)
20
7
5
58
34
(9)
(3)
(2)
(26)
(15)
* Includes recipients of blood and blood products and unknown factors.
On the basis of the 1987 CDC criteria (see text).
t
em
1996;23 (August)
Fever and HIV Infection
Table 2. Summary of 213 diagnosed causes of 191 episodes of
fever involvingHIV-infectedsubjects admittedto Boston City Hospital during the periods of study.
Cause of fever
Bacterial
Pneumonia
Skin and soft-tissue infection
Sinusitis
Endocarditis
Osteomyelitis or septic arthritis
Catheter sepsis
Urinary tract infection
Salmonellosis
Clostridium difficile colitis
Mycobacterial
Disseminated Mycobacterium avium complex
infection
Disseminated Mycobacterium tuberculosis
infection
Mycobacterium chelonae pneumonia
Parasitic
Pneumocystis carinii pneumonia
Cerebral toxoplasmosis
Viral
Disseminated varicella infection
Cytomegalovirus retinitis
Herpes simplex (primary or esophagitis)
Acute hepatitis B
Enteroviral meningitis
Fungal
Cryptococca1 meningitis
Candida endocarditis
Candida esophagitis
Tumor
Lymphoma
Kaposi's sarcoma (visceral)
Carcinoma
Other
Pancreatitis
Acalculous cholecystitis or sclerosing cholangitis
Drug fever
Lymphoid interstitial pneumonia
Gouty arthritis
Pneumothorax
No. of episodes
(% of all causes)
127 (60)
73
18
7
10
2
323
fever; nine patients with PCP had been prescribed prophylaxis.
Disseminated MAC infection accounted for another 6% of the
other causes of fever; ten of these infections were diagnosed
by blood culture and two by bone marrow aspiration. Both
patients whose blood yielded Mycobacterium tuberculosis had
no evidence of pulmonary parenchymal disease. Except for one
episode of PCP diagnosed in a patient with a CD4 lymphocyte
count of 240/mm 3 , the 59 AIDS-defining opportunistic infections were seen solely in subjects with baseline CD4 lymphocyte counts of <200/mm3 .
6
9
1
1
15 (7)
12
2
1
34 (16)
31
Admissions Involving Multiple Diagnoses
Of the 22 episodes in which there were two identified causes
of fever, 5 each were due to two bacterial infections, 4 had
two nonbacterial causes, and 13 had one of each (figure 1). No
significant differences were found in demographic information,
CD4 lymphocyte count, admission laboratory data, or outcome
measures between episodes in which two fever sources were
identified and those involving only one diagnosis.
3
8 (4)
2
2
2
1
1
9 (4)
6
1
2
7 (3)
Outcomes
Of the 10 episodes of fever lasting > 14 days, 3 were attributed to bacterial infections (lung abscess, sinusitis, and empyema), 6 had nonbacterial causes (PCP, pancreatitis, non-Hodgkin's lymphoma, acalculous cholecystitis, and disseminated
60
5
1
1
13 (6)
5
3
2
1
1
1
n =73
50
;;g
~
(/)
c:
40
0
13
~
.s
l1i
"a5
t5
co
30
20
co
10
NOTE. For 29 fevers, no source was identified; for 22, two sources were
identified.
stream infections were caused by bacterial species, and
S. aureus, S. pneumoniae, and other streptococcal species accounted for > 50% of the isolates.
Other Infectious Causes
Other infectious agents comprised 73 (34%) of the 213
causes of fever. Fifty-nine (81%) met the 1987 CDC criteria
for AIDS-defining opportunistic infections (table 2) [21]. PCP
accounted for 42% of the nonbacterial, or other, causes of
Figure 2. Summaryof 127 episodes of fever attributedto bacterial
infections,of which 83 (65%) were bacteriologically confirmed. Pneumonia accounted for 73 (57%) of the 127 bacterial infections, followed by skin and soft-tissue infection (14%), possible or probable
endocarditis (8%), and urinary tract infection (7%). The black portion
of each bar defines the percentage of infections that were associated
with bacteremia. The number of episodes in which each type of
infection was diagnosed (n) is indicated at the top of each bar
("other" includes mycobacterial, fungal, and viral infections as well
as all noninfectious etiologies, as listed in the text).
Barat et al.
324
Table 3. Pathogens isolated (and the primary site of infection) in
81 cases of culture-confirmed bacterial infection causing fever in
patients admitted to Boston City Hospital.
Organism(s): site or type of infection
Gram-positive bacteria (n = 58)
Streptococcus pneumoniae: pneumonia
Staphylococcus aureus:
Endocarditis
Bone and joint
Catheter sepsis
Pneumonia
Skin or soft tissue
Sinusitis
Streptococcus species*
Endocarditis
Skin or soft-tissue
Pneumonia
Staphylococcus epidermidis: catheter sepsis
Enterococcus faecalis: urinary tract
Bacillus cereus: endocarditis
Clostridium difficile: colitis
Gram-negative bacteria (n = 21)
Escherichia coli: urinary tract
Haemophilus influenzae (non-type b): pneumonia
Pseudomonas aeruginosa:
Pneumonia
Sinusitis
Skin or soft tissue
Moraxella catarrhalis: pneumonia
Klebsiella pneumoniae: urinary tract
Proteus mirabilis: urinary tract
Acinetobacter baumannii: catheter sepsis
Eikenella corrodens: skin or soft tissue
Salmonella enteritidis group G: bacteremia
Mixed (n = 2)
Bacteroides and Streptococcus species: skin or soft tissue
E. coli and Klebsiella oxytoca: skin or soft tissue
No. of
episodes
of fever
27
20
6
em 1996;23 (August)
was due to nonbacterial or other causes (n = 69), they were
significantly more likely (P < .05) to report a history of injection drug use; have an elevated admission leukocyte count,
higher mean absolute CD41ymphocyte count (137 ± 182/mm 3
vs. 64 ± 107/mnr'), shorter duration of fever, and decreased
hospital stays; and demonstrated a trend toward lower hospital
mortality. As shown in figure 4, nonbacterial or other causes
of fever, usually opportunistic infections, were more commonly
noted in patients with a CD4 count of < 100/mm 3 than in
patients with bacterial infections.
2
4
4
3
1
7
3
3
I
1
1
1
1
6
5
4
2
1
1
1
1
1
1
1
1
Discussion
Although fever may occur at any stage of HIV infection, it
is most commonly a symptom in people with advanced HIV
infection, and it may be caused by a wide spectrum ofinfectious
agents, tumors, or commonly prescribed drugs [17,20,25,26].
We were able to identify a cause in 87% of the 220 episodes
of fever in this study of HIV-infected inpatients, which is consistent with the 83% reported by Sepkowitz and co-workers
[20] with regard to a cohort of outpatients.
Bacterial pneumonia, the single most common diagnosis,
accounted for more than one-half of the bacterial infections
identified and one-third of all causes overall. S. pneumoniae
was the etiologic agent in 66% ofthe culture-confirmed cases of
pneumonia. Thirty percent of the patients with pneumococcal
pneumonia in our study had bacteremia, a rate that is consistent
with that in other reports [5, 6, 8, 9]. Our data support earlier
observations from this hospital and other centers, underscoring
the increased susceptibility of HIV-infected patients to bacterial
pneumonia [4, 5, 8-12].
Our findings, however, differ from those of Sepkowitz and
co-workers [20], who reported a 20% incidence of pneumonia
in their longitudinal study of outpatients, predominantly homo-
NOTE. Forty-five bacterial infections were not bacteriologically confirmed.
* Includes j3-hemolytic streptococci (n = 5) and viridans streptococci
(n = 2).
tuberculosis), and the remaining patient had cerebral toxoplasmosis and central venous catheter sepsis.
Ten subjects died during hospitalization: 3 had bacterial infections (sinusitis, pneumonia, and empyema/cellulitis), 6 had
nonbacterial conditions (2, non-Hodgkin's lymphoma; 2, PCP;
and 1 each, candidal endocarditis and pancreatitis), and 1 died
with unexplained fever. Four study subjects were febrile at the
time of their death (one each had pneumonia and pancreatitis,
and two had lymphoma).
Bacterial Infections vs. Other Causes
As shown in table 4, when patients with fever due to bacterial
infection (n = 109) were compared with patients whose fever
C. neoformans
C. tropicalis
M. avium complex (n = 10)
M. tuberculosis (n =2)
Figure 3. Summary of organisms isolated from blood cultures following 43 admissions of Hlv-infected patients with fever. Grampositive bacteria accounted for 25 (58%) of the pathogens isolated
and 86% of the bacterial isolates. S. pneumoniae and S. aureus were
the most common organisms isolated. Ten episodes of M avium
complex infection, two episodes of M tuberculosis bacteremia, and
two episodes of fungemia were identified.
em 1996;23 (August)
325
Fever and HIV Infection
Table4. Comparison of 109 HIV-infected subjects with fever due to bacterial infection and 69 with fever due to other conditions.
No. (%) of patients with fever due
to indicated cause
Variable
Age (y)
18-40
>40
Sex
Male
Female
Race
White
Nonwhite
Risk-related behavior
Injection drug use
Other
Medication on admission
Trimethoprim-sulfamethoxazole
Dapsone
Ciprofloxacin
Rifampin
Laboratory values
WBC count, > 11,000/mm3
CD4 lymphocyte count, < 100/mm 3
Outcome
Fever, ~3 d
Hospital stay, ~7 d
Death
Bacterial
Other*
(n == 109)
(n == 69)
P value
RR (95% CI)
80 (73)
29 (27)
51 (74)
18 (26)
.9
0.99 (0.83-1.19)
73 (67)
36 (33)
54 (78)
15 (22)
.1
0.86 (0.71-1.03)
27 (25)
82 (75)
14 (20)
55 (80)
.5
1.22 (0.69- 2.16)
79 (72)
30 (28)
38 (55)
31 (45)
.02
1.30 (1.02-1.66)
(12)
(10)
(7)
(7)
.2
.3
.008
.03
0.55 (0.21 ~ 1.46)
1.54 (0.67 - 3.51)
0.13 (0.20-1.06)
7 (6)
17 (16)
0
1 (1)
8
7
5
5
24 (22)t
4 (6)+
49 (52)
48 (76)
.004
.002
3.74 (1.36-10.32)
0.68 (0.54-0.86)
76 (73)
52 (48)
3 (3)
34 (51)
13 (19)
6 (9)
.003
.001
.09
1.44 (1.11-1.86)
2.53 (1.49-4.29)
0.32 (0.08-1.22)
* Nonbacterial causes include viral (n == 5), fungal (n == 7), and mycobacterial (n == 8) infection; Pneumocystis carinii pneumonia (n = 31); toxoplasmosis (n = 1);
tumors (n == 7); and noninfectious conditions (n == 10).
t n == 95.
+ n == 63.
sexual men, and found no cases attributable to S. pneumoniae
or H influenzae. This disparity probably reflects differences in
both study design and study populations. Our study population
was primarily nonwhite; injection drug use was included as
a risk-related behavior; and hospitalization was required for
inclusion in this study. Our subjects who reported a history of
injection drug use were significantly more likely to have a
bacterial infection, especially pneumonia, than were patients
with other risk-related behaviors (P < .02), a finding that confirms the recent report by Hirschtick and co-workers [16].
S. aureus, the second most common bacterial species isolated from our study patients, accounted for 10 (40%) of the
25 cases of gram-positive bacteremia, all of which were
secondary to endocarditis and central venous catheter infections. High rates of S. aureus infections in HIV-infected
patients have also been reported by other investigators [4,
13-15, 20, 27, 28].
Our data, and those of others, suggest that most bacterial
infections occur in the setting of advanced immunosuppression
(CD4 lymphocyte count, <200/mm3 ) [4-15, 20, 27, 28]. The
increased susceptibility of HIV-infected individuals to bacterial
infections is probably related to defects in lymphocytes, macrophages, and neutrophil function [26-30]. In those with
advanced HIV disease, impairments of chemotactic and phagocytic activity of neutrophils and macrophages have been demonstrated. Defects in antibody production and in the activity
of both classical and alternative complement pathways have
also been reported [29-33]. In addition, the risk of bacterial
infections and colonization in patients with AIDS may be altered by the use of prophylactic agents against opportunistic
infections [27, 28, 34].
Some investigators have suggested that bacterial infections
may accelerate the progression of HIV disease [14, 35, 36].
Selwyn and co-workers reported that injection drug users have
a substantial rate of pre-AIDS morbidity from pyogenic bacterial infections, such as pneumonia and sepsis [14]. Serious
bacterial infection or pulmonary tuberculosis accelerated the
rate of progression to AIDS by 3.7-fold when CD4lymphocyte
counts were >500/mm3 and by 220-fold when CD4+ lymphocyte counts were < 150/mm3 • The exact mechanism for this is
unclear, but it may be related to a direct effect or upregulation,
either through classical antigen presentation to the CD4 lym-
Barat et al.
326
80
~
n
=48
60
~
>
.S!
'0 40
til
Q)
~
:::l
0
en
20
o
<100
100-199
200-500
>500
CD4 cell count (lmm 3 )
Figure 4. Summary of the frequency of bacterial infection (solid
bars) vs. other sources of fever (hatched bars), as related to absolute
CD4+ T lymphocyte counts (zmrrr'), Frequencies were defined as the
percentages of all bacterialinfections (n = 95) and of all other sources
of fever (n = 63) among persons whose CD4 lymphocyte count was
determined (see text). Bacterial infectionswere commonover a wide
spectrumof CD4 counts. Other sources,which includedopportunistic
infections, were more common when the CD4 count was < 1OO/mm3 •
Thirteen episodes of fever caused by both a bacterial infection and
another source were excluded. Other causes included mycobacterial,
fungal, and viral infections, as well as all noninfectious etiologies(as
listed in the text). The number at the top of each bar is the number
of episodes.
phocyte or indirectly via superantigens, cytokines, and other
immune mediators [35-38].
In our study, other (nonbacterial) infections accounted for
34% of the causes of fever identified. Fifty-nine (82%) of these
were AIDS-defining opportunistic infections that met the 1987
CDC criteria, and 76% occurred in patients whose CD4 lymphocyte count was < 100/mm3 • Most of these were PCP or
disseminated MAC infections. Almost one-third of the patients
with PCP had been prescribed prophylaxis, a proportion suggesting the possibility of poor compliance or drug failure [39].
Of the 12 patients with mycobacterial bacteremia, 10 had MAC
infection and 2 had M tuberculosis infection with no evidence
of pulmonary disease. More than one-half of the patients with
disseminated MAC infection had another fever source identified.
If lysis-centrifugation blood cultures had not been included in
the admission evaluation of these patients, these mycobacterial
infections might have remained undetected [3, 40].
Compared with patients with nonbacterial (other) causes of
fever, the group with bacterial infection was significantly more
likely to report a history of injection drug use and to have a
significantly higher admission leukocyte count and baseline
CD4 lymphocyte count, have more rapid resolution of fever,
and have a 9% shorter hospital stay. The crude mortality rate
for the 220 episodes of fever was 4.5%. In our comparative
analysis, death was the outcome of 3% of the 109 episodes
of fever due to bacterial infections and 6% of those due to
nonbacterial causes.
CID 1996;23 (August)
Given the high rate of bacterial infections in people with AIDS,
future studies should focus on the utility of specific prophylactic
regimens, particularly against infection with S. pneumoniae,
H influenzae, and S. aureus. Trimethoprim/sulfamethoxazole,
when administered as prophylaxis for PCP, decreases the rate of
bacterial colonization and infection and therefore is preferred to
aerosolized pentamidine [16, 28, 34, 41]. However, a high rate
of side effects and poor compliance with the use of oral medication
in some risk groups remain problems [42]. In addition, the use
of rifabutin or clarithromycin for prophylaxis against MAC infection may also reduce the rates of bacterial infection, but there
remain concerns over the emergence of antibiotic-resistant strains
of pneumococcus and other bacteria [43, 44].
The use of effective bacterial vaccines to prevent infection
would be ideal, but to date the immune response of AIDS patients
to pneumococcal polysaccharide and H infiuenzae type b conjugate vaccines has been unsatisfactory [4, 8, 11]. It is noteworthy
that the H influenzae type b vaccine would not have protected
against the nontypeable strains that accounted for all H influenzae
infections in this study. All of the clinic patients in our program
were offered pneumococcal vaccine at the time of our study, but
recent concerns that pneumococcal vaccine may increase plasma
levels of HIV-RNA have led us to reconsider the risk-benefit of
routine vaccination [45].
Data from this study may not be generalizable because of
the high proportions of injection-drug-using and homeless study
patients, who may be less compliant with medications and who
may be prescribed aerosolized pentamidine as prophylaxis for
PCP, a less-effective agent than trimethoprim-sulfamethoxazole
in preventing bacterial infection. In addition, this study was conducted over two different time periods, each including fall and
winter seasons, when more bacterial infections occur. Finally,
admission to the hospital was required for enrollment in the study,
and with the changing health care delivery systems in the United
States, in the future similar patients might be treated on an outpatient basis.
Clinicians frequently feel overwhelmed when faced with possible causes of fever in HIV-infected patients. Our study indicates
that the cause of the patient's fever is usually identified and that
bacterial pathogens, particularly those causing pneumonia, are
a common source of fever in patients with AIDS who require
hospitalization. In contrast to opportunistic infections, such as
PCP and MAC infection, which usually occurred in patients with
CD4 lymphocyte counts of < 100/mm 3, bacterial infections occurred as both early and late complications of HIV infection.
However, clinicians must realize that the relationship between
bacterial infections and HIV infection is a dynamic process that
may change with the population at risk, the natural history of
the disease, and the use of antibiotics for prophylaxis against
opportunistic pathogens, as well as the evolution of treatment and
antimicrobial resistance.
Acknowledgments
The authors thank Drs. Peter Rice, Brant Viner, and Judith
Steinberg for their assistance in the development of the study, Dr.
em 1996;23
(August)
Fever and HIV Infection
Meg Sullivan for her careful review and comments, and Maria
Tetzaguic for her assistance in preparation of the manuscript.
327
21. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR Morb Mortal
Wkly Rep 1987;36(suppl IS):3S-15S.
22. von Reyn CF, Levy BS, Arbeit RD, Friedland G, Crumpacker CS. Infective
References
1. Follansbee SE, Busch DF, Wofsy CB, et al. An outbreak of Pneumocystis
carinii pneumonia in homosexual men. Ann Intern Med 1982; 96(part
1):705-13.
2. Mildvan D, Mathur D, Enlow RW, et al. Opportunistic infections and
immune deficiency in homosexual men. Ann Intern Med 1982; 96(part
1):700-4.
3. Horsburgh CR Jr. Mycobacterium avium complex infection in the acquired
immunodeficiency syndrome. N Engl J Med 1991;324:1332~8.
4. Witt DJ, Craven DE, McCabe WR. Bacterial infections in adult patients
with the acquired immune deficiency syndrome (AIDS) and AIDSrelated complex. Am J Med 1987;82:900-6.
5. Polsky B, Gold JWM, Whimbey E, et al. Bacterial pneumonia in patients
with the acquired immunodeficiency syndrome. Ann Intern Med 1986;
104:38-41.
6. Whimbey E, Gold JWM, Polsky B, et al. Bacteremia and fungemia in
patients with the acquired immunodeficiency syndrome. Ann Intern Med
1986; 104:511-4.
7. Krumholz HM, Sande MA, Lo B. Community-acquired bacteremia in
patients with acquired immunodeficiency syndrome: clinical presentation, bacteriology, and outcome. Am J Med 1989; 86:776-9.
8. Simberkoff MS, El Sadr W, Schiffman G, Rahal JJ Jr. Streptococcus
pneumoniae infections and bacteremia in patients with acquired immune
deficiency syndrome, with report of a pneumococcal vaccine failure.
Am Rev Respir Dis 1984; 130:1174-6.
9. Janoff EN, Breiman RF, Daley CL, Hopewell PC. Pneumococcal disease
during HIV infection: epidemiologic, clinical, and immunologic perspectives. Ann Intern Med 1992; 117:314-24.
10. Chaisson RE. Bacterial pneumonia in patients with human immunodeficiency virus infection. Semin Respir Infect 1989; 4: 133--8.
11. Steinhart R, Reingold AL, Taylor F, Anderson G, Wenger JD. Invasive
Haemophilus influenzae infections in men with HIV infection. JAMA
1992;268:3350-2.
12. Schlamm HT, Yancovitz SR. Haemophilus injiuenzae pneumonia in young
adults with AIDS, ARC, or risk of AIDS. Am J Med 1989; 86: 11-4.
13. Jacobson MA, Gellermann H, Chambers H. Staphylococcus aureus bacteremia and recurrent staphylococcal infection in patients with acquired
immunodeficiency syndrome and AIDS-related complex. Am J Med
1988; 85: 172-6.
14. Selwyn PA, Alcabes P, Hartel D, et al. Clinical manifestations and predictors of disease progression in drug users with human immunodeficiency virus infection. N Engl 1 Med 1992;327:1697-703.
15. Nichols L, Balogh K, Silverman M. Bacterial infections in the acquired
immune deficiency syndrome: clinicopathologic correlations in a series
of autopsy cases. Am 1 Clin Pathol 1989; 92:787 -90.
16. Hirschtick RE, Glassroth J, lordan MC, et al. Bacterial pneumonia in
persons infected with the human immunodeficiency virus. N Engl 1
Med 1995;333:845-51.
17. Genne D, Chave 1-P, Glauser M-P. Fievre d'origine indeterminee dans un
collectif de patients HIV positifs. Schweiz Med Wochenschr 1992; 122:
1797-802.
18. Miralles P, Moreno S, Perez-Tascon M, Cosin 1, Diaz MD, Bouza E.
Fever of uncertain origin in patients infected with the human immunodeficiency virus. Clin Infect Dis 1995;20:872-5.
19. Bissuel F, Leport C, Perronne C, Longuet P, Vilde JL. Fever of unknown
origin in HlV-infected patients: a critical analysis of a retrospective
series of 57 cases. 1 Intern Med 1994;236:529-35.
20. Sepkowitz KA, Telzak EE, Carrow M, Armstrong D. Fever among outpatients with advanced human immunodeficiency virus infection. Arch
Intern Med 1993; 153:1909-12.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
endocarditis: an analysis based on strict case definitions. Ann Intern
Med 1981;94(part 1):505-18.
Norusis MJ. SPSS/PC Plus Statistics 4.0. Chicago: SPSS, 1990.
Rosner B. Fundamentals of biostatistics. 4th ed. Belmont, California: Wadsworth Publishing, 1995.
Cooper DA, Gold 1, MacLean P, et al. Acute AIDS retrovirus infection:
definition of a clinical illness associated with seroconversion. Lancet
1985; 1:537-40.
Fox R, Eldred LJ, Fuchs El, et al. Clinical manifestations of acute infection
with human immunodeficiency virus in a cohort of gay men. AIDS
1987; 1:35-8.
Weber Dl, Becherer PR, Rutala WA, Samsa GP, Wilson MB, White GC
II. Nosocomial infection rate as a function of human immunodeficiency
virus type I status in hemophiliacs. Am J Med 1991;91(suppl 3B):
206S-12S.
Weinke T, Schiller R, Fehrenback FJ, Pohle HD, Association between
Staphylococcus aureus nasopharyngeal colonization and septicemia in
patients infected with the human immunodeficiency virus. Eur J Clin
MicrobioIInfect Dis 1992; 11:985--9.
Lane HC, Masur H, Edgar LC, Whalen G, Rook AR Fauci AS. Abnormality of B-cell activation and immunoregulation in patients with the acquired immunodeficiency syndrome. N Engl J Med 1983; 309:453-8.
Yarchoan R, Redfield RR, Broder S. Mechanisms of B cell activation in
patients with acquired immunodeficiency syndrome and related disorders: contribution of antibody-producing B cells. of Epstein-Barr virusinfected B cells, and of immunoglobulin production induced by human
T cell lymphotropic virus, type III/lymphadenopathy-associated virus.
1 Clin Invest 1986;78:439-47.
Bender BS, Bohnsack IF, Sourlis SH, Frank MM, Quinn TC. Demonstration of defective C3-receptor-mediated clearance by the reticuloendothelial system in patients with acquired immunodeficiency syndrome. J
Clin Invest 1987;79:715-20.
Perricone R, Fontana L, de Carolis C, Carini C. Sirianni MC, Aiuti F.
Evidence for activation of complement in patients with AIDS related
complex (ARC) and/or lymphadenopathy syndrome (LAS). Clin Exp
Immunol 1987; 70:500-7.
Tausk FA, McCutchan JA, Spechko P, Schreiber RD, Gigli 1. Altered
erythrocyte C3b receptor expression, immune complexes, and complement activation in homosexual men in varying risk groups for acquired
immune deficiency syndrome. J Clin Invest 1986; 78:977 -82.
Craven DE, Fagan M, Steger KA, Gunn 1, Rolitsky C. Risk factors for
nasopharyngeal (NP) colonization with Staphylococcus aureus in HIVinfected outpatients [abstract no S51]. In: Program and abstracts of the
3rd Annual Meeting of the Society of Health Care Epidemiologists of
America. Thorofare, New Jersey: SLACK, 1993.
Johnson HM, Russell lK, Pontzer CH. Superantigens in human disease.
Sci Am 1992;266(April):92-101.
Brinchmann· IE, Gaudernack G, Thorsby E, Vartdal F, Staphylococcal
exotoxin superantigens induce human immunodeficiency virus type I
expression in naturally infected CD4+ T cells. J Virol 1992;66:
5924-8.
Pantaleo G, Graziosi C, Fauci AS. Immunopathogenesis of human immunodeficiency virus infection. N Engl 1 Med 1993;328:327-35.
Fauci AS. The human immunodeficiency virus: infectivity and mechanisms
of pathogenesis. Science 1988; 239:617 - 22.
Samet IH, Libman H, Steger KA, et al. Compliance with zidovudine
therapy in patients infected with human immunodeficiency virus, type
1: a cross-sectional study in a municipal hospital clinic. Am J Med
1992;92:495-502.
Barber TW, Craven DE, McCabe WR. Bacteremia due to Mycobacterium
tuberculosis in patients with human immunodeficiency virus infection:
328
Barat et al.
a report of nine cases and a review of the literature. Medicine (Baltimore) 1990;69:375-83.
41. Hardy WD, Feinberg J, Finkelstein OM, et al. A controlled trial oftrimethoprim-sulfamethoxazole or aerosolized pentamidine for secondary prophylaxis of Pneumocystis carinii pneumonia in patients with the acquired immunodeficiency syndrome-AIDS Clinical Trials Group
Protocol 021. N Engl J Med 1992;327:1842-8.
42. Bezanson G, Burbridge S, Haldane 0, Yoell C, Marrie T. Diverse populations of Legionellapneumophila present in the water of geographically
clustered institutions served by the same water reservoir. J Clin Microbiology 1992;30:570-6.
eID 1996;23 (August)
43. Breiman RF, Butler JC, Tenover FC, Elliott JA, Facklam RR. Emergence
of drug-resistant pneumococcal infections in the United States. JAMA
1994;271: 1831-5.
44. Friedland IR, McCracken GH Jr. Management of infections caused by
antibiotic-resistant Streptococcus pneumonia. N Engl J Med 1994; 331:
377-82.
45. Janoff EN, Swindells S, Brichacek B, Stevenson M. Increased HIV-l
burden with immune activation following immunization with pneumococcal vaccine [abstract 1237]. In: Program and abstracts of the 35th
Interscience Conference on Antimicrobial Agents and Chemotherapy
(San Francisco). Washington, DC: American Society for Microbiology,
1995.