Ebook Infectious diseases in critical care: Part 2

Chapter 27

27 Influenza
P.R. Brookmeyer, K.F. Woeltje

27.1
Introduction

27.2
History

Influenza infections account for significant morbidity
and mortality both in the United States and worldwide.
Approximately 5 – 15 % of the world’s population develops the disease annually. In the United States, 114,000
hospitalizations and 36,000 deaths are thought to occur
annually [1], with an estimated annual economic impact of 3 – 5 billion dollars [2]. Complications of influenza include primary and secondary pneumonias, respiratory failure and rarely myositis and neurologic
failures. These complications often lead to ICU admission, especially in the elderly or immunocompromised
population.
Superimposed on these annual epidemics are periodic pandemics, the most famous being the “Spanish
Influenza” of 1918 – 1919, in which at least 20 million
and perhaps as many as 100 million persons succumbed worldwide [3]. Based on conservative attack
and mortality rates, it is estimated that in the United
States alone the next influenza pandemic may result in
314,000 – 734,000 hospitalizations, and claim between
89,000 and 207,000 lives, with an economic impact of
70 – 170 billion dollars [4]. In the new pandemic, it is
projected that the ICU capacity in the United States will
be overwhelmed, requiring the painful decision to
withhold care from patients unlikely to survive, focusing on patients most likely to respond to ventilatory
and other therapy.

The influenza virus has likely been causing annual epidemics and periodic pandemics since antiquity. One of
the first references to influenza in the “modern literature” appears to be Sydenham’s account in 1679 [5]. In
a classic review of historical pathology by Hirsch, 299
outbreaks of influenza occurring at an average interval
of 2.4 years were calculated between 1173 and 1875 [6].
Industrialization and the increased pace of transportation resulted in increasingly rapid spread of severe pandemic influenza. This culminated in the 1918 – 1919
“Spanish Influenza.” This famous pandemic was notable for its surprisingly heavy toll on young adults, with
mortality rates in some areas reaching 5 – 10 %. In the
United States, draconian infection control measures included closing public schools, creating quarantines,
and travel passes. At least three additional somewhat
milder pandemics occurred throughout the remainder
of the 20th century (Fig. 27.1).

1917 H1N1

1957

27.3
Virology
The influenza virus is a member of the Orthomyxoviridae family, a family which includes influenza A, B, C,
Thogoto virus, and the infectious salmon anemia virus.
This family is characterized by a host derived envelope,
a negative sense single stranded, segmented RNA genome, and envelope glycoproteins important in viral
entry and exit from cells. The morphology of the three

1977 H1N1

1957 H2N2 1968

1968 H3N2

Fig. 27.1. Influenza A antigenic shifts


27.3 Virology

subtypes of influenza is similar, with an 80 – 120 nm viron size, 9 – 12 structural proteins, and 7 – 8 gene segments. On the surface of the influenza virus are spikelike projections of glycoproteins that possess either
hemagglutinin or neuraminidase activity, both of
which are critical to viral replication. The hemagglutinin facilitates entry of the virus into host cells by attachment to sialic-acid receptors. A major function of
the neuraminidase is to catalyze the cleavage of glycosidic linkages to sialic acid, which allows the completed
virion to be released from infected cells [7]. There are at
least 16 antigenetically diverse hemagglutinins and 9
distinct neuraminidases in influenza A, the majority of
which exist in non-human hosts [8]. Influenza A viruses are typically designated HxNy where the x and y
represent which hemagglutinin and neuraminidase, respectively, the virus carries. Thus influenza A H3N2
possesses a type 3 hemagglutinin and a type 2 neuraminidase. The numbering scheme is arbitrary and
carries no intrinsic meaning; the numbers only represent a way to distinguish between types of the molecules. In contrast, influenza B has only one known
hemagglutinin and only one neuraminidase. Other viral proteins include the Matrix (M) protein, which controls nuclear transport, the Nucleoprotein (NP), a regulator of transcription, and Matrix 2 (M2) protein, an
ion channel required for uncoating.
Influenza is classified into types A, B and C based on
differences in viral proteins. Influenza C is somewhat
morphologically distinct, and is classified in a different
genus from influenza A and B. It infects both humans
and swine, but tends to cause only mild disease without
season variation [9]. In contrast, both influenza A and
B are major causes of disease. Influenza B infects only
humans, typically causing severe disease in the elderly
or high risk patients. It rarely causes epidemics, and
does not cause pandemics. Influenza A infects many
hosts, including humans, birds, swine, horses, and marine mammals. It is a common cause of both annual epidemics and periodic pandemics.

27.3.1
Antigenic Variation
While infection with influenza results in the development of both humoral and cell mediated protective immunity, individuals may be re-infected periodically.
This is secondary to changes in influenza antigens resulting in virus subtypes to which humans have little or
no resistance. Through these changes, influenza has remained a significant pathogen over the ages despite the
advent of vaccines. The changes occur via changes in
the surface glycoproteins of the virus, neuraminidase
and hemagglutinin. Two types of antigenic change are
described, known as antigenic drift and antigenic shift.

27.3.1.1
Antigenic Drift
Antigenic drift refers to the minor antigenic changes
which occur in the hemagglutinin and neuraminidase
proteins. The mechanism of antigenic drift is the gradual accumulation of amino acid substitutions due to
point mutations in the hemagglutinin and neuraminidase genes [10, 11]. As mutations accumulate, antibodies generated by exposure to previous strains do
not neutralize current strains to the same extent, resulting in only limited or partial immunity to the new
strains. It is felt that decreased recognition of the new
strains acts as a type of natural selection; new strains
with less immune recognition become the predominant strain in annual epidemics. Antigenic drift is present in both the influenza A and B subtypes.
27.3.1.2
Antigenic Shift
Antigenic shift occurs only in influenza A. Compared
with previous strains, the predominant circulating virus possesses a different hemagglutinin, neuraminidase, or both. There is little or no antibody recognition
of these new stains, thereby creating strains that may
become a source of epidemic and pandemic influenza.
There is a strong association between antigenic shifts
with the occurrence of pandemics. The severe pandemics of 1918 – 1919 (shift to H1N1) and 1957 (shift to
H2N2) were associated with shifts of both the hemagglutinin and neuraminidase [12, 13]. The less extensive
pandemic of 1968 was associated with only a shift to a

new hemagglutinin (shift to H3N2) [14]. Interestingly,
the “pseudo-pandemic” of 1977, which involved an influenza A virus which had shifted back to H1N1, affected primarily younger individuals, born after the H1N1
virus had last circulated [15].
Antigenic shift can occur through a variety of mechanisms. Non-human influenza is selective in its tropism, and cannot easily replicate in humans [16]. However, avian influenza viruses may replicate in non-avian, non-human reservoirs (like swine). A pig that was
co-infected with both avian and human strains of influenza might result in a genetic reassortment that produces a novel virus capable of replication in and transmission between humans [17]. This reassortment process may happen frequently, but may result in viruses
with decreased pathogenicity or limited tropism in humans, and therefore severe pandemics do not begin.
Alternatively, mutations may occur directly in a
non-human virus, such as an avian virus, that allow the
virus to readily spread from person to person [18]. This
process may occur partially, so that spread from animals to humans is possible, but human-to-human
spread does not occur. An example is H5N1 avian influ-

285


286

27 Influenza

enza. Beginning in late 2003 an epizootic developed in
Southeast Asia, which by the spring of 2006 had become a panzootic in wild birds and domestic poultry
involving parts of Europe and Africa as well. Between
December 2003 and March 2006, a total of 186 persons
had cases of H5N1 influenza confirmed by the World
Health Organization, of whom 105 (56 %) died. Almost
all patients who developed the infection appear to have
acquired it directly from sick birds, presumably because the virus had restricted tropism and was not able
to spread readily from person to person [19]. At the
time this chapter was written the H5N1 avian influenza
panzootic was still spreading.

27.4
Epidemiology
In temperate regions influenza spread occurs annually
with the peak epidemic during winter months. Conversely, in tropical regions outbreaks of influenza may
occur year round. In annual influenza epidemics between 5 % and 15 % of the population may develop disease. While attack rates are greatest in the young, influenza-associated mortality is highest in the elderly and
immunocompromised. Risk factors for influenza-associated complications include chronic lung, heart and
renal disease [20, 21]. The entire epidemic appears to
take approximately 5 – 6 weeks to circulate through the
community. How influenza persists between the annual
epidemics is poorly understood.
Epidemic influenza occurs annually. However, an influenza pandemic occurs every several decades and involves the entire world. Influenza strains causing pandemic influenza are usually the result of antigenic shift,
with little immunity in the populace. While past pandemics such as the 1918 pandemic took many months
to spread throughout the world, the rapid pace of modern travel would likely allow a new pandemic to spread
much more rapidly, allowing little time for initially unaffected regions to prepare.

27.5
Transmission and Pathophysiology
Influenza spreads rapidly in communities. The mechanism of spread from person to person is primarily
droplet via small particle sized aerosols [22]. Once the
virus is deposited on the respiratory epithelium, the influenza virus attaches to ciliated columnar epithelial
cells via the hemagglutinin molecule. The cells are then
invaded and viral replication occurs. Released viruses
then infect large numbers of adjacent epithelial cells,
and therefore within a few replication cycles large numbers of cells may be infected. The incubation period

from exposure to the onset of illness appears to range
from 1 to 3 days, with the average period 2 days. Adults
can be infectious from the day before symptoms begin
through approximately 5 days after illness onset. Children can be infectious for 10 days or more, and young

children can shed virus for several days before their illness onset. Severely immunocompromised persons can
shed virus for weeks or months [23]. Immune responses to influenza infection include both nonspecific
and specific immunity. Nonspecific defenses include
nonspecific mucoproteins which bind virus and the
mechanical apparatus of the muco-ciliary apparatus.
Patients with defective muco-ciliary apparatuses, such
as smokers, tend to have higher attack rates and more
severe complications of influenza infection. Specific
defenses include both humeral and cell mediated responses. Infection with influenza results in long-lived
resistance to re-infection with the same virus subtype.
However, because of antigenic shift and drift, there is
only limited protection against new subtypes. A good
illustration of the long lived immunity to specific viruses is the 1977 reemergence of the H1N1 subtype,
where people alive during the 1918 pandemic were
largely immune and not affected.
Antibody responses to the influenza virus are typically directed against the hemagglutinin, neuraminidase, structural proteins M and NP, and to some degree
to the M2 protein. Antibodies responses have variable
cross protection within viral subtypes depending on
the amount of change of the antigen resulting from antigenic shift or drift. Antibodies to hemagglutinin appear most important in protecting against disease and
future infection with the same subtype. Antibodies to
neuraminidase reduce efficient release of virus and decreases plaque size in in-vitro assays. Peak antibodies
are formed approximately 4 – 7 weeks after infection,
then slowly decline. There appears to be a significant
mucosal response to the hemagglutinin antigen, with
nasal secretions containing IgG and IgA.

27.6
Clinical Disease
The clinical features of an uncomplicated influenza are
nondescript, and virtually indistinguishable from other respiratory viral infections. Influenza is characterized by an abrupt onset of headache, fevers, often high

grade, dry cough, myalgia, malaise and anorexia. The
cough is variable, often initially nonproductive, then
productive of small amounts of mucous, usually nonpurulent. Duration of fevers average 3 days, with a
range of 4 – 8 days. Cough and weakness (“post-influenza asthenia”) may persist for weeks after fever and
upper respiratory tract symptoms have resolved. Physical exam usually reveals flushing, tachycardia, and oc-


27.6 Clinical Disease

casionally tachypnea. The pulmonary exam is generally unremarkable in uncomplicated cases. Early in the
illness even otherwise healthy people may appear quite
ill, and during times of epidemic both physician practices and emergency rooms are often swamped with influenza patients, which potentiates the spread to noninfected patients.
27.6.1
Complications
The most common complication of influenza is pneumonia. Pneumonia can either be primary influenza
pneumonia or a secondary bacterial pneumonia. Primary influenza pneumonia was first well documented
in the influenza pandemic of 1957 – 1958 [24]. It is
thought to be a major cause of death during the earlier
pandemic of 1918 – 1919. Symptoms include high fever,
dyspnea, hypoxemia, and respiratory distress. Chest
radiographs are similar to other viral pneumonias, revealing scant bilateral interstitial infiltrates. Primary
influenza pneumonia has become increasingly rare in
the current interpandemic era.
Secondary bacteria pneumonias are similar to noninfluenza associated pneumonias. Up to 25 % of all
mortality from influenza and a large proportion of ICU
admission secondary to influenza are due to secondary
bacterial pneumonias [25]. S. pneumonia is the most
common pathogen associated with post-influenza
pneumonia, accounting for up to 48 % in some series.
S. aureus, an otherwise uncommon cause of community-acquired pneumonia, is the second most common

organism isolated in this setting (19 %). Other more
typical pneumonia pathogens, such as Haemophilus influenza, are common as well [26]. Secondary pneumonias often develop as the patient is improving from the
primary influenza infection, with the patient improving briefly, then becoming again febrile, now with
worsening respiratory status and purulent secretion.
Some patients may have features of both viral and bacterial pneumonia. While influenza usually does not require ICU care, high risk patients with severe pneumonia may require intubation and ICU level care.
Non-pneumonia complications of influenza have also been reported. An important complication of influenza is myositis with elevated muscle enzymes. This
must be differentiated from the myalgias, which are
very common with the influenza syndrome. Other
complications include pericarditis, myocarditis, and
CNS complications, the most common of which appears to be a Guillain-Barre type syndrome [27]. Finally, Reye’s syndrome has been reported in children infected with influenza B and receiving aspirin [28].

27.6.2
Diagnosis
In times of a confirmed epidemic, when influenza is
widespread in the community, a clinical definition based
on fever greater than 37.8 °C, and two of four symptoms:
cough, myalgia, sore throat and headache, was found to
have a sensitivity of 77.6 % and specificity of 55 %, for the
diagnosis of influenza [29, 30]. However, at the beginning
of epidemics, with sporadic cases, and with atypical presentation, the clinical laboratory must be utilized to differentiate influenza from other respiratory viruses.
Available tests include viral culture, a rapid diagnosis using viral antigens, and the investigational PCR tests.
Viral culture is the gold standard for laboratory diagnosis. Virus can be easily isolated by nasal swabs,
throat cultures, and sputum or bronchoalveolar lavage
samples. One study concluded that sputum and nasal
aspirates had the highest positive predictive value, and
throat swabs the worst; however, this study did not include bronchoalveolar lavage specimens [31]. After collection and transport in viral transport medium, the
specimens are inoculated into specific cell cultures,
where virus is detected by cytopathic effect [32]. Less
commonly, embryonated eggs can be used for virus
propagation, followed by characterization of the virus

by hemagglutination inhibition. Unfortunately, viral
culture takes up to 72 h to see a cytopathic effect, but
has the benefit of allowing for sub-typing of viral
strains, which is critical in the assessment of the current year’s vaccine and development of the next.
As rapid diagnosis of influenza is very important for
treatment and infection control, a number of commercial rapid diagnostic tests have recently been developed. These tests can yield results in as little as 30 min.
They differ in the types of influenza viruses they can
detect and whether they can distinguish between influenza types. Different tests can detect: (1) only influenza
A viruses; (2) both influenza A and B viruses, but not
distinguish between the two types; or (3) both influenza A and B and differentiation between the two [33].
These tests are based on the immunologic detection of
viral antigens via immunofluorescence or enzyme immunoassays. The reported sensitivities of these rapid
diagnostic methods range from 40 % to 80 % [34].
PCR has also been used for diagnosis, though usually
in a research setting. Some authors have suggested that
PCR may be more sensitive than viral culture, as it can
detect virons which have lost replicative viability [35].
Unfortunately, PCR is expensive, and labor intensive, and
currently tends to be confined to research institutions.
Serological diagnosis of influenza is possible, but
can be difficult to interpret as most people have been
previously infected. Acute and convalescent specimens,
which reveal a fourfold rise in titers, are considered diagnostic.

287


288

27 Influenza

27.6.3
Treatment
While prevention of influenza is by the far the best
measure to combat influenza, four antiviral drugs in
two mechanistic classes are currently available and
FDA approved for the treatment of influenza. These
drugs, when used in the first 24 – 48 h of illness, appear
to shorten duration of symptoms for between 1 and
2 days [36, 37]. The M2 inhibitors amantidine and rimantidine have been used since the 1960s, but are only
active against influenza A. The M2 inhibitors target the
M2 ion channel, which is important in replication of the
viron. The major side effects of amantidine are central
nervous system symptoms such insomnia, impaired
thinking, dizziness and lightheadedness, resulting in
discontinuation rates of up to 13 %. Ramantidine appears to have far fewer symptoms, and discontinuation
rates of about 6 % have been reported [38]. In recent
years an increasing M2 channel inhibitor resistance has
surfaced. During the 2005 – 2006 influenza year, CDC
testing of 120 influenza A (H3N2) viruses isolated from
patients in 23 states revealed resistance rates of 91 %.
Therefore, during this season, the CDC has recommended against the use of M2 inhibitors in the treatment or prevention of influenza A [39]. Continuation of
this resistance trend appears likely in the future.
Neuraminidase inhibitors, including inhaled zanamivir and oral oseltamivir, are newer potent agents, active and approved against both influenza A and B. The
neuraminidase inhibitors inhibit the functioning of the
viral neuraminidase, which cleaves sialic acid containing receptors, allowing release of completed viron from
the infected cell. Oseltamivir is generally well tolerated,
and major side effects are limited to nausea and vomiting, which typically do not require drug cessation. Zanamivir is supplied as a dry powder for inhalation, and
has been linked to bronchospasm and decrease in peak
flows in asthmatics [40], as well as gastrointestinal upset. The manufacturer has released a warning advising

patients with COPD or asthma to have a fast acting inhaler available prior to administration.

27.7
Prevention
27.7.1
Vaccination
Vaccination is by far the best method for prevention of
influenza. Influenza is unique among vaccine preventable illnesses because its high rate of mutation requires
development and implementation of a new vaccine annually. Worldwide surveillance and a degree of luck are
required to select the proper antigenic variants of influenza to include in the vaccine months before the start of

the annual flu season [41]. In the United States there are
currently two licensed vaccines, a trivalent inactivated
vaccine (TIV), and a trivalent live-attenuated influenza
vaccine (LAIV).
The inactivated vaccine was first licensed in 1943,
and now usually contains three influenza antigenic
strains – two type A, and one type B. After the likely
predominant strains are identified, the viruses are
grown in embryonated chicken eggs. They are then inactivated, purified, split into viral fragments, and finally combined into vaccine. Nearly 6 months after identification of target strains is required for vaccine production. Therefore if the educated guesses regarding the
dominant strains are incorrect there is no time to develop alternative vaccines. When there is a good match
between vaccine and epidemic virus, levels of protection from influenza infection range from 70 % to 90 %
[42], although it is typically less in elderly and chronically ill patients. Patients who do get infected with influenza despite having been vaccinated tend to have
less severe disease, and have lower mortality rates. The
inactivated vaccine is well tolerated; contraindications
are limited to allergies to eggs and a history of a severe
adverse reaction. Individuals with a febrile infection
should not be vaccinated until its resolution, since they
may have a decreased immune response to the vaccine.
The live attenuated influenza vaccine was licensed in

2003. Although it is a live viral vaccine, the virus is cold
adapted, so that it only replicates at the lower temperatures found in the anterior nares [43]. While both the
inactivated and live vaccines induce systemic antibody
responses, the cold adapted vaccine additionally confers a significant specific mucosal antibody response
(IgA). The cold adapted vaccine is currently only FDA
approved for those between 5 and 49 years of age. Contraindications include immunosuppression, HIV infection, malignancy, leukemia, or lymphoma, and those
between age 5 and 17 receiving aspirin products, because of the association of Reye syndrome with aspirin
and wild-type influenza infection [44]. The live attenuated vaccine can be given to healthcare workers. Work
restrictions are not necessary after this vaccine except
for those caring for immunocompromised patients
who require a protective environment (e.g., bone-marrow transplant patients) [45].
Influenza vaccine is recommended for patients at increased risk for complications, including those older
than 50, and those with chronic pulmonary or cardiac
disease, diabetes, renal disfunction, or immunosuppression (see Table 27.1). Vaccination is also strongly
recommended for all healthcare workers.
During the 2004 – 2005 influenza season, manufacture problems resulted in large shortages of the killed
vaccine, resulting in rationing of vaccine. The CDC has
recommended a triage system to identify those at highest risk who should receive vaccination priority in


27.8 Infection Control
Table 27.1. Priority groups for the inactivated influenza vaccine
in case of shortages (adapted from [54])
Tier Priority group
1A
1B

1C

Persons aged & 65 years with comorbid conditions

Residents of long-term-care facilities
Persons aged 2 – 64 years with comorbid conditions
Persons aged > 65 years without comorbid conditions
Children aged 6 – 23 months
Pregnant women
Healthcare personnel
Household contacts and out-of-home caregivers of
children aged < 6 months

2

Household contacts of children and adults at increased risk for influenza-related complications
Healthy persons aged 50 – 64 years

3

Persons aged 2 – 49 years without high-risk conditions

Table 27.2. CDC recommendations for influenza vaccination
(adapted from [54])
Persons at increased risk for complications
Persons aged & 65 years
Residents of nursing homes and other chronic care
Adults and children who have chronic pulmonary or cardiovascular system diseases, including asthma (hypertension is excluded)
Adults and children with chronic metabolic diseases (including diabetes mellitus), renal dysfunction, hemoglobinopathies, or immunosuppression
Adults and children who have any condition (e.g., cognitive
dysfunction, spinal cord injuries, seizure disorders, or
other neuromuscular disorders) that can compromise respiratory function or the handling of respiratory secretion
Children and adolescents (aged 6 months–18 years) who are
receiving long-term aspirin

Women who will be pregnant during the influenza season
Children aged 6 – 23 months
Persons aged 50 – 64 years
Vaccination is recommended for all persons aged
50 – 64 years
Persons who can transmit influenza to those at high risk
Healthcare workers including physicians, nurses, and other
personnel
Employees of assisted living and other residences for persons in groups at high risk
Persons who provide home care to persons in groups at
high risk; and household contacts (including children) of
persons in groups at high risk
Household contacts of children aged 0 – 23 months

times of shortages (see Table 27.2). New vaccine development and production techniques, such as acellular
vaccines, that allow for rapid production and deployment need to be developed in order to avoid future
shortages. These methods would also allow rapid vaccine development during the influenza seasons when
antigen matches are poor. In the setting of a vaccine
shortage, consideration could also be given to using the
LAIV in an expanded patient population (although this
would be an off-label use) [46].

27.7.2
Antiviral Prophylaxis
All of the antiviral medicines used for therapy have also
been used as post-exposure prophylaxis during times
when influenza is circulating in the community. However, because of the rapid development of resistance in
the H3N2 influenza virus noted during the 2005 – 2006
influenza season, the M2 inhibitors amantadine and
rimantadine are no longer recommended for prophylaxis. Among neuraminidase inhibitors, zanamivir has

not been FDA approved for prophylaxis. As antiviral
prophylaxis is expensive, and not without side effects,
prophylaxis must not be used in place of vaccination.
Additionally, all individuals who are initiated on antiviral prophylaxis should also receive the influenza vaccine. The Advisory Committee on Immunization Practices recommends consideration of antiviral prophylaxis for patients at high risk of complications who have
not received vaccination, those who are unlikely to respond to vaccination and healthcare workers who have
not received vaccination, during times when influenza
is active in the community [47]. Duration of prophylaxis is controversial and depends of the aim. As a bridge
to vaccination, antiviral drugs should be continued for
2 weeks after vaccination. In “seasonal prophylaxis,”
where the individual cannot receive or is not expected
to amount an immune response to the vaccination, prophylaxis should be initiated upon widespread reports
of influenza in the community and should continue for
4 – 6 weeks [48]. Antiviral drugs can also be used as
post-exposure prophylaxis, where drugs are given for
7 – 10 days after contact with an infected person [49].
This will not protect against influenza contracted from
outside the contact after the prophylactic period, and
may be best suited to times of sporadic cases. Many anecdotal reports also support the use of antiviral drugs
in aborting epidemics in nursing homes, and could be
extrapolated to outbreaks in intensive care units [50].

27.8
Infection Control
Patients with influenza should be placed in isolation to
prevent nosocomial spread of the disease. There have
also been several well documented cases of intra-ICU
spread of influenza [51]. The Centers for Disease Control and Prevention (CDC) recommend that patients
with known or suspected influenza be placed in “Droplet Precautions.” [52]. Patients should be placed in a
private room if possible; otherwise cohorting of influenza patients is acceptable. Healthcare workers should
wear a surgical or procedure mask when entering the

room (or working within 0.9 m of the patient). The
mask should be removed upon leaving the room, and

289


290

27 Influenza

hand hygiene should be implemented. Patients should
stay in their rooms to the extent possible. If a patient
with known or suspected influenza must travel to a
procedure, a surgical or procedure mask should be
placed on the patient prior to leaving the room. Negative pressure rooms and N-95 respirators are not recommended for routine influenza patients. ICUs should
have policies to exclude visitors who have febrile respiratory symptoms. Healthcare workers with febrile respiratory illnesses should likewise not come to work,
thereby avoiding the risk of spreading influenza to patients and coworkers.
If there is suspicion of nosocomial acquisition of influenza in an ICU, an investigation should be conducted
by the hospital’s infection control program. Surveillance
for possible additional patients with influenza who may
have gone unrecognized should be conducted. ICU personnel should also be surveyed to determine who might
have served as a source. Good infection control practices
should be reinforced, especially the prompt isolation of
patients (using droplet precautions) as soon as influenza
is even suspected. Patients and HCW in the ICU who
have not been vaccinated should be offered the flu vaccine. If additional nosocomial cases of influenza occur
despite infection control measures, or if the outbreak is
due to a strain of influenza that is a poor match to the
current vaccine, strong consideration should be given to
administering chemoprophylaxis to non-infected ICU

patients for at least 2 weeks [53]. Active surveillance for
additional cases of influenza should continue for at least
2 weeks after the last diagnosed case.

References
1. Thompson WW, Shay D, Weintraub E, et al. (2003) Mortality associated with influenza and respiratory syncytial virus in the United States. JAMA 289(2):179 – 186
2. Schoenbaum SC (1987) Economic impact of influenza. The
individual’s perspective. Am J Med 82(6A):26 – 30
3. Johnson NP, Mueller J (2002) Updating the accounts: global mortality of the 1918 – 20 “Spanish” influenza pandemic.
Bull Hist Med 76(1):105 – 115
4. Meltzer MI, Cox NJ, Fukuda K (1999) The economic impact of pandemic influenza in the United States: priorities
for intervention. Emerg Infect Dis 5(5):659 – 671
5. Treanor JJ (2005) Influenza virus. In: Mandell GL, Bennett
JE, Dolin R (eds) Principles and practice of infectious diseases. Churchill Livingstone, Philadelphia, PA, p 2060
6. Hirsch A (1883) Handbook of geographical and historical
pathology, 2nd edn. New Sydenham Society, London
7. Gubareva LV, Kaiser L, Hayden FG (2000) Influenza virus
neuraminidase inhibitors. Lancet 355(9206):827 – 835
8. Nicholson KG, Wood JM, Zambon M (2003) Influenza.
Lancet 362(9397):1733 – 1745
9. Zambon MC (1999) Epidemiology and pathogenesis of influenza. J Antimicrob Chemother 44 Suppl B:3 – 9
10. Luther P, Bergmann KC, Oxford JS (1984) An investigation
of antigenic drift of neuraminidases of influenza A (H1N1)
viruses. J Hyg (Lond) 92(2):223 – 229

11. Webster RG, Kendal AP, Gerhard W (1979) Analysis of antigenic drift in recently isolated influenza A (H1N1) viruses using monoclonal antibody preparations. Virology
96(1):258 – 264
12. Taubenberger JK, Reid AH, Lourens RM, et al. (2005)
Characterization of the 1918 influenza virus polymerase
genes. Nature 437(7060):889 – 893

13. Kilbourne ED (1959) Studies on influenza in the pandemic
of 1957 – 58. III. Isolation of influenza A (Asian strain) viruses from influenza patients with pulmonary complications. Details of virus isolation and characterization of isolates, with quantitative comparison of isolation methods. J
Clin Invest 38:266 – 274
14. Schulman JL, Kilbourne ED (1969) Independent variation
in nature of the hemagglutinin and neuraminidase antigens of influenza virus: distinctiveness of the hemagglutinin antigen of Hong Kong-68 virus. Proc Natl Acad Sci U S
A 63(2):326 – 333
15. Kilbourne ED (2006) Influenza pandemics of the 20th century. Emerg Infect Dis 12(1):9 – 14
16. Beare AS, Webster RG (1991) Replication of avian influenza viruses in humans. Arch Virol 119(1 – 2):37 – 42
17. Scholtissek C, Rohde W, Von Hoyningen V, Rott R (1978)
On the origin of the human influenza virus subtypes H2N2
and H3N2. Virology 87(1):13 – 20
18. Belshe RB (2005) The origins of pandemic influenza – lessons from the 1918 virus. N Engl J Med 353(21):2209 – 2211
19. Beigel JH, Farrar J, Han AM, et al. (2005) Avian influenza A
(H5N1) infection in humans. N Engl J Med 353(13):
1374 – 1385
20. Glezen WP, Keitel WA, Taber LH, et al. (1991) Age distribution of patients with medically-attended illnesses caused
by sequential variants of influenza A/H1N1: comparison to
age-specific infection rates, 1978 – 1989. Am J Epidemiol
133(3):296 – 304
21. Baker WH, Mullooly JP (1980) Impact of epidemic type A
influenza in a defined adult population. Am J Epidemiol
112(6):798 – 811
22. Alford RH, Kasel JA, Gerone PJ (1966) Human influenza
from aerosol inhalation. Proc Soc Exp Biol Med 122(3):
800 – 804
23. accessed
3/27/2006
24. Louria DB, Blumenfeld HL, Ellis JT, et al. (1959) Studies on
influenza in the pandemic of 1957 – 1958. II. Pulmonary
complications of influenza. J Clin Invest 38(1 Part 2):213 –

265
25. Simonsen L (1999) The global impact of influenza on morbidity and mortality. Vaccine 17(Suppl 1):3 – 10
26. Oliveira EC, Marik PE, Colice G (2001) Influenza pneumonia: a descriptive study. Chest 119(6):1717 – 1723
27. Hayase Y, Tobita K (1997) Influenza virus and neurological
diseases. Psychiatry Clin Neurosci 51(4):181 – 184
28. Nishida N, Chiba T, Ohtani M, Yoshioka N (2005) Sudden
unexpected death of a 17-year-old male infected with the
influenza virus. Leg Med (Tokyo) 7(1):51 – 57
29. Boivin G, Hardy I, Tellier G, Maziade J (2000) Predicting
influenza infections during epidemics with use of a clinical
case definition. Clin Infect Dis 31(5):1166 – 1169
30. Zambon M, Hays J, Webster A, et al. Relationship of clinical diagnosis to confirmed virological, serologic or molecular detection of influenza. Arch Intern Med 161(17):
2116 – 2122
31. Covalciuc KA, Webb KH, Carlson CA (1999) Comparison
of four clinical specimen types for detection of influenza A
and B viruses by optical immunoassay (FLU OIA Test) and
cell culture methods. J Clin Microbiol 37(12):3971 – 3974
32. Treanor JJ (2005) Influenza virus. In: Mandell GL, Bennett


References

33.
34.

35.
36.

37.

38.

39.
40.
41.
42.

43.
44.

JE, Dolin R (eds) Principles and practice of infectious diseases. Churchill Livingstone, Philadelphia, PA, p 2071
/>accessed on 3/7/2006
Treanor JJ (2005) Influenza virus. In: Mandell GL, Bennett
JE, Dolin R (eds) Principles and practice of infectious diseases. Churchill Livingstone, Philadelphia, PA, pp 2070 –
2071
Newton DW, Mellen CF, Baxter BD, et al. (2002) Practical
and sensitive screening strategy for detection of influenza
virus. J Clin Microbiol 40(11):4353 – 4356
Nicholson KG, Aoki FY, Osterhaus AD, et al. (2000) Efficacy and safety of oseltamivir in treatment of acute influenza: a randomised controlled trial. Lancet 355(9218):1845 –
1850
Hayden FG, Osterhaus ADME, Treanor JJ, et al. (1997) Efficacy and safety of the neuraminidase inhibitor zanamivir
in the treatment of influenza virus infections. N Engl J Med
337(13):927 – 928
Dolin R, Reichman RC, Madore HP, et al. (1982) A controlled trial of amantadine and rimantadine in the prophylaxis of influenza A infection. N Engl J Med 307(10):580 –
584
accessed on 3/7/
2006
Neuraminidase inhibitors for treatment of influenza A and
B infections. MMWR Morb Mortal Wkly Rep 48(49):1139
Treanor J (2004) Weathering the influenza vaccine crises.

N Engl J Med 351(20):2037 – 2040
Boyce TG, Gruber WC, Coleman-Dockery SD, et al. (1999)
Mucosal immune response to trivalent live attenuated intranasal influenza vaccine in children. Vaccine 18(1 – 2):
82 – 88
Maassab HF, DeBorde DC (1985) Development and characterization of cold-adapted viruses for use as live virus
vaccines Vaccine 3(5):355 – 69
Prescribing Information for FluMist™, Influenza Virus
Vaccine Live, Intranasal, 2004 – 2005 Formula. MedImmune Vaccines, Inc. Gaithersburg, MD

45. Talbot TT, Bradley SF, Cosgrove SE (2005) Influenza vaccination of healthcare workers and vaccine allocation for
healthcare workers during vaccine shortages. Infect Control Hosp Epidemiol 26(11):882 – 890
46. Advisory Committee on Immunization Practices (ACIP)
(2004) Prevention and control of influenza. Recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 53(RR06):1
47. Cosgrove SE, Fishman NO, Talbot TR, Woeltje KF, Schaffner W, Fraser VJ, McMillan JA, Perl TM (2005) Strategies
for use of a limited influenza vaccine supply. JAMA 293(2):
229 – 232
48. Hayden FG, Atmar RL, Schilling M, et al. (1999) Use of the
selective oral neuraminidase inhibitor oseltamivir to prevent influenza. N Engl J Med 341(18):1336 – 1343
49. Welliver R, Monto AS, Carewicz O, et al. (2001) Effectiveness of oseltamivir in preventing influenza in household
contacts: a randomized controlled trial. JAMA 285(6):
748 – 754
50. Schilling M, Povinelli L, Krause P, et al. (1998) Efficacy of
zanamivir for chemoprophylaxis of nursing home influenza outbreaks. Vaccine 16(18):1771 – 1774
51. Oliveira EC, Lee B, Colice GL (2003) Influenza in the intensive care unit. J Intensive Care Med 18(2):80 – 91
52. Garner JS, Hospital Infection Control Practices Advisory
Committee (1996) Guideline for isolation precautions in
hospitals. Infect Control Hosp Epidemiol 17:53 – 80
53. Centers for Disease Control and Prevention. Prevention
and control of influenza: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR
53(RR-6):1 – 40

54. Centers for Disease Control and Prevention (CDC) Tiered
use of inactivated influenza vaccine in the event of a vaccine shortage. MMWR Morb Mortal Wkly Rep 54(30):
749 – 750

291


Chapter 28

28 Bloodstream Infection in the Intensive Care Unit
J. Valles

28.1
Introduction
Nosocomial infections occur in 5 – 10 % of patients admitted to hospitals in the United States [1]. The endemic rates of nosocomial infections vary markedly between hospitals and between areas of the same hospital.
Patients in intensive care units (ICUs), representing
8 – 15 % of hospital admissions, suffer a disproportionately high percentage of nosocomial infections compared with patients in non-critical care areas [2 – 7].
Wenzel et al. [3] reported that patients admitted to
ICUs account for 45 % of all nosocomial pneumonias
and bloodstream infections, although critical care
units comprise only 5 – 10 % of all hospital beds. Severity of underlying disease, invasive diagnostic and therapeutic procedures, contaminated life-support equipment, and the prevalence of resistant microorganisms
are critical factors in the high rate of infection in ICUs
[8].
Donowitz et al. [5] reported a threefold increase in
the risk of nosocomial infection for ICU patients when
compared with ward patients (18 % vs. 6 %; p < 0.001);
and bloodstream infections were 7.4 times as likely to
occur in ICU patients as in ward patients, with an infection rate in the ICU of 5.2 episodes per 100 admissions
compared with 0.7 episodes per 100 admissions in a
general ward (p < 0.001). Trilla et al. [9], in a study of

the risk factors for nosocomial bloodstream infection
in a large Spanish university hospital, found that
among other variables, the admission to an ICU was
linked with a marked increase in the risk of nosocomial
bloodstream infection (OR = 2.37; CI 95 %: 1.67 – 3.38;
p = 0.02).
On the other hand, 40 % of patients admitted to the
ICU present infections acquired in the community, and
17 % of them present bacteremia [10]. The incidence
rate of patients with community-acquired bacteremia
admitted in a general ICU is about 9 – 10 episodes per
1,000 admissions [11, 12], representing 30 – 40 % of all
episodes of bacteremia in the ICU (Fig. 28.1).
The aim of this chapter is to discuss the clinical importance of bloodstream infection in the ICU, including nosocomial and community-acquired episodes.

(%) 70
60
50
40
30
20
10
0

94

95

ICU-BI

96

97

99

N-BI

2000 2001 2002 2003 2004

C-BI

years

Fig. 28.1. Distribution of bacteremias in the medical-surgical
ICU of Hospital Sabadell (period 1994 – 2004). ICU-BI intensive care unit-acquired bloodstream infection, N-BI nosocomial (outside ICU)-acquired bloodstream infection, C-BI community-acquired bloodstream infection

28.2
Pathophysiology of Bloodstream Infection
Invasion of the blood by microorganisms usually occurs
via one of two mechanisms: drainage from the primary
focus of infection via the lymphatic system to the vascular system, or direct entry from needles (e.g., in intravenous drug users) or other contaminated intravascular
devices such as catheters or graft material. The presence
of bloodstream infection represents either the failure of
an individual’s host defenses to localize an infection at
its primary site or the failure of a physician to remove,
drain, or otherwise sterilize that focus. Ordinarily, host
defenses respond promptly to a sudden influx of microorganisms, particularly by efficient phagocytosis by
macrophages or the mononuclear phagocytic system
that helps clear the blood within minutes to hours.

Clearance may be less efficient when microorganisms
are encapsulated, or it may be enhanced if the host has
antibodies specific for the infecting organism. Clearance of the bloodstream is not always successful. Examples of this problem are bloodstream infections associated with intravascular foci and endovascular infections
and episodes that occur in individuals whose host defense mechanisms either are too impaired to respond
efficiently or are simply overwhelmed [13].


28.3 Definitions

For that reason, the presence of living microorganisms in blood is of substantial clinical importance; it is
an indicator of disseminated infection and, as such,
generally indicates a poorer prognosis than that associated with localized disease.

28.3
Definitions
Nosocomial bloodstream infection in the ICU is defined in a patient with a clinically significant blood culture positive for a bacterium or fungus that is obtained
more than 72 h after admission to the ICU or previously, if it is directly related to a invasive manipulation on
admission to the ICU (e.g., urinary catheterization or
insertion of intravenous line). By contrast, a community-acquired bacteremia is defined when the infection
develops in a patient prior to hospital and ICU admission, or if this episode of bacteremia develops within
the first 48 h of hospital and ICU admission, and it is
not associated with any procedure performed after
hospital or ICU admission. These definitions from the
Centers for Disease Control and Prevention (CDC) consider that infections that are not nosocomial infections
are community-acquired by default [14]. However,
there are patients residing in the community, who are
receiving care at home, living in nursing homes and rehabilitation centers, receiving chronic dialysis, and receiving chemotherapy in physicians’ offices who may
present bloodstream infections. These infections have
traditionally been categorized as community-acquired
infections. For this reason, recently a new classification

scheme for bloodstream infection has been proposed
that distinguishes among patients with community-acquired, healthcare-associated, and nosocomial infections. Healthcare-associated bloodstream infection has
been defined when a positive blood culture is obtained
from a patient at the time of hospital admission or
within 48 h of admission if the patient fulfilled any of
the following criteria: (1) received intravenous therapy
at home, received wound care or specialized nursing
care or had self-administered intravenous medical
therapy; (2) attended a hospital hemodialysis clinic or
received intravenous chemotherapy; (3) was hospitalized in an acute care hospital for 2 or more days in the
90 days before the bloodstream infection; or (4) resided
in a nursing home or long-term care facility [15].
Bloodstream infections may be classified as primary
or secondary according to the source of the infection
[14]. Primary bloodstream infection occurs without
any recognizable focus of infection with the same organism at another site at the time of positive blood culture, and secondary bloodstream infections are infections that developed subsequent to a documented infection with the same microorganism at another site.

Episodes secondary to intravenous or arterial lines
have traditionally been classified as primary bacteremias; however, if local infection (defined as redness,
tenderness, and pus) is present at the site of an intravascular line, and if the semiquantitative (yielding > 15
colonies) or quantitative culture of a segment catheter
is positive to the same strain as in the blood cultures,
they may be classified as secondary bacteremias. According to this definition, in the absence of an identified source, primary bacteremias should be designated
bacteremias of unknown origin [16 – 19].
According to clinical patterns of bacteremia, it may
also be useful to categorize bloodstream infection as
transient, intermittent, or continuous [13]. Transient
bacteremia, lasting minutes to hours, is the most common and occurs after manipulation of infected tissues
(e.g., abscesses); during certain surgical procedures;
when procedures are undertaken that involve contaminated or colonized mucosal surfaces (e.g., gastrointestinal endoscopy); and, predictably, at the onset of acute

bacterial infections such as pneumonia, meningitis,
and complicated urinary infections. Intermittent bacteremia is that which occurs, clears, and then recurs in
the same patient due to the same microorganism. Classically, this type of bacteremia is associated with undrained closed space infections, such as intra-abdominal abscesses. Continuous bacteremia is characteristic
of infective endocarditis as well as other endovascular
infections such as arterial graft infections, and suppurative thrombophlebitis associated with intravenous
line infections commonly seen in critically ill patients.
Bloodstream infections may also be categorized as
unimicrobial or polymicrobial depending on the number of microorganisms isolated during a single bacteremic episode.
Blood cultures which are found to be positive in the
laboratory but which do not truly reflect bloodstream
infection in the patient have been termed contaminant
bloodstream infections or, more recently, pseudobloodstream infections [16]. Several techniques are available
to assist the clinician and microbiologist in interpreting
the clinical importance of a positive blood culture. The
categorical decision to consider the bloodstream infection as true infection or a contaminant should take into
account, at least: the patient’s clinical history, physical
findings, body temperature at the time of the blood culture, leukocyte count and differential cell counts, the
identity of microorganism isolated and the result of cultures of specimens from other sites. Indeed, the type of
microorganism isolated may have some predictive value: common blood isolates that always or nearly always
(> 90 %) represent true infection include S. aureus, E.
coli and other members of the Enterobacteriaceae, Pseudomonas aeruginosa, Streptococcus pneumoniae, and
Candida albicans. Other microorganisms such as Corynebacterium spp., Bacillus spp., and Propionibacterium

293


294

28 Bloodstream Infection in the Intensive Care Unit

acnes rarely (< 5 %) represent true bloodstream infection. More problematic are the viridans group streptococci which represent true bloodstream infection in
28 % of cases, enterococci in 78 %, and coagulase-negative staphylococci (CNS) in 15 % [20, 21].
The number of positive blood cultures out of the total number performed is frequently used to determine
the clinical significance of the isolate, but recent data
suggest that this technique is flawed. Mirret and colleagues [22] examined the significance of CNS in blood
cultures. For conventional two-bottle culture sets, 49 %
of those classified as significant infections and 68 %
classified as contaminants grew in one bottle, whereas
51 % of pathogens and 68 % of contaminants grew in
both bottles. The degree of overlap is so great that it is
difficult to predict the clinical significance based on the
number of positive bottles. It is important to note that
although coagulase-negative staphylococci have frequently been considered as contaminants in the past,
recent studies have shown that even a single blood-culture positive for these microorganisms is frequently associated with clinically relevant episodes of bloodstream infections [23 – 25].
When a culture is unexpectedly positive (in the absence of signs or symptoms) or when only one of several cultures is positive for a microorganism, it can often
be dismissed as a contaminant. Every positive blood
culture, however, should be carefully evaluated before
being dismissed as insignificant [16].

28.4
Epidemiology
Nosocomial infection in ICU patients is a frequent
event with potentially lethal consequences. Because patients in ICUs are severely ill and undergo invasive procedures, they suffer a disproportionate percentage of
nosocomial infections [5, 7, 26 – 28]. Compared with
patients in general medical/surgical wards, who have
been found to have an overall risk of 6 % of acquiring
an infection during their hospital stay, the risk in critically ill patients in the ICU is around 18 % [5]. The nosocomial infection rates among ICU patients are as
much as 5 – 10 times higher than those recorded for patients admitted to other wards, meaning that nearly
25 % of all hospital-acquired infections occur in ICU
patients [29]. Nosocomial infections are more common

in ICUs because of the severity of the underlying disease, the duration of hospital stay, the use of invasive
procedures, contaminated life-support equipment, and
the prevalence of multiply resistant microorganisms.
Data from the European Prevalence of Infection in Intensive Care study (EPIC) collected in 1992 shown that
on the day of study a total of 21 % of patients admitted
to the ICU had an infection acquired in the ICU [30].

Other infections
Surgical wound infection
Urinary tract infection
Bacteremia
Lower respiratory tract
0

10

20

ICU

30

40

50%

All hospital

Fig. 28.2. Distribution of nosocomial infections in the ICU versus the whole hospital (NNIS) (from ref. [29], with permission)

Patients in the ICU not only have higher endemic rates of
nosocomial infection than patients in general wards, but
the distribution of their nosocomial infections also differs. The two most important nosocomial infections in
general wards are urinary tract infections and surgical
wound infections, whereas in the ICU lower respiratory
tract and bloodstream infections are the most frequent
[29] (Fig. 28.2). This distribution is related to the widespread use of mechanical ventilation and intravenous
catheters. Data compiled through the National Nosocomial Infections Surveillance System (NNIS) of the Centers for Disease Control and Prevention in the USA revealed that bloodstream infections accounted for almost
20 % of nosocomial infections in ICU patients, 87 % of
which were associated with a central line [31].
Despite the higher incidence of nosocomial bloodstream infection in ICUs, few studies have adequately
analyzed this infection in this selected population. The
studies conducted in critically ill patients in recent
years show that the incidence rate of nosocomial bloodstream infection in the ICU ranges from 27 to 67 episodes per 1,000 admissions [18, 19, 32, 33] (Table 28.1),
depending on the type of ICU (surgical or medical or
coronary care unit), the severity of patients, the use of
invasive devices and the length of ICU stay. These infection rates among ICU patients are as much as 5 – 10
times higher than those recorded for patients admitted
to general wards.

Table 28.1. Rates of nosocomial bloodstream infection in the ICU
Year Type of ICU
1994 Medical-surgical ICU
1994 Surgical ICU
1996 Adult ICUs Multicenter
study
1997 Adult ICUs Multicenter
study
a

ENBI/1000a Reference
67.2
26.7
41
36

Rello [18]
Pittet [32]
Brun-Buisson
[33]
Vall´es [19]

Episodes of nosocomial bloodstream infection per 1,000
admissions


28.5 Microbiology

A few epidemiologic studies focusing solely on community-acquired BSI on admission to the ICU are available. Data from a recent multicenter study reported a
community-acquired bloodstream infections rate of
10.2 episodes per 1,000 ICU admissions [34].

28.5
Microbiology
28.5.1
Nosocomial Bloodstream Infection
The spectrum of microorganisms that invade the
bloodstream in patients with nosocomial infections
during their stay in the ICU has been evaluated in several recent studies. Although almost any microorganism can produce bloodstream infection, staphylococci
and gram-negative bacilli account for the vast majority

of cases. However, among the staphylococci, coagulasenegative staphylococci (CNS) have recently become a
clinically significant agent of bloodstream infection in
the ICU [18 – 21]. The ascendance of this group of
staphylococci has increased the interpretative difficulties for clinicians, since a high number of CNS isolations represent contamination rather than true bloodstream infection. The increased importance of CNS
bloodstream infection seems to be related to the high
incidence of utilization of multiple invasive devices in
critically ill patients and to the multiple antimicrobial
therapy used for gram-negative infections in ICU patients, which results in selection of gram-positive microorganisms. The change in the spectrum of organisms causing nosocomial bloodstream infection in an
adult ICU is confirmed in the recent study by Edgeworth and colleagues [35], which analyzed the evolution of nosocomial bloodstream infection over 25 years
in the same ICU. Between 1971 and 1990, the frequency
of isolation of individual organisms changed little, with
S. aureus, P. aeruginosa, E. coli, and K. pneumoniae species predominating. However, between 1991 and 1995,
Table 28.2. Microorganisms
causing nosocomial bloodstream infection in adult
ICUs

CNS coagulase-negative
staphylococci

the number of bloodstream infections doubled, largely
due to the increased isolation of CNS, Enterococus spp.,
and intrinsically antibiotic-resistant gram-negative organisms, particularly P. aeruginosa and Candida spp.
Currently, the leading pathogens among cases of
nosocomial bloodstream infection in the ICU are
gram-positive microorganisms, representing nearly
half of the organisms isolated [18, 19, 32, 36] (Table 28.2). Coagulase-negative staphylococci (CNS), S.
aureus and enterococci are the most frequent grampositive bacteria in all studies, and CNS is isolated in
20 – 30 % of all episodes of bloodstream infection.
Gram-negative bacilli are responsible for 30 – 40 % of
bloodstream infection episodes, and the remaining

cases are mostly due to Candida spp. Polymicrobial episodes are relatively common, representing about 10 %.
Anaerobic bacteria are isolated in fewer than 5 % of
cases.
Among gram-positive bloodstream infections, the
incidence of the pathogens is similar in the different
ICUs, CNS being the most frequently isolated organism, and S. aureus the second commonest pathogen in
all studies. Only the incidence of strains with antibiotic
resistance such as methicillin-resistant Staphylococcus
aureus (MRSA) or vancomycin-resistant enterococci
(VRE) differs substantially according to the characteristics of individual institutions, and depending on
whether they become established as endemic nosocomial pathogens in the ICU. On the other hand, the
gram-negative species isolated from nosocomial
bloodstream infections in the ICUs of different institutions show marked variability. The relative contribution of each gram-negative species to the total number
of isolates from blood varies from hospital to hospital
and over time. The antibiotic policy of the institution
may induce the appearance of highly resistant microorganisms and the emergence of endemic nosocomial
pathogens, in particular Pseudomonas spp, Acinetobacter spp., and Enterobacteriaceae with extended-spectrum beta-lactamase (ESBL).

Reference

Gram-positive
Gram-negative
microorganisms microorganisms

Fungi

Rello [18]

44.1 % CNS
S. aureus

Enterococci

40.5 % P. aeruginosa
E. coli
Enterobacter spp.

5.4 % Candida spp.

Pittet [32]

51.0 % CNS
S. aureus
Enterococci

39.0 % Enterobacter spp. 4.8 % Candida spp.
Klebsiella spp.
S. marcescens

21 %

Vall´es [19]

49.8 % CNS
S. aureus
Enterococci

32.6 % P. aeruginosa
A. baumannii
K. pneumoniae

12.7 %

Jamal [36]

46.8 % CNS
S. aureus
Enterococci

36.6 % Enterobacter spp. 17.6 % Candida spp
S. marcescens
K. pneumoniae

4.4 % Candida spp.

Polymicrobial episodes
9.9 %

9.8 %

295


296

28 Bloodstream Infection in the Intensive Care Unit
Table 28.3. Microorganisms and sources of community-acquired bacteremias admitted in the ICU
Reference

Sources

Microorganisms

Forgacs Pulmonary
[11]
Genitourinary
Endocarditis
Biliary tract
Other
Unknown origin

38.5 %
23.0 %
8.0 %
5.9 %
11.1 %
20.0 %

S. pneumoniae
E. coli
S. aureus
Other GNB
Other GPC
Other

32.3 %
27.2 %
13.5 %
14.2 %
8.2 %
14.2 %

Vall´es
[12]

20.0 %
20.1 %
19.8 %
10.3 %
29.2 %

E. coli
S. pneumoniae
S. aureus
Other GNB
Other GPC
Other

28.1 %
17.9 %
14.9 %
18.6 %
9.5 %
11.07 %

Pulmonary
Abdominal
Genitourinary
Other
Unknown origin

The incidence of polymicrobial and anaerobic bloodstream infections depends on the incidence of surgical
patients in each ICU, because in two-thirds of these
bacteremic episodes the origin is an intra-abdominal
infection.
28.5.2
Community-Acquired Bloodstream Infection
In the bacteremic episodes acquired in the community
and admitted in the ICU, the incidence of gram-positive is similar to that of gram-negative microorganisms
and near to 10 % are polymicrobial episodes. E. coli, S.
pneumoniae and S. aureus are the leading pathogens,
and the prevalence of these microorganisms is related
to the main sources of bacteremia found in these patients, such as urinary, pulmonary tract, and unknown
origin [11, 12, 34] (Table 28.3).

28.6
Sources
According to a more recent analysis, the vast majority
(70 %) of nosocomial bloodstream infections in the
ICU are secondary bacteremias, including the bloodstream infections related to an intravascular catheterinfection, and the remaining 30 % are bacteremias of
unknown origin. Table 28.4 summarizes the sources of
nosocomial bacteremias in the ICU in several recent series [18, 19, 32, 35]. As shown, intravascular catheterrelated infections and respiratory tract infections are
the leading sources of secondary episodes.
The source of nosocomial bloodstream infections
varies according to microorganism. Coagulase-negative staphylococci and Staphylococcus aureus commonly complicate intravenous-related infections, whereas
gram-negative bacilli are the main etiology for secondary bloodstream infections following respiratory tract,
intra-abdominal and urinary tract infections. Among

Table 28.4. Major sources of nosocomial bloodstream infection
in the ICUs
Type of infection

Rello
[18]
(%)

Pittet
[32]
(%)

Vall´es
[19]
(%)

Edgeworth
[35] (%)

Intravenous catheter
Respiratory tract
Intra-abdominal
infection
Genitourinary tract
Surgical wound or soft
tissue
Other
Unknown origin

35
10
9

18
28
NA

37.1
17.5
6.1

62
3
6.9

5.4
8

5.9
2.4

2.4
3

14.5
20

2.9
28.1

–
22.4

3.6
8
7
27

bacteremias of unknown origin, most are caused by
gram-positive microorganisms, mainly CNS, and they
may originate in device-related infections not diagnosed at the time of the development of the bloodstream infection.
Among community-acquired bloodstream infections, lower respiratory tract, intra-abdominal and
genitourinary infections represent more than 80 % of
episodes of bacteremia admitted in the ICU (Table 28.3). Near to 30 % of episodes are of unknown origin including mainly meningococcal and staphylococcal infections [11, 12, 34].

28.7
Systemic Response to Bloodstream Infection
The host reaction to invading microbes involves a rapidly amplifying polyphony of signals and responses
that may spread beyond the invaded tissue. Fever or hypothermia, chills, tachypnea, and tachycardia often
herald the onset of the systemic inflammatory response
to microbial invasion, also called sepsis. However, the
interchangeable use of terms such as “bloodstream infection,” “sepsis,” and “septicemia” has led to confusion.
A recent definition of bloodstream infection classifies patients with severe infection and its sequelae [37].
Bloodstream infection and fungemia have been simply
defined as the presence of bacteria or fungi in blood
cultures, and four stages of increasing severity of systemic response have been described: the systemic inflammatory response syndrome (SIRS), which is identified by a combination of simple and readily available
clinical signs and symptoms (i.e., fever or hypothermia, tachycardia, tachypnea, and changes in blood leukocyte count); sepsis, in patients in whom the SIRS is
caused by documented infection; severe sepsis when
patients have a dysfunction of the major organs; and
septic shock, which describes patients with hypotension and organ dysfunction in addition to sepsis. As
sepsis progresses to septic shock, the risk of death in-

28.8 Risk Factors for Nosocomial Bloodstream Infection in the ICU

creases substantially. Early sepsis is usually reversible,
whereas many patients with septic shock succumb despite aggressive therapy.
The presence of organisms in the blood is one of the
most reliable criteria for characterizing a patient presenting with SIRS as having sepsis or one of its more severe presentations, such as severe sepsis or septic shock.
In a recent multicenter study, Brun-Buisson and colleagues [33] analyzed the relationship between bloodstream infection and severe sepsis in adults in ICUs and
general wards in 24 hospitals in France. In this study, of
the 842 episodes of clinically significant bloodstream
infection recorded, 162 (19 %) occurred in patients
hospitalized in ICUs. Three hundred and seventy-seven
episodes (45 %) of bloodstream infection were nosocomial, and their incidence was 12 times greater in ICUs
than in wards. The frequency of severe sepsis during
bloodstream infection differed markedly between
wards and ICUs (17 % vs. 65 %, p < 0.001). The nosocomial episodes acquired in the ICU represented an incidence rate of 41 episodes per 1,000 admissions and the
incidence rate of severe sepsis among patients with
nosocomial bloodstream infection in the ICU was 24
episodes per 1,000 admissions.
Another recent multicenter study reported by our
group [19] analyzed exclusively nosocomial bloodstream infections acquired in adult ICUs of 30 hospitals
in Spain, and classified their systemic response according to new definitions as sepsis, severe sepsis and septic
shock. Among 590 episodes of nosocomial bloodstream, the host reaction was classified as sepsis in 371
episodes (62.8 %), severe sepsis in 109 episodes (18.5 %),
and septic shock in the remaining 110 (18.6 %). The systemic response differed markedly according to source
of bloodstream infection (Table 28.5). The episodes of
bloodstream infection associated with intravascular
catheters showed the lowest rate of septic shock
(12.8 %), whereas the episodes of bloodstream infection secondary to lower respiratory tract, intra-abdominal or genitourinary tract infections showed the highest incidence of severe sepsis and septic shock. In the
study by Brun-Buisson et al. [33], in patients hospitalized in ICUs, intravascular catheter-related bloodstream infection was also associated with a lower risk
of severe sepsis (OR=0.2; 95 % CI: 0.1 – 0.5; p < 0.01).

Table 28.5. Distribution of
systemic response according
to source of 590 episodes of
ICU nosocomial bloodstream infection

The systemic response may differ according to the
microorganism causing the episode of bloodstream infection. Gram-negative and Candida spp. have been associated with a higher incidence of severe sepsis and
septic shock in our multicenter study [19], whereas
CNS was the microorganism causing the lowest incidence of septic shock. The multicenter study of BrunBuisson et al. [33] analyzed ICU bloodstream infections separately and found the episodes caused by CNS
to be also associated with a reduced risk of severe sepsis
(OR = 0.2; p = 0.02) relative to other organisms.
These results suggest that the source of infection and
probably the type of microorganism causing the episode of bloodstream infection, especially if a species
other than CNS is involved, may be important in the development of severe sepsis and septic shock.
Among community-acquired episodes the incidence
of severe sepsis and septic shock is higher than in nosocomial episodes, in part because the severity of systemic response is the motive for ICU admission. In the multicenter French study, a 74 % of community-acquired
episodes presented severe sepsis or septic shock at admission in the ICU [33]. In a multicenter Spanish study
carried out in 30 ICUs, the incidence of severe sepsis
and septic shock was also 75 %. In this study, gramnegative microorganisms and the urinary and intraabdominal infections were associated more frequently
with septic shock [34].

28.8
Risk Factors for Nosocomial Bloodstream
Infection in the ICU
The conditions that predispose an individual to bloodstream infection include not only host underlying conditions but therapeutic, microbial and environmental
factors as well. The illnesses that have been associated
with an increased risk of bloodstream infection include
hematologic and nonhematologic malignancies, diabetes mellitus, renal failure requiring dialysis, chronic hepatic failure, immune deficiency syndromes, and conditions associated with the loss of normal skin barriers
such as serious burns and decubitus ulcers. In the ICU,
therapeutic maneuvers associated with an increased

Source
Intravenous catheter
Lower respiratory tract
Intra-abdominal infection
Urinary tract
Surgical wound and soft tissue
Other
Unknown
Total

Sepsis

Number (%) of episodes
Severe sepsis Septic shock

Total

158 (68.5)
53 (51.5)
12 (33.3)
23 (65.7)
7 (50)
11 (64.7)
115 (69.3)
371 (62.8)

41 (18.7)
27 (26.2)
9 (25)

5 (14.3)
2 (14.3)
4 (23.5)
21 (12.6)
109 (18.5)

219 (37.1)
103 (17.5)
36 (6.1)
35 (5.9)
14 (2.4)
17 (2.9)
166 (28.1)
590 (100)

28 (12.8)
23 (22.3)
15 (41.7)
7 (20)
5 (35.7)
2 (11.8)
30 (18.1)
110 (18.6)

297


298

28 Bloodstream Infection in the Intensive Care Unit

risk of nosocomial bloodstream infection include procedures such as placement of intravascular and urinary
catheters, endoscopic procedures, and drainage of intra-abdominal infections.
Several risk factors have been associated with the acquisition of bloodstream infection by specific pathogens. Coagulase-negative staphylococci are mainly associated with central venous line infection and with the
use of intravenous lipid emulsions. Candida spp. infections are related to the exposure to multiple antibiotics,
hemodialysis, isolation of Candida species from sites
other than the blood, azotemia, and the use of indwelling catheters [38]. In a recent analysis of risk factors for
nosocomial candidemia in ICU patients with nosocomial bloodstream infections, we found that exposure to
more than four antibiotics during the ICU stay (OR:
4.10), parenteral nutrition (OR: 3.37), previous surgery
(OR: 2.60) and the presence of solid malignancy (OR:
1.57) were the variables that were independently associated with the development of Candida spp. infection
[39].

28.9
Prognosis
28.9.1
Nosocomial Bacteremia
The crude mortality associated with bacteremic sepsis
averages 35 % (range 20 – 50 % [17, 40, 41]. The mortality directly attributable to the nosocomial bloodstream
infection averaged 27 % (range 14 – 38 %) [42]. Although one-third of the deaths occur within the first
48 h after the onset of symptoms, mortality can occur
14 or more days later. Late deaths are often due to poorly controlled infection, complications during the stay in
the ICU, or failure of multiple organs [43]. Nosocomial
bloodstream infection is associated with higher crude
mortality rates than community-acquired infection
[16, 41]. In a study, Bueno-Cavanillas et al. [44] analyzed the impact of nosocomial infection on the mortality rate in an ICU. In that study, overall crude relative
risk of mortality was 2.48 (95 % CI=1.47 – 4.16) in patients with a nosocomial infection compared with noninfected patients. When the type of infection was evaluated, the risk of mortality for patients with bloodstream infection was 4.13 (95 % IC=2.11 – 8.11).
The risk of dying is influenced by the prior clinical
condition of the patient and the rate at which complications develop. Analysis using prognostic stratification

systems (such as the APACHE scoring system) indicate
that factoring in the patients’ age and certain physiologic variables results in more accurate estimates of the
risk of dying. Variables associated with the high carefatality rates include acute respiratory distress syndrome (ARDS), disseminated intravascular coagula-

tion (DIC), renal insufficiency, and multiple organ dysfunction (MOD). Microbial variables are less important, although high care-fatality rates have been observed for patients with bloodstream infection due to
Pseudomonas aeruginosa, Candida spp. and for patients with polymicrobial bloodstream infection.
In another study of bloodstream infection in an
adult ICU of a teaching hospital in the UK over a 12year period, Crowe and colleagues [45] analyzed 315
episodes of bloodstream infection, of which 82 % were
hospital-acquired, and found an overall mortality related to bloodstream infection of 44.4 %. They also observed that ICU stay was longer in bacteremic patients
(12 days) than non-bacteremic patients (3 days).
The crude mortality from bloodstream infection is
often 35 – 60 %, ranging from 12 % to 80 %. The attributable mortality defines the mortality directly associated with the episode of bloodstream infection, and excludes the mortality attributable to underlying conditions. It averages 26 %, but varies according to the specific microorganisms involved: CNS averaged 13.6 %;
enterococci, 31 %; and Candida spp. 38 % [23, 46, 47].
Pittet et al. in 1994 [32] analyzed the attributable
mortality, excess length of stay and extra costs due to
nosocomial bloodstream infection in a surgical ICU. In
this case-control study, the crude mortality rate was
50 %, differing significantly from that of the matched
controls (15 %, p < 0.01). In consequence, the attributable mortality associated with nosocomial bloodstream infection was 35 %. These authors also observed
that median length of hospital stay for cases was 14 days
longer than for controls. Furthermore, nosocomial
bloodstream infection was associated with a doubling
of time of SICU stays, and consequently with a significant economic burden.
This study demonstrates that nosocomial bloodstream infections cause excess mortality and significantly prolong ICU and hospital stay among critically
ill patients.
In another study of nosocomial bloodstream infection in a medical-surgical ICU reported by Rello et al.
[18], the overall mortality was 31.5 %, and 65.7 % of all
deaths were directly attributable to infection. Bloodstream infections from intra-abdominal, lower respiratory tract or unknown origin were associated with a
poor prognosis. A logistic regression analysis defined

intra-abdominal origin (p = 0.01, OR:15.7) and presence of shock (p < 0.004, OR: 3.3) as independently influencing the risk of death.
In a more recent study, Pittet et al. [48] analyzed the
importance of preexisting co-morbidities for the prognosis of bloodstream infection in critically ill patients.
The study was performed in a surgical ICU, and the authors analyzed 176 patients with bloodstream infection, of whom 125 (71 %) were nosocomially acquired.
The mean total length of ICU stay of bacteremic pa-


28.9 Prognosis

60%

Percent of patients who survive

70%
2
rr2 = 0.92

Death rate

50%
40%
30%
20%

100%

epsis
evere sepsis
eptic shoc

90%
0%
70%
60%
50%
40%
30%
20%
10%
0%
0

10%

5

10

15

20

25

30

35

40

45

50

55

60

No. of days of survival after bacteremia

Fig. 28.4. Survival after nosocomial bloodstream infection according to systemic response

0%
0

1
2
3
4
Number of preexisting comorbidities

>4

Fig. 28.3. Importance of preexisting co-morbidities for prognosis of septicemia in critically ill patients (from ref. [48], with
permission)

tients was also four times longer than that of non-bacteremic patients (17.6 days vs. 4.3 days). The overall
mortality rate of non-bacteremic was 8.8 %, whereas
that of bacteremic patients was 44.3 %. Thus, bacteremic patients had a fivefold increased risk of dying
when compared with non-bacteremic patients

(RR = 5.03, CI 95 % 4.17 – 6.07, p < 0.0001). In this study
they found a close correlation between the number of
co-morbidities and fatality rates (Fig. 28.3). In addition, APACHE II 20 was also identified as an independent predictor of mortality.
A number of factors have been suspected as being
associated with mortality in bloodstream infection.
The most widely recognized prognostic factors are age,
severity of the patient’s underlying disease, and the appropriateness of antimicrobial therapy. Among other
factors potentially related to the outcome of bloodstream infection, a multiple source of infection, secondary infection, bloodstream infection caused by
some difficult-to-treat organisms such as Pseudomonas
or Serratia spp., polymicrobial bloodstream infection,
and factors related to host response such as the occurrence of hypotension, shock, or organ failure have all
been described as prognostically important. In a
French multicenter study of bloodstream infection and
severe sepsis in ICUs and wards of 24 hospitals, BrunBuisson et al. [33] reported that bloodstream infection
due to E. coli or CNS was associated with a lower risk of
severe sepsis and death, whereas S. aureus and grampositive organisms other than CNS were associated
with an increased risk of death. The results of that study
emphasize the impact of end-organ dysfunction (i.e.,
severe sepsis and septic shock) on prognosis in bloodstream infection.
In the multicenter study on nosocomial bloodstream infection carried out by our group [19] in 30
Spanish ICUs, crude mortality was 41.6 %, and 56 % of

all deaths were directly attributable to the bloodstream
infection. The crude mortality was correlated to the severity of systemic response; it was as high as 80 %
among patients with septic shock, compared with 26 %
among patients whose bacteremic episodes were manifested exclusively as sepsis. The cumulative probability
of survival stratified according to the grade of systemic
response is shown in Fig. 28.4. In addition, bloodstream infections originating in the abdomen or respiratory tract were associated with the highest mortality
(p = 0.04).
Because crude mortality cannot differentiate between mortality directly related to bloodstream infection and mortality attributable to underlying conditions, we were aware that different factors may influence the prognosis if we considered directly related

mortality or crude mortality. For this reason we performed a double multivariate analysis with different
dependent variables: one, related mortality, and the
other, crude mortality. In the related mortality analysis,
in addition to the level of systemic response and associated complications, we found that the type of microorganisms involved and the source of bloodstream infection played an important role in the prognosis. In the
crude mortality analysis, we found that in addition to
the systemic response and associated complications,
mechanical ventilation at the time of development of
bloodstream infection, chronic hepatic failure, and
APACHE II > 15 at the time of diagnosis of bloodstream
infection were chosen as factors by the statistical model; this seems to indicate that underlying diseases and
the severity of patient’s conditions markedly influence
crude mortality among ICU patients with nosocomial
bloodstream infection. On the other hand, the immediate prognosis after an episode of nosocomial bloodstream infection (related mortality) correlated with
level of systemic response, type of microorganism involved and the different sources of bloodstream infection.
Pittet et al. [49] recently conducted a large cohort
study to determine prognostic factors of mortality in

299


300

28 Bloodstream Infection in the Intensive Care Unit

ICU patients with positive blood cultures. They analyzed 173 patients with bacteremia, of whom 53.1 %
were nosocomially acquired. Among patients with bacteremic sepsis, 75 died (43 %); in 81 % of them, the
cause of death was considered to be directly or indirectly related to the infection. In this study, the best two independent prognostic factors were the APACHE II
score at the onset of sepsis (OR, 1.13; CI 95 % 1.08 – 1.17;
p < 0.001) and the number of organ dysfunctions developing thereafter (OR, 2.39; CI 95 % 2.02 – 2.82;
p < 0.001). This study suggests that in ICU patients with

positive blood cultures outcome can be predicted by
the severity of illness at onset of sepsis and the number
of vital organ dysfunctions developing subsequently.
28.9.2
Community-Acquired Bacteremia
Patients admitted in the ICU with community-acquired
bacteremia present a crude mortality near to 40 %,
compared with a mortality of 18 % in bacteremic patients admitted in general wards [12, 34, 50]. This elevated mortality in part is due to the severity of systemic
response that presents in these patients and that is the
cause of admission in the ICU [12, 34]. In addition to
the severity of systemic response (severe sepsis and
septic shock) and associated complications, the appropriateness of empiric antimicrobial treatment is the
most important variable influencing the outcome of
these patients [12, 34]. The incidence of inappropriate
antibiotic treatment community-acquired bacteremias
admitted in the ICU in two studies range between 15 %

Fig. 28.5. Survival rate according to the presence of shock and
initial antibiotic treatment. Log-rank test: p < 0.001. Group A
septic shock + delayed antibiotic treatment; Group B septic
shock + appropriate antibiotic treatment; Group C no septic
shock + delayed antibiotic treatment; Group D no septic shock
+ appropriate antibiotic treatment. (From ref. [34], with permission)

and 20 % and the mortality among patients with empiric inappropriate antibiotic treatment was more than
70 % [12, 34, 51]. The correlation between survival
time, systemic response to community-acquired
bloodstream infection, and delayed antibiotic treatment is shown as Kaplan-Meier curves in Fig. 28.5.

28.10

Conclusions
1. Nosocomial bloodstream infections occur two to
seven times more often in intensive care unit (ICU)
patients than in ward patients. Recent studies have
shown that the incidence rate ranges between 26
and 67 episodes per 1,000 ICU admissions, depending on the type of ICU.
2. Patients with nosocomial ICU bloodstream infection have a higher prevalence of intravenous lines
and respiratory sources of infection than ward patients in whom urinary tract infection is the most
prevalent source of bloodstream infection.
3. Gram-positive microorganisms are the most prevalent cause of nosocomial bloodstream infection in
ICU patients. This high incidence is related to the
high prevalence of bloodstream infection associated
with intravascular catheters in critically ill patients,
and to the multiple antibiotic therapy used for gramnegative infections in ICU patients, which results in
the selection of gram-positive microorganisms.
4. Currently, gram-negative microorganisms cause
between 30 % and 40 % of ICU-acquired bloodstream infections, and multiresistant organisms,
such as P. aeruginosa, Serratia spp, or A. baumannii, are the most frequently isolated pathogens.
5. Approximately 40 % of ICU patients with nosocomial bloodstream infection show a severe systemic
response, such as severe sepsis or septic shock, associated with high mortality.
6. The attributable mortality from nosocomial bloodstream infections is high in critically ill patients,
and the infection is associated with excessively
long ICU and hospital stays, and a significant economic burden.
7. The incidence rate of community-acquired bacteremia in adult ICUs is 10 episodes/1,000 admissions.
S. pneumoniae, S. aureus and E. coli represent more
than 80 % of microorganisms causing communityacquired bacteremia in the critically ill patients.
Most episodes are associated with severe sepsis or
septic shock, and they are associated with a high
mortality, and in the majority of cases directly
related with the infection. The severity of systemic

response and the appropriateness of empiric antibiotic treatment significantly influence the prognosis of these patients.


References

References
1. Wenzel RP (1990) Organization for infection control. In:
Mandell GL, Douglas RG Jr, Bennett JE (eds) Principles
and practice of infectious diseases, 3rd edn. Churchill Livingstone, New York, pp 2176 – 2180
2. Weinstein RA (1991) Epidemiology and control of nosocomial infections in adult intensive care units. Am J Med
91(Suppl 3B):179S – 184S
3. Wenzel RP, Thompson RL, Landry SM, et al. (1983) Hospital-acquired infections in intensive care unit patients: an
overview with emphasis on epidemics. Infect Control 4:
371 – 5
4. Maki DG (1989) Risk factors for nosocomial infection in
intensive care: “devices vs nature” and goals for the next
decade. Arch Intern Med 149:30 – 35
5. Donowitz LG, Wenzel RP, Hoyt JW (1982) High risk of hospital-acquired infection in the ICU patient. Crit Care Med
10:355 – 7
6. Brown RB, Hosmer D, Chen HC, et al. (1985) A comparison of infections in the different ICUs within the same hospital. Crit Care Med 13:472 – 476
7. Daschner F (1985) Nosocomial infections in intensive care
units. Intensive Care Med 11:284 – 287
8. Massanari RM, Hierholzer WJ Jr (1986) The intensive care
unit. In: Bennett JV, Brachman PS (eds) Hospital infections, 2nd edn. Little, Brown and Company, Boston,
pp 285 – 298
9. Trilla A, Gatell JM, Mensa J, et al. (1991) Risk factors for
nosocomial bacteremia in a large Spanish teaching hospital: A case-control study. Infect Control Hosp Epidemiol
12:150 – 156
10. Ponce de Le´on-Rosales S, Molinar-Ramos F, Dom´ınguezCherit G, Rangel-Frausto S, V´azquez-Ramos V (2000)
Prevalence of infections in intensive care units in Mexico:

A multicenter study. Crit Care Med 28:1316 – 1321
11. Forgacs IC, Eykyn SJ, Bradley RD (1986) Serious infection
in the intensive therapy unit: a 15-year study of bacteraemia. Q J Med 60:773 – 779
12. Vall´es J, Ochagav´ıa A, Ru´e M, D´ıaz E, et al. (2000) Critically ill patients with community-acquired bacteremia: Characteristics and prognosis. Intensive Care Med 26(Suppl.3):
S222
13. Reimer LG, Wilson ML, Weinstein MP (1997) Update on
detection of bacteremia and fungemia. Clin Microbiol Rev
10:444 – 465
14. Garner JS, Jarvis WR, Emori TG, et al. (1988) CDC definitions for nosocomial infections. Am J Infect Control 16:
128 – 140
15. Friedman ND, Kaye KS, Stout JE, et al. (2002) Health careassociated bloodstream infections in adults: A reason to
change the accepted definition of community-acquired infections. Ann Intern Med 137:791 – 797
16. Pittet D (1993) Nosocomial bloodstream infections. In:
Wenzel RP (ed) Prevention and control of nosocomial infections, 2nd edn. Williams & Wilkins, Baltimore, MD,
pp 512 – 555
17. Gatell JM, Trilla A, Latorre X, et al. (1988) Nosocomial bacteremia in a large Spanish teaching hospital: analysis of
factors influencing prognosis. Rev Infect Dis 10:203 – 210
18. Rello J, Ricart M, Mirelis B, et al. (1994) Nosocomial bacteremia in a medical-surgical intensive care unit: epidemiologic characteristics and factors influencing mortality in
111 episodes. Intensive Care Med 20:94 – 98
19. Vall´es J, Le´on C, Alvarez-Lerma F, et al. (1997) Nosocomial
bacteremia in critically ill patients: a multicenter study
evaluating epidemiology and prognosis. Clin Infect Dis
24:387 – 395

20. Towns ML, Quartey SM, Weinstein MP, et al. (1993) The
clinical significance of positive blood cultures: a prospective, multicenter evaluation, abstr. C-232. In: Abstracts of
the 93rd General Meeting of the American Society for Microbiology 1993. American Society for Microbiology,
Washington, DC
21. Weinstein MP, Towns ML, Quartey SM, et al. (1997) The
clinical significance of positive blood cultures in the 1990s:

a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults. Clin Infect Dis 24:584 – 602
22. Mirret S, Weinstein MP, Reimer LG, et al. (1994) Interpretation of coagulase-negative staphylococci in blood cultures: does the number of positive bottles help? Abstr. C69. In: Abstracts of the 93rd General Meeting of the American Society for Microbiology 1994. American Society for
Microbiology, Washington, DC
23. Martin MA, Pfaller MA, Wenzel RP (1989) Coagulase-negative staphylococci bacteremia. Mortality and hospital
stay. Ann Intern Med 110:9 – 16
24. Ponce de Le´on S, Wenzel RP (1984) Hospital-acquired
bloodstream infections with Staphylococcus epidermidis.
Am J Med 77:639 – 644
25. Dom´ınguez-de Villota E, Algora-Weber A, Millan I, et al.
(1987) Early evaluation of coagulase-negative staphylococci in blood samples of intensive care unit patients. A
clinically uncertain judgement. Intensive Care Med 13:
390 – 394
26. Craven DE, Kunches LM, Lichtenberg DA, et al. (1988) Nosocomial infections and fatality in medical and surgical intensive care unit patients. Arch Intern Med 148:1161 – 1168
27. Constantini M, Donisi PM, Turrin MG, et al. (1987) Hospital-acquired infections surveillance and control in intensive care services. Results of an incidence study. Eur J Epidemiol 3:347 – 355
28. Daschner FD, Frey P, Wolff G, et al. (1982) Nosocomial infections in intensive care wards: A multicenter prospective
study. Intensive Care Med 8:5 – 9
29. Trilla A (1994) Epidemiology of nosocomial infections in
adult intensive care units. Intensive Care Med 20:S1–S4
30. Vincent JL, Bihari DJ, Suter PM, et al. (1995) The prevalence of nosocomial infection in intensive care units in Europe. JAMA 274:634 – 644
31. Richards MJ, Edwards JR, Culver DH, et al. (1999) Nosocomial infections in medical intensive care units in the United States. Crit Care Med 27:887 – 892
32. Pittet D, Tarara D, Wenzel RP (1994) Nosocomial bloodstream infection in critically ill patients. Excess length of
stay, extra costs, and attributable mortality. JAMA 271:
1598 – 1601
33. Brun-Buisson C, Doyon F, Carlet J, et al. (1996) Bacteremia
and severe sepsis in adults: A multicenter prospective survey in ICUs and wards of 24 hospitals. Am J Respir Crit
Care Med 154:617 – 624
34. Vall´es J, Rello J, Ochagav´ıa A, Garnacho J, Alcal´a MA, and
Spanish Collaborative Group for Infections in Intensive
Care Units of Sociedad Espa˜nola de Medicina Intensiva,
Cr´ıtica y Unidades Coronarias (2003) Community-acquired bloodstream infection in critically ill adult patients:

Impact of shock and inappropriate antibiotic therapy on
survival. Chest 123:1615 – 1624
35. Edgeworth JD, Treacher DF, Eykyn SJ (1999) A 25-year
study of nosocomial bacteremia in an adult intensive care
unit. Crit Care Med 27:1421 – 1428
36. Jamal WY, El-Din K, Rotimi VO, et al. (1999) An analysis of
hospital-acquired bacteraemia in intensive care unit patients in a university hospital in Kuwait. J Hosp Infect
43:49 – 56

301


302

28 Bloodstream Infection in the Intensive Care Unit
37. Bone RC, Balk RA, Cerra FB, et al. (1992) Definitions for
sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 101:1644 – 1655
38. Wenzel RP (1995) Isolation of Candida species from sites
other than the blood. Clin Infect Dis 20:1531 – 1534
39. D´ıaz E, Villagr´a A, Mart´ınez M, et al. (1998) Nosocomial
candidemia risk factors. Intensive Care Med 24 (Suppl 1):
S143
40. Leibovici L, Konisberger H, Pitlik SD (1992) Bacteremia
and fungemia of unknown origin in adults. Clin Infect Dis
14:436 – 439
41. Weinstein MP, Reller LB, Murphy JR, et al. (1983) The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bloodstream infection and
fungemia in adults. I. Laboratory and epidemiologic observations. Rev Infect Dis 5:35 – 53
42. Wenzel RP (1988) The mortality of hospital-acquired
bloodstream infections: need for a new vital statistic? Int J
Epidemiol 17:225 – 227

43. Munford RS (1994) Sepsis and septic shock. In: Isselbacher
KJ, Braunwald E, Wilson JD, Martin JB, Fauci AS, Kasper
DL (eds) Harrison’s principles of internal medicine, 13rd
edn. McGraw-Hill, New York, pp 511 – 515
44. Bueno-Cavanillas A, Delgado-Rodr´ıguez M, L´opez-Luque
A, et al. (1994) Influence of nosocomial infection on mortality rate in an intensive care unit. Crit Care Med 22:
55 – 60

45. Crowe M, Ispahani P, Humphreys H, et al. (1998) Bacteraemia in the adult intensive care unit of a teaching hospital in
Nottingham, UK, 1985 – 1996. Eur J Clin Microbiol Dis
17:377 – 384
46. Landry SL, Kaiser DL, Wenzel RP (1989) Hospital stay and
mortality attributed to nosocomial enterococcal bacteremia: a controlled study. Am J Infect Control 17:323 – 329
47. Wey SB, Motomi M, Pfaller MA, et al. (1988) Hospital-acquired candidemia: the attributable mortality and excess
length of stay. Arch Intern Med 148:2642 – 2645
48. Pittet D, Thi´event B, Wenzel RP, et al. (1993) Importance of
pre-existing co-morbidities for prognosis of septicemia in
critically ill patients. Intensive Care Med 19:265 – 272
49. Pittet D, Thi´event B, Wenzel RP, et al. (1996) Bedside prediction of mortality from bacteremic sepsis. A dynamic
analysis of ICU patients. Am J Respir Crit Care Med
153:684 – 693
50. Cart´on JA, Garc´ıa-Velasco G, Maradona JA, P´erez F, Asensi
V, Arribas JM (1988) Bacteriemia extrahospitalaria en
adultos. An´alisis prospectivo de 333 episodios. Med Clin
(Barc) 90:525 – 530
51. Ibrahim EH, Sherman G, Ward S, Fraser VJ, Kollef MH
(2000) The influence of inadequate antimicrobial treatment of bloodstream infections on patient outcomes in the
ICU setting. Chest 118:146 – 155

Chapter 29

Bloodstream Infections in Patients
with Total Parenteral Nutrition Catheters
R. Sierra, A. Ram´ırez

29.1
Introduction
Vascular access is an essential procedure in the management of critically ill patients, especially the insertion of central venous catheters (CVCs). The most commonly used CVCs are noncuffed percutaneously inserted catheters placed in the femoral, internal jugular or
subclavian veins [1 – 3]. Unfortunately, these intravascular devices are associated with the risk of complications. Potential CVC-related complications include
chiefly arterial puncture, pneumothorax, hemothorax,
thrombosis, hematoma, and infectious complications.
Among the most important life-threatening complications of intravascular devices are catheter-related
bloodstream infections (CRBSIs) [2 – 7], which represent a major cause of nosocomial infection in intensive
care units (ICUs) [8 – 10]. The National Nosocomial Infections Surveillance (NNIS) System reported in 2004
[11] a CRBSI mean rate in United States ICUs of 4.85
CRBSI cases per 1,000 central line-days (mean value of
pooled means from different types of surveyed ICU).
Mean rates of the other two main sources of nosocomial infection in US ICUs were 4.9 urinary catheter-associated urinary tract infections per 1,000 urinary-catheter-days, and 11.1 ventilator-associated pneumonias
per 1,000 ventilator-days. Twenty-five percent of bloodstream infections that occur in the ICU are secondary
to catheter-related infections (CRIs). In addition, up to
80 % of primary bacteremia may be linked to CRIs [9,
12, 13].
Attributable mortality from CRIs in critically ill patients has been found high in some studies [14 – 16],
though this finding is controversial [8]. Between 2,400
and 20,000 deaths are estimated to be produced by CRIs
yearly in the USA [1, 17, 18], giving mortality rates
ranging from 14 % to 28 % [6, 10, 14, 19 – 24]. Nevertheless, mortality rates from CRBSI are relatively low, if
they are compared with the mortality from other infectious foci [9]. CRBSI is also associated with an excess of
length of stay both in ICUs and hospital, further increasing cost [8, 9, 14, 17, 22, 24 – 29].

29.2
Definitions
29.2.1
CVCs
CVCs may be classified according to the insertion
length, e.g., (1) short-term catheters, in place < 10 days,
and (2) long-term catheters, in place > 10 days [6].
However, other researchers have defined short-term
catheters as those with placement duration < 7 days,
and long-term catheters as those in place > 7 days [30].
29.2.2
Exit Site Infection
Exit site infection is considered when local signs occur,
such as tenderness, skin erythema, induration within
2 cm of the catheter exit site (with or without fever), or
cellulitis along the subcutaneous tract, in the absence
of pus at the exit site. Except for the presence of pus,
these signs lack specificity and may be caused by host
immune response against the CVC, or by the administered fluid as well. The presence of pus is usually a diagnostic sign of infection, even when a culture from the
catheter tip is not available [30 – 35].
29.2.3
Colonization
Colonization occurs when a positive culture from either catheter tip, subcutaneous segment of the catheter,
or catheter hub is obtained with a result of & 15 colonyforming units (cfu)/ml [6, 30, 36].
29.2.4
Catheter-Related Bloodstream Infection
Catheter-related bloodstream infection is defined when
signs of systemic infection (i.e., sepsis) are associated
with positive blood cultures which have been obtained

by any diagnostic method. Matched microorganisms
should be isolated in the catheter tip, and in blood cultures from the peripheral vein. Furthermore, other apparent sources of infection should not occur [30].

29


304

29 Bloodstream Infections in Patients with Total Parenteral Nutrition Catheters

Diagnosing CRBSI by coagulase-negative staphylococci requires microbial growth to be obtained in at
least two peripheral-blood samples [37].
29.2.5
Infusate-Related Infection
Infusate-related infection is present when there are
signs of systemic infection, in the absence of other apparent infectious sources. In addition, the same microorganism should grow in both peripheral-blood samples, and in the fluids administered. Cultures of the
catheter tip are not required to be positive [30].

29.3
Etiology
Catheter-related infection is caused mainly by microorganisms from the skin flora. However, in the hospital
setting, a normal flora is usually replaced by pathogenic bacteria. Patients who are receiving antimicrobial
therapy are often colonized by gram-negative bacilli,
Staphylococcus aureus or fungi. Besides, microorganisms from the airways are frequently isolated in patients with tracheostomy. Microorganism types which
are isolated from catheters appear to be related to insertion sites. Aerobic gram-negative bacilli, Candida species, and anaerobes are isolated in the inguinal region
more frequently.
The most frequent pathogens related to the etiology
of CRBSI are coagulase-negative staphylococci, Staphylococcus aureus, enterococci, aerobic gram-negative
bacilli, and Candida spp. (especially C. albicans). The
microorganism most commonly isolated in catheterrelated sepsis is Staphylococcus epidermidis, which

seems to be associated with a lower mortality rate than
other pathogens. A higher rate of mortality has been
found associated with Staphylococcus aureus CRI. Antimicrobial treatment of these pathogens may be difficult
because many isolates are increasingly becoming resistant to oxacillin and other antibiotics [1, 38, 39].

29.4
Risk Factors
Multilumen central venous catheters are associated
with a greater risk of CRI when compared with the risk
from single-lumen catheters, since multilumen catheters are more frequently manipulated so increasing the
chance of a breakdown in protective barriers [40 – 46].
Heavy cutaneous colonization is also a major risk
factor for CRI [6]. CRI rate was decreased in patients
who received chlorhexidine gluconate for insertion-site
skin disinfection, compared with those who received

povidone-iodine. Such a practice constitutes a simple
measure for reducing the occurrence of CRI [47].
Femoral vein insertion site is considered to be associated with the highest rate of microbial colonization,
since this skin zone usually has a heavier cutaneous colonization. Colonization risk is lower for the jugular site
[6, 48]. Infection occurs more frequently in the jugular
vein than in the subclavian vein. It may be favored by
neck movements, which make dressing care of catheters difficult. Infection risk is lower for subclavian vein
insertion sites [49 – 51].
The longer the catheter is in place the higher the
probability of CRI occurrence [34]. CRI is also more
frequent in patients in whom two or more catheters
have been inserted [9].
Recent studies carried out on hematology-oncological patients have shown an association between fibrin
deposition, catheter-related thrombosis and infection

[52 – 56], but these findings have not been confirmed in
other studies [3, 56].
Administration of blood products through CVCs is
another risk factor for CRBI, although thrombocytopenia during catheterization may provide some protection against CRBI [57, 58].
Parenteral nutrition (PN) was identified as an independent risk factor for CRI in hospitalized patients,
particularly those in the ICU, which is probably explained by hyperglycemia. The pathogenic role of hyperglycemia in other patients groups is uncertain [35,
59 – 63].
ICU admission when nursing staff are less available
has also been identified as a risk factor for CRI.
Unstable clinical status has not been demonstrated
to be a risk factor for CRI [35].
Malnutrition appears not to be a risk factor for CRI
but influences clinical outcome, and is associated with
more complications, increased mortality rates, and increased hospital length of stay and costs [64, 65].

29.5
Pathogenesis
CRBSI principally occurs by two routes, extraluminally
and intraluminally. The extraluminal route occurs
when there is concordance among isolates from catheter segments, skin, and blood cultures. The intraluminal route occurs when isolates from a hub, or infusate
fluids, and blood cultures are concordant. The route of
infection is considered as being indeterminate when
both routes are possible [6].
CRBSI often occurs following catheter colonization
[1, 20]. Pathogens firstly have to gain access to the intraluminal or extraluminal surface of the catheter [6].
Intravascular devices cause a local inflammatory response in the site of insertion, and then several proteins


29.6 Diagnosis

covering the catheter [66 – 71] favor the adherence of
microorganisms by diverse mechanisms [66, 67].
Microorganisms gain access into the body through
one of the three following mechanisms:
1. At the time of insertion or later, the skin flora invades the percutaneous tract through the insertion
site, involving initially the external surface of the
catheter (extraluminal colonization). This mechanism is regarded to be the major mechanism in
short-term nontunneled catheter-associated infections.
2. Microorganisms contaminate the catheter hub and
lumen (intraluminal colonization). This mechanism results from frequent manipulations, or when
the catheter is inserted over a percutaneous guidewire. When epidemic CRBSI occurs, a contaminated infusion should also be considered.
3. Microorganisms may occasionally be carried hematogenously to the intravascular device from a
remote source of infection. This mechanism is not
frequent [1, 6, 48, 72 – 76].
Most infections associated with short-term catheters
are caused by skin flora surrounding the insertion site
which gains access via an extraluminal route, and occasionally intraluminally. With long-term catheters, there
is a predomination of intraluminal colonization with
contamination of the hub and afterwards of the lumen.
The intraluminal route commonly predominates when
the placement is longer than 1 – 2 weeks [6]. The mechanism of infection that is attributed to CVCs inserted in
old sites over a guidewire appears to be no different
from that of catheters inserted in de novo sites [6, 48].
After microorganisms gain access to the intravascular device, they can adhere to it, and produce extracellular polymer substances (“slime”), which facilitate further adhesion to CVC surfaces. These polymers develop
into a matrix which leads to biofilm formation. Infection is derived from the microbes’ ability to adhere,

proliferate, and elaborate biofilm. These actions allow
sustained infection, and hematogenous dissemination
[1, 6, 48].
The microorganisms commonly associated with

biofilm formation in catheters are: coagulase-negative
staphylococci, Staphylococcus aureus, Enterococcus faecalis, Klebsiella pneumonia, Pseudomonas aeruginosa,
and Candida albicans [1, 77].
All catheters develop biofilms in vivo. Initially, this
effect is not significant; however, when the catheter has
been in place for a long time, biofilms can become a
persistent source of infection, and may oppose host defenses by decreasing the effect of antibiotics [1, 78]. In
addition there is decreased diffusion of antibiotics in
biofilms, and other mechanisms which favor resistance
occurrence. Biofilm-associated pathogens require a
greater concentration of antibiotics to be eliminated
since they have decreased antimicrobial susceptibility
[1, 77, 79].

29.6
Diagnosis
The diagnosis of CRI is often based on the exclusion of
the presence of other inflammatory sources [34].
CRBSI diagnostic methods may be categorized into two
groups, those with catheter removal, and those without
catheter removal. The most common methods are
those with catheter removal and catheter-tip culturing
when CRBSI is suspected (Table 29.1). Nevertheless,
most of the catheters are not usually infected and replacement may increase the risk of complications and
cost [76, 80 – 83]. The methods for diagnosing CRI
without catheter removal are listed in Table 29.2.
Subcutaneous segment cultures appear not to be
useful for diagnosing CRBSI [84].

Table 29.1. Diagnostic methods with catheter removal [1, 2, 34]

Description

Diagnostic
cut-off value

Pooled sensitivity (95 % CI)

Pooled specificity (95 % CI)

Qualitative catheter
segment culture

Catheter segment is immersed in a broth
media, and then incubated for 24 – 72 h. This
method is not recommended since it has a
poor specificity

Any growth

0.87 (0.79 – 0.96)

0.75 (0.72 – 0.78)

Semiquantitative catheter segment culture
(Maki method)

The most used method to diagnose CRBSI.
The catheter tip is rolled 4 times across an
agar plate, then incubated, and observed after
an overnight period

& 15 cfu/ml

0.83 (0.79 – 0.87)

0.86 (0.85 – 0.87)

Quantitative catheter
segment culture

It requires vortex or sonicating catheter samples in broth, or flushing the broth through
the catheter, and then plating on blood agar
serial dilutions

& 103 cfu/ml

0.82 (0.78 – 0.86)

0.89 (0.87 – 0.91)

305


306

29 Bloodstream Infections in Patients with Total Parenteral Nutrition Catheters
Table 29.2. Diagnostic methods without catheter removal [1, 2, 34]
Description
Qualitative blood cul& 1 blood samples for cultures are drawn
ture through catheter

from the catheter
Quantitative blood cul- A blood sample for culture is drawn from
the catheter, and processed by pour-plate
ture through catheter
or lysis-centrifugation technique
Concomitant blood samples for cultures
Paired qualitative
blood cultures
are both drawn through the catheter, and
percutaneously

Diagnostic
cut-off value

Pooled sensitivity (95 % CI)

Pooled specificity (95 % CI)

Any growth

0.91 (0.84 – 0.98)

0.86 (0.83 – 0.89)

0.84 (0.80 – 0.89)

0.90 (0.88 – 0.92)

& 100 cfu

Microorganisms
0.79 (0.74 – 0.84)
are & 5-fold greater in central blood
sample
0.89 (0.86 – 0.92)
Differential time to
Concomitant blood samples for cultures Central blood
positivity
are both drawn through the catheter, and sample turns pospercutaneously, and then are monitored itive 120 min
before
continuously
Acridine orange leuko- 1 ml of blood is aspirated from the cathe- Any microorgan- 0.87 (0.80 – 0.94)
cyte cytospin (AOLC) ter, then the cells are lysed with sterile
ism is visualized
water, centrifuged, stained with acridine
orange, and observed. Simple and rapid
test. It allows an early targeted antimicrobial therapy, and is recommended as the
first line investigation of CRBSI [76, 85]

29.7
Management
Catheter removal whenever a CRI is suspected is the
common approach to managing these frequent nosocomial infections. However, many catheters are removed unnecessarily, since in many cases they are not
associated with infection. Besides, CVC reinsertion
may be further associated with complications [1].
Antibiotic therapy is empirically initiated by the intravenous route. The choice of a given antibiotic regime
usually depends on illness severity, patient risk factors,
and likely pathogens associated with the intravascular
device.
Vancomycin is recommended in hospitals where

there are frequently methicillin-resistant Staphylococci
(MRSA). Oxacillin should be used in the absence of epidemic, or endemic, MRSA flora.
In addition, empiric treatment with an antipseudomonal beta-lactamic agent should be considered in immunocompromised, or seriously ill, patients, to cover
enteric gram-negative bacteria and Pseudomonas spp.
When fungemia is suspected, then amphotericin B or
intravenous fluconazole should be used. Caspofungin
or voriconazole are alternative therapies when candidiasis is suspected in an unstable patient. If the clinical
status of the patient has been stabilized, switching to
oral agents can be considered [30].
Catheters may not have to be removed initially, particularly if the microorganism isolated is coagulasenegative staphylococci [30, 86].
If severe sepsis is not in evidence (i.e., the presence
of hypotension, hypoperfusion, or organ failure) and

0.99 (0.98 – 1.0)

0.83 (0.79 – 0.87)

0.93 (0.89 – 0.97)

no infection signs are observed at the insertion site, the
catheter should be removed only when either: (a) cultures of blood drawn from the catheter yield positive
results, (b) there is persistent fever, or (c) the results of
peripheral blood cultures are negative because the
catheter was not cultured.
Whenever patients exhibit a serious illness, sepsis,
or signs of infection at the exit site, the catheter should
be removed.
For treatment purposes, patients with non-tunneled
catheters and CRBSI may be distributed into two
groups: complicated CRBSI (with septic thrombosis,

endocarditis, osteomyelitis, or emboli) or non-complicated CRBSI.
In the case of a peripheral blood culture negative result, and the catheter culture reveals significant growth
of S. aureus or C. albicans (either febrile patients with
valvular heart disease or neutropenic patients), then
the patient should be observed and peripheral blood
cultures repeated. Some authors advise the delivery of a
short course (5 – 7 days) of antibiotic therapy [1, 30].
CRI caused by coagulase-negative staphylococcus
must receive a 5 – 7 day course of antimicrobial therapy,
combined with catheter removal. A course of 10 –14 days
of local antibiotic lock (ABL) may be applied if the catheter is not removed. The catheter should be removed for
pathogens other than coagulase-negative staphylococci,
and patients should receive 10 – 14 days of antimicrobial
therapy. A course of 4 – 6 weeks should be considered in
the case of persistent bacteremia or fungemia after catheter removal, or if there is evidence of complicated infection (except in cases of osteomyelitis, which requires
6 – 8 weeks of therapy). The antimicrobial treatment for


29.8 Bloodstream Infections in Patients with Total Parenteral Nutrition Catheters

Candida spp. should last up to 14 days after the last positive blood culture.
Streptokinase in combination with antimicrobial
therapy has not been demonstrated to be beneficial for
the treatment of CRI [30].
In the case of persistent bacteremia, fungemia, or
when clinical improvement after 3 days of appropiate
antibiotic therapy and catheter withdrawal is lacking,
endocarditis should be ruled out with transesophageal
echocardiography. If the results of such a test are negative, then aggressive workup for septic thrombosis or
for another metastasic infection should ensue [1, 30].

In cases of tunneled CVCs or implantable devices it
is important to confirm that a related infection has occurred. Catheters must be removed in cases of complicated infections, CRI by Candida spp., tunnel infection,
port abscess, and when following an initially maintained catheter there is clinical deterioration or persistent bacteremia.
The treatment regime is similar to that of non-tunneled catheters, in the case of pathogens other than
Candida spp., and those mentioned above. However, if
the catheter has to be retained, systemic antibiotic therapy should be combined with ABL for 10 – 14 days (Table 29.3) [30]. ABL has been used to decrease the duration of systemic antibiotic treatment, and to maintain a
high antibiotic concentration within the CVC. ABL
comprises a mixture of 0.3 ml (40 mg) of teicoplanin
(400 mg per 3 ml) and 0.2 ml of sodium heparin at
500 IU per 5 ml, although other antibiotics or antifungal agents can be also used. When CRBSI is confirmed,
this 0.5-ml lock is injected into the catheter, and left for
12 h. Later, this small volume is aspirated before initiating PN. ABL is administered for 12 – 15 days, in combination with short-duration systemic antibiotherapy
(usually a glycopeptide plus an aminoglycoside). Systemic antiobiotherapy is administered in general for
the first 5 days [39].
Table 29.3. Management of tunneled CVCs [30]
Evidence
level
Ensure that the CVC is really the source of infec- IIIB
tion
The CVC should be removed in case of complicated infections

IIB

For salvage of the CVC in patients with uncom- IIB
plicated infections, ABL should be used for
2 weeks with standard systemic antibiotic therapy in the absence of tunnel or pocket infection
Tunneled catheter pocket infections or port
abscess require removal of catheter and usually
7 – 10 days of appropriate antibiotic therapy

IIIC

Antibiotic lock therapy is recommended for
treatment when the catheter is retained

IIIB

ABL success depends on antibiotic concentrations
within the catheter [87]. High antibiotic concentrations
augment antimicrobial efficacy and lessen the secondary effects of systemic antibiotic treatment.
The ABL method is recommended and supported by
findings from in vitro models which have shown reductions in staphylococcal, gram-negative and fungal colonization rates. Some trials have also demonstrated clinical efficacy for CRBSI, especially for non-tunneled catheters [39]. However, ABL is not recommended in longterm PN, because ABL appears not to prevent a second
or third episode of CRI by the same bacterial strain but
with an increase in teicoplanin resistance [38, 88, 89].

29.8
Bloodstream Infections in Patients
with Total Parenteral Nutrition Catheters
Parenteral nutrition is indicated when gut function is
altered, and enteral nutrition is not suitable. PN serves
to prevent the adverse effects of malnutrition, and its
use is not exclusive to hospitalized patients. Delivery of
PN to outpatients is known as home PN (HPN). PN is
not indicated for unstable patients. The impact of PN
on mortality and morbidity is a controversial issue, because of the occurrence of frequent complications related to PN use. CRI constitutes a major complication
derived from PN, and represents the main cause for readmission to hospital in HPN patients [39, 90 – 96].
Subclavian vein access is a common approach for
delivering PN, whether subcutaneously or not [97].
Subclavian vein access is preferred for infection control
purposes. Frequency of mechanical complications may

be decreased by using bedside ultrasound for catheter
placement [31].
Peripherally inserted central catheters (PICCs) can
also be used for delivering PN. PICCs are small-size
catheters inserted into the subclavian vein through the
basilic or cephalic vein. PICCs are associated with fewer
mechanical complications during the insertion procedure than other venous access, but are long-term catheters in HPN patients. The use of such catheters appears
to be more associated with increased risk of phlebitis,
thrombosis or sepsis when compared with that of CVCs
[98 – 103]. A higher frequency of CRI cases related to
PICCs used for HPN may be related to a higher exposure of the arms to microbes than the chest wall surface. It is crucial not only to use a sterile technique during insertion, but also to deliver proper catheter care
[97]. HPN patients should report to their healthcare
provider any changes in their catheter site, and any new
discomfort, and as well as avoiding submerging the
catheter under water. Showering can be allowed whenever the catheter and connecting device are protected
with an impermeable cover during the shower [31].

307


308

29 Bloodstream Infections in Patients with Total Parenteral Nutrition Catheters

By tunneling CVC appears to reduce the CRI risk.
This measure should be considered when circumstances make it not feasible to cannulate a subclavian
vein [6]. For patients requiring frequent or continuous
venous access, a PICC or tunneled CVC is usually employed. However, a totally implantable access device is
the recommended approach for patients who require
long-term, intermittent vascular access [31].

Candida spp. and Malassetia spp. are more frequently isolated in PN patients with CRI than in patients with CVCs not used to PN. Certain Candida spp.,
in the presence of glucose-containing fluids, may also
produce slime, which may explain the elevated rate of
CRBSI caused by fungal pathogens found among patients receiving PN [31].
Increased blood glucose levels have been related to
higher infection rates in hospitalized patients [104], especially in critically ill patients [105]. Hyperglycemia in
PN patients can be explained by the intense activation
of contraregulatory hormones, and cytokine responses, which are both associated with circumstances
such as severe disease, and excessive administration of
glucose. Patients with PN exhibit frequently sustained
hyperglycemia, and often receive insulin. Hyperglycemia impairs immune response as well, reducing neutrophil chemotaxis and phagocytosis, which can increase risk of infection onset [59, 105, 106]. Tight control of glycemia may reduce mortality rates significantly in surgical ICU patients [105]; however, such intensive insulin therapy has been demonstrated to reduce
only morbidity, but not mortality, rates, in patients in
the medical ICU [107].
Possibly the contamination by particulates, such as
undetected trace elements, could also favor CRI occurrence [108, 109].
A high CRI incidence rate occurring in PN patients
could favor the use of antiseptic- or antibiotic-impregnated catheters [50, 110 – 113]. The use of antibioticimpregnated catheters is associated with lower colonization rates. However, such CRBSI incidence rates appear to be no different when they are compared with
those of non-impregnated catheters [114 – 124]. Antimicrobial-impregnated catheters have been demonstrated to reduce the risk of CRBSI only among patients
whose catheters were used for delivering total PN [116,
125]. Minocycline plus rifampin-impregnated catheters were demonstrated to be effective only against
staphylococci strains (S. aureus and S. epidermidis).
However, colonization frequency by Candida spp. is
higher than in non-impregnated catheters [116, 121,
126]. Utilization of miconazole plus rifampin-impregnated catheters is associated with lower rates of CRI
when compared to standard catheters. These special
catheters may be effective on prevention of CRI by Candida spp., although it has not yet been demonstrated
[10].

Tubing used for administering total PN, or lipid
emulsions, should be replaced within 24 h after initiating the infusion.

For infection control purposes, all CVCs must have
the least number of ports or lumens needed for the
management of the patient, and should be removed as
soon as their use is no longer essential [31]. Catheter
colonization risk in PN patients is decreased when single-lumen catheters are inserted through the subclavian vein, are used exclusively for PN, and are cared for
by and under the control of a multidisciplinary team
[97, 127].

References
1. Lane RK, Matthay MA (2002) Central line infections. Curr
Opin Crit Care 8:441 – 8
2. Safdar N, Fine JP, Maki DG (2005) Meta-analysis: methods
for diagnosing intravascular device-related bloodstream
infection. Ann Intern Med 142:451 – 466. Erratum in: Ann
Intern Med 142:803
3. Abdelkefi A, Torjman L, Ladeb S, et al. (2005) Randomized
trial of prevention of catheter-related bloodstream infection by continuous infusion of low-dose unfractionated
heparin in patients with hematologic and oncologic disease. J Clin Oncol 23:7864 – 7870
4. Maki DG (1989) Pathogenesis, prevention and management of infections due to intravascular devices used for infusion therapy. In: Waldwogel F (ed) Infections associated
with indwelling medical devices. American Society for Microbiology, Washington, DC, pp 161 – 177
5. Maki DG, Goldman DA, Rhame FS (1973) Infection control in intravenous therapy. Ann Intern Med 79:867 – 887
6. Safdar N, Maki DG (2004) The pathogenesis of catheter-related bloodstream infection with noncuffed short-term
central venous catheters. Intensive Care Med 30:62 – 67
7. Valles J, Leon C, Alvarez-Lerma F (1997) Nosocomial bacteremia in critically ill patients: a multicenter study evaluating epidemiology and prognosis. Clin Infect Dis 24:
387 – 395
8. Rello J, Ochagavia A, Sabanes E, et al. (2002) Evaluation of
outcome of intravenous catheter-related infections in critically ill patients. Am J Respir Crit Care Med 162:1027 – 1030
9. Brun-Buisson C, Doyon F, Sollet JP, et al. (2004) Prevention
of intravascular catheter-related infection with newer
chlorhexidine-silver sulfadiazine-coated catheters: a randomized controlled trial. Intensive Care Med 30:837 – 843

10. Yucel N, Lefering R, Maegele M, et al. (2004) Reduced colonization and infection with miconazole-rifampicin modified central venous catheters: a randomized controlled
clinical trial. J Antimicrob Chemother 54:1109 – 1115
11. National Nosocomial Infections Surveillance System
(2004) National Nosocomial Infections Surveillance
(NNIS) System Report, data summary from January 1992
through June 2004, issued October 2004. Am J Infect Control 32:470 – 485
12. Maki DG (1981) Nosocomial bacteremia. An epidemiologic overview. Am J Med 70:719 – 732
13. Richards MJ, Edwards JR, Culver DH, et al. (2000) Nosocomial infections in combined medical-surgical intensive
care units in the United States. Infect Control Hosp Epidemiol 21:510 – 515
14. Pittet D, Tarara D, Wenzel RP (1994) Nosocomial bloodstream infection in critically ill patients. Excess length of