(Chest. 2001;120:2059-2093.)
© 2001
American College of Chest Physicians
Infection Control in the ICU*
Philippe Eggimann, MD and
Didier Pittet, MD, MS
*
From the Medical Intensive Care Unit (Dr. Eggimann) and the Infection Control Program (Dr. Pittet), Department of Internal Medicine, University of Geneva Hospitals, Geneva, Switzerland.
Correspondence to: Didier Pittet, MD, MS, Infection Control Program, Department of Internal Medicine, University Hospitals of Geneva, 1211 Geneva 14, Switzerland; e-mail: didier.pittet{at}hcuge.ch
 |
Abstract
|
|---|
Nosocomial infections (NIs) now concern 5 to 15% of hospitalized
patients and can lead to complications in 25 to 33% of those patients
admitted to ICUs. The most common causes are pneumonia related to
mechanical ventilation, intra-abdominal infections following trauma or
surgery, and bacteremia derived from intravascular devices. This
overview is targeted at ICU physicians to convince them that the
principles of infection control in the ICU are based on simple concepts
and that the application of preventive strategies should not be viewed
as an administrative or constraining control of their activity but,
rather, as basic measures that are easy to implement at the bedside. A
detailed knowledge of the epidemiology, based on adequate surveillance
methodologies, is necessary to understand the pathophysiology and the
rationale of preventive strategies that have been demonstrated to be
effective. The principles of general preventive measures such as the
implementation of standard and isolation precautions, and the control
of antibiotic use are reviewed. Specific practical measures, targeted
at the practical prevention and control of ventilator-associated
pneumonia, sinusitis, and bloodstream, urinary tract, and surgical site
infections are detailed. Recent data strongly confirm that these
strategies may only be effective over prolonged periods if they can be
integrated into the behavior of all staff members who are involved in
patient care. Accordingly, infection control measures are to be
viewed as a priority and have to be integrated fully into the
continuous process of improvement of the quality of
care.
Key Words: bloodstream infection • critical care • epidemiology • nosocomial infection • prevention • ventilator-associated pneumonia
|
Introduction
|
|---|
According
to the Institute of Medicine1
in Washington, DC,
preventable adverse events in the United States, including
hospital-acquired infections, are responsible for 44,000 to 98,000
deaths annually and represent a cost of $17 to $29 billion. As precise
epidemiologic data about these events are sparse, this estimation was
extrapolated from two studies only.2
3
4
5
This report has
generated a considerable debate in the medical
literature.6
7
8
9
Nevertheless,
data10
11
12
have suggested that the likelihood of
the occurrence of these events may increase by 6% for each day spent
in the hospital, and they were found to be more frequent among patients
in ICUs.
During the last decade, the growing emphasis on outpatient medical
management has resulted in a marked reduction of beds in many
health-care institutions, and this policy has been responsible for an
increasing severity of illness among hospitalized patients. Data from
the Centers for Disease Control and Prevention (CDC) National
Nosocomial Infection Surveillance (NNIS) system show a 17% increase in
the number of ICU beds at the 117 participating hospitals from 1988
through 1995, as compared with a slight decrease in the total bed
capacity.13
Although representing only 5 to 15% of
hospital beds, ICUs accounted for 10 to 25% of health-care costs,
corresponding to 1 to 2% of the gross national product of the United
States.14
Nosocomial infections (NIs) affect > 2 million persons annually in
the United States and concern 5 to 35% of patients who are admitted to
ICUs.15
They are viewed as an inexorable tribute to pay to
the more aggressive management of the population, characterized by the
use of sophisticated technologies and invasive devices. The
pathophysiology of NIs includes colonization of the host by potentially
dangerous pathogens, such as microorganisms from exogenous or
endogenous sources, including resistant strains such as
methicillin-resistant Staphylococcus aureus (MRSA),
vancomycin-resistant enterococci (VRE), azole-resistant Candida spp,
and extended-spectrum ß-lactamase (ESBL) Gram-negative pathogens.
Ventilator-associated pneumonia, catheter-related bloodstream
infections, surgical site infections (SSIs), and urinary
catheter-related infections account for > 80% of
NIs.16
17
The Study on the Efficacy of Nosocomial Infection
Control15
18
19
from the CDC has suggested that at least
one third of NIs are preventable through infection control programs,
which have been implemented in most centers during the last 2 decades.
Risk factors are well-identified and have been the target of efficient
preventive measures. This may explain why NI rates are now included in
the criteria used for assessing the quality of patient care in many
institutions. Control and prevention include general measures such as
hand hygiene, isolation and restriction of antibiotic use, and more
specific measures that have been demonstrated to be efficient in
reducing particular types of NIs.20
21
22
23
24
25
26
|
Definitions
|
|---|
NI schematically encompasses any infection that is neither present
nor incubating on hospital admission. Precise definitions have been
largely debated in the literature, but those proposed by the CDC in
198827
28
have been validated and are now widely used.
Minor adaptations are generally proposed for specific populations, but
infections are considered to be hospital-acquired if they develop at
least 48 h after hospital admission without proven prior
incubation. If infections occur up to 3 days after hospital discharge
or within 30 days of an operative procedure, they are attributed to the
admitting hospital or ward, or to the surgical procedure, respectively
(Table 1
).24
25
29
30
31
32
A specific terminology is used to describe the epidemiology of NIs. The
prevalence of infected patients is defined as the number of patients
with an active infection divided by the total number of patients who
are present at the time of the survey. The prevalence of infection is
the number of active infections divided by the total number of patients
who are present at the time of the survey. The incidence of infected
patients is defined as the number of patients who developed any
infection divided by the total number of patients at risk who are
hospitalized in the ward concerned during a determined period of time.
Once infected, patients cannot be considered at risk of infection. The
incidence of infection is defined as the number of infectious episodes
divided by the total number of patients who were hospitalized in the
concerned ward during a determined period of time. The
incidence-density of infection/infected patients refers to the number
of infectious episodes/infected patients per 1,000 patient-days at
risk. The latter is the most appropriate way to express infection rates
and to measure the impact of preventive strategies. However, this
approach mandates the prospective surveillance of all patients who are
at risk for NIs with individual records of events considered both in
the numerator and the denominator.33
34
|
Epidemiology of NIs
|
|---|
Epidemiologic data collected from surveillance activities are used
to determine NI rates. Benchmarking then may be used to monitor their
evolution and to detect any unusual variation that may be potentially
suspect of outbreaks or high endemic rates of NI. Importantly, NI rates
vary widely according to the type of ICU and the population served.
They may also vary with the type of surveillance (Table 2
).22
24
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
A prevalence of 20.6% was reported by Vincent et al16
in
the European Prevalence of Infection in Intensive Care study, which
included 10,038 patients from 1,417 European ICUs in 1992. Pneumonia
was the most common NI (46.9%), followed by lower respiratory tract
infection other than pneumonia (17.8%), urinary tract infection (UTI)
(17.6%), and laboratory-confirmed bloodstream infection
(12%).16
Importantly, NIs are easier to compare if they are
presented as incidence densities related to device use (eg,
endotracheal tube, central venous catheter [CVC], or urinary
catheter) [Table 3 ].24
37
39
40
42
44
50
51
52
53
54
55
56
57
58
An incidence of 9.2%,
corresponding to an incidence density of 23.7 episodes per
1,000 patient-days, was reported for the 164,034 patients in 119 ICUs
surveyed from 1986 through 1990 in the NNIS system.59
Data
collected from 112 medical ICUs between 1992 and 1997 indicated that
NIs developed in 7.8% of hospitalized patients (14,177 of 181,993
patients), corresponding to an incidence density of 19.8
episodes per 1,000 patient-days. UTIs (31%) were the most common, with
95% occurring in catheterized patients. Pneumonia, which was
ventilator-associated in 86% of cases, represented 27% of all NIs,
and bloodstream infections represented 19% (laboratory-confirmed,
18.2%, and clinical sepsis, 0.8%), of which 87% were found to be
catheter-related.35
NI device-related rates
(ie, catheter-related UTI, central venous catheter-related
bloodstream infections, and ventilator-associated pneumonia) were 5.5,
4.0, and 7.1, respectively, episodes per 1,000 device-days for coronary
ICUs, 6.4, 5.3, and 6.8, respectively, for medical ICUs, 4.8, 6.9, and
4.0, respectively, for pediatric ICUs, and 4.6, 5.1, and 12.5,
respectively, for surgical
ICUs.48
50
Comparable incidences of NIs have
been reported in ICUs from other developed
countries.17
42
60
61
Moreover, preliminary data from the
NNIS system suggest that risk-adjusted NI rates decreased over time for
these three infections that are continuously monitored in
ICUs.50
|
Impact of NIs
|
|---|
A significant correlation was found between the prevalence rate of
ICU-acquired infection and mortality rate. In the European Prevalence
of Infection in Intensive Care study, laboratory-proven bloodstream
infection (odds ratio [OR], 1.73; 95% confidence interval [CI],
1.25 to 2.41), pneumonia (OR, 1.91; 95% CI, 1.6 to 2.29), and clinical
sepsis (OR, 3.75; 95% CI, 1.71 to 7.18) were independently associated
with an increased mortality rate. Additional independent predictors of
death were an acute physiology and chronic health evaluation (APACHE)
II score > 20 (OR, 15.6; 95% CI, 9.3 to 26), prolonged (
21 days)
ICU stay (OR, 2.52; 95% CI, 1.99 to 3.18), age > 60 years (OR for
age 60 to 69 years, 1.7; 95% CI, 1.07 to 2.71; OR for age 70
years, 2.08; 95% CI, 1.31 to 3.31), the presence of organ failure on
hospital admission (OR, 1.68; 95% CI, 1.45 to 19.5), and cancer as
comorbidity (OR, 1.48; 95% CI, 1.23 to 1.79).16
The analysis of the impact of NIs on health care revealed that they are
responsible for a significant increase in mortality, morbidity, length
of hospital and ICU stay, and resource utilization in almost all of the
groups of patients studied (Table 4
).22
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
This impact is determined by the attributable part of these parameters.
Accordingly, the attributable mortality of NI is defined as the
difference in the death rate of patients and noninfected patients in a
series adjusted for the presence of other confounding factors. Several
epidemiologic methods may be used to determine the mortality, or any
other parameter of the impact of a NI. Direct estimation is a
simple method in which an experienced clinician subjectively estimates
whether the death of a patient is related to the NI or not. This
technique systematically underestimates the attributable part of the
mortality. The appropriateness of the evaluation protocol is another
direct method that is used to estimate the prolongation of the length
of hospital stay. Based on standardized criteria, the patient is
evaluated daily to determine whether the stay in the hospital is
related to the underlying disease and/or to the presence of an NI.
Another method compares two groups of patients, one with a specified NI
and one without a specified NI. Differences are expected to be
attributable to the NI. However, this technique does not take into
consideration potential confounding parameters that may exist between
the two groups of patients. This effect can be attenuated by including
factors potentially related to death in multivariate analysis.
Nonetheless, these adjustments are generally insufficient and the
attributable part is often overestimated. The so-called case-controlled
studies (ie, those called, more appropriately, historical
cohort studies with matching on potential confounders) are considered
to be the best way to determine the impact of NIs. Infected and
noninfected patients are carefully matched for several confounding
factors related to the parameter investigated (eg, age,
severity of underlying disease, associated comorbidities, and time of
exposure to risk factors). Biased evaluations of the impact are minimal
with this methodologic approach, except when case and control patients
are matched too closely using variables that predict or confound the
outcome of interest.65
Crude mortality rates are particularly high in critically ill patients,
but the attributable mortality varies according to the type of
infection. The differences reported between the studies may be related
to some confusion between the associated and the attributable parts
(Table 4)
. In addition, some methodological bias also may play a role.
Insufficient matching criteria (eg, low case/control ratio
or few and irrelevant matching parameters) may overestimate the impact,
but overmatching abolishes differences between case patients and
control subjects. Cost-effectiveness analysis is based on these data,
which imply that the controversies in the recent literature regarding
the attributable mortality of NIs concerns not only epidemiologists
but, also, ICU physicians who have to select and implement preventive
strategies.65
80
|
Risk Factors
|
|---|
Independent risk factors for NIs have been identified in several
studies (Table 5
).16
42
56
64
81
82
Among them, the severity of underlying
illness assessed by scoring systems such as APACHE II/III or simplified
acute physiologic score II are the most widely used. However, these
scores were designed to predict mortality and are less consistent
predictors of NIs.61
83
These general scores also may be
of limited value in the field of sepsis. In a series84
of
88 consecutive patients with septic shock, we found a low predictive
value for APACHE II and simplified acute physiologic II scores. A
prolonged length of stay, mechanical ventilation, and the use of
vascular accesses also were identified. Apart from the overall risk
factors for NIs, more specific risk factors have been delineated from
numerous studies designed to identify those associated with specific
infections.
Understaffing and overcrowding in ICUs have been
reported85
86
87
to increase the risk of human errors,
iatrogenic complications, and even death. They have also played an
important role in several outbreaks and are to be considered as
potential risk factors for the acquisition of NIs.88
89
90
91
Fridkin et al92
reported an outbreak of catheter-related
bloodstream infections that apparently were associated with total
parenteral nutrition in critically ill surgical patients. After
adjustment for confounding parameters (ie, type of
nutrition, mechanical ventilation, and duration of hospitalization),
the patient-to-nurse ratio was found to be a major independent risk
factor. As compared with a patient-to-nurse ratio of 1, the relative
risks (RRs) were 3.95 (95% CI, 1.07 to 14.5), 16.6 (95% CI, 1.15 to
211), and 61.5 (95% CI, 1.23 to 3,074) for ratios of 1.2, 1.5, and 2,
respectively.92
In our sophisticated ICU environments,
many factors contribute to the development of NIs, but complex, careful
investigation may identify precise factors that may be simple to
correct. We highlighted the importance of understaffing and
overcrowding during an outbreak of serious Enterobacter
cloacae infections in a neonatal ICU.93
Molecular
studies demonstrated that eight patients (5.73 episodes per 1,000
patient-days as compared with 0.86 episodes per 1,000 patient-days for
the preceding 21-month period), representing 13.3% of infants who were
hospitalized over a 2-month period, were infected by three epidemic
clones. Cross-transmission was facilitated by understaffing (57% of
required personnel) and overcrowding (166% of theoretical capacity)
with an increased risk of E cloacae carriage during the
outbreak period as compared with the control period (OR, 5.97; 95% CI,
2.2 to 16.4). The use of multiple-dose vials for caffeine and
budesonide inhalation spray therapy was also independently associated
with E cloacae carriage (OR, 16.3; 95% CI, 1.8 to 
). The
outbreak was stopped after a decrease in workload, reinforcement of
single-dose medication, and increased compliance with hand hygiene
before IV line handling, which rose from 25% to 70%.
|
Pathophysiology of NIs
|
|---|
The colonization of the host by potentially pathogenic
microorganisms is a prerequisite for the further development of most
NIs and may occur from exogenous or endogenous sources. As a
consequence of the severity of the underlying diseases with possibly
impaired host defenses, and in the presence of risk factors, critically
ill patients are particularly susceptible to a rapid colonization by
endemic pathogens of the hospital flora.
The endemic transmission of exogenous staphylococci and other potential
pathogens by the hands of health-care workers (HCWs) is
well-documented.91
94
95
96
97
Goldmann et al98
reported the presence of Gram-negative bacilli on the hands of 75% of
neonatal ICU personnel. A report from the National Epidemiology
of Mycoses Survey with surveillance cultures systematically performed
on the hands of HCWs from 13 ICUs showed that 33% of patients (range,
18 to 58%) in adult ICUs and 29% of patients (range, 8 to 62%) in
pediatric ICUs were positive for Candida spp over an 18-month
period.99
Importantly, the hands of HCWs are only
transiently contaminated and, as discussed later, appropriate hand
hygiene measures are sufficient to remove the organisms and to stop the
transmission.
Many NIs are believed to arise from the endogenous flora of the skin,
oropharyngeal, or GI tracts due to treatments such as chemotherapy,
corticosteroid therapy, or antibiotic therapy, and also by the use of
invasive devices such as intravascular or urinary catheters and
nasogastric or endotracheal tubes. This flora also is responsible for
the majority of surgical wound infections.
|
Microbiology
|
|---|
A continuous shift toward more resistant strains of bacteria has
been reported for several decades. Concern has focused on MRSA, VRE,
ESBLs, fluoroquinolone-resistant Pseudomonas aeruginosa, and
fluconazole-resistant Candida spp.100
101
These pathogens
have become the leading causes of NIs, particularly in ICUs where most
were found to have a certain specificity according to the type of
ICU.13
102
103
The predominant pathogens reported in the
ICUs participating in the NNIS and in European countries are
coagulase-negative staphylococci (CoNS), S aureus, P
aeruginosa, entercococci, and Candida spp (Table 6
).16
35
37
60
104
The factors responsible for this evolution are not fully understood,
but antibiotic pressure certainly plays a major role.105
Studies106
107
108
109
110
have repetitively demonstrated that
antibiotic exposure, particularly to cephalosporins, constitutes an
independent risk factor for colonization and infection with both
resistant Gram-positive cocci and Gram-negative bacilli in ICUs. This
selective pressure was recently emphasized by Harbarth et
al111
in their elegant analysis of the impact of
cephalosporin-based prophylaxis in a cohort of 2,641 consecutive
patients who had been referred for heart surgery over a 5-year period.
As compared to short-term prophylaxis, prolonged prophylaxis
(ie, > 48 h) was not associated with a decreased risk of
SSI but was clearly correlated with an increased risk of colonization
with resistant microorganisms.
A further relationship between antibiotic resistance and antibiotic use
in ICUs is strongly suggested for some pathogens by a prospective
survey in 41 hospitals included in phase 2 of the Intensive Care
Antimicrobial Resistance Epidemiology project.103
Average
antimicrobial use, which was expressed as the daily defined dose per
1,000 patient-days, revealed that first-generation and third-generation
cephalosporins and parenteral vancomycin were the most commonly used
agents in the ICUs included in the project. The demographics of these
hospitals were similar to the 221 other institutions participating in
the NNIS system, and susceptibility could be analyzed for 290,045
isolates collected over a 12-month period. The highest resistance rates
occurred among isolates from ICU patients, followed in decreasing order
by those from non-ICU patients and outpatients. These organisms
included the following: methicillin-resistant CoNS (resistance rates,
75%, 60.4%, and 44.5%, respectively); MRSA (resistance rates,
35.2%, 31.9%, and 17.7%, respectively); VRE (resistance rates,
13.0%, 11.8%, and 2.5%, respectively); piperacillin-resistant
P aeruginosa (resistance rates, 12.2%, 8.3%, and 6.0%,
respectively); and ceftazidime-resistant, cefotaxime-resistant, or
ceftriaxone-resistant Enterobacter spp (resistance rates, 25.0%,
22.3%, and 10.1%, respectively). All these stepwise decreases were
statistically significant. In contrast, this was not the case for
penicillin-resistant pnemococci (resistance rates, 9.5%, 10.4%, and
9.8%, respectively) or for fluoroquinolone-resistant P
aeruginosa (resistance rates, 16.4%, 17.6%, and 20.0%). Apart
from fluoroquinolones, which may have a similar exposure in both parts
of the hospital, for each of the antimicrobial groups used at higher
levels in ICUs there was a correspondingly higher rate of resistant
pathogens among isolates from the ICU compared with non-ICU patients.
Several reports112
also have demonstrated the spread of
antibiotic resistance from ICUs to other hospital wards.
S aureus and CoNS
Currently, > 60% of CoNS isolates and nearly 20%
of S aureus isolates from ICUs are resistant not only
to methicillin, but also to several other agents such as
aminoglycosides, tetracyclines, and
quinolones.103
113
114
115
Although not associated with
higher mortality rates, compared with infections due to
methicillin-sensitive S aureus, bacteremia due to MRSA may
be more difficult to treat.30
The proportion of cases in
which MRSA is responsible for NIs in critically ill patients reported
to the NNIS system increased from < 30% in 1989 to up to 40%
in 1997.60
MRSA already accounts for 30% to
> 50% of cases in some European ICUs, particularly in southern
Europe and the Mediterranean area.116
117
Infection
control measures rely on the interruption of cross-transmission by
appropriate hand hygiene measures, isolation precautions, and the
reduction of selective pressure by inappropriate antibiotic
use.113
117
118
119
Vancomycin-intermediate and glycopeptide-intermediate S
aureus have emerged.120
121
122
123
Routine
disk-diffusion for the determination of antibiotic resistance does not
correctly identify these strains, which have to be suspected on an
epidemiologic basis or in patients with staphylococcal infections and a
poor response to despite adequate glycopeptide
therapy.124
The precise mechanism responsible for the
emergence of these strains has not been fully
elucidated.125
The vanA, vanB, and
vanC genes, which are responsible for
glycopeptide-resistance acquisition among enterococci, were not
isolated from these strains, suggesting a different mechanism of
resistance. Epidemiologic data suggest that the increased use of
glycopeptides in hospitalized patients may play a role in this
evolution.120
121
122
Infection control measures rely on the
strict application of all the guidelines recommended for the prevention
and control of MRSA.126
127
VRE
The rate of VRE infection increased from 0.5% in 1989 to 22% in
1997 among ICU patients with NIs reported to the NNIS, and bacteremia
due to enterococci may be particularly difficult to
treat.128
129
Risk factors associated with the acquisition
of gentamicin resistance by enterococci in a general hospital reported
by Axelrod and Talbot130
included length of stay, mean
duration of antibiotic therapy received, and admission to an ICU. GI
colonization with VRE and the use of antimicrobial agents active
against anaerobes were found by Edmond et al131
to be risk
factors for the development of VRE bacteremia. This was recently
confirmed by Donskey et al132
who found that
antianaerobic agents promoted high-density colonization with VRE. In an
accompanying editorial, Wenzel and Edmond133
highlighted
the importance of these findings, which support the concept of
antibiotic pressure (ie, the crude relationship between the
extent of antibiotic use and the selection of resistant strains). VRE
may be found in the stool samples of as many as 47% of asymptomatic
patients after antibiotic administration.134
ESBLs
Outbreaks of NIs caused by multiresistant Enterobacteriaceae have
been reported.135
136
137
Brun-Buisson et al112
described an outbreak caused by Klebsiella pneumoniae that
successively involved three ICUs in the same hospital. The resistance
was plasmid-mediated. In a prospective study on the colonization of
critically ill patients with ESBLs over a six-month period, De Champs
et al138
identified prolonged ICU stay as a significant
risk factor and reported a decrease in the number of colonized patients
after a change in the antibiotic policy.
Other Gram-Negative Pathogens
The proportion of other Gram-negative bacilli, such as P
aeruginosa resistant to third-generation cephalosporins or to
carbapenems, has remained stable at around 15% in most centers. The
NNIS system has reported35
that the incidence of
fluoroquinolone-resistant P aeruginosa has increased from
5% in 1989 to up to 15% in 1997 among ICU patients with NIs.
Ventilator-associated pneumonia due to these microorganisms has already
been reported139
in some European centers to be associated
with worse outcome.
Candida spp
In the United States, the rate of severe fungal infections
increased from 2.0 to 3.8 episodes per 1,000 hospital admissions
between 1980 and 1990 in 115 participating hospitals in the NNIS
system, with Candida spp responsible for 78% of those
episodes.140
During the same period of time, the incidence
of candidemia increased fivefold in medical centers having > 500 beds
and 2.2-fold in those with < 200 beds. Candida was responsible for
7.2% of bloodstream infections (10.2% in ICUs), preceded by
enterococci, S aureus, and CoNS.141
Epidemiologic data from 1992 to 1997 indicate that fungal infections
accounted for 12% of NIs.35
A 20-fold increase in the
rate of candidemia was reported in a single institution where NIs were
prospectively surveyed from 1981 through 1990.142
However,
recent data suggest that this incidence may be stable in some other
institutions.143
144
The emergence of serious infections related to Candida
glabrata and Candida krusei, which are mostly resistant
to triazoles (fluconazole and itraconazole), was
reported145
146
147
148
by bone-marrow transplant centers and
some ICUs, where the proportion of these strains may represent > 50%
of isolates from colonized patients. However, no such evolution has
been reported23
149
150
in other institutions where the
use of triazole prophylaxis was restricted to high-risk patients. The
importance of these findings has to be balanced by the observation that
the reduction of infections related to Candida albicans is
largely superior to the increase of those related to intrinsically
resistant strains of non-albicans Candida spp.151
152
Data
from a surveillance program, which was designated to monitor the
epidemiology of pathogens in 72 medical centers worldwide, indicate
that C albicans remained largely predominant in the late
1990s.153
154
In fact, 97% of strains from European
medical centers were susceptible to fluconazole; 86.5% were highly
susceptible (minimum inhibitory concentration needed to kill 50% of
isolates [MIC50], < 8 µg/mL), 10.6% were
dose-related susceptible (MIC50, between 8 and 32
µg/mL), and 84% were susceptible to itraconazole (60.6% were highly
susceptible [MIC50, < 8µµg/mL]; and
23.5% were dose-related susceptible [MIC50, 8
to 32 µg/mL]). These data confirmed those obtained in US medical
centers where 75% of strains were hospital-acquired, including 44%
from ICU patients.154
|
Surveillance of NIs
|
|---|
The surveillance of NIs was recognized to be a major component of
infection control in the late 1970s. The Study on the Efficacy of
Nosocomial Infection Control18
showed that NI rates
decreased on average 32% in hospitals where surveillance programs were
implemented, compared with an increase of 18% in other institutions
over a 5-year period. The four key elements for successful prevention
were the following: the presence of at least one epidemiologist for
1,000 beds; one specialized trained nurse for 250 beds; the existence
of a planned surveillance system; and restitution of NI rates. Such
programs were rapidly imposed in the United States as important
criteria for hospital accreditation.155
Although less
widespread than in the United States, infection control programs also
were shown to be effective in Europe.156
157
Surveillance includes the following several distinct components:
epidemiologic surveillance and intervention; administrative controls
for medical equipment, for health-care personnel, and for patients; and
engineering controls (Table 7
). These have to be viewed as tools that have to be appropriately
selected to solve specific problems.15
158
Epidemiologic surveillance is defined as the continuous collection,
tabulation, analysis, and dissemination of all information on the
occurrence of NIs in a specified ward and/or hospital.159
Several concepts have been developed, and the major advantages and
disadvantages of specific tools are presented in Table 8
. Total surveillance with the meticulous collection of clinical and
microbiological data for each hospitalized patient is labor-intensive,
time-consuming, and not always feasible on a practical
basis.60
At the other end of the spectrum, the
computerized surveillance of data from the microbiology laboratory
alone gives limited information, which may be pertinent to a
specific problem. Other types of computerized systems may be
extremely helpful and may facilitate the rapid identification and
handling of specific problems. For example, we implemented a fully
computerized automatic alert system to identify at the time of hospital
admission any patient in whom MRSA has been identified previously by
the microbiology laboratory either during a previous hospital stay or
during ambulatory care.29
This automatic alert system is
now used to detect other resistant organisms and carriers.
In practical terms, a combined approach allows for the optimal use of
resources.158
Continuous monitoring of different
infections or microorganisms is mandatory to detect outbreaks that
requires both specific and emergency measures.160
The
surveillance of defined infections in particular wards or units may be
useful for particular epidemiologic profiles and may help to design
targeted programs to reduce the number of
NIs.23
24
118
161
Administrative controls are guidelines
that must be checked and executed by HCWs (Table 7)
. However, some
controls are effective only if appropriate changes are incorporated
into routine activities. We experienced a cluster of invasive pulmonary
aspergillosis in nonimmunocompromised critically ill patients
associated with room air-filter replacement.162
Such fatal
infections could have been prevented by the development and the
application of guidelines for this procedure.
|
Control and Prevention of NIs
|
|---|
Prevention plays a major role in the control of NIs, and consensus
conference and expert panels have established numerous guidelines both
in the United States and in European
countries.100
163
164
165
166
These guidelines concern three main
approaches, which can be schematized as follows. First, methods and
techniques are needed to prevent cross-contamination and to control the
potential sources of pathogens that could be transmitted from patient
to patient or from HCW to patient. These methods and techniques include
appropriate protocols for cleansing, disinfecting, and caring for
various pieces of equipment and devices. Second, guidelines are needed
for the appropriate use of surgical antibiotic prophylaxis or empirical
therapy among selected groups of patients. Third, strategies to limit
the emergence of resistant microorganisms need to be developed. In
addition, specifically targeted measures against various types of NIs
also have been proposed.
Isolation Precautions
More than 50% of patients who are admitted to ICUs already have
been colonized at the time of admission with the microorganism
responsible for subsequent infection; some patients will acquire it
from the environment. The CDC164
has published guidelines
on isolation precautions to minimize the risk of transmission of
infectious agents from colonized/infected patients to other patients or
HCWs. In brief, these guidelines are based on the application of the
concepts of standard precautions (Table 9
). Microorganisms may be transmitted by airborne droplet nuclei, by
large-particle droplets, or by direct contact. Additional specific
precautions are recommended accordingly (Table 10
).
However, despite the fact that the use of guidelines has
become a popular approach to improve the process of care, efforts to
implement them in clinical practice often have been
unsuccessful.167
Most requirements regarding infection
control measures are unpopular and require restrictive procedures for
which compliance is difficult to maintain, and it has been suggested
that noncompliance is connected with the yearning of human beings for
liberty.168
This is the case in the particular field of
the MRSA pandemic, despite the fact that infection control measures
have been proved to be efficacious and cost-effective.169
It has been shown that noncompliance may be related to several aspects
of human behavior, including the false perception of an invisible risk,
the underestimation of individual responsibility in the epidemiology of
the institution, passive attitudes regarding the increasing complexity
of the process of care, and the negative impact of the socioeconomic
constraints that are responsible for understaffing.168
Local factors have to be taken into account to help to incorporate
changes in the behavior of both the patients and the
HCWs.168
170
As discussed in specific sections
below, we have observed a strong positive impact in our institution
after applying these concepts to the hospital-wide promotion of a
bedside hand disinfection technique and to the implementation of an
educational program targeted at vascular access care in the medical
ICU.24
25
Standard Precautions
The key role of HCWs hands in the transmission of pathogens from
patient to patient was demonstrated > 150 years ago by Ignaz
Semmelweis. This obstetrician from Vienna was able to dramatically
reduce the mortality related to puerperal fever by implementing
systematic hand disinfection in chlorinated lime before examining
patients.171
Since then, routine hand washing before and
after patient contact remains the most important infection control
measure.172
173
The endemic transmission of exogenous staphylococci and other potential
pathogens by the hands of HCWs is
well-documented.91
94
95
96
97
This phenomenon is of particular
concern in the ICU where patient care necessitates frequent contact.
Goldmann et al98
reported the presence of Gram-negative
bacilli on the hands of 75% of neonatal ICU personnel. As already
mentioned, data have shown that one third to two thirds of the hands of
HCWs in ICUs were found to be colonized by Candida spp.99
We have demonstrated174
that bacterial contamination of
the hands increases linearly with time on ungloved hands during patient
care (16 colony-forming units [CFU] per minute; 95% CI, 11 to 25
CFU/min). Higher contamination was documented with direct patient
contact such as respiratory care, handling of body fluid secretions,
and interruption in the sequence of patient care (ie, the
HCW left the patient’s bedside to accomplish another task such as
answering a telephone and then returned to resume care). We found that
the method of hand cleansing before care affected the amount of
bacterial contamination; in particular, the absence of hand
disinfection before patient care was associated with an increase of 68
CFU (increase, 16 to 119 CFU), independent of the type of care provided
and the hospital location.174
Updated guidelines for hand washing and/or hand disinfection were
published by the Healthcare Infection Control Practices Advisory
Committee (HICPAC)175
in 1995
(http://www.cdc.gov/ncidod/hip/sterile/sterile.htm). However, low-level
compliance with hand hygiene has been systematically reported,
particularly in ICUs where it does not exceed
40%.118
176
177
178
Several reasons have been suggested for
such a low level of compliance, including the lack of priority over
other required procedures, insufficient time, inconvenient placement of
hand-washing facilities, allergy or intolerance to hand-hygiene
solutions, and lack of leadership from senior medical
staff.177
179
180
181
We have reported174
that
compliance was inversely proportional to the number of opportunities
per hour of patient care. In addition, those HCWs who do wash
frequently and vigorously risk skin damage, which, ironically, results
in the shedding of more organisms into the environment.182
Attempts to improve compliance with hand hygiene have been associated
with some improvement.43
183
Only a few interventions have
been associated with a sustained effect.25
184
185
186
The
main parameters associated with successful improvement have been
extensively discussed elsewhere (http://infection.thelancet.com), and
examples based on published interventions are given herein.
Experience reported187
with alcohol-based handrubs
suggested that hand disinfection reduces hand contamination more than
hand washing. In a study published by Doebbeling et al,43
a hand-disinfection system using an antimicrobial agent (chlorhexidine)
reduced the rate of NIs more effectively than one using alcohol and
soap. This improvement was essentially explained by a better compliance
with hand-hygiene instructions when chlorhexidine was
used.43
We observed that the promotion of hand
disinfection with an alcohol-based hand-rub solution, which was
distributed widely as disposable individual pocket bottles as well as
placed at the patient bedside, may significantly improve the compliance
of ICU staff for whom almost two thirds of their work time
theoretically could be required for optimal adherence to infection
control guidelines on hand hygiene practice.188
This was
also the case in a French medical ICU178
where the
increase in compliance to hand hygiene measures from 42.4 to 60.9% was
essentially attributed to the availability of an alcohol solution for
handrubs. However, the effect of this punctual intervention was not
sustained, and compliance decreased over a 3-month period from 60.9 to
51.3%. At our institution, the promotion of an elementary bedside
hand-disinfection technique by a hospital-wide campaign resulted in a
sustained improvement in compliance with hand hygiene from 48 to 66%
over 4 years. During the same period, the prevalence of overall NIs and
MRSA transmission decreased from 16.9 to 9.9% and from 2.16 to 0.93
episodes per 10,000 patient-days, respectively. Considering the
hypothesis that only 25% of the reduction in the infection rates could
be attributed to the improved compliance in hand hygiene practice, this
intervention might have prevented > 900 NIs and, thus, was largely
cost-effective.25
Behavioral changes may have played a key
role in the success of this intervention, based on a multimodal and
multidisciplinary approach including communication and education tools
such as "Talking Walls" (widely exhibited cartoon posters, which
are available at www.hopisafe.ch), active participation and positive
feedback at both the individual and institutional levels, and the
systematic involvement of institutional
leaders.185
189
190
191
Other requirements for standard precautions are listed in Table 9 .
Gloves should be used for any anticipated contact with blood, mucous
membranes, nonintact skin, secretions, and moist body substances of all
patients.192
However, gloves may have small and/or
inapparent defects or may be torn during use so that hands may become
contaminated.193
194
195
196
Doebbeling et al197
showed not only that washing gloved hands was ineffective for
decontamination but, also, that 5 to 10% of hands were contaminated
after glove removal. This explains why the gloves themselves may be
potentially responsible for the unrecognized cross-transmission of
pathogens if they are not changed between patient contacts and if hands
are not scrupulously washed or disinfected before and after
degloving.198
199
In addition to gloves and gowns, masks
must be used to protect mucous membranes of the eyes, nose, and mouth
during procedures and patient-care activities that are likely to
generate splashes or sprays of blood, body fluid secretions, and
excretions.164
The simultaneous use of goggles or a mask
that includes a transparent eyeshade are strongly recommended for the
respiratory care of patients receiving mechanical ventilation
(eg, mouth care, suction or aspiration in the endotracheal
tube, or aerosol therapy).
Transmission-Based Precautions
In addition to standard precautions, transmission-based
precautions include specific measures according to the mode of
transmission of the microorganisms. Although all theoretical
requirements for an ideal isolation system would be practically
unfeasible, appropriate isolation remains the cornerstone of infection
control measures to prevent the transmission of microorganisms from
and/or to the patients. Recommendations for patient placement,
including isolation in special rooms, are included in the requirements
for transmission-based precautions (Tables 10
and 11
).164
170
200
Source isolation would prevent the
transmission of microorganisms from the patient.
Airborne Precaution:
In addition to standard precautions, airborne precautions prevent the
transmission of microorganisms transmitted by the inhalation of droplet
nuclei or contaminated dust particles. Droplet nuclei are < 5 µm in
size and can remain suspended in the air for long periods and can
travel long distances. This is the case for patients with pulmonary and
laryngeal tuberculosis, varicella and disseminated zoster, acute viral
hemorrhagic fever, or measles, who should be placed in a private room
with negative air pressure in relation to the surrounding area with at
least six air changes per hour and with an appropriate discharge of air
before it is circulated to other areas in the hospital.201
The door of the room should be kept closed. An isolation room with an
anteroom is sometimes used, however, it is unknown whether the anteroom
adds to the effectiveness of the isolation. The main role of the
anteroom is to allow air pressure differentials to be maintained at the
time of door opening. When an isolation room with an anteroom is used,
the two doors should not be opened at the same time. In addition, the
efficacy of such engineering controls applied to the air pressure has
to be monitored. Inappropriate outward airflow was observed in 38% of
140 respiratory isolation rooms in the state of New York from 1992 to
1998. Multiple factors were identified as being associated with the
malfunction of these sophisticated rooms, including an unbalanced
ventilation system, a shared anteroom, a turbulent airflow pattern, and
automated control system inaccuracies. All the factors were
detected by a simple visible smoke test, which should be included in
the list of controls in the charge of infection control
programs.202
Specifications for the ventilation of the
room, such as negative pressure with external extraction of the
contaminated air after adequate filtration for the patients infected or
colonized by airborne-transmitted agents.203
When such
isolation rooms are unavailable, the patient should be placed in a
private room or placed in a cohort with another patient infected by the
same organism. In these situations, however, a consultation with the
infection control team is advised. Airborne precautions require
respiratory protection for any HCWs or visitors with high-efficiency
masks (dust masks) that have been approved by the National Institute
for Occupational Safety and Health (N-95
standard).170
203
This also has to be applied to
the patient during transport and/or movements outside his isolation
room.
Droplet Precaution:
In addition to the standard precautions, droplet precautions prevent
the transmission of microorganisms transmitted by large particles
(ie, those particles > 5 µm in size) containing
infecting microorganisms that are produced during coughing, sneezing,
and talking, or during invasive procedures such as bronchoscopy and
suctioning. They can also be deposited on the mucous membranes of the
host’s eyes, nose, and mouth. This is the case for Haemophilus
influenzae type B, meningococci, multidrug-resistant
pneumococci or any other multidrug-resistant organisms in the
respiratory tract (eg, MRSA, ESBLs, or Gram-negative
bacteria), pharyngeal diphtheria, Mycoplasma pneumoniae,
and some viral diseases (Table 10) . However, a close contact of < 60
cm to 1 m is necessary for transmission to occur since respiratory
droplets do not last very long in the air and usually travel short
distances. In addition to the standard precautions, a mask is
recommended when an HCW is working within 60 cm to 1 m of the
patient. Droplet precautions require the patient to be placed in a
private room or to be placed in a room with another patient infected by
the same organism. Special air handling and ventilation are
unnecessary, and the door may remain open. When these measures are not
possible, a spatial separation of at least 60 cm to 1 m between
the patient and other patients or visitors should be observed.
Contact Precaution:
In addition to standard precautions, contact precautions prevent the
transmission of epidemiologically important microorganisms
(ie, MRSA, ESBLs, Gram-negative bacteria, VRE, or
Clostridium difficile) that can be transmitted by
physical direct or indirect contact with the patient or his direct
environment. The patient is to be placed in a private room or in a room
with another patient infected by the same organism. For any contact
with the patient, HCWs should wear gloves and gowns, which should be
removed before leaving the room, and this should be followed by
systematic hand disinfection measures. Patient-care
devices, including stethoscopes and blood-pressure cuffs, should not be
used for other patients without rigorous cleansing and disinfection.
Protective isolation measures for immunosuppressed patients
such as those who have undergone transplantation or who are
deeply neutropenic, have been published. 201
203
204
In
addition to standard precautions, they include contact precautions as
well as the placement of the patient in a private room with filtrated
air instilled in positive pressure.201
203
204
Private rooms with specific ventilation specifications probably could
improve the efficacy of airborne droplet and contact precautions, but
that kind of specification is particularly difficult to obtain in most
ICUs. In addition, some authors205
206
207
have
pointed out that, apart from the practical difficulties involved in
introducing this isolation measure, additional difficulties also may be
associated with some psychological stress that has also to be taken
into account.205
206
207
However, because aggressive
support for organ failure in a critically ill patient must be
considered as an absolute priority, isolation precautions often are
imposed as secondary management objectives.
Patients who are readmitted to the hospital are at particularly high
risk for carrying and transmitting resistant microorganisms that were
acquired during a prior hospitalization. Those with suspected
infections should be appropriately segregated at the time of hospital
admission. When a private room is not available, patients infected or
colonized by the same microorganism can share a room. This situation,
which is referred to as cohorting, can be safely used provided that the
patients are not infected with other potentially transmissible
pathogens and that the likelihood of reinfection with the same
microorganism is minimal.
Control of Antimicrobial Use
As previously discussed, the use of antimicrobial agents has been
shown to be one of the major determinants in the shift toward resistant
strains.166
Accordingly, most experts in infectious
diseases and infection control now recommend a strict limitation of
antibiotic use.208
209
Several strategies targeted at the
use of antimicrobial agents have been suggested to control the
emergence of resistance. They include the following: an optimal use of
antimicrobial agents; strict control, removal, or restriction of the
agents; use of antimicrobial agents in combination; and cycling of the
available agents.210
Antimicrobial use can be divided into the following three categories:
definite therapy for proven infections; prophylaxis for specific
infections; and empirical therapy for suspicion of infection (with the
latter representing the large majority of cases). Considering the high
mortality and morbidity associated with NIs, most intensivists
systematically apply the concept of early empirical broad-spectrum
antimicrobial coverage for critically ill patients in whom the
development of an NI is suspected.208
The selection of antimicrobial agents to be prescribed to critically
ill patients is crucial. In a surveillance study of 2,000 consecutive
ICU patients, Kollef et al22
evaluated the treatment
administered to 655 patients with either community-acquired infections
or NIs. Inadequate antimicrobial treatment was prescribed in 45% of
patients with NIs that developed following therapy for a
community-acquired infection, in 34% of patients with NIs alone, and
in 17% of patients with community-acquired infections (p < 0.0001).
The mortality rate of patients receiving inadequate therapy (52%) was
significantly higher than that for those receiving adequate treatment
(12%) [adjusted OR, 4.26; 95% CI, 3.52 to 5.15; p < 0.001].
Prior administration of antibiotics (adjusted OR, 3.39; 95% CI, 2.88
to 4.23; p < 0.001), the presence of bloodstream infection (adjusted
OR, 1.88; 95% CI, 1.52 to 3.32; p = 0.003), an increasing APACHE II
score (adjusted OR, 1.04; 95% CI, 1.03 to 1.05; p = 0.002), and
decreasing patient age (adjusted OR, 1.01; 95% CI, 1.01 to 1.02;
p = 0.012) were independently associated with inadequate
antimicrobial prescriptions.22
These data confirmed
previous observations made in both critically ill and neutropenic
cancer patients.211
212
213
214
215
216
217
218
This conflict of interest is responsible for a vicious circle in which
microorganisms could potentially emerge as the true winners and has
stimulated the development of new strategies targeted at a better use
of antimicrobial agents.219
Guidelines for the systematic
evaluation of fever in critically ill patients have been
developed.220
221
They facilitate the early recognition of
NIs, which must be based on a high index of suspicion. Additional
guidelines222
223
224
225
for the administration of empirical
antimicrobial therapy may help in choosing appropriate agents. The
implementation of such general recommendations in both surgical and
medical ICUs has been reported to reduce costs without adversely
affecting patients’ outcomes.36
45
226
Methods for an
optimal coverage of pathogens that may be potentially resistant to
empirical antimicrobial therapy would include the selection of a new
class of antimicrobial agents or the routine administration of combined
agents from different classes. It should be mentioned that the efficacy
of a combination of aminoglycoside with ß-lactam remains
controversial. Based on an in vitro synergetic effect, its
clinical utility was demonstrated only for tuberculosis and HIV
infections. In addition, most new-generation agents already cover a
very broad spectrum. Accordingly, most experts do not systematically
recommend such combinations as initial empirical therapy for any
suspected infections.214
220
227
228
229
230
231
Any empirical treatment has to be reevaluated after 48 to 72 h. By
taking into account the results of the initial cultures and the
clinical evolution, the spectrum can usually be narrowed without
compromising patient outcome. This strategy was recently applied to the
management of ventilator-associated pneumonia by Fagon et
al.26
They compared noninvasive vs invasive diagnostic
techniques as standard management in a series of 413 consecutive
patients suspected of developing such a complication. The invasive
workup consisted of bronchoscopy with direct examination, and empirical
therapy was started if results of testing were positive. Further
treatment was started, adjusted, or discontinued according to the
results of quantitative cultures obtained from protected-brush
specimens or BAL fluid. The invasive approach resulted in the
treatment of 52% of patients (107 of 204 patients) with antibiotics
(44% of patients [90 of 204 patients] did not receive antibiotics),
compared with the noninvasive approach in which 91% of patients (191
of 209 patients) were treated with antibiotics (7% of patients [18 of
209 patients] did not receive antibiotics). In addition, the
former strategy was associated with a significant reduction in the
number of antibiotic-free days at day 7 (2.2 vs 5.0, respectively;
p < 0.001) and at day 28 (7.5 vs 11.5, respectively; p < 0.001).
Furthermore, the mortality rate was markedly reduced at day 14 (26% vs
16%, respectively; p = 0.022). This invasive diagnostic strategy may
become the standard of care for diagnosing ventilator-associated
pneumonia and should be considered as part of an antibiotic control
strategy in the ICU.232
This may also contribute to
limiting the selective pressure of antimicrobial agents on ward
microorganisms.
The inappropriate use of antibiotics, related to either too generous or
too restrictive use, has stimulated the application of computerized
antimicrobial guidelines. Automatic stop orders after 72 h of
empiric
TOP
Abstract
Introduction
