Special Articles

Practice parameters for hemodynamic support of sepsis in adult

patients: 2004 update
Steven M. Hollenberg, MD; Tom S. Ahrens, DNS, RN, CCRN, CS; Djillali Annane, MD, PhD;
Mark E. Astiz, MD, FCCM; Donald B. Chalfin, MD, MS, FCCM, FCCP; Joseph F. Dasta, MSc, FCCM;
Stephen O. Heard, MD, FCCM; Claude Martin, MD, FCCM; Lena M. Napolitano, MD, FCCM;
Gregory M. Susla, PharmD, FCCM; Richard Totaro, MB, BS, FRACP, FJFICM;
Jean-Louis Vincent, MD, PhD, FCCM; Sergio Zanotti-Cavazzoni, MD

Objective: To provide the American College of Critical Care
Medicine with updated guidelines for hemodynamic support of
adult patients with sepsis.
Data Source: Publications relevant to hemodynamic support of
septic patients were obtained from the medical literature, sup-
plemented by the expertise and experience of members of an
international task force convened from the membership of the
Society of Critical Care Medicine.
Study Selection: Both human studies and relevant animal
studies were considered.
Data Synthesis: The experts articles reviewed the literature
and classified the strength of evidence of human studies accord-
ing to study design and scientific value. Recommendations were
drafted and graded levels based on an evidence-based rating
system described in the text. The recommendations were de-
bated, and the task force chairman modified the document until
<10% of the experts disagreed with the recommendations.
Conclusions: An organized approach to the hemodynamic sup-
port of sepsis was formulated. The fundamental principle is that
clinicians using hemodynamic therapies should define specific
goals and end points, titrate therapies to those end points, and
evaluate the results of their interventions on an ongoing basis
by monitoring a combination of variables of global and regional
perfusion. Using this approach, specific recommendations for
fluid resuscitation, vasopressor therapy, and inotropic therapy
of septic in adult patients were promulgated. (Crit Care Med
2004; 32:1928 –1948)
KEY WORDS: sepsis; hemodynamic support; fluid resuscitation;
vasopressor therapy; inotropic therapy

S hock occurs when the circu-
latory system fails to main-
tain adequate cellular perfu-
sion. Shock is a syndrome
that may arise from any of several ini-
tiating causes; as this syndrome
progresses, a common pattern compris-
ing an array of symptoms, signs, and
laboratory abnormalities that result
from hypoperfusion emerges. If shock
is not reversed, irreversible cellular
damage may ensue.
The American College of Critical Care Medicine
(ACCM), which honors individuals for their achieve-
ments and contributions to multidisciplinary critical
care medicine, is the consultative body of the Society
of Critical Care Medicine (SCCM) that possesses rec-
ognized expertise in the practice of critical care. The
College has developed administrative guidelines and
clinical practice parameters for the critical care prac-
titioner. New guidelines and practice parameters are
continually developed, and current ones are system-
atically reviewed and revised.
Copyright © 2004 by the Society of Critical Care
Medicine and Lippincott Williams & Wilkins
DOI: 10.1097/01.CCM.0000139761.05492.D6
Septic shock results when infectious
agents or infection-induced mediators
in the bloodstream produce hemody-
namic decompensation. Septic shock is
primarily a form of distributive shock
and is characterized by ineffective tis-
sue oxygen delivery and extraction as-
sociated with inappropriate peripheral
vasodilation despite preserved or in-
creased cardiac output (1). In septic
shock, a complex interaction between
pathologic vasodilation, relative and ab-
solute hypovolemia, myocardial dys-
function, and altered blood flow distri-
bution occurs due to the inflammatory
response to infection. Even after the
restoration of intravascular volume,
microcirculatory abnormalities may
persist and lead to maldistribution of
cardiac output (2). About half of the
patients who succumb to septic shock
die of multiple organ system failure (1).
Most of the rest have progressive hypo-
tension with low systemic vascular re-
sistance refractory to vasopressor
agents (3). Although myocardial dys-
function is not uncommon, death from
myocardial failure is rare (1).
Cellular dysfunction in sepsis is the
final outcome of a process with multi-
ple stimuli. Prominent mechanisms in-
clude cellular ischemia, disruption of
cellular metabolism by the effects of
inflammatory mediators, and toxic ef-
fects of free radicals (3). Activation of
caspases and induction of heat shock
proteins may lead to apoptotic cell
death. In early shock, compensatory
mechanisms are activated in an attempt
to restore pressure and flow to vital
organs. When these compensatory
mechanisms begin to fail, damage to
cellular membranes, loss of ion gradi-
ents, leakage of lysosomal enzymes,
proteolysis due to activation of cellular
proteases, and reductions in cellular
energy stores occur and may result in
cell death (3). Once enough cells from
vital organs have reached this stage,
shock can become irreversible, and
death can occur despite eradication of
the underlying septic focus.

1928 Crit Care Med 2004 Vol. 32, No. 9

 Therapy of septic shock may be viewed
as having three main components. The
initial priority in managing septic shock
is to maintain a reasonable mean arterial
pressure and cardiac output to keep the
patient alive. Then the nidus of infection
must be identified and eliminated, using
antimicrobial therapy in all cases and
surgical drainage whenever indicated.
Another therapeutic goal is to interrupt
the pathogenic sequence leading to septic
shock. While these latter goals are being
pursued, adequate organ system perfu-
sion and function must be maintained,
guided by cardiovascular monitoring.
The purpose of this practice parameter is
to provide guidelines for hemodynamic
support in sepsis to maintain adequate
organ system and cellular perfusion.
PURPOSE AND STRUCTURE OF
PRACTICE PARAMETERS FOR
HEMODYNAMIC SUPPORT IN
SEPSIS
These practice parameters were devel-
oped by a panel convened by the Ameri-
can College of Critical Care Medicine of
the Society of Critical Care Medicine, and
updated by a similar panel, to assist
health care providers in the management
of hemodynamic support for patients
with sepsis and septic shock. These
guidelines are intended for adult patients
and do not cover all conceivable clinical
scenarios. Nonetheless, they do represent
an attempt to review the state of knowl-
edge concerning hemodynamic therapy
of sepsis and to supplement specific ther-
apeutic recommendations with guide-
lines about how to optimize therapy and
how to evaluate the results of therapeutic
interventions. The information and rec-
ommendations are predicated upon an
expert-based review of the available sci-
entific data, clinical investigations, and
outcomes research. Where such data are
unavailable or limited in scope, consen-
sus was attained by considering published
expert opinion and discussion among a
wide range of experts. The citations of
human studies have been annotated into
levels of scientific support as per Co-
chrane group recommendations (4) as
follows:
Level I: large, randomized trials with
clear-cut results; low risk of false-
positive (␣) error or false-negative (␤)
error
Level II: small, randomized trials with
uncertain results; moderate to high
Crit Care Med 2004 Vol. 32, No. 9
risk of false-positive (␣) error and/or
false-negative (␤) error
Level III: nonrandomized, contempo-
raneous controls
Level IV: nonrandomized, historical
controls and expert opinion
Level V: case series, uncontrolled stud-
ies, and expert opinion
The strength of the recommendations
has been graded as modified from the
guidelines of Evidence-Based Medicine
Working Group as follows: (5)
A: Supported by at least two level I
investigations
B: Supported by only one level I inves-
tigation
C: Supported by level II investigations
only
D: Supported by at least one level III
investigation
E: Supported by level IV or level V
investigations only
Hemodynamic therapy of sepsis has
been considered in each of three catego-
ries: fluid resuscitation, vasopressor ther-
apy, and inotropic therapy. Since the ini-
tial formulation of the guidelines, a
randomized, double-blind, placebo-con-
trolled, multiple-center trial of recombi-
nant human activated protein C has been
completed (6). Although this trial showed
that treatment with recombinant acti-
vated protein C is effective in patients
with septic shock, activated protein C is
not a hemodynamic therapy per se, nor
was hemodynamic instability a requisite
for inclusion in the trial. Thus, consider-
ation of activated protein C and other
therapies not directed at hemodynamic
stabilization is outside the scope of these
practice parameters.
An algorithm outlining an approach to
hemodynamic support of patients with
septic shock based on the recommenda-
tions in these parameters is shown in
Figure 1.
BASIC PRINCIPLES
Septic shock requires early, vigorous
resuscitation. An integrated approach di-
rected at rapidly restoring systemic oxy-
gen delivery and improving tissue oxy-
genation has been demonstrated to
improve survival significantly in septic
shock (7). Although the specific approach
that is used may vary, there are critical
elements that should be incorporated in
any resuscitative effort. Therapy should
be guided by parameters that reflect the
adequacy of tissue and organ perfusion.
Fluid infusion should be vigorous and
titrated to clinical end points of volume
repletion. Systemic oxygen delivery
should be supported by ensuring arterial
oxygen saturation, maintaining adequate
concentrations of hemoglobin, and using
vasoactive agents directed to physiologic
and clinical end points.
Patients with septic shock should be
treated in an intensive care unit. Contin-
uous electrocardiographic monitoring
should be performed for detection of
rhythm disturbances, and pulse oximetry
is useful to detect fluctuations in arterial
oxygenation. Urine output is monitored
continuously as well. Laboratory mea-
surements such as arterial blood gases,
serum electrolytes, complete blood
counts, coagulation variables, and lactate
concentrations should be done early and
repeated as indicated.
In shock states, estimation of blood
pressure using a cuff is commonly inac-
curate, and use of an arterial cannula
provides a more appropriate and repro-
ducible measurement of arterial pressure
(3). These catheters also allow beat-to-
beat analysis so that decisions regarding
therapy can be based on immediate and
reproducible blood pressure information.
Such monitoring facilitates the adminis-
tration of large quantities of fluids and
potent vasopressor and inotropic agents
to critically ill patients (3).
Although patients with shock and
mild hypovolemia may be treated suc-
cessfully with rapid fluid replacement,
right heart catheterization may be useful
to provide a diagnostic hemodynamic as-
sessment in patients with moderate or
severe shock. In addition, because hemo-
dynamics can change rapidly in sepsis,
and because noninvasive evaluation is
frequently incorrect in estimating filling
pressures and cardiac output, pulmonary
artery catheterization is often useful for
monitoring the response to therapy.
Goals and End Points of
Hemodynamic Support in Septic
Patients
Shock represents the failure of the cir-
culatory system to maintain adequate de-
livery of oxygen and other nutrients to
tissues, causing cellular and then organ
dysfunction. Thus the ultimate goals of
hemodynamic therapy in shock are to
1929
Figure 1. Suggested algorithm for hemodynamic support of adult patients with severe sepsis and septic shock. SBP, systolic blood pressure; MAP, mean
arterial pressure; ICU, intensive care unit; BP, blood pressure; HR, heart rate; Hgb, hemoglobin. *Adequate cardiac filling pressures can be assessed by
response of cardiac output (CO) to increases of pulmonary artery occlusion pressure. Maximal benefit is usually achieved at pulmonary artery occlusion
pressure 12–15 mm Hg. Variation in arterial pressure with respiration can also be used to identify patients who would benefit from increased fluid
administration. ⫹Cardiac output can be assessed by echocardiography or by measuring cardiac index and/or mixed-venous oxygen saturation with a
pulmonary artery catheter. ‡Perfusion can be assessed using a combination of clinical and laboratory variables, as described in the text. 㛳A corticotropin
stimulation test is recommended. See text for details.

1930 Crit Care Med 2004 Vol. 32, No. 9

restore effective tissue perfusion and to
normalize cellular metabolism.
In hypovolemic, cardiogenic, and ex-
tracardiac obstructive shock, hypoten-
sion results from a decrease in cardiac
output, with consequent anaerobic tissue
metabolism. Septic shock, the prototypi-
cal form of distributive shock, is different
and more complicated. In septic patients,
tissue hypoperfusion results not only
from decreased perfusion pressure attrib-
utable to hypotension but also from ab-
normal shunting of a normal or increased
cardiac output (3). Cellular alterations
may also occur. Hemodynamic support of
sepsis thus requires consideration of both
global and regional perfusion.
The practical import of the complex-
ity of hemodynamics in sepsis is that
the goals of therapy are much more
difficult to define with certainty than in
other forms of shock in which global
hypoperfusion is the dominant pathol-
ogy. In cardiogenic shock, for example,
the goal of therapy is to increase car-
diac output, although the degree of hy-
poperfusion may vary in different or-
gans. Indexes of regional perfusion
usually correlate well with indexes of
global perfusion, and both can be used
to monitor the effects of therapy. In
sepsis, maldistribution of a normal car-
diac output can impair organ perfusion,
and maldistribution of blood flow
within organs due to perturbation of
resistance vessel tone or microvascular
obstruction can exacerbate organ dys-
function. To add to the complexity, me-
diators of sepsis can perturb cellular
metabolism, leading to inadequate uti-
lization of oxygen and other nutrients
despite adequate perfusion. One would
not expect such abnormalities to be
corrected by hemodynamic therapy.
Despite the complexity of the patho-
physiology of sepsis, an underlying ap-
proach to the hemodynamic support of
sepsis can be formulated, with the under-
standing that the basic principles of the
approach are more important than the
specific recommendations, which will
certainly change as our understanding of
sepsis improves. For example, although
which variables most accurately reflect
the effects of therapy in septic patients
may be uncertain, it should be apparent
that therapeutic efficacy should be as-
sessed by monitoring a combination of
variables. Similarly, although specific end
points may be arguable, the idea that
clinicians should define specific goals and
end points, titrate therapies to those end
Crit Care Med 2004 Vol. 32, No. 9
points, and evaluate the results of their
interventions on an ongoing basis re-
mains a fundamental principle.
An important recent trial supports
early goal-directed therapy in sepsis. A
total of 263 patients with severe sepsis or
septic shock were randomized to receive
either 6 hrs of early goal-directed therapy
or standard therapy in the emergency de-
partment before admission to the inten-
sive care unit (7). The resuscitation strat-
egy involved rapid administration of
intravenous fluids targeted to a central
venous pressure of 8 –12 mm Hg, correc-
tion of anemia to a hematocrit ⱖ30%,
vasopressor agents as necessary to main-
tain mean arterial pressure ⱖ65 mm Hg,
and administration of dobutamine in an
attempt to achieve a central venous oxy-
gen saturation ⱖ70%. Patients assigned
to early goal-directed therapy had a sig-
nificantly higher central venous oxygen
saturation, lower lactate concentration
and base deficit, and significantly lower
Acute Physiology and Chronic Health
Evaluation II scores, indicating less se-
vere organ dysfunction (7). More impor-
tantly, in-hospital mortality rate was sig-
nificantly decreased in the group
assigned to early goal-directed therapy,
from 46.5% to 30.5% (p ⫽ .009) (7). This
study provides strong support for the no-
tion that therapy for sepsis should be
initiated as early as possible and should
be directed toward clearly defined goals.
Indexes of Global Perfusion
Bedside clinical assessment provides a
good indication of global perfusion. Sep-
tic shock is by definition characterized by
hypotension, which generally refers to a
mean arterial pressure below 60 –70 mm
Hg in adults. Mean arterial pressure is
preferable to systolic pressure because of
its closer relationship to the autoregula-
tory limits of organ blood flow (3). In
interpreting any given level of arterial
pressure, however, the chronic level of
pressure must be considered. Hypoten-
sion is usually accompanied by tachycar-
dia.
Indications of decreased perfusion in-
clude oliguria, clouded sensorium, de-
layed capillary refill, and cool skin. Some
caution is necessary in interpreting these
signs in septic patients, however, since
organ dysfunction can occur in the ab-
sence of global hypoperfusion.
In most forms of shock, elevated
blood lactate concentrations reflect an-
aerobic metabolism due to hypoperfu-
sion, but the interpretation of blood
lactate concentrations in septic patients
is not always straightforward. Some
studies in animal models of sepsis have
found normal high-energy phosphate
concentrations (8) but others have not
(9); the differences may relate to the
severity of the septic model, with more
severe sepsis being associated with de-
pletion of adenosine triphosphate de-
spite maintenance of systemic oxygen
delivery and tissue oxygenation. A num-
ber of studies have indicated that in-
creasing either global (10) or regional
(11) oxygen delivery fails to alter ele-
vated lactate concentrations in patients
with sepsis. A number of studies have
suggested that elevated lactate may re-
sult from cellular metabolic alterations
rather than from global hypoperfusion
in sepsis (12, 13). Accelerated glycolysis
with high pyruvate production (14), in-
hibited pyruvate dehydrogenase, and
decreased clearance by the liver may
contribute to elevated lactate concen-
trations. Nonetheless, although lactate
concentrations should not be consid-
ered to represent tissue hypoxia in the
strict sense, the prognostic value of el-
evations of blood lactate has been well
established in septic shock patients
(15–17). The trend of lactate concentra-
tions is a better indicator than a single
value (15, 16). It is also of interest to
note that blood lactate concentrations
are a better prognostic indicator than
oxygen-derived variables (calculated ox-
ygen delivery and consumption) (18).
Mixed venous oxyhemoglobin satu-
ration (SvO2) can be measured in pa-
tients with a right heart catheter in
place, either intermittently by sampling
blood from the pulmonary artery port
or continuously using a fiberoptic
oximeter. SvO2 is dependent on cardiac
output, oxygen demand, hemoglobin,
and oxygen saturation. SvO2 reflects the
balance between oxygen delivery and
consumption and can decrease when
oxygen delivery falls in relation to the
oxygen requirements of the tissues. The
normal SvO2 value is 70 –75% in criti-
cally ill patients, but SvO2 can be ele-
vated in septic patients due to maldis-
tribution of blood flow, and so values
must be interpreted in the context of
the wider hemodynamic picture. None-
theless, If SvO2 remains low despite
achievement of other end points of re-
suscitation, this suggests increased ox-
ygen extraction and therefore poten-
tially incomplete resuscitation. A
1931
recent study showed that monitoring of
central venous oxygen saturation
(ScvO2) can be a valuable guide to early
resuscitation (7).
Indexes of Regional Perfusion
Adequacy of regional perfusion is usu-
ally assessed clinically by evaluating in-
dexes of organ function, such as myocar-
dial ischemia, decreased urine output,
increased blood urea nitrogen and creat-
inine, an abnormal sensorium, increased
serum concentrations of transaminases,
lactic dehydrogenase, and bilirubin, and
prolonged clotting tests (3). Methods of
measuring regional perfusion more di-
rectly have been under investigation,
with a focus on the splanchnic circula-
tion, for several reasons. First, the hepa-
tosplanchnic circulation may be compro-
mised early in acute circulatory failure.
Measurements of oxygen saturation in
the hepatic vein have revealed oxygen de-
saturation in a subset of septic patients,
suggesting that hepatosplanchnic oxygen
supply may be inadequate in some pa-
tients, even when more global variables
appear adequate (19). Second, the gut
(especially the stomach) may be accessi-
ble to monitoring systems. Third, the
countercurrent flow in the gut microcir-
culation increases the risk of mucosal
hypoxia. Finally, the gut may have a
higher critical oxygen delivery threshold
than other organs (20), and gut ischemia
increases intestinal permeability.
Gastric tonometry is a method to as-
sess regional perfusion in the gut that
employs a balloon in the stomach to mea-
sure intramucosal PCO2. Gastric mucosal
PCO2 is influenced directly by systemic
arterial PCO2, however, and so use of gas-
tric-arterial PCO2 difference has been pro-
posed as the primary tonometric variable
of interest, although even this measure is
not a simple measure of gastric mucosal
hypoxia (21). Despite its complexity,
tonometry is a reasonably good predictor
for the ultimate outcome of critically ill
patients (22–26). Its utility to guide ther-
apy in patients with sepsis and septic
shock, however, has not been proven.
More recently, capnography in the sub-
lingual area, a technique that is less in-
vasive and easier to use, has been shown
to yield tissue PCO2 measurements that
correlate with those obtained by gastric
tonometry (27).
1932
FLUID RESUSCITATION IN
SEPSIS
Goals and Monitoring of Fluid
Resuscitation
Septic shock is characterized by de-
creased effective capillary perfusion re-
sulting from both global and distributive
abnormalities of systemic and microcir-
culatory blood flow. An important factor
contributing to the impairment in tissue
perfusion is hypovolemia (13, 28 –30).
The initial phases of experimental and
clinical septic shock present as a low car-
diac output syndrome with low filling
pressures and evolve to a hyperdynamic
state only after volume repletion (13, 28).
Increased blood and plasma volumes are
associated with increased cardiac output
and enhanced survival from septic shock
(31). Failure to appreciate the degree of
underlying hypovolemia may result in a
low cardiac output.
Large fluid deficits exist in patients
with septic shock. Up to 6 –10 L of crys-
talloid solutions or 2 to 4 L of colloid
solutions may be required for initial re-
suscitation in the first 24 hrs (7, 32).
Volume repletion in patients with septic
shock produces significant improvement
in cardiac function and systemic oxygen
delivery, thereby enhancing tissue perfu-
sion and reversing anaerobic metabolism
(33). Despite sepsis-induced myocardial
depression, cardiac index will usually im-
prove by 25– 40% during fluid resuscita-
tion (34). In approximately 50% of septic
patients who initially present with hypo-
tension, fluids alone will reverse hypoten-
sion and restore hemodynamic stability
(35).
In sepsis, increases in interstitial fluid
volume may already exist and venous ca-
pacitance changes play a major role in
contributing to hypovolemia, and so re-
pleting the interstitial space, which may
have a role in hemorrhagic shock, does
not appear to be as important. Intravas-
cular volume can be repleted through the
use of packed red cells, crystalloid solu-
tions, and colloid solutions.
The goal of fluid resuscitation in sep-
tic shock is restoration of tissue perfusion
and normalization of oxidative metabo-
lism. Increasing cardiac output and oxy-
gen delivery is dependent on expansion of
blood and plasma volume. Fluid infusion
is best initiated with predetermined bo-
luses (250 –500 mL every 15 mins) ti-
trated to clinical end points of heart rate,
urine output, and blood pressure. Pa-
tients who do not respond rapidly to ini-
tial fluid boluses or those with poor phys-
iologic reserve should be considered for
invasive hemodynamic monitoring. Fill-
ing pressures should be increased to a
level associated with maximal increases
in cardiac output. In most patients with
septic shock, cardiac output will be opti-
mized at pulmonary artery occlusion
pressures between 12 and 15 mm Hg
(34). Increases above this range usually
do not significantly enhance end-diastolic
volume or stroke volume and increase
the risk for developing pulmonary edema.
If only central venous pressure is avail-
able, levels of 8 –12 mm Hg should be
targeted (7).
In patients requiring mechanical venti-
lation, changes in arterial pressure during
mechanical breaths may also serve as a
useful indicator of underlying hypovolemia
(36 –38). The effects of increased pleural
pressure on ventricular filling are accentu-
ated in preload-deficient states, resulting in
cyclic decreases in systolic arterial pressure
and widening of the arterial pulse pressure.
When these changes are present, they ap-
pear to be predictive of fluid responsiveness
in septic patients with circulatory failure
(37, 38). These measurements require that
the patient have minimal or absent sponta-
neous respiratory efforts, which may neces-
sitate the use of neuromuscular blocking
agents (37, 38).
Far more important than the specific
method of monitoring is the use of that
method in a dynamic fashion. Evaluation
of the response to fluid infusion is much
more useful than one measurement at a
single time point. This is particularly true
in unstable patients, since cardiac and
vascular compliance may change over
time.
Resuscitation should be titrated to end
points of oxygen metabolism and organ
function. Associations have been ob-
served between improved survival and in-
creased levels of central venous oxygen
saturation, systemic oxygen delivery, re-
versal of lactic acidosis, and increases in
gastric intramucosal pH (7, 18, 24, 39).
However, the specific choice of end
points remains controversial.
Fluid Resuscitation Therapies
Crystalloids. The crystalloid solutions
used most commonly for resuscitation
are 0.9% sodium chloride (normal saline)
and lactated Ringer’s solution. The lac-
tate content of Ringer’s solution is rap-
idly metabolized during resuscitation and
Crit Care Med 2004 Vol. 32, No. 9
does not significantly affect the use of
arterial lactate concentration as a marker
of tissue hypoperfusion (40).
The volume of distribution of normal
saline and Ringer’s lactate is the extracel-
lular compartment. Under ideal condi-
tions, approximately 25% of the infused
amount will remain intravascular while
the rest is distributed to the extravascular
space. Clinically, 100 –200 mL of intra-
vascular volume expansion can be ex-
pected after the infusion of 1 L of isotonic
crystalloids (41, 42). Resuscitation from
septic shock frequently requires crystal-
loid volumes ranging from 6 to 10 L
during the initial 24-hr period, which re-
sults in significant hemodilution of
plasma proteins and decreases in colloid
osmotic pressure.
Hypertonic saline solutions have a so-
dium content ranging from 400 to 2400
mOsm/L. Hypertonic solutions have po-
tentially advantageous physiologic effects
including improved cardiac contractility
and precapillary vasodilation (43). The
primary risk when using these fluids is
iatrogenically induced hypertonic states
due to sodium load. Experience with hy-
pertonic solutions in septic shock is lim-
ited.
Colloids. Many different colloidal solu-
tions are available, including plasma pro-
tein fraction, albumin, gelatins, dextrans,
and hydroxyethyl starch. The principal
solutions used in clinical resuscitation
are albumin and hydroxyethyl starch.
Albumin is a naturally occurring
plasma protein that accounts for approx-
imately 80% of the plasma colloid os-
motic pressure in normal subjects. Hu-
man serum albumin is available in the
United States in 5% and 25% solutions;
other concentrations are available in Eu-
rope. The 5% solution, rather than the
25% solution, should be used for initial
resuscitation. After 1 L of 5% albumin,
plasma volume expansion ranges from
500 to 1000 mL (41, 42). Mobilization of
extravascular volume is required for ef-
fective increases in intravascular volume
when using 25% albumin. If fluid is suc-
cessfully mobilized from the interstitial
space, a 100-mL aliquot can produce in-
creases of 400 –500 mL in the intravascu-
lar volume 1 hr after infusion (42). In the
setting of increased vascular permeability
such as septic shock, significantly smaller
amounts of fluid may be mobilized.
The recently completed Saline versus
Albumin Fluid Evaluation (SAFE) trial
randomized 6,997 critically ill patients to
resuscitation with albumin or saline.
Crit Care Med 2004 Vol. 32, No. 9
There was no difference in 28-day mor-
tality rate (20.9% with albumin vs. 21.1%
with saline) (44).
Hydroxyethyl starch is a synthetic col-
loid formed from hydroxyethyl-substi-
tuted branched-chain amylopectin. It is
available in the United States a 6% solu-
tion of normal saline with a colloid os-
motic pressure of approximately 30
mOsm/L (45). One liter of hydroxyethyl
starch solution expands plasma volume
by 700 mL to 1 L with as much as 40% of
maximum volume expansion persisting
for 24 hrs (41)
There have been reports suggesting
that hydroxyethyl starch molecules may
adversely affect renal function by causing
tubular injury (46, 47). In patients with
sepsis, resuscitation with hydroxyethyl
starch solution, as compared with gela-
tin, resulted in significantly higher se-
rum creatinine concentrations without
associated differences in the need for re-
nal replacement (47). Studies in other
groups of patients have not observed dif-
ferences in renal function when hydroxy-
ethyl starch solution was compared with
other fluids (48 –51). Importantly, these
studies were done with a variety of hy-
droxyethyl starch solutions, each with
different physical properties that may
have different effects on renal tubular
cells. Additional investigations are re-
quired to reconcile these divergent obser-
vations.
Hydroxyethyl starch can cause dose-
dependent decreases in factor VIII activity
and prolongation of partial thromboplas-
tin time. Although these changes appear
to be primarily dilutional, there have
been reports of increased bleeding, pri-
marily in patients undergoing cardiac
surgery (52). However, only minor clot-
ting abnormalities and no increased inci-
dence of bleeding have been noted in
patients with hypovolemic and septic
shock (52).
Efficacy
Patients with septic shock can be suc-
cessfully resuscitated with either crystal-
loid or colloids. Increases in cardiac out-
put and systemic oxygen delivery are
proportional to the expansion of intravas-
cular volume achieved. When crystalloids
and colloids are titrated to the same level
of filling pressure, they are equally effec-
tive in restoring tissue perfusion (32).
Resuscitation with crystalloid solutions
will require two to four times more vol-
ume than colloids and may require
slightly longer periods to achieve desired
hemodynamic end points. Colloid solu-
tions are much more expensive than crys-
talloid solutions. Five percent albumin
and 6% hydroxyethyl starch solution are
equivalent with respect to the amount of
fluid required during resuscitation.
Complications
The major complications of fluid re-
suscitation are pulmonary and systemic
edema. These complications are related
to three principal factors: a) increases in
hydrostatic pressures; b) decreases in col-
loid osmotic pressure; and c) increases in
microvascular permeability associated
with septic shock. The controversy con-
cerning crystalloid and colloid resuscita-
tion revolves around the importance of
maintaining plasma colloid osmotic pres-
sure. Large volume crystalloid resuscita-
tion results in significant decreases in
plasma colloid osmotic pressure, whereas
plasma colloid osmotic pressure is main-
tained with colloid infusion (32). In ex-
perimental studies, decreases in plasma
colloid osmotic pressure increase ex-
travascular fluid flux in the lungs and
lower the level of hydrostatic pressure
associated with lung water accumulation
(53, 54). Some, but not all, clinical re-
ports have observed a correlation be-
tween decreases in the colloid osmotic
pressure-pulmonary artery occlusion
pressure gradient and the presence of
pulmonary edema (55–57). Several clini-
cal studies have randomized subjects to
crystalloid or colloid infusion and exam-
ined the development of pulmonary
edema with mixed results, demonstrating
either no differences between solutions
or an increased incidence of pulmonary
edema with crystalloids (32, 58, 59). Ex-
perimental reports in septic models dem-
onstrate no increase in extravascular
lung water when hydrostatic pressures
are maintained at low levels, indicating
that in sepsis the primary determinant of
extravascular fluid flux appears to be mi-
crovascular pressure rather than colloid
osmotic pressure (60). Together, these
data suggest that when lower filling pres-
sures are maintained there is no signifi-
cant difference in the development of pul-
monary edema with crystalloids or
colloids. However, if higher filling pres-
sures are required to optimize cardiac
performance in patients with ventricular
dysfunction, colloids may mitigate
against extravascular fluid flux (32).
1933
 The acute respiratory distress syn-
drome occurs in 30 – 60% of patients with
septic shock. Of concern has been the
possibility that in the setting of increased
microvascular permeability, colloid parti-
cles could migrate into the interstitium
where they would favor fluid retention in
the lung and worsen pulmonary edema. A
number of studies, including a variety of
models of increased microvascular per-
meability, as well as clinical studies in
patients with septic shock and the acute
respiratory distress syndrome, have not
found evidence of increased lung water or
compromised lung function with colloids
(32, 61– 63).
Systemic edema is a frequent compli-
cation of fluid resuscitation. The relative
roles of increased microvascular perme-
ability, increases in hydrostatic pressure,
and decreases in plasma colloid osmotic
pressure in the development of this com-
plication during sepsis are unclear. Tis-
sue edema may reduce tissue oxygen ten-
sions by increasing the distance for
diffusion of oxygen into cells. During ex-
perimental peritonitis, crystalloid ther-
apy was associated with increased endo-
thelial cell swelling and decreased
systemic capillary cross-sectional area
when compared with colloid infusion
(64). In contrast, other studies compar-
ing the impact of large volume crystalloid
infusion on skeletal muscle and intestinal
oxygen metabolism have observed no im-
pairment of oxidative metabolism despite
significant edema formation (60, 65). The
integrity of the gastrointestinal mucosa
as a barrier to bacterial translocation also
does not appear to be affected by de-
creases in colloid osmotic pressure and
the development of tissue edema follow-
ing crystalloid resuscitation. A compari-
son of crystalloid and colloid resuscita-
tion in thermal injury found that the
extent of resuscitation and not the choice
of fluids was the major determinant of
bacterial translocation (66).
Finally, there have been multiple
meta-analyses of the clinical studies com-
paring crystalloids with colloids, which
have examined the effect of resuscitation
with these solutions on mortality rate.
The results have been conflicting, with
some of the reports suggesting differ-
ences in mortality rate favoring crystal-
loids, whereas others have shown no dif-
ferences (67– 69). These differences
reflect the poor quality of many of the
underlying studies, the heterogeneity in
patient populations, and the fact that
1934
none of the clinical studies was ever de-
signed with mortality as an end point.
Transfusion Therapy
The optimal hemoglobin and hemato-
crit for patients with septic shock is un-
certain. This is a major clinical issue
since hemoglobin concentrations usually
range between 8 and 10 g/dL in patients
with septic shock. The decrease in hemo-
globin is related to several factors includ-
ing ineffective erythropoiesis and he-
modilution. Decreases in hemoglobin in
the range of 1–3 g/dL can be expected
during resuscitation of septic shock with
either crystalloids or colloids (32).
In most patients, this degree of ane-
mia is well tolerated because the associ-
ated decrease in blood viscosity decreases
afterload and increases venous return
thereby increasing stroke volume and
cardiac output. The decrease in blood vis-
cosity may also compensate for other
rheologic changes that occur in patients
with septic shock and may enhance mi-
crovascular blood flow. However, several
factors may affect the ability of the pa-
tient to tolerate the decrease in hemato-
crit and should be considered. Cardiac
dysfunction will limit the increase in car-
diac output in response to decreased vis-
cosity and may result in inadequate levels
of systemic oxygen delivery. In markedly
hypermetabolic states, the increase in
cardiac output may not be adequate to
compensate for the decrease in arterial
oxygen content, potentially compromis-
ing systemic oxygen metabolism. The in-
ability to extract oxygen, related either to
anatomical abnormalities such as in cor-
onary artery diseases or physiologic ab-
normalities in sepsis, may result in
greater dependence on oxygen content to
maintain oxidative metabolism (70, 71).
To date, studies examining the effects
of transfusing critically ill patients with
hemoglobin concentrations in the range
of 8 –10 g/dL have not demonstrated any
consistent benefit in tissue perfusion.
The majority of trials have demonstrated
no significant increase in systemic oxy-
gen consumption when the major effect
of transfusion therapy is to increase oxy-
gen content (72–74). Other studies sug-
gest that increasing oxygen content by
transfusion therapy is not as effective in
restoring splanchnic perfusion as it is in
increasing cardiac output (75). Indeed,
the transfusion of aged, more rigid, red
cells has been associated with decreased
gastric intramucosal pH and may accen-
tuate the rheologic abnormalities seen in
sepsis (76). Blood transfusion may also
have immunosuppressive effects (77).
Moreover, a study randomizing critically
ill patients to transfusion thresholds of 7
or 10 g/dL failed to demonstrate any dif-
ferences in clinically significant out-
comes (78).
Accordingly, the optimal hemoglobin
for patients with hemodynamically signif-
icant sepsis has not been defined. Most
patients will tolerate hemoglobin concen-
trations in the range of 8 –10 g/dL. Some
patients, however, may have clinical vari-
ables that suggest a need for increased
oxygen delivery, including excessive
tachycardia, cardiac dysfunction, signifi-
cant underlying cardiac or pulmonary
disease, severe mixed venous oxygen de-
saturation, or failure to clear lactic aci-
dosis. Patients with sepsis and hemody-
namic instability tend to be in the second
category. Such patients were excluded
from the randomized trials of transfusion
thresholds and may benefit from higher
hemoglobin concentrations. Although no
data exist to support transfusion to a pre-
defined threshold, most experts recom-
mend maintenance of hemoglobin con-
centrations in the 8 –10 g/dL range in
patients with sepsis and hemodynamic
instability.
Vasopressor Therapy
Goals and Monitoring of Vasopressor
Therapy. When fluid administration fails
to restore an adequate arterial pressure
and organ perfusion, therapy with vaso-
pressor agents should be initiated (79).
Vasopressor therapy may also be required
transiently to maintain perfusion in the
face of life-threatening hypotension, even
when adequate cardiac filling pressures
have not yet been attained. Potential
agents include dopamine, norepineph-
rine, phenylephrine, epinephrine, and va-
sopressin.
Arterial pressure is the end point of
vasopressor therapy, and the restoration
of adequate pressure is the criterion of
effectiveness. Blood pressure, however,
does not always equate to blood flow, and
the precise level of mean arterial pressure
to aim for is not necessarily the same in
all patients. Animal studies suggest that
below a mean arterial pressure of 60 mm
Hg, autoregulation in the coronary, re-
nal, and central nervous system vascular
beds is compromised. When organ auto-
regulation is lost, organ flow becomes
linearly dependent on pressure (80, 81).
Crit Care Med 2004 Vol. 32, No. 9
Thus, maintenance of a mean arterial
pressure of 60 mm Hg is usually required
to maintain and optimize flow (82– 84).
Loss of autoregulation can occur at dif-
ferent levels in different organs, however,
and thus some patients may require
higher blood pressures to maintain ade-
quate perfusion. In addition, the degree
to which autoregulation is intact in septic
patients is uncertain. It is important to
supplement end points such as blood
pressure with assessment of regional and
global perfusion by a combination of the
methods outlined previously.
Particular attention should be paid to
certain peripheral circulations during va-
sopressor infusion. Vasopressor therapy
to augment renal perfusion pressure has
been shown to increase urine output
and/or creatinine clearance in a number
of open-label clinical series; the targeted
mean blood pressure varied but was as
high as 75 mm Hg (85–95). However,
significant improvements in renal func-
tion with an increase in renal perfusion
pressure have not been demonstrated in
prospective, randomized studies. A recent
study compared vasopressor therapy tar-
geted to 65, 75, and 85 mm Hg in patients
with septic shock and found no signifi-
cant effect on systemic oxygen metabo-
lism, skin microcirculatory blood flow,
urine output, or splanchnic perfusion
(96). Vasopressors should be titrated to
the minimum level required to optimize
urine flow; in some patients this can be
achieved with a mean arterial pressure of
60 or 65 mm Hg.
The gastrointestinal tract, particularly
perfusion of the splanchnic bed and the
integrity of the gut mucosa, occupies a
key position in the pathogenesis of mul-
tiple organ failure in sepsis. The effects of
vasopressor agents on splanchnic circu-
lation may play a role in their selection
for a given patient.
Whether a potent vasopressor also has
positive inotropic effects is of clinical im-
portance in patients with low cardiac out-
put (97). If vasopressor infusion impairs
stroke volume, addition of an inotropic
agent such as dobutamine should be con-
sidered (91).
Individual Vasopressor Agents
Dopamine. Dopamine is the natural
precursor of norepinephrine and epi-
nephrine. Dopamine possesses several
distinct dose-dependent pharmacologic
effects. At doses ⬍5 ␮g·kg⫺1·min⫺1, the
predominant effect of dopamine is to
Crit Care Med 2004 Vol. 32, No. 9
stimulate dopaminergic DA1 and DA2 re-
ceptors to cause vasodilation in the renal,
mesenteric, and coronary beds. Infusion
of low doses of dopamine increases glo-
merular filtration rate, renal blood flow,
and sodium excretion (98, 99). At doses of
5–10 ␮g·kg⫺1·min⫺1, ␤1-adrenergic ef-
fects predominate, increasing cardiac
contractility and heart rate. Dopamine
causes the release of norepinephrine
from nerve terminals, which contributes
to its effects on the heart. At doses ⬎10
␮g·kg ⫺1 ·min ⫺1 , ␣ 1 -adrenergic effects
predominate, leading to arterial vasocon-
striction and an increase in blood pres-
sure. It should be recognized, however,
that there is a great deal of overlap in
these effects, particularly in critically ill
patients.
The hemodynamic effects of dopamine
in patients with septic shock have been
reported in a number of open labeled
trials. Dopamine has been shown to pro-
duce a median increase in mean arterial
pressure of 24% in patients who re-
mained hypotensive after optimal fluid
resuscitation (29, 100 –111). Dopamine
increases mean arterial pressure and car-
diac output, primarily due to an increase
in stroke volume, and to a lesser extent to
an increase in heart rate (29, 100 –111).
The median dose of dopamine required to
restore blood pressure was 15
␮g·kg⫺1·min⫺1. In most studies central
venous, pulmonary artery, and pulmo-
nary occlusion pressures, systemic vascu-
lar resistance index, and pulmonary ar-
tery resistance index were unchanged. In
patients with elevated pulmonary artery
occlusion pressures, dopamine may fur-
ther increase occlusion pressure by in-
creasing venous return. Patients receiv-
ing dopamine infusion rates ⬎20
␮g·kg⫺1·min⫺1 did show increases in
right heart pressures as well as heart rate.
Dopamine has been shown to improve
right ventricular contractility in patients
with underlying right ventricular failure
(112).
Dopamine increases pulmonary shunt
fraction, probably due to the increase in
cardiac output, which can reopen vessels
in poorly ventilated areas of the lung,
(104, 110). PaO2, however, remains rela-
tively constant, which may be due to he-
modynamic improvement and/or an in-
creased mixed venous oxygen saturation
(104, 105, 110).
Dopamine has been shown to increase
oxygen delivery, but its effects on calcu-
lated or measured oxygen consumption
have been mixed (100 –102). Oxygen ex-
traction ratio typically decreases, sug-
gesting no improvement in tissue oxy-
genation (100, 102). This may be due to a
failure to improve microcirculatory flow
in vital organs or lack of a meaningful
tissue oxygen debt in some patients
(102).
The effect of dopamine on splanchnic
perfusion as assessed by gastric tonomet-
ric variables has also been mixed. In-
creases in splanchnic blood flow have
been reported but have not always been
associated with increases in splanchnic
oxygen consumption or effects on gastric
intramucosal pH (100, 103, 113, 114).
One pilot study reported that despite an
increase in both systemic oxygen delivery
and systemic oxygen consumption with
dopamine, gastric intramucosal pH was
reduced (101). The authors speculated
that dopamine might have redistributed
blood flow within the gut, reducing mu-
cosal blood flow and increasing mucosal
oxygen debt. Decreased gastric mucosal
blood flow was reported with dopamine in
another study, but gastric PCO2, gastric-
arterial PCO2 difference, and calculated
intramucosal pH were unchanged (115).
In laboratory animals and healthy vol-
unteers, low doses of dopamine increase
renal blood flow and glomerular filtration
rate and inhibit proximal-tubular resorp-
tion of sodium, which result in natriure-
sis (116). With this physiologic rationale,
low-dose dopamine is commonly admin-
istered to critically ill patients in the be-
lief that it reduces the risk of renal failure
by increasing renal blood flow. This issue
has now been addressed by an adequately
powered randomized clinical trial, which
enrolled 328 critically ill patients with
early renal dysfunction (urine output
⬍0.5 mL·kg⫺1·hr⫺1 over 4 hrs, creatine
⬎150 ␮mol/L or an increase of ⬎80
␮mol/L over 24 hrs) (117). Patients were
randomized to low (“renal”) dose dopa-
mine (2 ␮g·kg⫺1·min⫺1) or placebo, and
the primary end point was peak serum
creatinine. No difference was found in
either the primary outcome (peak serum
creatinine 245 vs. 249 ␮mol/L, p ⫽ .92),
other renal outcomes (increase in creat-
inine, need for renal replacement), urine
output (increased in both groups, per-
haps due to furosemide administration),
time to recovery of normal renal func-
tion, or secondary outcomes (survival to
either intensive care unit or hospital dis-
charge, intensive care unit stay, hospital
stay, arrhythmias) (117). Thus, the avail-
able data do not support administration
1935
of low doses of dopamine solely to main-
tain renal function.
In summary, dopamine appears to be
very effective in increasing mean arterial
pressure in patients who remain hypoten-
sive after optimal volume expansion.
Since mean arterial pressure increases
primarily as a result of increasing cardiac
index, dopamine may be particularly use-
ful in patients who are hypotensive with
compromised cardiac function or cardiac
reserve. The major undesirable effects of
dopamine are tachycardia and arrhyth-
mogenesis, both of which are more
prominent than with other vasopressor
agents. Other side effects include in-
creased pulmonary artery occlusion pres-
sure, increased pulmonary shunt, and the
potential for decreased prolactin release
and consequent immunosuppression
(118).
Norepinephrine. Norepinephrine is a
potent ␣-adrenergic agonist with less
pronounced ␤-adrenergic agonist effects.
Norepinephrine usually causes a clini-
cally significant increase in mean arterial
pressure attributable to its vasoconstric-
tive effects, with little change in heart
rate or cardiac output, leading to in-
creased systemic vascular resistance.
Norepinephrine generally increases car-
diac output by 10 –20% and increases
stroke volume by 10 –15% (85, 86, 89, 91,
93, 119). Clinical studies have reported
either no change (85, 86, 89, 93, 112) or
modest increases (1–3 mm Hg) (92, 94,
101, 103, 106) in pulmonary artery occlu-
sion pressure. Mean pulmonary arterial
pressure is either unchanged (86, 89, 91,
94, 106) or increased slightly (93, 94, 106,
112). The combination of norepinephrine
with dobutamine may be attractive in the
setting of sepsis. In one study, addition of
norepinephrine in patients with septic
shock unresponsive to dobutamine sig-
nificantly improved both mean arterial
pressure and cardiac output (120).
Norepinephrine is more potent than
dopamine and may be more effective at
reversing hypotension in septic shock pa-
tients. In open labeled trials, norepineph-
rine has been shown to increase mean
arterial pressure in patients who re-
mained hypotensive after fluid resuscita-
tion and dopamine (86, 89, 91, 93, 94,
103, 106, 112, 121). Reported doses have
ranged from 0.01 to 3.3 ␮g·kg⫺1·min⫺1
(91, 93). Thus, large doses of the drug
may be required in some patients with
septic shock, possibly due to ␣-receptor
down-regulation in sepsis (122).
In the only randomized trial compar-
1936
ing vasopressor agents, 32 volume-
resuscitated patients with hyperdynamic
sepsis syndrome were prospectively ran-
domized to receive either dopamine or nor-
epinephrine to achieve and maintain nor-
mal hemodynamic and oxygen transport
parameters for ⱖ6 hrs (106). Dopamine
administration (10 –25 ␮g·kg⫺1·min⫺1) re-
sulted in successful treatment in only 31%
of patients whereas norepinephrine admin-
istration (1.5 ⫾ 1.2 ␮g·kg⫺1·min⫺1) was
successful in 93% (p ⬍ .001). Of the 11
patients who did not respond to dopamine,
ten responded when norepinephrine was
added (106).
In patients with hypotension and hy-
povolemia, that is, during hemorrhagic
or hypovolemic shock, the vasoconstric-
tive effects of norepinephrine can have
detrimental effects on renal hemodynam-
ics, with the potential for renal ischemia
(123–125). The situation may differ in
hyperdynamic septic shock (92). Norepi-
nephrine has a greater effect on efferent
than afferent renal arteriolar resistance
and increases the filtration fraction. Sev-
eral studies have shown increases in
urine output, creatinine clearance, and
osmolar clearance in patients with septic
shock treated with norepinephrine alone
or norepinephrine added to dobutamine
(29, 85, 88, 92, 94, 101, 106, 112). These
studies support the hypothesis that in
fluid-resuscitated patients with septic
shock, norepinephrine may optimize re-
nal blood flow and renal vascular resis-
tance (85, 92, 94).
Although early studies in patients with
only mildly elevated serum lactate con-
centrations showed no significant
changes over a relatively short period of
time (1–3 hrs) with norepinephrine, (89,
101, 103, 112) in a later study in which
initial lactate concentrations were ele-
vated (4.8 ⫾ 1.6 mmol/L), a statistically
and clinically significant decrease (2.9 ⫾
0.8 mmol/L) was observed at the end of
the 6-hr study period (106). The results of
these studies suggest that the use of nor-
epinephrine does not worsen and can
even improve tissue oxygenation of pa-
tients with septic shock.
Results of studies of the effects of nor-
epinephrine on splanchnic blood flow in
patients with septic shock have been
mixed. In one study, the effect of norepi-
nephrine on splanchnic blood flow was
unpredictable (103), whereas another
study showed that septic patients who
switched from dobutamine to norepi-
nephrine or from dobutamine and nor-
epinephrine to norepinephrine alone had
a decrease in cardiac output and a parallel
decrease in splanchnic blood flow (100).
In these studies, however, splanchnic ox-
ygen consumption remained unchanged
(100, 103, 126). One pilot study found
that gastric mucosal pHi was significantly
increased during a 3-hr treatment with
norepinephrine whereas it was signifi-
cantly decreased during treatment with
dopamine (101). A more recent study
compared the effects of norepinephrine,
epinephrine, and dopamine in 20 patients
with septic shock (127). In the ten pa-
tients with moderate shock, no differ-
ences in splanchnic blood flow or gastric-
arterial PCO2 difference were observed
(127). In the ten with severe shock, car-
diac index was higher and the effects of
norepinephrine and dopamine were sim-
ilar, but splanchnic blood flow was lower
despite a higher cardiac index with epi-
nephrine than with norepinephrine
(127).
In summary, the clinical experience
with norepinephrine in septic shock pa-
tients strongly suggests that this drug
can successfully increase blood pressure
without causing a deterioration in car-
diac index and organ function (128). Used
in doses of 0.01–3 ␮g·kg⫺1·min⫺1, nor-
epinephrine reliably improves hemody-
namic variables in most patients with
septic shock. The effect of the drug on
oxygen transport variables cannot be de-
termined fully from the available data.
However, other clinical variables of pe-
ripheral perfusion, such as urine flow and
lactate concentration, are significantly
improved in most studies. Unfortunately,
only one report was controlled (106), and
whether using norepinephrine in septic
shock patients affects mortality rate as
compared with dopamine or epinephrine
still requires a prospective clinical trial.
The available data do not support a det-
rimental effect of norepinephrine, how-
ever. In a recent multivariate analysis in-
cluding 97 septic shock patients,
mortality rate was favorably influenced by
the use of norepinephrine as part of the
hemodynamic management; use of high-
dose dopamine, epinephrine, or dobut-
amine had no significant effect (129).
When the use of norepinephrine is con-
templated, it should be used early and not
withheld as a last resort (130).
Phenylephrine. Phenylephrine, a se-
lective ␣-1 adrenergic agonist, has been
used by rapid intravenous administration
to treat supraventricular tachycardia by
causing a reflex vagal stimulation to the
heart resulting from a rapid increase in
Crit Care Med 2004 Vol. 32, No. 9
blood pressure. It is also used intrave-
nously in anesthesia to increase blood
pressure. Its rapid onset, short duration,
and primary vascular effects make it an
attractive agent in the management of
hypotension associated with sepsis. How-
ever, there are concerns about its poten-
tial to reduce cardiac output and lower
heart rate in these patients.
Unfortunately, only a few studies have
evaluated the use of phenylephrine in hy-
perdynamic sepsis. As such, guidelines on
its clinical use are limited. One study in
normotensive hyperdynamic septic pa-
tients showed that short-term adminis-
tration of phenylephrine at a dosage of 70
␮g/min increased mean arterial pressure,
cardiac output, and stroke volume (131).
In a dose-response study, phenylephrine
administered to normotensive hyperdy-
namic septic patients in incremental
doses of 0.5– 8 ␮g·kg⫺1·min⫺1 increased
mean arterial pressure, systemic vascular
resistance, and stroke index, whereas no
change was seen in cardiac index (132).
Heart rate was slightly but significantly
lower, with a decrease ranging from 3 to
9 beats/min. This study found no statisti-
cally significant changes in either oxygen
delivery or consumption overall, but a
clinically significant (⬎15%) increase in
oxygen consumption was seen in eight of
ten patients in at least one dosage.
Only one study has evaluated the ef-
fects of phenylephrine in treating hypo-
tension associated with sepsis (95). In a
small study of 13 patients with hyperdy-
namic septic shock (baseline cardiac in-
dex 3.3 L·min⫺1·m⫺2) receiving either
low-dose dopamine or dobutamine, who
remained hypotensive despite fluid ad-
ministration (mean arterial pressure 57
mm Hg), phenylephrine was begun at 0.5
␮g·kg⫺1·min⫺1 and was titrated to main-
tain a mean arterial pressure ⬎70 mm
Hg. Patients required phenylephrine for
an average of 65 hrs, and the maximum
dosage in each patient averaged 3.7
␮g·kg ⫺1 ·min ⫺1 (range 0.4 –9.1
␮g·kg⫺1·min⫺1). Phenylephrine resulted
in an increase in mean arterial pressure,
systemic vascular resistance, cardiac in-
dex, and stroke index. There was no
change in heart rate. A significant in-
crease in urine output without a change
in serum creatinine was observed during
phenylephrine therapy.
The limited information available with
phenylephrine suggests that this drug
can increase blood pressure modestly in
fluid-resuscitated septic shock patients.
In addition, phenylephrine therapy does
Crit Care Med 2004 Vol. 32, No. 9
not impair cardiac or renal function.
Phenylephrine may be a good choice
when tachyarrhythmias limit therapy
with other vasopressors. An increase in
oxygen consumption and delivery may
occur during therapy.
Epinephrine. In patients unresponsive
to volume expansion or other catechol-
amine infusions, epinephrine can in-
crease mean arterial pressure, primarily
by increasing cardiac index and stroke
volume with more modest increases in
systemic vascular resistance and heart
rate (90, 133–135). The dose-response re-
lationship is more predictable in some
studies (134) than others (90, 133). In
patients with right ventricular failure,
epinephrine increases right ventricular
function by improving contractility
(136). Epinephrine can increase oxygen
delivery, but oxygen consumption may be
increased as well (133–137).
Epinephrine decreases splanchnic
blood flow, with transient increases in
arterial, splanchnic, and hepatic venous
lactate concentrations, decreases in pHi,
and increases in PCO2 gap (100, 121, 138).
These effects may be due to a reduction in
splanchnic oxygen delivery to a level that
impairs nutrient blood flow and results in
a reduction in global tissue oxygenation,
(100, 121) and may potentially be re-
versed by the concomitant administra-
tion of dobutamine (121). Alternatively,
CO2 production secondary to the thermo-
genic effect of epinephrine may play a
role. These studies have been limited by
the concurrent use of other cat-
echolamines. Two more recent studies,
however, found increased gastric muco-
sal perfusion with epinephrine compared
with norepinephrine alone, to an extent
similar to that of norepinephrine in com-
bination with dobutamine (139, 140). In a
recent study of 20 patients with septic
shock, dopamine was replaced by either
norepinephrine or epinephrine. In ten
patients with severe shock (mean arterial
pressure ⬍65 mm Hg despite high-dose
dopamine), epinephrine increased global
oxygen delivery and consumption but
caused a lower absolute and fractional
splanchnic blood flow and lower indocya-
nine green clearance, thus validating the
adverse effects of epinephrine alone on
the splanchnic circulation (127).
Epinephrine administration has been
associated with increases in systemic and
regional lactate concentrations (121, 133,
137). Despite respiratory compensation
and decreased arterial PCO2, the increase
in plasma lactate was associated with de-
creases in arterial pH and base excess
(137). The monitoring periods were
short, and so it is unclear if these in-
creases are transient; in the one longer
study, arterial lactate and pHi returned to
normal values within 24 hrs (121). Other
adverse effects of epinephrine include in-
creases in heart rate, but electrocardio-
graphic changes indicating ischemia or
arrhythmias have not been reported in
septic patients (133, 134). Epinephrine
has had minimal effects on pulmonary
artery pressures and pulmonary vascular
resistance in sepsis (133, 134).
In summary, epinephrine clearly in-
creases blood pressure in patients unre-
sponsive to traditional agents. However,
because of its effects on gastric blood flow
and its propensity to increase lactate con-
centrations, its use should be limited to
patients who fail to respond to traditional
therapies for increasing or maintaining
blood pressure.
Corticosteroids. Corticosteroids exert
important actions on various elements of
the cardiovascular system including the
capillaries, the arterioles, and the myo-
cardium. Topical glucocorticoids con-
strict the dermal vessels, provoking
blanching (141), although the mecha-
nisms of this vasoconstriction remain
poorly understood. Corticosteroids may
up-regulate the sympathetic nervous sys-
tem and the renin-angiotensin system
(142, 143) and also enhance vascular re-
sponses to norepinephrine and angioten-
sin II, possibly through stimulation of the
phosphoinositide signaling system in
smooth muscle cells (144). Glucocorti-
coids also inhibit nitric oxide production
by inducible nitric oxide synthase (145).
Corticosteroids may potentiate catechol-
amine activity by several mechanisms: in-
creasing phenylethanolamine N-methyl-
transferase activity and epinephrine
synthesis (146), inhibiting catecholamine
reuptake in neuromuscular junctions and
decreasing their metabolism (147), in-
creasing binding capacity and affinity of
␤-adrenergic receptors in arterial smooth
muscle cells (148), and potentiating re-
ceptor G coupling and catecholamine-
induced cyclic adenosine monophosphate
synthesis (149). Corticosteroids also in-
crease angiotensin II type I receptor ex-
pression in vascular smooth muscles
(150) and significantly enhance central
pressor effects of exogenous angiotensin
II (151).
Numerous studies in various animal
models, in healthy volunteers challenged
with lipopolysaccharide, and in patients
1937
consistently show that corticosteroids en-
hance vascular responsiveness to vasoac-
tive agents. In a rodent endotoxemia
model, pretreatment with dexametha-
sone prevented endotoxin-induced vascu-
lar hyporesponsiveness to norepineph-
rine (152), probably by inhibiting
extracellular release of lipocortin-1 (153).
In a rodent model of hypotensive and
hypokinetic septic shock, dexamethasone
administration resulted in a complete re-
versal of hypotension, improvement in
aortic blood flow, and reduced plasma
lactate and nitrite/nitrate concentrations
without affecting myocardial ␤-adrener-
gic receptor numbers or myocardial cy-
clic adenosine monophosphate, suggest-
ing an effect on inducible nitric oxide
synthase rather than on adrenergic re-
ceptor sensitivity (154). In healthy volun-
teers, the profound reduction in venous
contractile response to norepinephrine
induced by local instillation of endotoxin
was completely prevented with pretreat-
ment with 100 mg of oral hydrocortisone
(155).
In another experiment, hydrocorti-
sone given before or concurrent with an
intravenous endotoxin challenge in 23
healthy subjects prevented the decreases
in blood pressure and increases in heart
rate and circulating epinephrine concen-
trations (156). In nine patients with sep-
tic shock, the relationship between the
cortisol response to corticotropin (ACTH;
defined by an increment in plasma corti-
sol concentrations of less than 9 ␮g/dL
(250 nmol/L) after an intravenous bolus
of 250 ␮g of ACTH) and pressor response
to norepinephrine was investigated; five
of the nine had such an impaired re-
sponse (157). These five patients had a
profound decrease in pressor response to
incremental dose of norepinephrine (dif-
ference of 7 mm Hg at a norepinephrine
infusion rate of 0.05 ␮g·kg⫺1·min⫺1 and
20 mm Hg at a rate of 1.5 ␮g·kg⫺1·min⫺1,
p ⫽ .028) when compared with the other
four. In addition, the maximal increase in
mean arterial pressure during norepi-
nephrine infusion correlated positively
with the maximal increment in plasma
cortisol concentrations after cortico-
tropin (r ⫽ .783, p ⫽ .013). In the pa-
tients with a cortisol response to ACTH
⬍9 ␮g/dL, a single intravenous bolus of
50 mg of hydrocortisone normalized the
pressor response to norepinephrine. An-
other study investigated the effects of a
single intravenous bolus of 50 mg of hy-
drocortisone on phenylephrine-mean ar-
terial pressure response curves in 12 pa-
1938
tients with septic shock and 12 healthy
volunteers (158). Septic patients had de-
creased maximum responses to phenyl-
ephrine compared with healthy volun-
teers, an effect that was mostly reversed
by hydrocortisone. In this experiment,
the effects of hydrocortisone did not cor-
relate with circulating catecholamines
concentrations, plasma renin activity, or
cyclic guanosine monophosphate con-
centrations.
Prolonged treatment (ⱖ5 days) with
intravenous hydrocortisone at stress
doses (around 300 mg daily) increases
mean systemic arterial pressure and sys-
temic vascular resistance with no signif-
icant change in pulmonary hemodynam-
ics and cardiac index. In three phase II
(159 –161) and one phase III trial (162) of
patients with vasopressor-dependent sep-
tic shock, prolonged treatment with a low
dose of corticosteroids was associated
with a significant reduction in the dura-
tion of shock.
In patients with septic shock, the ef-
fects of corticosteroids on systemic and
pulmonary hemodynamics vary accord-
ing to concomitant vasopressor therapy.
In septic shock not treated by vasocon-
strictors, intravenous administration of
50 mg of hydrocortisone had little effect
on systemic blood pressure (157, 158). In
one randomized, placebo-controlled,
double-blind trial, the hemodynamic ef-
fects of 100 mg of hydrocortisone every 8
hrs for 5 days were investigated in 41
septic shock patients (159). In this study,
mean arterial pressure increased in the
hydrocortisone group (⫹10% at peak ef-
fects) and decreased (⫺7% at peak ef-
fects) in the placebo group, with no effect
on cardiac output. In a second random-
ized, placebo-controlled, double-blind
trial, the hemodynamic effects of a con-
tinuous infusion of hydrocortisone (0.18
mg·kg⫺1·hr⫺1) for 6 days were investi-
gated in 40 hyperdynamic septic shock
patients (160). In this study, as compared
with placebo, hydrocortisone increased
mean arterial pressure from study day 1
(⫹7 mm Hg) to study day 5 (⫹8 mm Hg).
Cardiac output decreased in the hydro-
cortisone group (⫺25%) compared with
the placebo group (⫹6%), and systemic
vascular resistance was increased (by 260
and 369 dynes·sec/cm5·m2 at study day 1
and 5, respectively). In a third random-
ized, placebo-controlled trial, the hemo-
dynamic effects of a continuous infusion
of hydrocortisone (10 mg/hr) were inves-
tigated in 40 septic shock patients (163).
As compared with placebo, hydrocorti-
sone significantly improved mean arterial
pressure (⫹14 mm Hg vs. ⫹ 1 mm Hg at
peak effect), again with a slight decrease
in cardiac output in the hydrocortisone
group (⫺11%) as compared with the pla-
cebo group (⫹9%) and an increase in
systemic vascular resistance (⫹237 vs. ⫹
10 dynes·sec/cm5·m2 at peak effect).
The favorable effect of corticosteroids
on vascular responsiveness to vasopressor
agents is manifested at the bedside by a
shortening of the time on vasoconstrictor
drugs. Five randomized, placebo-con-
trolled, double-blind trials have investi-
gated the effects of prolonged (⬎3 days)
treatment with moderate doses of hydro-
cortisone (200 –300 mg per day) on shock
duration and vasopressor withdrawal in
septic shock. In one study, hydrocorti-
sone significantly reduced the amount of
catecholamine needed to maintain ade-
quate systemic hemodynamics on day
one (⫺40% from baseline vs. ⫹ 43%) and
shortened mean time to cessation of va-
sopressor therapy (4 days vs. 13) (159). At
study day 7, the rate of shock reversal was
68% in the hydrocortisone group and
21% in the placebo group. In a second
study, at study day 7, the rate of shock
reversal was 85% in the hydrocortisone
group and 60% in the placebo group. The
median time to vasopressor therapy with-
drawal was 2 days in the treated group
and 7 days in the placebo group (160). In
a third study, hydrocortisone increased
the rate of shock reversal at study day 3
compared with placebo (70% vs. 33%)
and decreased the median time to cessa-
tion of vasopressor therapy (3 days vs. 5
days) (161). In a fourth study, at study
day 3, the rate of shock reversal was 70%
in the hydrocortisone group and 30% in
the placebo group (163). Finally, in the
fifth study, a phase III randomized place-
bo-controlled, double-blind trial of 300
septic shock patients, hydrocortisone
combined with fludrocortisone signifi-
cantly reduced the time on vasopressors
in nonresponders to ACTH (increment in
plasma cortisol concentrations of ⬍9
␮g/dL [248 nmol/L]) with a median time
to vasopressor therapy withdrawal of 7 vs.
10 days (log rank p ⫽ .009). The rate of
shock reversal at study day 7 was 70% vs.
50% (162). These effects were not seen in
septic shock patients who had a cortisol
increment of ⬎9 ␮g/dL (248 nmol/L) in
response to ACTH.
High doses of corticosteroids (i.e., 30
mg/kg of methylprednisolone or equiva-
lent) once to four times had no effect on
survival in severe sepsis or septic shock
Crit Care Med 2004 Vol. 32, No. 9
(164). By contrast, in septic shock, low
doses ranging from 200 to 300 mg daily
of corticosteroids given for a prolonged
period of time (⬎5 days) have been
shown to improve outcome in several
controlled clinical trials. In 18 critically
ill patients, compared with standard
treatment alone, 100 mg twice daily of
hydrocortisone dramatically improved in-
tensive care unit survival rate (90% vs.
12.5%) (165). In another study of 41 sep-
tic shock patients, as compared with pla-
cebo, a 100-mg bolus of hydrocortisone
every 8 hrs for 5 days improved the 28-
day survival rate (68% vs. 37%) (159).
Similar findings have been reported in
another study in abstract form (161). Fi-
nally, a phase III, multiple-center, place-
bo-controlled, randomized, double-blind
study evaluated the efficacy and safety of
a combination of hydrocortisone (50-mg
intravenous bolus four times per day) and
fludrocortisone (50 ␮g orally once a day)
given for 7 days in 300 patients with
septic shock (162). In this trial, nonre-
sponders to ACTH were more likely to
benefit from cortisol replacement, with
30-day survival rates 47% vs. 37% (log
rank p ⫽ .02), intensive care unit survival
rates 42% vs. 30% (log rank p ⫽ .01), and
hospital survival rates 39% vs. 28% (log
rank p ⫽ .02). Patients who responded
normally to ACTH (cortisol increment of
⬎9 ␮g/dL [250 nmol/L] after 250 ␮g of
corticotropin) had no benefit from corti-
costeroid therapy (1-month mortality
rates: 61% vs. 53%, log rank p ⫽ .81).
Vasopressin. Vasopressin is a peptide
hormone synthesized in the hypothala-
mus and then transported to and stored
in the pituitary gland. Vasopressin is re-
leased in response to decreases in blood
volume, decreased intravascular volume,
and increased plasma osmolality. Vaso-
pressin constricts vascular smooth mus-
cle directly via V1 receptors and also in-
creases responsiveness of the vasculature
to catecholamines (166). Vasopressin
may also increase blood pressure by inhi-
bition of vascular smooth muscle NO pro-
duction (167) and K ⫹ -ATP channels
(168).
Normal concentrations of vasopressin
have little effect on blood pressure in
physiologic conditions (166), but vaso-
pressin helps maintain blood pressure
during hypovolemia, (169) and seems to
restore impaired hemodynamic mecha-
nisms and also inhibit pathologic vascu-
lar responses in shock. Increased concen-
trations of vasopressin have been
documented in several types of shock
Crit Care Med 2004 Vol. 32, No. 9
(170, 171), but a growing body of evi-
dence indicates that this response is ab-
normal or blunted in septic shock. One
study found markedly increased concen-
trations of circulating vasopressin in 12
patients with cardiogenic shock but
much lower concentrations in 19 patients
with septic shock, concentrations that
were hypothesized to be inappropriately
low (172). One potential mechanism for
this relative vasopressin deficiency would
be depletion of pituitary stores, possibly
in conjunction with impaired synthesis.
Depletion of vasopressin stores in the
neurohypophysis evaluated by magnetic
resonance imaging has in fact been de-
scribed in a small group of septic shock
patients (173). A recent prospective co-
hort study of patients with septic shock
found that vasopressin concentrations
were almost always elevated in the initial
hours of septic shock and decreased af-
terward; one third of patients developed
relative vasopressin deficiency as defined
by the investigators (174).
Given this theoretical rationale, sev-
eral small observational studies have ex-
amined the effects of addition of vaso-
pressin to catecholamines in patients
with pressor-refractory septic shock. The
first report showed that with infusion of
low dose of vasopressin (0.04 units/min)
in five patients with septic shock, vaso-
pressin plasma concentrations reached
100 pg/mol (a concentration commensu-
rate with those in patients with normal
stress responses), and blood pressure in-
creased significantly (172). Discontinua-
tion of vasopressin was followed by a
marked decrease in the arterial pressure.
Similar findings were noted by the same
group with low-dose vasopressin admin-
istration in the 19 patients with sepsis
and low vasopressin concentrations in
the study cited previously (172). Other
studies have tested longer infusions. One
report examined the effects of infusion of
0.04 units/min of vasopressin for 16 hrs
in 16 patients with catecholamine-
refractory septic shock and found an in-
crease in mean arterial pressure in 14 –
16, with stable cardiac output, and
increased urine output in the ten patients
who were not anuric on study entry
(175). In another report, in 50 patients
with severe septic shock who had re-
ceived continuous vasopressin infusion
for 48 hrs, mean arterial pressure in-
creased by 18% in the 4 hrs after the
beginning of the infusion and then stabi-
lized at 24 and 48 hrs; catecholamine
doses were reduced by 33% at the 4th
hour (p ⫽ .01) and by 50% at the 48th
hour (166). Of note, five of the six pa-
tients with cardiac arrest during the
study had received vasopressin doses
⬎0.05 units/min (166).
Randomized studies of vasopressin
infusion have been small. In one, ten
patients with hyperdynamic septic
shock on catecholamines were random-
ized to a low dose of vasopressin (0.04
units/min) or placebo (176). The pa-
tients who received vasopressin had a
significant increase in systolic arterial
pressure (from 98 to 125 mm Hg, p ⬍
.05) with successful weaning of cat-
echolamines. No variation in arterial
pressure was noted in the placebo
group. The cardiac index did not differ
in the two groups. Before termination
of the study at 24 hrs, two of the five
patients in the placebo group died of
refractory hypotension; there were no
deaths during the study in vasopressin-
treated patients. Another group ran-
domized 24 patients with septic shock
on high-dose vasopressors to a 4-hr in-
fusion of either norepinephrine or va-
sopressin, with open-label titration of
vasopressors to maintain mean arterial
pressure (177). In the vasopressin
group, norepinephrine doses were sig-
nificantly reduced at the 4th hour (25
to 5 ␮g/min, p ⬍ .001). Vasopressin
doses varied between 0.01 and 0.08
units/min. In the norepinephrine
group, doses were not significantly
modified. Mean arterial pressure and
cardiac index were maintained in both
groups, and the gastric CO2 gradient
was unchanged as well. Urine output
and creatinine clearance increased sig-
nificantly in the vasopressin group but
did not vary in the norepinephrine
group (177).
All of the previously cited studies in-
fused arginine-vasopressin, the vasopressin
that is naturally present in humans. Lysine-
vasopressin or terlipressin, the vasopressin
present in the pig, has been evaluated in
patients with septic shock in one reported
study (178). Terlipressin administered as a
single bolus of 1 mg to eight patients with
septic shock refractory to catecholamines,
hydrocortisone, and methylene blue im-
proved blood pressure during the first 5 hrs
and enabled partial or total weaning of cat-
echolamines (178).
In summary, vasopressin plays an im-
portant role in normalizing blood pres-
sure in states of shock. There is evidence
that in septic shock, a relative deficiency
of vasopressin may contribute to persis-
1939
tent hypotension. Current evidence dem-
onstrates that in catecholamine-resistant
septic shock, the addition of low-dose va-
sopressin (0.01– 0.04 units/min) by con-
tinuous infusion to catecholamines can
be used to increase blood pressure and
decrease catecholamine doses. There is
concern, however, that vasopressin infu-
sion in septic patients may either de-
crease splanchnic perfusion or redistrib-
ute blood flow away from the splanchnic
mucosa (179, 180). Data examining out-
comes and clinical side effects are lim-
ited.
Complications of Vasopressor
Therapy
All of the catecholamine vasopressor
agents can cause significant tachycar-
dia, especially in patients who are inad-
equately volume resuscitated. Tachyar-
rhythmias can occur as well. In patients
with significant coronary atherosclero-
sis, vasopressor-induced coronary ar-
tery constriction may precipitate myo-
cardial ischemia and infarction; this is
of particular concern in patients treated
with vasopressin. In the presence of
myocardial dysfunction, excessive vaso-
constriction can decrease stroke vol-
ume, cardiac output, and oxygen deliv-
ery. Should this occur, the dose of
vasopressor should be lowered, or the
addition of an inotropic agent such as
dobutamine should be considered (91).
Vasopressors can also cause limb isch-
emia and necrosis.
Administration of vasopressors may
impair blood flow to the splanchnic sys-
tem, and this can be manifested by stress
ulceration, ileus, malabsorption, and
even bowel infarction (121, 137). Gut
mucosal integrity occupies a key position
in the pathogenesis of multiple organ
failure, and countercurrent flow in
splanchnic microcirculation gives the gut
a higher critical threshold for oxygen de-
livery than other organs. If possible, epi-
sodes of intramucosal acidosis, which
might be detected either by a decrease in
gastric mucosal pHi or an increase in
gastric mucosal PCO2, should be avoided,
although no prospective randomized con-
trolled trial has demonstrated a decrease
in mortality rate with pHi or gastric PCO2-
directed care in the management of pa-
tients with septic shock.
1940
INOTROPIC THERAPY IN
SEPSIS
Overview
Sepsis is characterized by a hyperdy-
namic state with normal to low blood
pressure, normal to high cardiac index,
and a low systemic vascular resistance (1,
29). Although cardiac output is usually
maintained in the volume-resuscitated
septic patient, a number of investigations
have demonstrated that cardiac function
is impaired (97, 181, 182). This myocar-
dial dysfunction is characterized by a de-
creased ejection fraction, ventricular di-
lation, impaired contractile response to
volume loading, and a low peak systolic
pressure/end-systolic volume ratio (a rel-
atively load-independent measure of ven-
tricular function) (183–185). The mech-
anism of this cardiac dysfunction is
complex. Myocardial ischemia is unlikely,
as coronary blood flow is normal and
there is no net lactate production across
the coronary vascular bed (186, 187). An-
imal studies of endotoxemia or bacterial
infection have suggested that myocardial
edema (188), alterations in sarcolemmal
or intracellular calcium homeostasis
(189), and uncoupling or disruption of
␤-adrenergic signal transduction may
contribute to the cardiac contractile dys-
function (190). A variety of inflammatory
mediators, including prostanoids (191),
platelet-activating factor (65), tumor ne-
crosis factor-␣, interleukin-1 and inter-
leukin-2, (192), and nitric oxide (193,
194) have been shown to cause myocar-
dial depression in a number of animal
models, possibly through the sphinomy-
elinase pathway (195). Although it is
clear that myocardial performance is al-
tered during sepsis and septic shock, end
points for cardiac resuscitation are un-
certain.
Inotropic therapy in septic shock is
complex, because different approaches
endeavor to achieve different goals. In
patients with decreased cardiac output,
the goals of therapy are straightforward
and are aimed at restoring normal phys-
iology. Because of the complexity of as-
sessment of clinical variables in septic
patients, measurement of cardiac output
is advisable. Such measurements need to
be interpreted in the clinical context; a
patient with preexisting cardiac disease
may have a limited ability to increase
cardiac output, and thus values within
the normal range or even slightly below
normal may actually represent a hyper-
dynamic response for that particular pa-
tient. Thus, other end points of global
perfusion should be followed as well.
When global hypoperfusion is manifested
by decreased mixed venous oxygen satu-
ration, this measure may be followed as
an index of the efficacy of inotropic ther-
apy. Similarly, although lactate produc-
tion in sepsis is complex, a decrease in
blood lactate concentrations concomitant
with increased cardiac output is a good
prognostic sign. To further complicate
matters, the pharmacokinetics and phar-
macodynamics of inotropic agents in sep-
tic patients can be quite complex and
variable (196, 197).
Some critically ill septic patients are
hypermetabolic and may require high
levels of oxygen delivery to maintain ox-
idative metabolism. Data from the 1980s
and early 1990s suggested that a linear
relationship between oxygen delivery and
oxygen consumption (“pathologic supply
dependency”) was common in septic pa-
tients, (33, 198) with the inference that
oxygen delivery was insufficient to meet
the metabolic needs of the patient. These
observations led to the hypothesis that
resuscitation to predetermined elevated
end points of cardiac index and oxygen
delivery and consumption (“hyperresus-
citation”) might improve patient out-
come. Retrospective analyses showed that
achievement of cardiac index ⬎4.5
L·min ⫺1 ·m ⫺2 , oxygen delivery ⬎600
mL·min⫺1·m⫺2, and oxygen consump-
tion ⬎170 mL·min⫺1·m⫺2 correlated
with improved survival (39). Other inves-
tigations, however, have challenged the
concept of pathologic supply-dependency
and hyperresuscitation (199 –202). Al-
though cardiac index and oxygen delivery
are correlated with outcome (39), it is
unclear if increases in these variables are
the cause of increased survival or repre-
sent the underlying physiologic reserve of
the patient. Randomized studies to test
the practice of routinely increasing oxy-
gen delivery to these predefined levels in
all critically ill patients have produced
conflicting results (199 –201, 203, 204),
and it is unclear if increases in cardiac
index and oxygen delivery are the cause of
increased survival or represent underly-
ing physiologic reserve of the patient.
Thus, a strategy of routinely increasing
oxygen delivery to predetermined ele-
vated end points of cardiac index and
oxygen delivery cannot be recommended
on the basis of current data (205). None-
theless, some clinicians believe that this
issue has not been settled definitively in
Crit Care Med 2004 Vol. 32, No. 9
Table 1. Summary of cardiac effects of inotropes used in sepsis and septic shock: Physiologic values are reported as percent change from baseline
Dose Range, Heart
Drug ␮g 䡠 kg⫺1 䡠 m⫺1 Rate
Isoproterenol 1.5 to 18 ␮g/min 11 to 20
Dopamine 2 to 55 1 to 23
Epinephrine 0.06 to 0.47 ⫺6 to 27
Norepinephrine 0.03 to 3.3 ⫺6 to 8
Dobutamine 2 to 28 9 to 23
Milrinonea 0.5 1 41 to 49
SVRI, systemic vascular resistance index; LVSWI, left ventricular stroke work index.
a
With other inotropes including dopamine, dobutamine, norepinephrine and/or epinephrine.
those patients with septic shock and ar-
gue that a subset of these patients may
benefit from therapy aimed at supranor-
mal oxygen delivery.
Uncertainty exists in regard to other
end points for inotropic therapy. Deficits
in oxygen delivery can clearly cause a
lactic acidosis, but the converse is not
true: elevated lactate concentrations in
patients with sepsis or septic shock do
not necessarily reflect deficits in oxygen
delivery (206). In adequately resuscitated
septic patients, mixed venous oxygen sat-
uration is usually normal or high, this
value correlates poorly with cardiac out-
put, and several studies have questioned
the value of SvO2 as the end point for
inotropic therapy in critically ill patients
(207, 208). Low mixed venous oxygen sat-
uration may indicate decreased global ox-
ygen delivery, however (207).
Despite seemingly adequate resuscita-
tion, some septic shock patients develop
multiple organ failure, resulting in death.
It has been argued that even after hypo-
tension has been corrected and global
oxygen delivery is adequate in patients
with septic shock, blood flow and tissue
perfusion can remain suboptimal. Gastric
tonometry and sublingual capnometry
monitor gastric and sublingual PCO2 as a
proxy for determining the adequacy of
gut perfusion. Although these monitors
serve as good predictors for the ultimate
outcome of critically ill patients (22–25),
their utility to guide therapy in patients
with sepsis and septic shock has not been
proven. In dysoxic states with normal or
elevated blood flow, these monitors gen-
erally fail to detect an elevated intramu-
cosal-arterial PCO2 gap (209). No current
evidence supports improved outcome
with empirical therapy to raise cardiac
output in patients with normal blood
pressure, but a subpopulation of patients
might have regional hypoperfusion that
would respond to additional therapy. One
Crit Care Med 2004 Vol. 32, No. 9
Cardiac Stroke
Index Volume Index SVRI
47 to 119 22 to 89 ⫺24 to ⫺44
4 to 44 7 to 32 ⫺6 to 18
24 to 54 12 ⫺7 to 34
⫺3 to 21 5 to 15 13 to 111
12 to 61 15 ⫺6 to ⫺21
47 ⫺30 to ⫺35
would want to titrate such therapy to an
index of regional perfusion, although the
precise end points are uncertain. In this
context, it is important to realize that
different interventions to increase oxygen
delivery, such as fluid resuscitation,
blood transfusion, or infusion of vasoac-
tive agents, can have different effects on
regional perfusion (101, 112, 121). Differ-
ent vasoactive agents have been shown to
have divergent effects on gastric intramu-
cosal pH.
An inotropic agent should be consid-
ered to maintain an adequate cardiac in-
dex, mean arterial pressure, SvO2, and
urine output. Cardiac output can be mea-
sured using a pulmonary artery catheter,
by echocardiography, with an esophageal
Doppler probe, or by pulse contour anal-
ysis. Regardless of the measurement
method chosen, clinicians should define
specific goals and desired end points of
inotropic therapy in septic patients and
titrate therapy to those end points. These
goals and end points should be refined at
frequent intervals as patients’ clinical sta-
tus changes.
Therapies and Efficacy
Most investigations evaluating inotro-
pic agents have been observational and
have used the baseline hemodynamic
characteristics of the patient as the con-
trols. The majority of these studies have
used heart rate, cardiac index or output,
and/or stroke volume or stroke volume
index as the outcome variables. Only a
few studies have assessed ventricular
function by reporting left (or right) ven-
tricular stroke work index. The results
are summarized in Table 1.
Individual Inotropic Agents
Isoproterenol. Isoproterenol is a ␤1-
and ␤2-adrenergic agonist. Few studies
LVSWI References
74 to 157 (29, 230)
5 to 91 (102, 104, 105, 107, 109, 110,
112–114, 138, 211–214)
32 to 95 (135, 136, 138)
42 to 142 (91, 94, 102, 104, 107, 113, 120)
23 to 58 (16, 92, 203, 215–218)
51 to 56 (228, 229)
have evaluated isoproterenol in sepsis
and septic shock. In septic shock patients
with a low cardiac index (mean ⬍2.0
L·min⫺1·m⫺2), isoproterenol (2– 8 ␮g/
min) will significantly increase cardiac
index without decreasing blood pressure
but at the expense of increasing heart
rate (29, 210). In patients with a normal
cardiac index, however, isoproterenol can
decrease blood pressure through its ␤2-
adrenergic effects. In addition, the chro-
notropic effects of ␤1-adrenergic stimula-
tion can precipitate myocardial ischemia.
Dopamine. Dopamine is an adrenergic
agonist with predominant dopaminergic
properties at doses ⬍5 ␮g·kg⫺1·min⫺1
and increased ␤ and ␣ activity at doses
⬎5 ␮g·kg⫺1·min⫺1. However, even at low
doses, significant ␣ and ␤ agonism may
occur since the pharmacokinetics of do-
pamine in critically ill patients is highly
variable (197).
In patients with severe sepsis and/or
septic shock, most studies have shown
that dopamine will increase cardiac index
with a range from 4% to 44%, left ven-
tricular stroke work index by 5–91%, and
right ventricular stroke work index by a
modest 5–10% (101, 103, 104, 106, 108,
109, 111–113, 137, 210 –213). These im-
provements in cardiac performance come
at the expense of an increase in the heart
rate of approximately 15% (range up to
23%). The greatest increase in these vari-
ables occurs at doses ranging from 3 to
12 ␮g·kg⫺1·min⫺1. At higher doses, the
rate of improvement in cardiac function
decreases. Although dopamine may in-
crease mesenteric blood flow, it may also
decrease mesenteric oxygen consumption
(113). It is not clear whether effects of
dopamine are superior to any other ad-
renergic agent.
Dobutamine. Dobutamine is a race-
mic mixture of two isomers, a D isomer
with ␤1- and ␤2-adrenergic effects, and an
L isomer with ␤1- and ␣1-adrenergic ef-
1941
fects; its predominant effect is inotropic
via stimulation of ␤1 receptors, with a
variable effect on blood pressure.
A number of studies have investigated
the effect of dobutamine on cardiac func-
tion during sepsis or septic shock at doses
ranging from 2 to 28 ␮g·kg⫺1·min⫺1 (91,
202, 214 –218). In these studies, increases
in cardiac index ranged from 12% to
61%. However, heart rate increases, often
significantly (9 –23%). Two studies re-
ported that left ventricular stroke work
index increased by 23–58% at mean do-
butamine doses of 5–12 ␮g·kg⫺1·min⫺1
(214, 215). Similar increases in right ven-
tricular stroke work were also observed in
these studies.
Although dobutamine does not influ-
ence the distribution of blood flow, ther-
apy is often aimed at increasing blood
flow to organs such as the gut or the
kidneys.
Epinephrine. Epinephrine stimulates
both ␣ and ␤ receptors. At low doses, the
␤-adrenergic effects predominate. A few
recent studies have examined the hemo-
dynamic effects of epinephrine in septic
shock at doses ranging from 0.1 to 0.5
␮g·kg⫺1·min⫺1 (134, 135, 137). The in-
crease in cardiac index varied from 24%
to 54%, and the heart rate response was
variable. Increases in left ventricular
stroke work index as high as 95% have
been noted (135). Other studies indicated
that lactic acidosis is increased and per-
fusion to the gut is altered with the use of
epinephrine (121, 127, 137).
Norepinephrine. Like epinephrine,
norepinephrine stimulates both ␣ and ␤
receptors; however, the ␣-adrenergic re-
sponse is the predominant effect. The ef-
fect of norepinephrine on cardiac index is
modest, with the majority of studies
showing no change or increases of up to
21% while heart rate is unaffected or
even decreases by up to 8% (91, 93, 101,
103, 106, 112, 119). However, several
studies have shown a marked increase in
left and right ventricular stroke work in-
dex due to increased blood pressure (93,
101, 112, 119).
Combination and Comparative Stud-
ies. A number of studies have investigated
catecholamine combinations (86, 94,
121, 126, 219 –222). The majority of these
studies did not study the catecholamine
combination in a standardized fashion,
thus limiting the conclusions that can be
drawn about the effects of these catechol-
amine combinations on cardiac function.
Patients who do not respond to dopamine
with an increase in cardiac index may
1942
reach the desired end point with a dopa-
mine/norepinephrine combination (106).
Dobutamine and norepinephrine appear
to be an effective combination to improve
cardiac index and blood pressure (91) In
addition, some (223) but not all (224,
225) studies have shown that dobutamine
or a dobutamine/norepinephrine combi-
nation will also enhance mesenteric per-
fusion.
A few investigations have been per-
formed comparing different inotropic
regimens (87–90, 101, 112, 121, 137).
Epinephrine appears to be as good if not
better at improving cardiac performance
than dopamine or a dobutamine/norepi-
nephrine combination (121, 137). How-
ever, with one exception (140), studies
have shown that when epinephrine is
compared with other adrenergic agents
or their combinations, there are in-
creases in arterial lactate and decreases in
gastric intramucosal pH, suggesting that
perfusion to regional vascular beds may
be impaired (121, 127, 137, 139, 226). In
several studies, dopamine increased car-
diac index and stroke volume index to a
greater extent than norepinephrine, but
increases in left and right ventricular
stroke volume index were about the same
with the two agents (101, 112). There was
less prominent tachycardia with norepi-
nephrine, and one unconfirmed pilot
study suggested that mesenteric perfu-
sion is impaired with dopamine com-
pared with norepinephrine (101, 112).
Phosphodiesterase Inhibitors. Phos-
phodiesterase inhibitors are vasodilators
with long half-lives, raising the potential
for prolonged decreases in blood pressure
when used in septic patients. There are a
few small studies of these agents in pa-
tients with sepsis, but meaningful con-
clusions cannot be made because of the
size of the studies and the concomitant
use of disparate adrenergic agents (227–
230).
Complications
In the septic patient who has been
inadequately volume resuscitated, all of
the inotropic agents can cause significant
tachycardia and other cardiac arrhyth-
mias (77). In patients with coexisting cor-
onary disease, the change in myocardial
oxygen consumption may precipitate
myocardial ischemia and infarction
(199). Excessive doses of catecholamines
can also result in myocardial band necro-
sis independent of the presence of coro-
nary disease.
Sole use of inotropic agents that also
have vasodilatory activity (e.g., isoproter-
enol, milrinone) is likely to reduce blood
pressure. These reductions can be long-
lasting with agents that have long half-
lives.
Administration of inotropic agents
that have pressor activity may impair
blood flow to other organ beds, such as
the splanchnic circulation (121, 137). Ef-
forts to ensure adequate volume resusci-
tation and to assess end-organ function
must be made.
RECOMMENDATIONS FOR
HEMODYNAMIC SUPPORT OF
SEPTIC PATIENTS
Basic Principles
1. Resuscitation of patients with sepsis
should be initiated expeditiously
and pursued vigorously. Measures
to improve tissue and organ perfu-
sion are most effective when applied
early.
2. Patients with septic shock should
be treated in an intensive care unit,
with continuous electrocardio-
graphic monitoring and monitoring
of arterial oxygenation.
3. Arterial cannulation should be per-
formed in patients with shock to
provide a more accurate measure-
ment of intra-arterial pressure and
to allow beat-to-beat analysis so
that decisions regarding therapy
can be based on immediate and re-
producible blood pressure informa-
tion.
4. Resuscitation should be titrated to
clinical end points of arterial pres-
sure, heart rate, urine output, skin
perfusion, and mental status, and
indexes of tissue perfusion such as
blood lactate concentrations and
mixed venous oxygen saturation.
5. Assessment of cardiac filling pres-
sures may require central venous or
pulmonary artery catheterization.
Pulmonary artery catheterization
also allows for assessment of pul-
monary artery pressures, cardiac
output measurement, and measure-
ment of mixed venous oxygen satu-
ration. Echocardiography may also
be useful to assess ventricular vol-
umes and cardiac performance.
Crit Care Med 2004 Vol. 32, No. 9

T
he fundamental
principle is that
clinicians using
hemodynamic therapies
should define specific goals
and end points, titrate ther-
apies to those end points,
and evaluate the results of
their interventions on an on-
going basis by monitoring a
combination of variables of
global and regional
perfusion.
Fluid Resuscitation
Recommendation 1—Level B. Fluid
infusion should be the initial step in he-
modynamic support of patients with sep-
tic shock. Initial fluid resuscitation
should be titrated to clinical end points.
Recommendation 2—Level B. Iso-
tonic crystalloids or iso-oncotic colloids
are equally effective when titrated to the
same hemodynamic end points.
Recommendation 3—Level D. Inva-
sive hemodynamic monitoring should be
considered in those patients not respond-
ing promptly to initial resuscitative ef-
forts. Pulmonary edema may occur as a
complication of fluid resuscitation and
necessitates monitoring of arterial oxy-
genation. Fluid infusion should be ti-
trated to a level of filling pressure asso-
ciated with the greatest increase in
cardiac output and stroke volume. For
most patients, this will be a pulmonary
artery occlusion pressure in the range of
12–15 mm Hg. An increase in the varia-
tion of arterial pressure with respiration
may also be used to identify patients
likely to respond to additional fluid ad-
ministration.
Recommendation 4 —Level C. Hemo-
globin concentrations should be main-
tained between 8 and 10 gm/dL. In pa-
tients with low cardiac output, mixed
venous oxygen desaturation, lactic acido-
sis, widened gastric-arterial PCO2 gradi-
ents, or significant cardiac or pulmonary
Crit Care Med 2004 Vol. 32, No. 9
disease, transfusion to a higher concen-
tration of hemoglobin may be desirable.
Vasopressor Therapy
Recommendation 1—Level C. Dopa-
mine and norepinephrine are both effec-
tive for increasing arterial blood pres-
sure. It is imperative to ensure that
patients are adequately fluid resuscitated.
Dopamine raises cardiac output more
than norepinephrine, but its use may be
limited by tachycardia. Norepinephrine
may be a more effective vasopressor in
some patients.
Recommendation 2—Level D. Phenyl-
ephrine is an alternative to increase blood
pressure, especially in the setting of
tachyarrhythmias. Epinephrine can be
considered for refractory hypotension, al-
though adverse effects are common, and
epinephrine may potentially decrease
mesenteric perfusion.
Recommendation 3—Level B. Admin-
istration of low doses of dopamine to
maintain renal function is not recom-
mended.
Recommendation 4 —Level C. Pa-
tients with hypotension refractory to cat-
echolamine vasopressors may benefit
from addition of replacement dose ste-
roids.
Recommendation 5—Level D. Low
doses of vasopressin given after 24 hrs as
hormone replacement may be effective in
raising blood pressure in patients refrac-
tory to other vasopressors, although no
conclusive data are yet available regard-
ing outcome.
Inotropic Therapy
Recommendation 1—Level C. Dobut-
amine is the first choice for patients with
low cardiac index and/or low mixed ve-
nous oxygen saturation and an adequate
mean arterial pressure following fluid re-
suscitation. Dobutamine may cause hy-
potension and/or tachycardia in some pa-
tients, especially those with decreased
filling pressures.
Recommendation 2—Level B. In pa-
tients with evidence of tissue hypoperfu-
sion, addition of dobutamine may be
helpful to increase cardiac output and
improve organ perfusion. A strategy of
routinely increasing cardiac index to pre-
defined “supranormal” levels (⬎4.5
L·min⫺1·m⫺2) has not been shown to im-
prove outcome.
Recommendation 3—Level C. A vaso-
pressor such as norepinephrine and an
inotrope such as dobutamine can be ti-
trated separately to maintain both mean
arterial pressure and cardiac output.
REFERENCES
1. Parrillo JE, Parker MM, Natanson C, et al:
Septic shock in humans: Advances in the
understanding of pathogenesis, cardiovas-
cular dysfunction, and therapy. Ann Intern
Med 1990; 113:227–242
2. Ince C, Sinaasappel M: Microcirculatory ox-
ygenation and shunting in sepsis and shock.
Crit Care Med 1999; 27:1369 –1377
3. Hollenberg SM, Parrillo JE: Shock. In: Har-
rison’s Principles of Internal Medicine.
Fourteenth Edition. Fauci AS, Braunwald
E, Isselbacher KJ, et al. (Eds). New York,
McGraw-Hill, 1997, pp 214 –222
4. Cook DJ, Guyatt GH, Laupacis A, et al:
Rules of evidence and clinical recommen-
dations on the use of antithrombotic
agents. Chest 1992 102(4 Suppl):
305S–311S
5. Evidence-Based Medicine Working Group.
Evidence-based medicine: A new approach
to teaching the practice of medicine. JAMA
1992; 268:2420 –2425
6. (I) Bernard GR, Vincent JL, Laterre PF, et
al: Efficacy and safety of recombinant hu-
man activated protein C for severe sepsis.
N Engl J Med 2001; 344:699 –709
7. (I) Rivers E, Nguyen B, Havstad S, et al:
Early goal-directed therapy in the treat-
ment of severe sepsis and septic shock.
N Engl J Med 2001; 345:1368 –1377
8. Hotchkiss RS, Karl IE: Reevaluation of the
role of cellular hypoxia and bioenergetic
failure in sepsis. JAMA 1992; 267:
1503–1510
9. Astiz M, Rackow EC, Weil MH, et al: Early
impairment of oxidative metabolism and
energy production in severe sepsis. Circ
Shock 1988; 26:311–320
10. (II) Hayes MA, Timmins AC, Yau EH, et al:
Oxygen transport patterns in patients with
sepsis syndrome or septic shock: Influence
of treatment and relationship to outcome.
Crit Care Med 1997; 25:926 –936
11. (V) Steffes CP, Dahn MS, Lange MP: Oxygen
transport-dependent splanchnic metabo-
lism in the sepsis syndrome. Arch Surg
1994; 129:46 –52
12. Bredle D, Samsel R, Schumacker P: Critical
O2 delivery to skeletal muscle at high and
low PO 2 in endotoxemic dogs. J Appl
Physiol 1989; 66:2553–2558
13. Rackow E, Astiz ME, Weil MH: Increases in
oxygen extraction during rapidly fatal septic
shock in rats. J Lab Clin Med 1987; 109:
660 – 664
14. Gore DC, Jahoor F, Hibbert JM, et al: Lactic
acidosis during sepsis is related to increased
pyruvate production, not deficits in tissue
oxygen availability. Ann Surg 1996; 224:
97–102
15. (V) Friedman G, Berlot G, Kahn RJ: Com-
1943
 bined measurements of blood lactate levels
and gastric intramucosal pH in patients
with severe sepsis. Crit Care Med 1995; 23:
1184 –1193
16. (V) Vincent JL, Dufaye P, Berre J: Serial
lactate determinations during circulatory
shock. Crit Care Med 1983; 11:449 – 451
17. (V) Weil MH, Afifi AA: Experimental and
clinical studies on lactate and pyruvate as
indicators of the severity of acute circula-
tory failure (shock). Circulation 1970; 41:
989 –1001
18. (V) Bakker J, Coffemils M, Leon M, et al:
Blood lactates are superior to oxygen-
derived variables in predicting outcome in
human septic shock. Chest 1992; 99:
956 –962
19. (V) De Backer D, Creteur J, Noordally O, et
al: Does hepato-splanchnic VO2/DO2 de-
pendency exist in critically ill septic pa-
tients? Am J Respir Crit Care Med 1988;
157:1219 –1225
20. Nelson D, Beyer C, Samsel R, et al: Patho-
logic supply dependence of systemic and
intestinal O2 uptake during bacteremia in
the dog. J Appl Physiol 1987; 63:1487–1489
21. Russell JA: Gastric tonometry: Does it
work? Intensive Care Med 1997; 23:3– 6
22. (V) Marik PE: Gastric intramucosal pH. A
better predictor of multiorgan dysfunction
syndrome and death than oxygen-derived
variables in patients with sepsis. Chest
1993; 104:225–229
23. (V) Maynard N, Bihari D, Beale R, et al:
Assessment of splanchnic oxygenation by
gastric tonometry in patients with acute
circulatory failure. JAMA 1993; 270:
1203–1210
24. (II) Gutierrez G, Palizas F, Doglio G, et al:
Gastric intramucosal pH as a therapeutic
index of tissue oxygenation in critically ill
patients. Lancet 1992; 339:195–199
25. (V) Doglio GR, Pusajo JF, Egurrola MA, et
al: Gastric mucosal pH as a prognostic in-
dex of mortality in critically ill patients. Crit
Care Med 1991; 19:1037–1040
26. (III) Weil MH, Nakagawa Y, Tang W, et al:
Sublingual capnometry: A new noninvasive
measurement for diagnosis and quantita-
tion of severity of circulatory shock. Crit
Care Med 1999; 27:1225–1229
27. (III) Marik PE: Sublingual capnography: A
clinical validation study. Chest 2001; 120:
923–927
28. Carroll G, Snyder J: Hyperdynamic severe
intravascular sepsis depends on fluid ad-
ministration in cynomolgus monkey. Am J
Physiol 1982; 243:R131–R141
29. (V) Winslow EJ, Loeb HS, Rahimtoola SH,
et al: Hemodynamic studies and results of
therapy in 50 patients with bacteremic
shock. Am J Med 1973; 54:421– 432
30. Greenfield LJ, Jackson RH, Elkins RC: Car-
diopulmonary effects of volume loading of
primates in endotoxin shock. Surgery 1974;
76:560 –570
31. (V) Weil MH, Nishijima H: Cardiac output
1944
in bacterial shock. Am J Med 1978; 64:
920 –923
32. (V) Rackow EC, Falk JL, Fein IA: Fluid re-
suscitation in shock: A comparison of car-
diorespiratory effects of albumin,
hetastarch and saline solutions in patients
with hypovolemic shock. Crit Care Med
1983; 11:839 – 850
33. (III) Haupt MT, Gilbert EM, Carlson RW:
Fluid loading increases oxygen consump-
tion in septic patients with lactic acidosis.
Am Rev Respir Dis 1985; 131:912–916
34. (III) Packman MJ, Rackow EC: Optimum
left heart filling pressure during fluid resus-
citation of patients with hypovolemic and
septic shock. Crit Care Med 1983; 11:
165–169
35. Sugerman H, Diaco J, Pollack T, et al: Phys-
iologic management of septicemic shock in
man. Surg Forum 1971; 22:3–5
36. (V) Michard F, Boussat S, Chemla D, et al:
Relation between respiratory changes in ar-
terial pulse pressure and fluid responsive-
ness in septic patients with acute circula-
tory failure. Am J Respir Crit Care Med
2000; 162:134 –138
37. (V) Tavernier B, Makhotine O, Lebuffe G, et
al: Systolic pressure variation as a guide to
fluid therapy in patients with sepsis-
induced hypotension. Anesthesiology 1998;
89:1313–1321
38. (V) Michard F, Teboul JL: Predicting fluid
responsiveness in ICU patients: A critical
analysis of the evidence. Chest 2002; 121:
2000 –2008
39. (II) Tuchschmidt J, Fried J, Astiz M, et al:
Elevation of cardiac output and oxygen de-
livery improves outcome in septic shock.
Chest 1992; 102:216 –220
40. (V) Canizaro PC, Prager MD, Shires GT: The
infusion of Ringer’s lactate solution during
shock. Changes in lactate, excess lactate,
and pH. Am J Surg 1971; 122:494 –501
41. (III) Lamke LO, Liljedahl SO: Plasma vol-
ume changes after infusion of various
plasma expanders. Resuscitation 1976;
5:93–102
42. (III) Shoemaker WC: Comparisons of the
relative effectiveness of whole blood trans-
fusions and various types of fluid therapy in
resuscitation. Crit Care Med 1976; 4:71–78
43. Holcroft JW: Hypertonic saline for resusci-
tation of the patient in shock. Adv Surg
2001; 35:297–318
44. (I) Finfer S, Bellomo R, Boyce N, et al: A
comparison of albumin and saline for fluid
resuscitation in the intensive care unit.
N Engl J Med 2004; 350:2247–2256
45. (II) Gan TJ, Bennett-Guerrero E, Phillips-
Bute B, et al: Hextend, a physiologically
balanced plasma expander for large volume
use in major surgery: A randomized phase
III clinical trial. Hextend Study Group.
Anesth Analg 1999; 88:992–998
46. (V) Cittanova ML, Leblanc I, Legendre C, et
al: Effect of hydroxyethylstarch in brain-
dead kidney donors on renal function in
kidney-transplant recipients. Lancet 1996;
348:1620 –1622
47. (II) Schortgen F, Lacherade JC, Bruneel F,
et al: Effects of hydroxyethylstarch and gel-
atin on renal function in severe sepsis: A
multicentre randomised study. Lancet
2001; 357:911–916
48. (V) Kumle B, Boldt J, Piper S, et al: The
influence of different intravascular volume
replacement regimens on renal function in
the elderly. Anesth Analg 1999; 89:
1124 –1130
49. (V) Shatney CH, Deepika K, Militello PR, et
al: Efficacy of hetastarch in the resuscita-
tion of patients with multisystem trauma
and shock. Arch Surg 1983; 118:804 – 809
50. (II) Boldt J, Muller M, Mentges D, et al:
Volume therapy in the critically ill: Is there
a difference? Intensive Care Med 1998; 24:
28 –36
51. (V) Dehne MG, Muhling J, Sablotzki A, et al:
Hydroxyethyl starch (HES) does not di-
rectly affect renal function in patients with
no prior renal impairment. J Clin Anesth
2001; 13:103–111
52. (V) de Jonge E, Levi M: Effects of different
plasma substitutes on blood coagulation: A
comparative review. Crit Care Med 2001;
29:1261–1267
53. Guyton AC, Lindsey AW: Effect of elevated
left atrial pressure and decreased plasma
protein concentration on the development
of pulmonary edema. Circ Res 1959;
7:649 – 657
54. Kramer GC, Harms BA, Bodai BI, et al:
Effects of hypoproteinemia and increased
vascular pressure on lung fluid balance in
sheep. J Appl Physiol 1983; 55:1514 –1522
55. (V) Rackow EC, Fein AI, Siegel J: The rela-
tionships of colloid osmotic-pulmonary ar-
tery wedge pressure gradient to pulmonary
edema and mortality in critically ill pa-
tients. Chest 1982; 82:433– 437
56. (V) Finley RJ, Holliday RL, Lefcoe M, et al:
Pulmonary edema in patients with sepsis.
Surg Gynecol Obstet 1975; 140:851– 857
57. (V) Sise MJ, Shackford SR, Peters RM, et al:
Serum oncotic-hydrostatic pressure differ-
ences in critically ill patients. Anesth Analg
1982; 61:496 – 498
58. (II) Virgilio RW, Rice CL, Smith DE, et al:
Crystalloid vs. colloid resuscitation. Is one
better? A randomized clinical study. Sur-
gery 1979; 85:129 –139
59. (II) Moss G, Lower R, Jilek J: Colloid or
crystalloid in the resuscitation of hemor-
rhagic shock: A controlled clinical trial.
Surgery 1981; 89:434 – 437
60. Rackow EC, Astiz ME, Schumer W, et al:
Lung and muscle water after crystalloid and
colloid infusion in septic rats: Effect on
oxygen delivery and metabolism. J Lab Clin
Med 1989; 113:184 –189
61. Nylander WAJ, Hammon JWJ, Roselli RJ, et
al: Comparison of the effects of saline and
homologous plasma infusion on lung fluid
balance during endotoxemia in unanesthe-
tized sheep. Surgery 1981; 90:221–228
Crit Care Med 2004 Vol. 32, No. 9
 62. (II) Metildi LA, Shackford SR, Virgilio RW,
et al: Crystalloid versus colloid in fluid re-
suscitation of patients with severe pulmo-
nary insufficiency. Surg Gynecol Obstet
1984; 158:207–212
63. (V) Appel P, Shoemaker WC: Evaluation of
fluid therapy in adult respiratory failure.
Crit Care Med 1981; 9:862– 869
64. (V) Morisaki H, Bloos F, Keys J, et al: Com-
pared with crystalloid, colloid therapy slows
progression of extrapulmonary tissue injury
in septic sheep. J Appl Physiol 1994; 77:
1507–1518
65. Baum TD, Wang H, Rothschild HR, et al:
Mesenteric oxygen metabolism, ileal muco-
sal hydrogen ion concentration and tissue
edema after crystalloid or colloid resuscita-
tion in porcine endotoxic shock: Compari-
son of Ringer’s lactate and 6% hetastarch.
Circ Shock 1990; 30:385–397
66. O’Brien R, Murdoch J, Kuehn R, et al: The
effect of albumin or crystalloid resuscita-
tion on bacterial translocation and endo-
toxin absorption following experimental
burn injury. J Surg Res 1992; 52:161–166
67. (V) Cochrane Injuries Group Albumin Re-
viewers. Human albumin administration in
critically ill patients: Systematic review of
randomised controlled trials. BMJ1998;
317:235–240
68. (V) Choi PT, Yip G, Quinonez LG, et al:
Crystalloids vs. colloids in fluid resuscita-
tion: A systematic review. Crit Care Med
1999; 27:200 –210
69. (V) Wilkes MM, Navickis RJ: Patient survival
after human albumin administration. A
meta-analysis of randomized, controlled tri-
als. Ann Intern Med 2001; 135:149 –164
70. (V) Nelson AH, Fleisher LA, Rosenbaum SH:
Relationship between postoperative anemia
and cardiac morbidity in high-risk vascular
patients in the intensive care unit. Crit Care
Med 1993; 216:860 – 866
71. Morisaki H, Sibbald W, Martin C, et al:
Hyperdynamic sepsis depresses the circula-
tory compensation of normovolemic ane-
mia in conscious rats. J Appl Physiol 1996;
80:656 – 664
72. (V) Conrad S, Dietch K, Hebert C: Effect of
red cell transfusion on oxygen consumption
following fluid resuscitation in septic
shock. Circ Shock 1990; 31:419 – 420
73. (V) Steffes C, Bender J, Levison M: Blood
transfusion and oxygen consumption in
surgical sepsis. Crit Care Med 1991; 19:
512–517
74. (V) Mink R, Pollack M: Effect of blood trans-
fusion on oxygen consumption in pediatric
septic shock. Crit Care Med 1990; 18:
1087–1091
75. (III) Silverman H, Tuma P: Gastric tonom-
etry in patients with sepsis. Effects of do-
butamine infusions and packed red blood
cell transfusions. Chest 1992; 102:184 –189
76. (V) Marik P, Sibbald W: Effect of stored-
blood transfusion on oxygen delivery in pa-
tients with sepsis. JAMA 1993; 269:
3024 –3029
Crit Care Med 2004 Vol. 32, No. 9
77. Bordin JO, Heddle NM, Blajchman MA: Bi-
ologic effects of leukocytes present in trans-
fused cellular blood products. Blood 1994;
84:1703–1721
78. (I) Hebert PC, Wells G, Blajchman M, et al:
A multicenter, randomized, controlled clin-
ical trial of transfusion requirements in
critical care. N Engl J Med 1999; 340:
409 – 417
79. Hollenberg SM, Ahrens TS, Astiz ME, et al:
Task Force of the American College of Crit-
ical Care Medicine. Practice parameters for
hemodynamic support of sepsis in adult pa-
tients. Crit Care Med 1999; 27:639 – 660
80. Bersten AD, Holt AW: Vasoactive drugs and
the importance of renal perfusion pressure.
New Horiz 1995; 3:650 – 661
81. Kirchheim HR, Ehmke H, Hackenthal E, et
al: Autoregulation of renal blood flow, glo-
merular filtration rate and renin release in
conscious dogs. Pflugers Arch 1987; 410:
441– 449
82. (V) Barry K, Mazze R, Schwarz F: Preven-
tion of surgical oliguria and renal hemody-
namic suppression by sustained hydration.
N Engl J Med 1967; 270:1371–1377
83. (V) Bush HL Jr, Huse JB, Johnson WC, et al:
Prevention of renal insufficiency after ab-
dominal aortic aneurysm resection by opti-
mal volume loading. Arch Surg 1981; 116:
1517–1524
84. Kelleher SP, Robinette JB, Conger JD: Sym-
pathetic nervous system in the loss of au-
toregulation in acute renal failure. Am J
Physiol 1984; 246:F379 –F386
85. (V) Desjars P, Pinaud M, Bugnon D, et al:
Norepinephrine therapy has no deleterious
renal effects in human septic shock. Crit
Care Med 1989; 17:426 – 429
86. (V) Desjars P, Pinaud M, Tasseau F, et al: A
reappraisal of norepinephrine therapy in
human septic shock. Crit Care Med 1987;
15:134 –137
87. (V) Bollaert PE, Bauer P, Audibert G, et al:
Effects of epinephrine on hemodynamics
and oxygen metabolism in dopamine-
resistant septic shock. Chest 1990; 98:
949 –953
88. (V) Fukuoka T, Nishimura M, Imanaka H, et
al: Effects of norepinephrine on renal func-
tion in septic patients with normal and el-
evated serum lactate levels. Crit Care Med
1989; 17:1104 –1107
89. (V) Hesselvik JF, Brodin B: Low dose nor-
epinephrine in patients with septic shock
and oliguria: Effects on afterload, urine
flow, and oxygen transport. Crit Care Med
1989; 17:179 –180
90. (V) Lipman J, Roux A, Kraus P: Vasocon-
strictor effects of adrenaline in human sep-
tic shock. Anaesth Intensive Care 1991; 19:
61– 65
91. (III) Martin C, Saux P, Eon B, et al: Septic
shock: A goal-directed therapy using vol-
ume loading, dobutamine and/or norepi-
nephrine. Acta Anaesthesiol Scand 1990;
34:413– 417
92. (V) Martin C, Eon B, Saux P, et al: Renal
effects of norepinephrine used to treat sep-
tic shock patients. Crit Care Med 1990; 18:
282–285
93. (V) Meadows D, Edwards JD, Wilkins RG, et
al: Reversal of intractable septic shock with
norepinephrine therapy. Crit Care Med
1988; 16:663– 667
94. (V) Redl-Wenzl EM, Armbruster C, Edel-
mann G, et al: The effects of norepinephrine
on hemodynamics and renal function in
severe septic shock states. Intensive Care
Med 1993; 19:151–154
95. (V) Gregory JS, Bonfiglio MF, Dasta JF, et
al: Experience with phenylephrine as a
component of the pharmacologic support of
septic shock. Crit Care Med 1991; 19:
1395–1400
96. (III) LeDoux D, Astiz ME, Carpati CM, et al:
Effects of perfusion pressure on tissue per-
fusion in septic shock. Crit Care Med 2000;
28:2729 –2732
97. (V) Parker MM, Shelhamer JH, Bacharach
SL, et al: Profound but reversible myocar-
dial depression in patients with septic
shock. Ann Intern Med 1984; 100:483– 490
98. (V) Hoogenberg K, Smit AJ, Girbes ARJ:
Effects of low-dose dopamine on renal and
systemic hemodynamics during incremen-
tal norepinephrine infusion in healthy vol-
unteers. Crit Care Med 1998; 26:260 –265
99. (V) Lherm T, Troche G, Rossignol M, et al:
Renal effects of low-dose dopamine in pa-
tients with sepsis syndrome or septic shock
treated with catecholamines. Intensive Care
Med 1996; 22:213–219
100. (V) Meier-Hellmann A, Bredle DL, Specht
M, et al: The effects of low-dose dopamine
on splanchnic blood flow and oxygen utili-
zation in patients with septic shock. Inten-
sive Care Med 1997; 23:31–37
101. (II) Marik PE, Mohedin M: The contrasting
effects of dopamine and norepinephrine on
systemic and splanchnic oxygen utilization
in hyperdynamic sepsis. JAMA 1994; 272:
1354 –1357
102. (III) Hannemann L, Reinhart K, Grenzer O,
et al: Comparison of dopamine to dobut-
amine and norepinephrine for oxygen deliv-
ery and uptake in septic shock. Crit Care
Med 1995; 23:1962–1970
103. (III) Ruokonen E, Takala J, Kari A, et al:
Regional blood flow and oxygen transport in
septic shock. Crit Care Med 1993; 21:
1296 –1303
104. (III) Jardin F, Gurdjian F, Desfonds P, et al:
Effect of dopamine on intrapulmonary
shunt fraction and oxygen transport in se-
vere sepsis with circulatory and respiratory
failure. Crit Care Med 1979; 7:273–277
105. (III) Jardin F, Eveleigh MC, Gurdjian F, et
al: Venous admixture in human septic
shock. Comparative effects of blood volume
expansion, dopamine infusion and isopro-
terenol infusion in mismatching of ventila-
tion and pulmonary blood flow in peritoni-
tis. Circulation 1979; 60:155–159
106. (II) Martin C, Papazian L, Perrin G, et al:
Norepinephrine or dopamine for the treat-
1945
 ment of hyperdynamic septic shock. Chest
1993; 103:1826 –1831
107. (II) Regnier B, Safran D, Carlet J, et al:
Comparative haemodynamic effects of do-
pamine and dobutamine in septic shock.
Intensive Care Med 1979; 5:115–120
108. (V) Samii K, Le Gall JR, Regnier B, et al:
Hemodynamic effects of dopamine in septic
shock with and without acute renal failure.
Arch Surg 1978; 113:1414 –1416
109. (V) Drueck C, Welch GW, Pruitt BA Jr: He-
modynamic analysis of septic shock in ther-
mal injury: Treatment with dopamine. Am
Surg 1978; 44:424 – 427
110. (V) Regnier B, Rapin M, Gory G, et al: Hae-
modynamic effects of dopamine in septic
shock. Intensive Care Med 1977; 3:47–53
111. (V) Wilson RF, Sibbald WJ, Jaanimagi JL:
Hemodynamic effects of dopamine in criti-
cally ill septic patients. J Surg Res 1976;
20:163–172
112. (II) Schreuder WO, Schneider AJ, Groen-
eveld ABJ, et al: Effect of dopamine vs nor-
epinephrine on hemodynamics in septic
shock. Chest 1989; 95:1282–1288
113. (IV) Jakob SM, Ruokonen E, Takala J: Ef-
fects of dopamine on systemic and regional
blood flow and metabolism in septic and
cardiac surgery patients. Shock 2002; 18:
8 –13
114. (IV) Maynard ND, Bihari DJ, Dalton RN, et
al: Increasing splanchnic blood flow in the
critically ill. Chest 1995; 108:1648 –1654
115. (II) Neviere R, Chagnon JL, Vallet B, et al:
Dobutamine improves gastrointestinal mu-
cosal blood flow in a porcine model of en-
dotoxic shock. Crit Care Med 1997; 25:
1371–1377
116. Denton MD, Chertow GM, Brady HR: “Re-
nal-dose” dopamine for the treatment of
acute renal failure: Scientific rationale, ex-
perimental studies and clinical trials. Kid-
ney Int 1996; 50:4 –14
117. (I) Bellomo R, Chapman M, Finfer S, et al J:
Low-dose dopamine in patients with early
renal dysfunction: A placebo-controlled
randomised trial. Australian and New Zea-
land Intensive Care Society (ANZICS) Clin-
ical Trials Group. Lancet 2000; 356:
2139 –2143
118. Van den Berghe G, de Zegher F: Anterior
pituitary function during critical illness and
dopamine treatment. Crit Care Med 1996;
24:1580 –1590
119. (V) Martin C, Perrin G, Saux P, et al: Effects
of norepinephrine on right ventricular
function in septic shock patients. Intensive
Care Med 1994; 20:444 – 447
120. (III) Martin C, Viviand X, Arnaud S, et al:
Effects of norepinephrine plus dobutamine
or norepinephrine alone on left ventricular
performance of septic shock patients. Crit
Care Med 1999; 27:1708 –1713
121. (II) Levy B, Bollaert PE, Charpentier C, et
al: Comparison of norepinephrine and do-
butamine to epinephrine for hemodynam-
ics, lactate metabolism, and gastric tono-
metric variables in septic shock: A
1946
prospective, randomized study. Intensive
Care Med 1997; 23:282–287
122. Chernow B, Roth BL: Pharmacologic ma-
nipulation of the peripheral vasculature in
shock: Clinical and experimental ap-
proaches. Circulatory Shock 1986; 18:
141–155
123. Murakawa K, Kobayashi A: Effects of vaso-
pressors on renal tissue gas tensions during
hemorrhagic shock in dogs. Crit Care Med
1988; 16:789 –792
124. Conger JD, Robinette JB, Guggenheim SJ:
Effect of acetylcholine on the early phase of
reversible norepinephrine-induced acute
renal failure. Kidney Int 1981; 19:399 – 409
125. Schaer GL, Fink MP, Parrillo JE: Norepi-
nephrine alone versus norepinephrine plus
low-dose dopamine: Enhanced renal blood
flow with combination pressor therapy. Crit
Care Med 1985; 13:492– 496
126. (V) Reinelt H, Radermacher P, Fischer G, et
al: Effects of a dobutamine-induced in-
crease in splanchnic blood flow on hepatic
metabolic activity in patients with septic
shock. Anesthesiology 1997; 86:818 – 824
127. (V) De Backer D, Creteur J, Silva E, et al:
Effects of dopamine, norepinephrine, and
epinephrine on the splanchnic circulation
in septic shock: Which is best? Crit Care
Med 2003; 31:1659 –1667
128. Dasta JF: Norepinephrine in septic shock:
Renewed interest in an old drug. DICP Ann
Pharmacother 1990; 24:153–156
129. (V) Martin C, Viviand X, Leone M, et al:
Effect of norepinephrine on the outcome of
septic shock. Crit Care Med 2000; 28:
2758 –2765
130. (V) Moyer J, Skelton J, Mills L: Norepineph-
rine: effect in normal subjects: Use in treat-
ment of shock unresponsive to other mea-
sures. Am J Med 1953; 15:330 –343
131. (V) Yamazaki T, Shimada Y, Taenaka N, et
al: Circulatory responses to afterloading
with phenylephrine in hyperdynamic sepsis.
Crit Care Med 1982; 10:432– 435
132. (III) Flancbaum L, Dick M, Dasta J, et al: A
dose-response study of phenylephrine in
critically ill, septic surgical patients. Eur
J Clin Pharmacol 1997; 51:461– 465
133. (V) Wilson W, Lipman J, Scribante J, et al:
Septic shock: Does adrenaline have a role as
a first-line inotropic agent? Anesth Inten-
sive Care 1992; 20:470 – 474
134. (III) Moran JL, MS O’Fathartaigh, Peisach
AR, et al: Epinephrine as an inotropic agent
in septic shock: A dose-profile analysis. Crit
Care Med 1993; 21:70 –77
135. (III) Mackenzie SJ, Kapadia F, Nimmo GR,
et al: Adrenaline in treatment of septic
shock: Effects on haemodynamics and oxy-
gen transport. Intensive Care Med 1991;
17:36 –39
136. (V) Le Tulzo Y, Seguin P, Gacouin A, et al:
Effects of epinephrine on right ventricular
function in patients with severe septic
shock and right ventricular failure: A pre-
liminary study. Intensive Care Med 1997;
23:664 – 670
137. (II) Day NP, Phu NH, Bethell DP, et al: The
effects of dopamine and adrenaline infu-
sions on acid-base balance and systemic
haemodynamics in severe infection. Lancet
1996; 348:219 –223
138. (V) Zhou SX, Qiu HB, Huang YZ, et al:
Effects of norepinephrine, epinephrine, and
norepinephrine-dobutamine on systemic
and gastric mucosal oxygenation in septic
shock. Acta Pharmacol Sin 2002; 23:
654 – 658
139. (V) Duranteau J, Sitbon P, Teboul JL, et al:
Effects of epinephrine, norepinephrine, or
the combination of norepinephrine and do-
butamine on gastric mucosa in septic
shock. Crit Care Med 1999; 27:893–900
140. Seguin P, Bellissant E, Le Tulzo Y, et al:
Effects of epinephrine compared with the
combination of dobutamine and norepi-
nephrine on gastric perfusion in septic
shock. Clin Pharmacol Ther 2002; 71:
381–388
141. McKenzie A, Stoughton RB: Method for
comparing percutaneous absorption of ste-
roids. Arch Dermatol 1962; 86:608 – 610
142. Yard AC, Kadowitz PJ: Studies on the mech-
anism of hydrocortisone potentiation of va-
soconstrictor responses to epinephrine in
the anesthetized animal. Eur J Pharmacol
1972; 20:1–9
143. Grunfeld JP, Eloy L: Glucocorticoids mod-
ulate vascular reactivity in the rat. Hyper-
tension 1987; 10:608 – 618
144. Steiner A, Vogt E, Locher R, et al: Stimula-
tion of the phosphoinositide signalling sys-
tem as a possible mechanism for glucocor-
ticoid action in blood pressure control.
J Hypertens Suppl 1988; 6:S366 –S368
145. Rees DD, Cellek S, Palmer RMJ, et al: Dexa-
methasone prevents the induction by endo-
toxin of a nitric oxide synthase and the
associated effects on vascular tone: An in-
sight into endotoxin shock. Biochem Bio-
phys Res Commun 1990; 173:541–547
146. Kennedy B, Ziegler MG: Cardiac epineph-
rine synthesis. Regulation by a glucocorti-
coid. Circulation 1991; 84:891– 895
147. Dailey JW, Westfall TC: Effects of adrenal-
ectomy and adrenal steroids on norepineph-
rine synthesis and monamine oxidase activ-
ity. Eur J Pharmacol 1978; 48:383–391
148. Sakaue M, Hoffman BB: Glucocorticoids in-
duce transcription and expression of the
alpha 1B adrenergic receptor gene in DTT1
MF-2 smooth muscle cells. J Clin Invest
1991; 88:385–389
149. Haigh RM, Jones CT: Effect of glucocorti-
coids on alpha 1-adrenergic receptor bind-
ing in rat vascular smooth muscle. J Mol
Endocrinol 1990; 5:41– 48
150. Sato A, Suzuki H, Murakami M, et al: Glu-
cocorticoid increases angiotensin II type 1
receptor and its gene expression. Hyperten-
sion 1994; 23:25–30
151. Scheuer DA, Bechtold AG: Glucocorticoids
potentiate central actions of angiotensin to
increase arterial pressure. Am J Physiol
Crit Care Med 2004 Vol. 32, No. 9
 Regul Integr Comp Physiol 2001; 280:
R1719 –R1726
152. Paya D, Gray GA, Fleming I, et al: Effect of
dexamethasone on the onset and persis-
tence of vascular hyporeactivity induced by
E coli lipopolysaccharide in rats. Circ Shock
1993; 41:103–112
153. Wu CC, Croxtall JD, Perretti M, et al: Li-
pocortin 1 mediates the inhibition by dexa-
methasone of the induction by endotoxin of
nitric oxide synthase in the rat. Proc Natl
Acad Sci U S A 1995; 92:3473–3477
154. Mansart A, Bollaert PE, Seguin C, et al:
Hemodynamic effects of early versus late
glucocorticosteroid administration in ex-
perimental septic shock. Shock 2003; 19:
38 – 44
155. (V) Bhagat K, Collier J, Vallance P: Local
venous responses to endotoxin in humans.
Circulation 1996; 94:490 – 497
156. (V) Barber AE, Coyle SM, Marano MA, et al:
Glucocorticoid therapy alters hormonal and
cytokine responses to endotoxin in man.
J Immunol 1993; 150:1999 –2006
157. (III) Annane D, Bellissant E, Sebille V, et al:
Impaired pressor sensitivity to noradrena-
line in septic shock patients with and with-
out impaired adrenal function reserve. Br J
Clin Pharmacol 1998; 46:589 –597
158. (III) Bellissant E, Annane D: Effect of hy-
drocortisone on phenylephrine—Mean ar-
terial pressure dose-response relationship
in septic shock. Clin Pharmacol Ther 2000;
68:293–303
159. (II) Bollaert PE, Charpentier C, Levy B, et
al: Reversal of late septic shock with supra-
physiologic doses of hydrocortisone. Crit
Care Med 1998; 26:645– 650
160. (II) Briegel J, Forst H, Haller M, et al: Stress
doses of hydrocortisone reverse hyperdy-
namic septic shock: a prospective, random-
ized, double-blind, single-center study. Crit
Care Med 1999; 27:723–732
161. (V) Chawla K, Kupfer Y, Tessler S: Hydro-
cortisone reverses refractory septic shock.
Abstr. Crit Care Med 1999; 27:A33
162. (I) Annane D, Sebille V, Charpentier C, et al:
Effect of treatment with low doses of hydro-
cortisone and fludrocortisone on mortality
in patients with septic shock. JAMA 2002;
288:862– 871
163. (II) Keh D, Boehnke T, Weber-Cartens S, et
al: Immunologic and hemodynamic effects
of “low-dose” hydrocortisone in septic
shock: A double-blind, randomized, place-
bo-controlled, crossover study. Am J Respir
Crit Care Med 2003; 167:512–520
164. Cronin L, Cook DJ, Carlet J, et al: Cortico-
steroid treatment for sepsis: a critical ap-
praisal and meta-analysis of the literature.
Crit Care Med 1995; 23:1430 –1439
165. (V) McKee JI, Finlay WE: Cortisol replace-
ment in severely stressed patients. Lancet
1983; 1:484
166. (V) Holmes CL, Walley KR, Chittock DR, et
al: The effects of vasopressin on hemody-
namics and renal function in severe septic
Crit Care Med 2004 Vol. 32, No. 9
shock: A case series. Intensive Care Med
2001; 27:1416 –1421
167. Kusano E, Tian S, Umino T, et al: Arginine
vasopressin inhibits interleukin-1 beta-
stimulated nitric oxide and cyclic guanosine
monophosphate production via the V1 re-
ceptor in cultured rat vascular smooth
muscle cells. J Hypertens 1997; 15:627– 632
168. Wakatsuki T, Nakaya Y, Inoue I: Vasopressin
modulates K(⫹)-channel activities of cul-
tured smooth muscle cells from porcine
coronary artery. Am J Physiol 1992; 263:
H491–H496
169. (V) Abboud FM, Floras JS, Aylward PE, et al:
Role of vasopressin in cardiovascular and
blood pressure regulation. Blood Vessels
1990; 27:106 –115
170. Wang BC, Flora-Ginter G, Leadley RJ Jr, et
al: Ventricular receptors stimulate vaso-
pressin release during hemorrhage. Am J
Physiol 1988; 254:R204 –R211
171. Wilson MF, Brackett DJ, Tompkins P, et al:
Elevated plasma vasopressin concentrations
during endotoxin and E. coli shock. Adv
Shock Res 1981; 6:15–26
172. (III) Landry DW, Levin HR, Gallant EM, et
al: Vasopressin deficiency contributes to the
vasodilation of septic shock. Circulation
1997; 95:1122–1125
173. (V) Sharshar T, Carlier R, Blanchard A, et al:
Depletion of neurohypophyseal content of
vasopressin in septic shock. Crit Care Med
2002; 30:497–500
174. (V) Sharshar T, Blanchard A, Paillard M, et
al: Circulating vasopressin levels in septic
shock. Crit Care Med 2003; 31:1752–1758
175. (V) Tsuneyoshi I, Yamada H, Kakihana Y, et
al: Hemodynamic and metabolic effects of
low-dose vasopressin infusions in vasodila-
tory septic shock. Crit Care Med 2001; 29:
487– 493
176. (II) Malay MB, Ashton RC Jr, Landry DW, et
al: Low-dose vasopressin in the treatment of
vasodilatory septic shock. J Trauma 1999;
47:699 –705
177. (II) Patel BM, Chittock DR, Russell JA, et al:
Beneficial effects of short-term vasopressin
infusion during severe septic shock. Anes-
thesiology 2002; 96:576 –582
178. (V) O’Brien A, Clapp L, Singer M: Terlipres-
sin for norepinephrine-resistant septic
shock. Lancet 2002; 359:1209 –1210
179. (III) van Haren FM, Rozendaal FW, van der
Hoeven JG: The effect of vasopressin on
gastric perfusion in catecholamine-depen-
dent patients in septic shock. Chest 2003;
124:2256 –2260
180. (III) Klinzing S, Simon M, Reinhart K, et al:
High-dose vasopressin is not superior to
norepinephrine in septic shock. Crit Care
Med 2003; 31:2646 –2650
181. (V) Parker M, Suffredini A, Natanson C, et
al: Responses of left ventricular function in
survivors and nonsurvivors of septic shock.
J Crit Care 1989; 4:19 –25
182. Raper R, Sibbald M, Driedger A, et al: Rel-
ative myocardial depression in normoten-
sive sepsis. J Crit Care 1989; 4:9 –18
183. (V) Parker MM, McCarthy K, Ognibene FP,
et al: Right ventricular dysfunction and di-
latation, similar to left ventricular changes,
characterize the cardiac depression of septic
shock in humans. Chest 1990; 97:126 –131
184. Parker MM, Ognibene FP, Parrillo JE: Peak
systolic pressure/end-systolic volume ratio,
a load-independent measure of ventricular
function, is reversibly decreased in human
septic shock. Crit Care Med 1994; 22:
1955–1959
185. (V) Ognibene FP, Parker MM, Natanson C,
et al: Depressed left ventricular perfor-
mance. Response to volume infusion in pa-
tients with sepsis and septic shock. Chest
1988; 93:903–910
186. (V) Dhainaut JF, Huyghebaert MF, Monsal-
lier JF, et al: Coronary hemodynamics and
myocardial metabolism of lactate, free fatty
acids, glucose, and ketones in patients with
septic shock. Circulation 1987; 75:533–541
187. (V) Cunnion RE, Schaer GL, Parker MM, et
al: The coronary circulation in human sep-
tic shock. Circulation 1986; 73:637– 644
188. Gotloib L, Shostak A, Galdi P, et al: Loss of
microvascular negative charges accompa-
nied by interstitial edema in septic rats’
heart. Circ Shock 1992; 36:45–56
189. Liu MS, Wu LL: Heart sarcolemmal Ca2⫹
transport in endotoxin shock: II. Mecha-
nism of impairment in ATP-dependent
Ca2⫹ transport. Mol Cell Biochem 1992;
112:135–142
190. Gulick T, Chung MK, Pieper SJ, et al: In-
terleukin 1 and tumor necrosis factor in-
hibit cardiac myocyte ␤-adrenergic respon-
siveness. Proc Natl Acad Sci USA 1989; 86:
6753– 6757
191. Carli A, Auclair MC, Vernimmen C: Indo-
methacin suppresses the early cardiode-
pressant factor released by endotoxin in the
rat: Possible involvement of a prostacyclin-
related material. Adv Shock Res 1983; 10:
161–71
192. Finkel MS, Oddis CV, Jacob TD, et al: Neg-
ative inotropic effects of cytokines on the
heart mediated by nitric oxide. Science
1992; 257:387–389
193. Brady AJB, Poole-Wilson PA, Harding SE, et
al: Nitric oxide production within cardiac
myocytes reduces their contractility in en-
dotoxemia. Am J Physiol 1992; 263:
H1963–H1966
194. Schulz R, Panas DL, Catena R, et al: The
role of nitric oxide in cardiac depression
induced by interleukin-1 beta and tumour
necrosis factor-alpha. Br J Pharmacol 1995;
114:27–34
195. Favory R, Lancel S, Marchetti P, et al: En-
dotoxin-induced myocardial dysfunction:
Evidence for a role of sphingosine produc-
tion. Crit Care Med 2004; 32:495–501
196. Klem C, Dasta JF, Reilley TE, et al: Variabil-
ity in dobutamine pharmacokinetics in un-
stable critically ill surgical patients. Crit
Care Med 1994; 22:1926 –1932
197. (V) Juste RN, Moran L, Hooper J, et al:
Dopamine clearance in critically ill pa-
1947
 tients. Intensive Care Med 1998; 24:
1217–1220
198. (III) Gilbert EM, Haupt MT, Mandanas RY,
et al: The effect of fluid loading, blood trans-
fusion, and catecholamine infusion on oxy-
gen delivery and consumption in patients
with sepsis. Am Rev Respir Dis 1986; 134:
873– 878
199. (II) Hayes MA, Timmins AC, Yau EHS, et al:
Elevation of systemic oxygen delivery in the
treatment of critically ill patients. N Engl
J Med 1994; 330:1717–1722
200. (I) Gattinoni L, Brazzi L, Pelosi P, et al: A
trial of goal-oriented hemodynamic therapy
in critically ill patients. N Engl J Med 1995;
333:1025–1032
201. (II) Yu M, Levy MM, Smith P, : Effect of
maximizing oxygen delivery on morbidity
and mortality rates in critically ill patients:
A prospective, randomized, controlled
study. Crit Care Med 1993; 21:830 – 838
202. (III) Ronco JJ, Fenwick JC, Tweeddale MG,
et al: Identification of the critical oxygen
delivery for anaerobic metabolism in criti-
cally ill septic and nonseptic humans. JAMA
1993; 270:1724 –1730
203. (II) Boyd O, Grounds RM, Bennett ED: A
randomized clinical trial of the effect of
deliberate perioperative increase of oxygen
delivery on mortality in high-risk surgical
patients. JAMA 1993; 270:2699 –2707
204. (I) Sandham JD, Hull RD, Brant RF, et al: A
randomized, controlled trial of the use of
pulmonary-artery catheters in high-risk
surgical patients. N Engl J Med 2003; 348:
5–14
205. Heyland DK, Cook DJ, King D, et al: Maxi-
mizing oxygen delivery in critically ill pa-
tients: A methodologic appraisal of the evi-
dence. Crit Care Med 1996; 24:517–524
206. McCarter FD, Nierman SR, James JH, et al:
Role of skeletal muscle Na⫹-K⫹ ATPase
activity in increased lactate production in
sub-acute sepsis. Life Sci 2002; 70:
1875–1888
207. Jain A, Shroff SG, Janicki JS, et al: Relation
between mixed venous oxygen saturation
and cardiac index. Nonlinearity and nor-
malization for oxygen uptake and hemoglo-
bin. Chest 1991; 99:1403–1409
208. (V) Vaughn S, Puri VK: Cardiac output
changes and continuous mixed venous ox-
ygen saturation measurement in the criti-
cally ill. Crit Care Med 1988; 16:495– 498
1948
209. Dubin A, Murias G, Estenssoro E, et al:
Intramucosal-arterial PCO2 gap fails to re-
flect intestinal dysoxia in hypoxic hypoxia.
Crit Care 2002; 6:514 –520
210. (III) Loeb HS, Winslow EB, Rahimtoola SH,
et al: Acute hemodynamic effects of dopa-
mine in patients with shock. Circulation
1971; 44:163–173
211. (III) Nasraway SA, Rackow EC, Astiz ME, et
al: Inotropic response to digoxin and dopa-
mine in patients with severe sepsis, cardiac
failure, and systemic hypoperfusion. Chest
1989; 95:612– 615
212. (III) de la Cal MA, Miravalles E, Pascual T, et
al: Dose-related hemodynamic and renal ef-
fects of dopamine in septic shock. Crit Care
Med 1984; 12:22–25
213. (V) Juste RN, Panikkar K, Soni N: The ef-
fects of low-dose dopamine infusions on
haemodynamic and renal parameters in pa-
tients with septic shock requiring treat-
ment with noradrenaline. Intensive Care
Med 1998; 24:564 –568
214. (V) Jardin F, Sportiche M, Bazin M, et al:
Dobutamine: A hemodynamic evaluation in
human septic shock. Crit Care Med 1981;
9:329 –332
215. (II) De Backer D, Berre J, Zhang H, et al:
Relationship between oxygen uptake and
oxygen delivery in septic patients: effects of
prostacyclin versus dobutamine. Crit Care
Med 1993; 21:1658 –1664
216. (III) Vallet B, Chopin C, Curtis SE, et al:
Prognostic value of the dobutamine test in
patients with sepsis syndrome and normal
lactate values: A prospective, multicenter
study. Crit Care Med 1993; 21:1868 –1875
217. (III) Gutierrez G, Clark C, Brown SD, et al:
Effect of dobutamine on oxygen consump-
tion and gastric mucosal pH in septic pa-
tients. Am J Respir Crit Care Med 1994;
150:324 –329
218. (III) De Backer D, Moraine JJ, Berre J, et al:
Effects of dobutamine on oxygen consump-
tion in septic patients. Direct versus indi-
rect determinations. Am J Respir Crit Care
Med 1994; 150:95–100
219. Vincent JL, Roman A, Kahn RJ: Dobut-
amine administration in septic shock: Addi-
tion to a standard protocol. Crit Care Med
1990; 18:689 – 693
220. (III) Mira JP, Fabre JE, Baigorri F, et al:
Lack of oxygen supply dependency in pa-
tients with severe sepsis. A study of oxygen
delivery increased by military antishock
trouser and dobutamine. Chest 1994; 106:
1524 –1531
221. (III) Redl-Wenzl EM, Armbruster C, Edel-
mann G, et al: Noradrenaline in the “high
output-low resistance” state of patients with
abdominal sepsis. Anaesthesist 1990; 39:
525–529
222. (V) Tell B, Majerus TC, Flancbaum L: Do-
butamine in elderly septic shock patients
refractory to dopamine. Intensive Care Med
1987; 13:14 –18
223. (V) Joly LM, Monchi M, Cariou A, et al:
Effects of dobutamine on gastric mucosal
perfusion and hepatic metabolism in pa-
tients with septic shock. Am J Respir Crit
Care Med 1999; 160:1983–1986
224. (II) Levy B, Nace L, Bollaert PE, et al: Com-
parison of systemic and regional effects of
dobutamine and dopexamine in norepi-
nephrine-treated septic shock. Intensive
Care Med 1999; 25:942–948
225. (II) Lebuffe G, Levy B, Neviere R, et al:
Dobutamine and gastric-to-arterial carbon
dioxide gap in severe sepsis without shock.
Intensive Care Med 2002; 28:265–271
226. (III) Meier-Hellmann A, Reinhart K, Bredle
DL, et al: Epinephrine impairs splanchnic
perfusion in septic shock. Crit Care Med
1997; 25:399 – 404
227. (II) Barton P, Garcia J, Kouatli A, et al:
Hemodynamic effects of i.v. milrinone lac-
tate in pediatric patients with septic shock.
A prospective, double-blinded, randomized,
placebo- controlled, interventional study.
Chest 1996; 109:1302–1312
228. (II) Hernandez G, Gigoux J, Bugedo G, et al:
Acute effect of dobutamine and amrinone
on hemodynamics and splanchnic perfusion
in septic shock patients. Rev Med Chil 1999;
127:660 – 666
229. (III) Heinz G, Geppert A, Delle Karth G, et
al: IV milrinone for cardiac output increase
and maintenance: Comparison in nonhy-
perdynamic SIRS/sepsis and congestive
heart failure. Intensive Care Med 1999; 25:
620 – 624
230. Weisul JP, O’Donnell TF Jr, Stone MA, et al:
Myocardial performance in clinical septic
shock: Effects of isoproterenol and glucose
potassium insulin. J Surg Res 1975; 18:
357–363
Crit Care Med 2004 Vol. 32, No. 9
INPATIENT SAFETY I

Inpatient safety I

Challenges in the care of the acutely ill

J F Bion, J E Heffner

Health care providers, hospital administrators, and politicians face competing challenges to reduce clinical errors,
control expenditure, increase access and throughput, and improve quality of care. The safe management of the
acutely ill inpatient presents particular difficulties. In the first of five Lancet articles on this topic we discuss patients’
safety in the acute hospital. We also present a framework in which responsibility for improvement and better
integration of care can be considered at the level of patient, local environment, hospital, and health care system; and
the other four papers in the series will examine in greater detail methods for measuring, monitoring, and improving
inpatient safety.

“It may seem a strange principle to enunciate as the very
first requirement in a hospital that it should do the sick
no harm.” Florence Nightingale.
Notes on hospitals (London, 1859)
Patients’ safety has come to characterise the first decade
of the third millennium just as managed care and cost-
containment did the 1990s. According to the much-
quoted Institute of Medicine (IOM) report To err is
human,1 between 44 000 and 98 000 patients die every
year in the USA as a result of clinical errors; the lower
estimate makes clinical error the seventh most common
cause of death.2 Health care expenditure incurred as a
result of clinical errors is thought to be US$17–29 billion
in the USA,1 Au$4·7 billion in Australia,3 and £6 billion in
the UK.4,5 Those countries, and also Canada and
Denmark, are implementing systems for improving
patients’ safety,6–8 and evidence-based recommendations
to improve safety are emerging.9
Medical trainees quickly learn—and soon come to
accept—that some background rate of clinical error is an
unfortunate yet seemingly ineradicable feature of patient
care. What is new is the characterisation of clinical error
as an epidemic, the recognition that most errors go
undetected and unreported, and the realisation that the
public perceives clinical errors as neither acceptable nor
inevitable. In the USA, more than 40% of people report
personal or family experiences with clinical errors, and
62% favour public disclosure of serious hospital errors.10
Causation, taxonomy, and detection
Causation
The term medical error has served as a convenient half-
truth by which adverse clinical events arising from a
presumed chain of causation are attributed to the last link
of that chain, usually a doctor or nurse. Health-care
institutions have failed to scrutinise the primary elements
in a causation pathway11 or search for root causes,12 even
Lancet 2004; 363: 970–77
See Commentary page 913
University Department of Anaesthesia and Intensive Care
Medicine, Queen Elizabeth Hospital, Birmingham B15 2TH, UK
(J F Bion FRCA); and Center for Clinical Effectiveness and Patient
Safety, Medical University of South Carolina, Charleston, South
Carolina, USA (J E Heffner MD)
Correspondence: Dr J F Bion
(e-mail: j.f.bion@bham.ac.uk)
though other industries have been implementing error
analysis for decades. Error-reduction efforts in health care
have focused on the doctor-patient interface that might
seem to casual observers to be the most obvious cause of
medical accidents. Although failures by clinicians do
contribute to unsafe acts, efforts to correct the
performance of individual health-care providers seldom
contribute to a general improvement in patients’ safety
throughout a health-care system.
Studies of human error in industrial and transportation
accidents13 have refocused our understanding of clinical
errors as a systems problem that requires systems-oriented
solutions. Latent failures embedded in organisational
design set the stage for unsafe practices. Failures such as
poorly designed equipment and monitoring alarms, paper
records with illegible physician orders, and work
conditions that promote fatigue and inattention lie
dormant within organisations’ structures until a fatigued
clinician with poor handwriting, for example, or a high
patient turnover triggers error. Heinrich surveyed14
industrial accidents in the 1940s and estimated that for
every major adverse event there were 29 minor ones and
300 non-injury accidents (near-misses). If similar
proportions relate to health care, latent failures represent
a tremendous opportunity to improve patient safety.
Even though errors of omission outnumber errors of
commission by two to one,3 organisations respond more
readily to errors of commission—for example, by
addressing the risk of administering highly concentrated
potassium solutions intravenously while overlooking
failure to correct hypokalaemia, which affects 20% of
inpatients and promotes life-threatening cardiac
Search strategy
We focused our review on patients’ safety and the
management of the acutely ill hospital patient. To retrieve
information about the health care environment, we also
searched for publications about health care organisation and
emergency services. We used MEDLINE, EMBASE, and
Google, to access government publications and ‘‘grey’’
literature, using singly and in paired combination the terms
safety, medical error, postoperative complications,
emergency medical services, critical care, and intensive care.
We discussed the themes extracted from this process with
other authors in the series and with professional colleagues
worldwide.

970 THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet publishing Group.

arrhythmias.15 Health services need to find novel ways of
avoiding errors of omission. Routine necropsy provides
one opportunity for identifying diagnostic and therapeutic
errors but they are less common than they used to be.16
Although a systems approach has improved safety in the
aerospace and airline industries by identifying and
correcting latent errors, systems engineering remains
largely untested in health care.17
Definitions
Disagreement exists not only about the precise scale of
medical error,18–21 the degree of public disclosure required
to stimulate safer medical practices,10 and the extent to
which errors can be eliminated in complex systems,22 but
also about the terminology.17,23 The IOM used an
outcomes-based definition: “the failure of a planned
action to be completed as intended or the use of a wrong
plan to achieve an aim.”1 Others define error as latent
failures at the systems level that include human,
organisational, and technical constraints on
performance,24 whether there are adverse consequences or
not. “Lumpers” include any instance of underuse,
overuse, or misuse of medical care23 whereas “splitters”
confine the definition to failed processes that have been
proved to cause adverse outcomes.17
These competing taxonomies mesh poorly with the
definitions of health care quality on which efforts to
improve organisational performance are based.25–29
Definitions of quality incorporate concepts of causation,
attribution, preventability, and near-misses, which are
absent from definitions of medical error. Communication
is hampered too; so is research; and the identification and
correction of unsafe practices before an adverse event
occurs are delayed. The medical errors movement has yet
to take full advantage of the principles of outcomes
research. For instance, no investigators have
comprehensively and prospectively looked at the
frequency of clinical errors using a priori definitions to
establish causative links between errors and outcomes.17
Most of the studies used for the IOM report were not
designed primarily to detect errors, assessors were not
masked to outcomes, and the reliability of the measures
were not provided. Because trained investigators have
poor interobserver agreement in identifying clinical errors
from medical record reviews,21,30 clinicians justifiably
expect prospective, masked, outcome-based studies to
determine the dimensions of the problem, identify
methods for correcting it, and describe the quality of
acute care.31
Cultural context
The causes of error are diverse, often complex, and rarely
attributable to single actions, events, or individuals. The
causes are usually rooted in unsafe systems rather than
individual caregivers. The final common pathway,
however, is the interaction between practitioner and
patient. This relation has undergone a profound change in
the past 50 years. Previously paternalistic, personal,
trusting, limited in therapeutic potency and low in
expectation of results, patient-clinician interactions have
become more equal, transparent, team-based, and
contractual. Patients want doctors to communicate more
effectively and devote more time to them. Treatments
have become more effective and have a correspondingly
greater capacity to harm if misused.32
Increasingly, the public expects medical services to
deliver the anticipated results, and on time, and if the
outcome is not as anticipated, a culprit must be sought
and retribution exacted. In a telephone survey of members
THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
INPATIENT SAFETY I
of the public in the USA, 50% believed that suspension of
licences to practise would be an effective way to prevent
clinical error.10 Substantial cultural differences also seem
to exist between medical disciplines33 and between
countries34,35 in attitudes to error. A reduction in
professional authority might have contributed to a
lessening of physicians’ sense of personal responsibility for
the duration of their patients’ care. Methods for reducing
error need to take into account changing doctor-patient
relationships and cultural differences.
Comparisons with industry
The safety and error-prevention record for health care
services is often compared unfavourably with that of
aviation, banking, chemicals, manufacturing, and military
services in peacetime, the best of which have highly
developed strategies to protect workers and clients. These
strategies include a safety culture that emphasises the
importance of safe practices, commitment of management
to safety, and non-punitive and simplified reporting of
errors with feedback of error analyses. Aviation has
focused on the importance of working in teams, and
lessons learned from crew resource management are
starting to be applied to health care.36
Industries with successful safety records emphasise
standardisation of practices combined with flexibility to
address unique circumstances. They also invest in safety
training and research. This strategy has resulted in
substantial reductions in errors and adverse outcomes,
with some industries reporting no serious workforce
injuries for many years. The IOM has adopted the
viewpoint of groups that study “high reliability
organisations” (nuclear aircraft carriers, nuclear power
plants, and air-traffic control) that successful systems
engineering for high levels of safety is achievable.37–39
Health services have, therefore, been encouraged to
align their safety efforts with those of industry40 and aim for
an error rate of less than 3·4 per million events.41 This is
known as “six sigma quality” because it lies outside six
SDs of a normal distribution. Anaesthesia has made
substantial contributions to patients’ safety through
concerted efforts by practitioners and equipment
manufacturers to standardise processes. The IOM report1
states that deaths attributable to anaesthesia have fallen to
5·4 per million, which approaches five sigma, although the
validity of the data that support this observation has been
questioned.42 The persistence of a steady rate of fatalities in
aviation22 or road transport43 despite increased traffic
volume suggests that benefits of safety efforts could be
counterbalanced by competing risk factors, such as high
throughput, seriously ill patients, and complex
interventions.
Although industrial experience provides valuable insight
into potential solutions for clinical errors, the extent to
which it applies to health care remains uncertain. Health
care is characterised by highly complex processes, unique
problems and needs of individual patients, loosely knit
teams, multiple outcome measures, incomplete evidence
for many health care decisions, and varying layers of
responsibility. In hospitals, unpredictable workloads and
uncertainty about individual patients’ outcomes are
additional factors. A more appropriate industrial analogy
for safe care of the acutely ill patient would be the armed
forces contending with the uncertainties of warfare,
friendly fire, and “collateral damage”, a distinction which
governments seem willing to apply to safety of military
personnel.44 Health services should learn from other
industries but they need models that address the unique
challenges of the acute hospital setting.
971
INPATIENT SAFETY I
Error and environment
Although most large safety studies have been in acute care
settings, few distinguish between elective admissions and
emergency admissions or emergencies during planned
treatment. Elective admissions and procedures have well-
defined care pathways which facilitate analysis of
deviations; emergencies tend to be less predictable. In our
view emergency care needs to be seen as a separate entity
with a focus on patients’ safety that cuts across disciplines
and departments. In view of the pressures to accelerate
patient throughput, improved safety will also require better
integration of community health care with discharge
planning, especially since preventable adverse events might
affect 6% of patients after hospital discharge.45
Error in acute care
The risk of adverse events is higher for patients admitted to
emergency departments3, 46–48 and general medical wards
than for those admitted for elective surgery.4,49 Clinical
errors most commonly affect the elderly,3 who account for
most emergency admissions50 and have the highest risk
from emergency surgery.51 The risk of error increases when
such care is provided, as it often is, by inexperienced
clinicians and unsupervised trainees.52–54 The risk of an
adverse event increases by about 6% per day for patients
admitted as emergencies ,55 and is especially high for those
needing lifesaving invasive interventions.46
Critical illness increases the opportunity for clinical
error56–59 because of the complexity of patients’ problems
and the frequency of invasive interventions. The intensity
of monitoring in critical care units means that errors are
more likely to be detected. Iatrogenic complications are a
common cause of admission to intensive care,60 and
previously suboptimal care is associated with an increased
mortality in the intensive care unit.61 Cardiopulmonary
arrest in hospital is frequently preceded by warning signs62
and can be prevented by interventions to identify and
manage patients earlier.63 Premature discharge from
intensive care is associated with increased in-hospital
mortality.64 These factors confirm the need for a systems
approach to error prevention.
Effect of case-mix
The greater risk of clinical errors in emergency admissions
is especially alarming because of the increasing demand for
emergency care.65,66 In the USA, emergency admissions
constitute more than a third of all hospital admissions,
41% of admissions of children, and 55% of admissions of
patients older than 80 years. 54% have at least one
comorbid condition and a third have two or more.67 Age
and comorbid disease add to the risk of adverse events68
either from limited physiological reserve or because of
exposure to more interventions. In the UK, about 60% of
hospital admissions are emergencies, and the proportion
has been increasing yearly by 2·1% since 1989, when data
were first collated. For patients older than 65 years, the
annual increase is 3·3%.69
Changes in service provision
In many developed countries the need for emergency
services has been growing in parallel with pressures to
reduce length of stay and numbers of beds.70 For example,
day-cases account for an increasing proportion of elective
admissions. In the UK, the number of National Health
Service beds fell from 480 000 in 1948 to 190 000 in 1998,
throughput and occupancy have increased, and length of
stay has shortened.69 Many regions in the USA have
shortages of beds and face the need to replace old
hospitals, and increasingly crowded emergency depart-
972
For personal use. Only reproduce with permission from The Lancet publishing Group.
ments have a statutory obligation to provide care for the
46 million Americans who carry no health insurance.
These pressures can represent a substantial hindrance to
efforts to improve safety.
Poor integration of acute hospital services can also
contribute to error. Market forces in the USA promote
competition between neighbouring facilities, often
resulting in redundant services. In the UK, however, a
desire for convenient local access causes acute hospitals to
proliferate, preventing economies of scale when nearby
hospitals replicate specialist services, a tension reflected in
abrupt changes in governmental planning.71 This lack of
integration dilutes professional experience, including
operative procedures, and lowers competency and quality
of outcomes.
Working hours
To provide safe acute care means employing adequate
numbers of trained staff, which is increasingly difficult in
the European Union because the Working Time Directive,
from August, 2004, limits the working hours for trainee
doctors to 56 per week and to an average of 48 for all
employees by 2009. “Work” is defined as “required to be
present at the health centre” regardless of the activities
being undertaken. This ruling has important implications
for out-of-hours and emergency care72 since being asleep in
an on-call bedroom will still count as work. In the USA,
resident physicians in training have, since July, 2003, been
limited to an average of 80 h a week.73 Comparison of the
effects of the European and American regulations should
prove useful, in view of evidence that fatigue from long
hours of work impairs patient safety.74
Concerns have been expressed that these constraints
may adversely affect training and acquisition of skills.75
They will certainly strain services in countries that have a
physician shortage. It will not be possible to train enough
extra doctors in the UK to fill the gap created by the
European directive before 2015. Reliance on non-
physician care-givers may improve the delivery of
preventive medicine and enhance performance of specific
tasks but does not seem to improve acute care.76 Moreover,
the worldwide shortage of nurses77 restricts the large-scale
transfer of traditional physician duties. A more
constructive approach for maintenance of patients’ safety
in acute care would be to focus on models of team-working
rather than transferring work to other groups.
Discontinuities in patient care
Hospital medicine has become increasingly compart-
mentalised, with blurred borders of responsibility and
multiple transitions (“hand offs”) between different health
care providers and teams.78 The benefits of reduced fatigue
from restricted hours of work for medical staff may be
offset by the discontinuities in care and personal isolation
caused by shift-working, the greater risk of communication
failures, and constraints on the delivery of clinical
education. A clinical vignette illustrates the problems
(panel).
This patient did not have just one illness and nor did she
have a distinct medical error that resulted in a well-defined
adverse event. She had several comorbidities, was exposed
to high-risk interventions with known side-effects, and was
treated in an acute care environment with insufficient
safeguards against clinical error. Potentially avoidable
adverse events were caused by faulty actions and inactions,
suboptimal training, poor supervision, and inadequate
service provision at local and national levels. However,
establishing a root-cause would be difficult because of the
varied, diffuse, and overlapping failures.
THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com

A 56-year-old woman was admitted on a Saturday evening to
a university teaching hospital with progressive malaise,
diffuse abdominal pain, and diarrhoea 1 week after
chemotherapy for breast cancer. 10 years previously she had
had cancer in the other breast and had undergone
mastectomy with radiotherapy. She was on warfarin for an
axillary vein thrombosis caused by a Hickman catheter and
had a history of penicillin allergy. She was apyrexial,
tachypnoeic, hypotensive, and oliguric; total leucocyte count
of less than 0·1103/L3, platelets 70103/L3, serum
creatinine 170 mol/L, and international normalised ratio
6·2. Blood culture produced gram-negative rods within 12 h.
She was treated by the resident medical staff in the
emergency unit with piperacillin and gentamicin, inadequate
intravenous fluids, parenteral non-steroidal anti-inflammatory
analgesics, and diuretics. She had a head injury after falling
unobserved from the commode where she had been placed
by the nursing assistants, who then sat her in a chair despite
a documented systolic blood pressure of 70 mm Hg. There
were only two trained nurses on duty for this 28-bed ward.
She was eventually seen by a consultant 24 h after
admission, and was referred to intensive care in neutropenic
septic shock with multiple organ failure.
Because no bed was immediately available in the intensive
care unit, mechanical ventilation and cardiovascular support
were started in the ward. Preparations for transfer to another
hospital were cancelled when an appropriate bed did become
free. Shortly after admission she developed a transient rash
that was attributed to piperacillin, already discontinued. A CT
scan showed a small subdural haematoma that did not need
intervention other than correction of the coagulopathy. Renal
replacement treatment was needed. During her 4 weeks in
intensive care, she developed a ventilator-associated
pneumonia and became colonised with meticillin-resistant
Staphylococcus aureus. After being discharged to the ward,
she developed bacteraemia from a central venous catheter
that had been left in place several days longer than intended
and needed a brief re-admission to intensive care.
She returned home 2 months later, weak but able to look
after herself. Her husband confided that he was surprised by
the contrast between the quality of care in the intensive care
unit compared with that on the ward.
Vincent and colleagues79,80 have developed a structured
approach for analysis of such episodes, combining the
factors that affect clinical practice with Reason’s81
organisational model of error. We have adapted this
model, bringing in elements from the quality literature
(table), as a method for assessment safe care of the acutely
ill patient.
Responding to the challenge
Our perspective is that safety and reliability are the most
important components of quality in health care and that
services need extensive restructuring to achieve an
acceptable level of these characteristics of high quality
care. The restructuring demands a systems approach. In
the panel, we identify deficiencies in resources and
organisation, processes and delivery of care, monitoring,
clinical competence, communication, continuity of care,
therapeutic interventions including prescribing, leadership
and governance, and patient empowerment. We classify
these deficiencies as occurring at the four levels of patient,
microsystem, organisation, and environment, a
classification that allows consideration of quality and
safety issues from different perspectives. What follows is a
THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
INPATIENT SAFETY I
summary of some of these issues. The next four articles in
this Lancet series will go into more detail, looking at the
monitoring of and improvements in safety for the acute
inpatient, both from a local and a systems standpoint.
Resources and organisation
Emergency departments
Safe emergency care needs prompt access to initial
treatment for life-threatening emergencies, and hospitals
cannot provide that if their emergency departments are
overloaded by medical problems that could have been
managed by a family practitioner or a pharmacist. Patients
cannot obtain prompt treatment without either adequate
local facilities or an efficient ambulance service. In large
hospitals, parallel provision of primary and secondary care
within the emergency department improves efficiency and
focuses resources on sicker patients. Small and rural acute
hospitals will need local solutions to professional staffing
that include shared management of the emergency
department by community practitioners, consultation
available via telemedicine links, and appropriately staffed
rapid transport to take more seriously ill patients to better-
equipped centres. This hub-and-spoke integration is well-
accepted for paediatric emergency care.82,83 Appropriate
investment in training and career structures will also be
needed to encourage practitioners—medical and non-
medical—to develop team-based approaches to
emergency care as a method for reducing errors and
improving quality.84,85
Models for delivering acute care
Future clinicians who provide core services in the acute
hospital setting will need new skill sets, which will be
transdisciplinary and especially suited to the range of
health care needs of hospital patients. Physicians, nurses,
pharmacists, and other health-care providers will develop
team-based models of care that avoid gaps in knowledge
and services. These gaps constitute an important risk for
patient safety.86 Essential skill sets will draw on
anaesthesia, internal medicine, surgery, accident and
emergency medicine, basic sciences, and ethics whereas
behavioural competencies will learn from aviation and
crew resource management.86 Critical care medicine
illustrates this development. Intensivists (ie, intensive care
specialists) have become the general practitioners of acute
care. They have adopted multidisciplinary collaboration,
contributed to systems management within intensive care,
and emphasised intervention at the first signs of clinical
instability. Integrative models with well-organised systems
within intensive care units have reduced mortality,87 and
the appointment of intensivists is a key recommendation
of the Leapfrog Group in the USA for improving hospital
safety.88
In the USA, acute hospital care has been developed into
a professional track termed hospital medicine practised by
the “hospitalist”.89 Australian hospitals have extended
intensive care services and personnel into hospital wards
by creating medical emergency teams.90 The UK has
outreach care provided by critical-care-trained nurses for
patients admitted to the wards.91 Such outreach detects
early signs of clinical deterioration and improves
communication between ward admitting teams and staff
in intensive care units.92,93 Although there are no
comparative data to favour one model over another,
studies suggest that early intervention might reduce
cardiac arrest rates,63,94 though there is also the risk that
the point of death could merely be shifted to the ICU.63
Better integration should also extend into post-hospital
recovery. We are only just beginning to identify the safety
973
INPATIENT SAFETY I

Resources and organisational
structures
Processes and delivery of care
Information management and
communication
Monitoring and analysis tools
Competence, training,
education, and behaviour
Governance
PSDA=patient self-determination act.
Conceptual framework for evaluating safe care of the acutely ill patient
Level
Patient Microsystem Organisation Environment
Equity Outreach, medical emergency Evidence-based resource Patients’ safety agencies
Access teams, hospitalists management Clinical networks
Patient safety officer at Interhospital transport
hospital board level systems
Health-care funding
Competence Transdisciplinary team-working Mapping the patient Primary and secondary care
Compassion Policies, protocols and journey linkages
Timeliness procedures Admission, Best practice guidelines
Computerised prescribing discharge, and transfers
Infection control
Electronic patient records Continuity of care and information Responsiveness National framework
including prescribing transfer Intranet access Workplace-based access linking safety, clinical
(computerised patient to information practice and training
order entry, CPOE) Collaborative planning
Severity of illness and Audit and PSDA cycles, including Adverse event monitoring Benchmarking
early warning systems errors of omission Consumer surveys Target indicators
Questionnaires, feedback Adverse event reporting and Staff sickness and Serious adverse
Patient-centred outcomes near-misses turnover rates event databanks
Failure mode and effects Observational databases,
analysis standardised mortality ratios
Root cause analysis Control charts
Advance directives Workplace-based training and Investment in and Integrated transdisciplinary
Public education in assessment of knowledge, skills, appointment of appropriately competency-based training
healthcare outcomes attitudes, and behaviour skilled staff at undergraduate and
postgraduate level
Empowerment through Local leadership and responsibility Culture of safety and Professional self-regulation
education, participation for individual patient outcomes learning Continuing professional
and transparency Collaborative management development
in the front-line Hospital accreditation

issues of post-discharge care,45 though we know that the
physical consequences of critical illness persist.95
Processes and delivery of care
Providing evidence-based interventions in a coordinated
and prompt manner to patients with complex health
problems defies the human abilities of inpatient
practitioners. With over 30 000 new publications entering
the MEDLINE database each month, clinicians need
knowledge pathfinders to assist their decision-making. Safe
care of patients will need greater reliance on clinical
pathways, clinical practice guidelines, and decision support
tools. No longer “physician-centric”, transdisciplinary
teams will undertake their clinical responsibilities in an
integrated manner using guidelines and protocols. A team
approach using treatment algorithms has achieved more
rapid weaning of ventilator-dependent patients compared
with physician-dependent approaches.96 Clinical practice
guidelines, adapted to the local clinical setting, increase the
likelihood of desired health-care outcomes.97
Information management and communication
Failures of communication between health-care providers
and between clinicians and patients are common causes of
error and litigation. Training in communication skills will
be especially important for multidisciplinary teams
functioning in rapid-paced inpatient settings. The electronic
medical record and computerised physician order entry
(prescription system) offer opportunities to provide
clinicians with an electronic platform that embeds decision
support and knowledge resources to promote best practice.
Computerised prescribing should improve patient safety by
reducing documentation and by warning of potential drug
interactions, contraindications, dosing adjustments, and
allergies. Evidence is emerging for benefits from error
reduction and improved decision making,98,99 and computer-
assisted record keeping and prescribing are among the
interventions recommended by the Leapfrog Group.88
Information technology does need staged “bottom-up”
development, pilot testing, and appropriate implemen-
tation into existing hospital cultures. Greater dependency
on computerised systems creates new safety issues when
the system fails.100 However, health services for the most
part still depend on the oral transmission of information
supported by handwritten records. It does seem logical to
build on the progress with computerised prescribing, and
develop an electronic patient record with modules for
investigations, diagnosis, interventions, and outcomes.
Monitoring and analysis
Early warning systems
Traditional monitoring in the acute hospital often
identifies adverse events only after they occur. Reactive
systems, based on instructions to call a doctor if there are
major changes in vital signs, should be replaced by
proactive monitoring to identify early changes and
empower ward staff to call for help and initiate further
investigation to prevent or limit the magnitude of adverse
events.92,93,101
Adverse event and error monitoring
Adverse event reporting is essential for improving patient
safety but current methods are unsatisfactory.102 Major
events may be more reliably reported, but near-misses are
likely to be ignored, deferred, or forgotten in the
pressured environment of clinical work. Barriers include
cumbersome and non-standardised paper formats; a
litigious culture that deters open reporting and discussion;
difficulty with identifying and reporting errors of
omission; a long interval between intervention and
adverse outcome (for example, nosocomial infection);
deficiencies in information synthesis, analysis, and
feedback; and failure of institutions to address
improvements in processes of care. Whether incident
reporting should be process or outcomes based remains
unclear.

974 THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet publishing Group.

Hospitals need data collection systems that allow
providers to enter information anonymously and easily.
Intranet-based online systems with automated analysis
and reporting to directors of patient safety and hospital
executives should improve institutional responsiveness.103
Collation of institutional data by regulatory or
governmental agencies should make it easier to detect
patterns and recommend improvements. The US
University Health System Consortium has an internet-
based reporting system; the Joint Committee on
Accreditation of Healthcare Organisations encourages
voluntary reporting of sentinel events; and four agencies
within the US Department of Health and Human Services
are integrating their data systems on clinical errors or
adverse events. These developments will increase the
pressure for public disclosure of patient safety records but
such data must accurately reflect the quality of care and
the hospital’s case-mix. So far, no such data collection
systems exist.
Competency-based training
Many of our proposals need investment in new ways of
training health care personnel. Traditional teaching has
been specialty-based. Evidence of the trainee’s success in
making the transition to independent practitioner has
rested on the completion of training of the required
duration followed by a formal examination. This model
has not facilitated horizontal integration of trainees across
disciplines, and it has inhibited the sharing of knowledge
and skills, and impeded collaboration and team building.
The focus on knowledge assessment also produces a
barrier between the curriculum and desirable educational
outcomes. This is now starting to change with the
development of competency-based training.
Competency-based training replaces time-based
training. The process of defining areas of competence and
mapping it to the core curriculum promotes integration of
common elements across specialties. Patients’ safety will
become one such core competency. The UK has a
competency-based training programme for multidisci-
plinary intensive-care medicine,104 and proposals are being
developed for a similar approach across Europe. The US
Accreditation Council for Graduate Medical Education
has also restructured postgraduate medical education
around competency-based “outcomes”.
The same educational principles must be applied in the
undergraduate setting but undergraduate curricula
remain largely silent on patients’ safety.105 One method by
which undergraduates can learn the basic principles of
patients’ safety and team-working in acute care is by
provision of transdisciplinary peer-led tuition in
resuscitation and emergency care, training senior students
to act as tutors and role models in multidisciplinary
groups.106
Governance: taking responsibility for change
To achieve improvements in acute patient safety in the
current climate and across different health care systems
will take many years, but the process has started. Classen103
proposes four evolutionary stages: building awareness;
development of organisational learning; proactive
management of risk; and establishment and maintenance
of high reliability organisations.104 Progress through these
stages will need a multiplicity of approaches, from step-
change at a local level to strategic planning at national and
international levels. It will also need the establishment of
common goals and shared responsibilities as we build
partnerships between the public, practitioners, politicians,
administrators, insurers, and educators.
THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
INPATIENT SAFETY I
Health care providers cannot be passive in this process
but must lead it, if they are to show patients that they are
worthy of public trust, and they must lead by example and
from the front. This in turn requires commitment from
the public, from politicians and from administrators to
provide an infrastructure that facilitates safe care and to
support those responsible for delivering it.
Conflict of interest statement
None declared.
References
1 Kohn LT, Corrigan JM, Donaldson MS, eds. To err is human:
building a safer health system. Washington DC: National Academy
Press, 2000.
2 Medical Errors: the scope of the problem. Rockville, MD: Agency for
Healthcare Research and Quality. Available at http://www.ahrq.
gov/qual/errback.htm (accessed January, 2004).
3 Wilson RM, Runciman WB, Gibberd RW, Harrison BT, Newby l,
Hamilton JD. The quality in Australian health care study. Med J Aust
1995; 163: 458–71.
4 Vincent C, Neale G, Woloshynowych M. Adverse events in British
hospitals: preliminary retrospective record review. BMJ 2001; 322:
517–19.
5 National Audit Office 2000. http://www.nao.gov.uk/publications/
nao_reports/9900230 (accessed January, 2004).
6 Department of Health. An organisation with a memory. http://www.
doh.gov.uk/orgmemreport (accessed January, 2004).
7 Shaw C, Coles J. Reporting of adverse clinical incidents: international
views and experience. London: CASPE Research, 2001.
8 Baker GR, Norton P. Patient safety and healthcare error in the
Canadian healthcare system: a systematic review and analysis of
leading practices in Canada with reference to key initiatives
elsewhere—a report to Health Canada. http://www.hc-sc.gc.ca/
english/care/report/index.html (accessed January, 2004).
9 Shojania KG, Duncan BW, McDonald KM, et al, eds. Making health
care safer: a critical analysis of patient safety practices. Rockville, MD:
Agency for Healthcare and Research Quality, 2001.
10 Blendon RJ, DesRoches CM, Brodie M, et al. Views of practicing
physicians and the public on medical errors. N Engl J Med 2002; 347:
1933–40.
11 Reason J. Latent errors and system disasters. In: Reason J, ed. Human
error. Cambridge: Cambridge University Press, 1990.
12 Joint Commission on Accreditation of Healthcare Organizations
(JCAHO). Conducting root cause analysis in response to a sentinel
event. Illinois: JCAHO, 1996.
13 Leape LL. Error in medicine. JAMA 1994; 272: 1851–57.
14 Heinrich HW. Industrial accident prevention: a scientific approach.
New York and London: McGraw-Hill, 1941.
15 Cohn JN, Kowey PR, Whelton PK, Prisant LM. New guidelines for
potassium replacement in clinical practice: a contemporary review by
the national council on potassium in clinical practice. Arch Intern Med
2000; 160: 2429–36.
16 Shojania KS, Burton EC, McDonald KM, Goldman L. The
autopsy as an outcome and performance measure. Rockville, Agency
for Healthcare Research and Quality; 2002. Available at http://
www.ahcpr.gov/clinic/epcsums/autopsum.pdf (accessed January,
2004).
17 Hofer TP, Kerr EA. What is an error? Boston: Institute for Health
Improvement, 2000.
18 Leape LL. Institute of Medicine medical error figures not
exaggerated. JAMA 2000; 284: 95–97.
19 Brennan TA. The Institute of Medicine report on medical errors:
could it do harm? N Engl J Med 2000; 342: 1123–25.
20 McDonald CJ, Weiner M, Hui SL. Deaths due to medical errors are
exaggerated in Institute of Medicine report. JAMA 2000; 284: 93–95.
21 Hayward RA, Hofer TP. Estimating hospital deaths due to medical
errors: preventability is in the eye of the reviewer. JAMA 2001; 286:
415–20.
22 Perrow C. Normal accidents: living with high risk technologies.
Princeton: Princeton University Press, 1999.
23 Lee TH. A broader concept of medical errors. N Engl J Med 2002;
347: 1965–67.
24 McNutt RA, Abrams R, Aron DC, for the Patient Safety Committee.
Patient safety efforts should focus on medical errors. JAMA 2002;
287: 1997–2001.
25 Luck J, Peabody JW, Dresselhaus TR, Lee M, Glassman P. How well
does chart abstraction measure quality? a prospective comparison of
standardized patients with the medical record. Am J Med 2000; 108:
642–49.
975
INPATIENT SAFETY I
26 Hofer TP, Bernstein SJ, DeMonner S, Hayward RA. Discussion
between reviewers does not improve reliability of peer review of
hospital quality. Med Care 2000; 38: 152–61.
27 Camacho LA, Rubin HR. Assessment of the validity and reliability of
three systems of medical record screening for quality of care
assessment. Med Care 1998; 36: 748–51.
28 Busse DK, Wright DJ. Classification and analysis of incidents in
complex medical environments. Top Health Inf Manage 2000; 20:
1–11.
29 Shojania KG, Showstack J, Wachter RM. Assessing hospital quality:
a review for clinicians. Eff Clin Pract 2001; 4: 82–90.
30 Layde PM, Maas LA, Teret SP, et al. Patient safety efforts should
focus on medical injuries. JAMA 2002; 287: 1993–97.
31 Handler JA, Gillam M, Sanders AB, Klasco R. Defining, identifying,
and measuring error in emergency medicine. Acad Emerg Med 2000;
7: 1183–88.
32 http://www.resourcefulpatient.org/index (accessed January, 2004).
33 Sexton BJ, Thomas EJ, Helmreich RL. Error, stress and teamwork in
medicine and aviation: cross sectional surveys. BMJ 2000; 320:
745–49.
34 Belton KJ, Lewis SC, Payne S, Rawlins MD, Wood S. Attitudinal
survey of adverse drug reaction reporting by medical practitioners in
the United Kingdom. Br J Clin Pharmacol 1995; 39: 223–26.
35 Eland IA, Belton KJ, Van Grootheest AC, Meiners AP, Rawlins MD,
Stricker BH. Attitudinal surveys of voluntary reporting of adverse
drug reactions. J Clin Pharmacol 1999; 48: 623–27.
36 Pizzi L, Goldfarb NI, Nash DB. Crew Resource Management and its
Applications in Medicine. In: Shojania KG, Duncan BW,
McDonald KM, et al, eds. Making health care safer: a critical analysis
of patient safety practices. Available at http://www.ahcpr.gov/
clinic/ptsafety/chap44.htm (accessed January, 2004).
37 Pool R. Beyond engineering: how society shapes technology. New
York: Oxford University Press, 1997.
38 Reason J. Human error: models and management BMJ 2000; 320:
768–70.
39 Weick KE. Organizational culture as a source of high reliability.
Calif Manage Rev 1987; 29: 112–27.
40 Chassin MR, Galvin RW, for the National Roundtable on Health
Care Quality. The urgent need to improve health care quality:
Institute of Medicine National Roundtable on Health Care Quality.
JAMA 1998; 280: 1000–05.
41 Chassin MR. Is health care ready for six sigma quality? Milbank Q
1998; 76: 565–91.
42 Lagasse RS. Anesthesia safety: model or myth? a review of the
published literature and analysis of current original
data. Anesthesiology 2002; 97: 1609–17.
43 Adams J. Risk. London: UCL Press, 1995.
44 House of Lords. Debate on armed forces pension scheme.
http://www.parliament.the-stationery-office.co.uk/pa/ld199900/
ldhansard/pdvn/lds03/text/30520–03.htm (accessed January, 2004).
45 Forster AJ, Murff HJ, Peterson JF, Gandhi TK, Bates DW. The
incidence and severity of adverse events affecting patients after
discharge from the hospital. Ann Intern Med 2003; 138: 161–67.
46 Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events
and negligence in hospitalized patients. N Engl J Med1991; 324:
370–76.
47 Leape LL, Brennan TA, Laird N, et al. The nature of adverse events
in hospitalized patients. N Engl J Med 1991; 324: 377––84.
48 Wolff AM, Bourke J, Campbell IA, Leembruggen DW. Detecting and
reducing hospital adverse events: outcomes of the Wimmera clinical
risk management program. Med J Aust 2001; 174: 621–25.
49 Neale G, Woloshynowych M, Vincent C. Exploring the causes of
adverse events in NHS hospital practice. J R Soc Med 2001; 94:
322–30.
50 Department of Health. Shaping the future: long term planning for
hospitals and related services. Available at www.doh.gov.uk/
nationalbeds.htm (accessed January, 2004).
51 Bellomo R, Goldsmith D, Postoperative serious adverse events in a
teaching hospital: a prospective study. Med J Aust 2002; 176: 216–18.
52 Wu AW, Folkman S, McPhee SJ, Lo B. Do house officers learn from
their mistakes? JAMA 1991; 265: 2089–94.
53 Lesar TS, Briceland LL, Delcoure K, Parmalee CJ, Masta-Gornic V,
Pohl H. Medication prescribing errors in a teaching hospital. JAMA
1990; 263: 2329–34.
54 Wilson DG, McArtney RG, Newcombe RG, et al. Medication errors
in paediatric practice: insights from a continuous quality improvement
approach. Eur J Pediatr 1998; 157: 769–74.
55 Andrews LB, Stocking C, Krizek T, et al. An alternative strategy for
studying adverse events in medical care. Lancet 1997 349: 309–13.
56 Wright D, Mackenzie SJ, Buchan I, Cairns CS, Price LE. Critical
incidents in the intensive therapy unit. Lancet 338: 676–78.
57 Giraud T, Dhainaut JF, Vaxelaire JF, et al. Iatrogenic complications
976
For personal use. Only reproduce with permission from The Lancet publishing Group.
in adult intensive care units: a prospective two-center study.
Crit Care Med 1993; 21: 40–51.
58 Donchin Y, Gopher D, Olin M, et al. A look into the nature and
causes of human errors in the intensive care unit. Crit Care Med 1995;
23: 294–300.
59 Muckart DJ, Bhagwanjee S, Aitchison JM. Adverse events in a
surgical intensive care unit: a cause of increased mortality.
S Afr J Surg 1994; 32: 69–73.
60 Darchy B, Le Miere E, Figueredo B, Bavoux E, Domart Y. Iatrogenic
diseases as a reason for admission to the intensive care unit:
incidence, causes, and consequences. Arch Intern Med 1999; 159:
71–78.
61 McQuillan P, Pilkington S, Allan A, et al. Confidential inquiry into
quality of care before admission to intensive care. BMJ 1998; 316:
1853–58.
62 Schein RM, Hazday N, Pena M, Ruben BH, Sprung CL. Clinical
antecedents to in hospital cardiopulmonary arrest. Chest 1990; 98:
1388–92.
63 Buist MD, Moore GE, Bernard SA, Waxman BP, Anderson JN,
Nguyen TU. Effects of medical emergency team on reduction of
incidence of and mortality from unexpected cardiac arrests in
hospital: preliminary study. BMJ 2002; 324: 387–90.
64 Goldfrad C, Rowan K. Consequences of discharges from intensive
care at night. Lancet 2000; 355: 1138–42.
65 World Health Organisation report. Emergency medical services.
Available at http://www.who.dk/emergservices (accessed January,
2004)
66 McCaig LF, Ly N. National hospital ambulatory medical care survey:
2000 emergency department summary—advance data from vital and
health statistics. Hyattsville, MD: National Center for Health
Statistics, 2002.
67 Elixhauser A, Yu K, Steiner C, Bierman AS. Hospitalization in the
United States, 1997. Rockville: Agency for Healthcare Research and
Quality, 2000.
68 Rothschild JM, Bates DW, Leape LL. Preventable medical injuries in
older patients. Arch Intern Med 2000; 160: 2717–28.
69 Department of Health. Shaping the future: long term planning for
hospitals and related services. http://www.doh.gov.uk/
nationalbeds.htm (accessed January, 2004).
70 Organisation for Economic Co-operation and Development health
data 2000: a comparative analysis of 29 OECD countries.
http://www.oecd.org (accessed January, 2004).
71 Keeping the NHS local: a new direction of travel. http://www.doh.
gov.uk/configuringhospitals(accessed January, 2004).
72 Guidance on implementing the European working time directive for
doctors in training. http://www.doh.gov.uk/workingtime/
ewtdguidance.htm (accessed January, 2004).
73 Philibert I, Friedmann P, Williams WT. New requirements for
resident duty hours. JAMA 2002; 288: 1112–14.
74 Gaba DM, Howard SK. Fatigue among clinicians and the safety of
patients. N Engl J Med 2002; 347: 1249–55.
75 Steinbrook R. The debate over residents’ work hours.
N Engl J Med 2002; 347: 1296–1302.
76 Druss BG, Marcus SC, Olfson M, Tanielian T, Pincus HA. Trends
in care by nonphysician clinicians in the united states. N Engl J Med
2003; 348: 130–37.
77 Buchan J. Global nursing shortages. BMJ 2002; 324: 751–52.
78 Richard I Cook RI, Render M, Woods DD. Gaps in the continuity of
care and progress on patient safety. BMJ 2000; 320: 791–94.
79 Vincent C, Taylor-Adams S, Chapman EJ, et al. How to investigate
and analyse clinical incidents: clinical risk unit and association of
litigation and risk management protocol. BMJ 2000; 320: 777–81.
80 Vincent C. Understanding and responding to adverse events.
N Engl J Med 2003; 348: 1051–56.
81 Reason JT. Understanding adverse events: human factors. In:
Vincent CA, ed. Clinical risk management. London: BMJ, 1995:
31–54.
82 Pearson G, Shann F, Barry P, et al. Should paediatric intensive care
be centralised? Trent versus Victoria. Lancet 1997; 349: 1213–17.
83 National Coordinating Group on Paediatric Intensive Care. Report to
the chief executive of the NHS executive: paediatric intensive care—a
framework for the future. www.doh.gov.uk/pub/docs/doh/
paedsum.pdf (accessed January, 2004).
84 Risser DT, Rice MM, Salisbury ML, Simon R, Jay GD, Berns SD,
for the MedTeams Research Consortium. The potential for improved
teamwork to reduce medical errors in the emergency department.
Ann Emerg Med 1999; 34 373–83.
85 Morey JC, Simon R, Jay GD, et al. Error reduction and performance
improvement in the emergency department through formal teamwork
training: evaluation results of the MedTeams project.
Health Serv Res 2002; 37: 1553–81.
86 Pizzi L, Goldfarb NI, Nash DB. Crew resource management and its
THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com

applications in medicine. http://www.ahcpr.gov/clinic/ptsafety/
chap44.htm (accessed January, 2004).
87 Pronovost PJ, Jenckes MW, Dorman T, et al. Organizational
characteristics of intensive care units related to outcomes of
abdominal aortic surgery. JAMA 1999; 281: 1310–17.
88 Leapfrog patient safety standards. www.leapfroggroup.org (accessed
January, 2004).
89 Wachter RM, Goldman L. The emerging role of “hospitalists” in the
American health care system. N Engl J Med 1996; 335: 514–17.
90 Lee A, Bishop G, Hillman KM, Daffurn K. The medical emergency
team. Anaesth Intensive Care 1995; 23: 183–86.
91 O’Dea J, Pepperman M, Bion J. Comprehensive critical care: a
national strategic framework in all but name. Intensive Care Med 2003;
29: 1594.
92 Goldhill DR, Worthington L, Mulcahy A, Tarling M, Sumner A. The
patient-at risk team: identifying and managing seriously ill ward
patients. Anaesthesia 1999; 54: 853–60.
93 Stenhouse C, Coates S, Tivey M, Allsop P, Parker T. Prospective
evaluation of a modified Early Warning Score to aid earlier detection
of patients developing critical illness on a surgical ward. Br J Anaesth
2000; 84: 663.
94 Bellomo R, Goldsmith D, Uchino S, et al. A prospective before-
and-after trial of a third medical emergency team. MJA 2003; 179:
1–3.
95 Herridge MS, Cheung AM, Tansey CM, et al, for the Canadian
Critical Care Trials Group. One-year outcomes in survivors of the
acute respiratory distress syndrome. N Engl J Med 2003; 348:
683–93.
96 Ely EW, Meade MO, Haponik EF, et al. Ventilator weaning protocols
THE LANCET • Vol 363 • March 20, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet publishing Group.
INPATIENT SAFETY I
driven by nonphysician health-care professionals: evidence-based
clinical practice guidelines. Chest 2001; 120 (suppl): 454–63.
97 Grimshaw JM, Shirran L, Thomas R, et al. Changing provider
behavior: an overview of systematic reviews of interventions. Med Care
2001; 39 (suppl): II2–45.
98 Bates DW, Teich JM, Lee J, et al. The impact of computerised
physician order entry on medication error prevention.
J Am Med Inform Assoc 1999; 6: 313–21.
99 Dexter PR, Perkins S, Overhage JM, Maharry K, Kohler RB,
McDonald CJ. A computerized reminder system to increase the use of
preventive care for hospitalized patients. N Engl J Med 2001; 345:
965–70.
100 Kilbridge P. Computer crash: lessons from a system failure.
N Engl J Med 2003; 348: 881–82.
101 Buist MD, Jarmolowski E, Burton PR, Bernard SA, Waxman BP,
Anderson J. Recognising clinical instability in hospital patients before
cardiac arrest or unplanned admission to intensive care: a pilot study
in a tertiary-care hospital. Med J Aust 1999; 171: 22–25.
102 Leape L. Reporting of adverse events. N Engl J Med 2002; 347:
1633–38.
103 Classen DC, Kilbridge PM. The roles and responsibility of physicians
to improve patient safety within health care delivery systems.
Acad Med 2002; 77: 963–72.
104 Intercollegiate Board for Training in Intensive Care Medicine.
Available from http:// www.ics.ac.uk (accessed January, 2004).
105 Estrada CA. Medical errors must be discussed during medical
education. BMJ 2000; 321: 507.
106 Perkins GD, Hulme J, Shore HR, Bion JF. Basic life support training
for health care students. Resuscitation 1999; 41: 19–23.
977
 critical care review
Predicting Fluid Responsiveness in ICU
Patients*
A Critical Analysis of the Evidence
Frédéric Michard, MD, PhD; and Jean-Louis Teboul, MD, PhD

Study objective: To identify and critically review the published peer-reviewed, English-language
studies investigating predictive factors of fluid responsiveness in ICU patients.
Design: Studies were collected by doing a search in MEDLINE (from 1966) and scanning the
reference lists of the articles. Studies were selected according to the following criteria: volume
expansion performed in critically ill patients, patients classified in two groups (responders and
nonresponders) according to the effects of volume expansion on stroke volume or on cardiac
output, and comparison of responder and nonresponder patients’ characteristics before volume
expansion.
Results: Twelve studies were analyzed in which the parameters tested were as follows: (1) static
indicators of cardiac preload (right atrial pressure [RAP], pulmonary artery occlusion pressure
[PAOP], right ventricular end-diastolic volume [RVEDV], and left ventricular end-diastolic area
[LVEDA]); and (2) dynamic parameters (inspiratory decrease in RAP [⌬RAP], expiratory decrease
in arterial systolic pressure [⌬down], respiratory changes in pulse pressure [⌬PP], and respiratory
changes in aortic blood velocity [⌬Vpeak]). Before fluid infusion, RAP, PAOP, RVEDV, and
LVEDA were not significantly lower in responders than in nonresponders in three of five studies,
in seven of nine studies, in four of six studies, and in one of three studies, respectively. When a
significant difference was found, no threshold value could discriminate responders and nonre-
sponders. Before fluid infusion, ⌬RAP, ⌬down, ⌬PP, and ⌬Vpeak were significantly higher in
responders, and a threshold value predicted fluid responsiveness with high positive (77 to 95%)
and negative (81 to 100%) predictive values.
Conclusion: Dynamic parameters should be used preferentially to static parameters to predict
fluid responsiveness in ICU patients. (CHEST 2002; 121:2000 –2008)

Key words: arterial pressure; cardiac output; cardiac preload; fluid responsiveness; left ventricular end-diastolic area;
pulmonary artery occlusion pressure; right atrial pressure; right ventricular end-diastolic volume; stroke volume; volume
expansion

Abbreviations: ⌬down ⫽ expiratory decrease in arterial systolic pressure; LVEDA ⫽ left ventricular end-diastolic area;
PAOP ⫽ pulmonary artery occlusion pressure; PEEP ⫽ positive end-expiratory pressure; ⌬PP ⫽ respiratory changes in
arterial pulse pressure; RAP ⫽ right atrial pressure; ⌬RAP ⫽ inspiratory decrease in right atrial pressure;
RVEDV ⫽ right ventricular end-diastolic volume; ⌬Vpeak ⫽ respiratory changes in aortic peak velocity

V olume expansion is frequently used in critically ill

patients to improve hemodynamics. Because of
the positive relationship between ventricular end-
diastolic volume and stroke volume,1 the expected
hemodynamic response to volume expansion is an
*From the Medical ICU, CHU de Bicêtre, Assistance Publique-
Hôpitaux de Paris, Le Kremlin-Bicêtre, Université Paris XI,
France.
Manuscript received June 4, 2001; revision accepted December
19, 2001.
increase in right ventricular end-diastolic volume
(RVEDV), left ventricular end-diastolic volume,
stroke volume, and cardiac output. The increase in
end-diastolic volume as a result of fluid therapy
depends on the partitioning of the fluid into the
different cardiovascular compliances organized in
Correspondence to: Frédéric Michard, MD, PhD, Service de
Réanimation Médicale, CHU de Bicêtre, Université Paris XI, 78,
rue du Général Leclerc, 94275 Le Kremlin Bicêtre cedex, France;
e-mail: f.michard@wanadoo.fr

2000 Critical Care Reviews

series. The increase in stroke volume as a result of
end-diastolic volume increase depends on ventricu-
lar function since a decrease in ventricular contrac-
tility decreases the slope of the relationship between
end-diastolic volume and stroke volume.1 Therefore,
only 40 to 72% of critically ill patients have been
shown to respond to volume expansion by a signifi-
cant increase in stroke volume or cardiac output in
studies2–13 designed to examine fluid responsiveness.
This finding emphasizes the need for predictive
factors of fluid responsiveness in order to select
patients who might benefit from volume expansion
and to avoid ineffective or even deleterious volume
expansion (worsening in gas exchange, hemodilu-
tion) in nonresponder patients, in whom inotropic
and/or vasopressor support should preferentially be
used.
Bedside indicators of ventricular preload have
been proposed as predictors of fluid responsive-
ness.2– 4,6 –9,11–13 In this regard, a postal survey in
Germany showed that right atrial pressure (RAP)
and pulmonary artery occlusion pressure (PAOP) are
used by a majority of ICU physicians when deciding
to administer fluid,14 and several recommendations
support the use of cardiac filling pressures in order
to guide fluid therapy in critically ill patients.15,16
Other bedside indicators of ventricular preload,
namely RVEDV and left ventricular end-diastolic
area (LVEDA), have also been tested as predictors of
the hemodynamic effects of volume expansion in
critically ill patients.2– 4,6 –9,11,13
The respiratory changes in RAP, arterial pressure,
and aortic blood velocity, assumed to be dynamic
indicators of the sensitivity of the heart to changes in
preload induced by changes in pleural pressure, have
also been proposed to predict fluid responsiveness in
critically ill patients.5,9,10,12,13 Therefore, the aim of
the present study was to analyze the clinical studies
investigating predictive factors of fluid responsive-
ness in critically ill patients in order to assess the
value of each parameter tested.
Materials and Methods
Selection of Studies To Be Evaluated
We collected studies investigating the predictive factors of
fluid responsiveness in critically ill patients by doing a search in
MEDLINE (from 1966). Studies were selected according to the
following criteria: volume expansion performed in critically ill
patients, patients classified in two groups (responders and non-
responders) according to the effects of volume expansion on
stroke volume or on cardiac output, and comparison of responder
and nonresponder patients characteristics before volume expan-
sion. The reference lists of the selected articles were scanned for
additional studies. Of the 12 included studies,2–13 11 studies were
identified from the electronic database and 1 study was identified
from reference tracing.5 The main characteristics of these studies
are presented in Table 1.
Parameters Tested as Predictors of Fluid Responsiveness
Ten studies have investigated the value of ventricular
preload indicators in predicting fluid responsiveness. The
parameters tested were RAP in five studies,2– 4,8,12 PAOP in
nine studies,2– 4,6 –9,11,12 RVEDV in six studies,2– 4,6 – 8 and

Table 1—Main Characteristics of Clinical Studies Investigating the Predictive Factors of Fluid Responsiveness in
ICU Patients*

Patients, FC, Fluid Volume Speed of Definition of Rate of

Source No. No. Infused Infused, mL FC, min Response Response, % Parameters Tested

Calvin et al2 28 28 5% Alb 250 20–30 ⌬SV ⬎ 0% 71 RAP, PAOP, RVEDV

Schneider et al3 18 18 FFP 500 30 ⌬SV ⬎ 0% 72 RAP, PAOP, RVEDV
Reuse et al4 41 41 4.5% Alb 300 30 ⌬CO ⬎ 0% 63 RAP, PAOP, RVEDV
Magder et al5 33 33 9% NaCl 100–950 ⌬CO ⬎ 250 52 ⌬RAP
mL/min
Diebel et al6 15 22 R. lactate 300–500 ⌬CO ⬎ 10% 59 PAOP, RVEDV
Colloids 500
Diebel et al7 32 65 R. lactate 300–500 ⌬CO ⬎ 20% 40 PAOP, RVEDV
Wagner and 25 36 9% NaCl 938 ⫾ 480 7–120 ⌬SV ⬎ 10% 56 RAP, PAOP, RVEDV
Leatherman8 5% Alb, FFP 574 ⫾ 187
Tavernier et al9 15 35 HES 500 30 ⌬SV ⬎ 15% 60 PAOP, LVEDA, ⌬down
Magder and Lagonidis10 29 29 25% Alb 100 15 ⌬CO ⬎ 250 45 ⌬RAP
9% NaCl 150–400 mL/min
Tousignant et al11 40 40 HES 500 15 ⌬SV ⬎ 20% 40 PAOP, LVEDA
Michard et al12 40 40 HES 500 30 ⌬CO ⬎ 15% 40 RAP, PAOP, ⌬PP
Feissel et al13 19 19 HES 8 mL/kg 30 ⌬CO ⬎ 15% 53 LVEDA, ⌬Vpeak
Total 334 406 52
*FC ⫽ fluid challenge; Alb ⫽ serum albumin; FFP ⫽ fresh frozen plasma; NaCl ⫽ serum saline solution; R. lactate ⫽ Ringer’s lactate;
HES ⫽ hydroxyethylstarch; ⌬SV ⫽ volume expansion-induced changes in stroke volume; ⌬CO ⫽ volume expansion-induced changes in cardiac
output.

www.chestjournal.org CHEST / 121 / 6 / JUNE, 2002 2001

LVEDA in three studies9,11,13 (Table 1). In all studies, the RAP
and PAOP were measured at end-expiration without ventilator
disconnection or removal of positive end-expiratory pressure
(PEEP). In four studies,4,6 – 8 RVEDV was calculated from the
measurement of right ventricular ejection fraction and cardiac
output by using a fast-response thermistor pulmonary artery
catheter as follows: RVEDV ⫽ (cardiac output/heart rate)/
right ventricular ejection fraction. In two other studies,2,3
RVEDV was evaluated by cardiac scintigraphy. LVEDA was
measured by transesophageal echocardiography using the
transgastric short-axis view of the left ventricle.9,11,13
Five studies have investigated the value of dynamic parameters
in predicting fluid responsiveness. These parameters were the
inspiratory decrease in RAP (⌬RAP) in two studies,5,10 the
expiratory decrease in arterial systolic pressure (⌬down) in one
study,9 the respiratory changes in arterial pulse pressure (⌬PP) in
one study,12 and the respiratory changes in aortic blood velocity
(⌬Vpeak) in one study13 (Table 1). The ⌬RAP was calculated as
the difference between the expiratory and the inspiratory
RAP.5,10 The ⌬down was calculated as the difference between
the value of the systolic pressure during an end-expiratory pause
and the minimal value of systolic pressure over a single respira-
tory cycle.9 The ⌬PP was calculated as the difference between
the maximal and the minimal value of pulse pressure over a single
respiratory cycle, divided by the mean of the two values, and
expressed as a percentage.12 The ⌬Vpeak was calculated as the
difference between the maximal and minimal peak velocity of
aortic blood flow over a single respiratory cycle, divided by the
mean of the two values, and expressed as a percentage.13 Aortic
blood flow was measured by a pulsed-wave Doppler echocardio-
graphic beam at the level of the aortic valve.13
of volume expansion was based on criteria listed in
Table 2. Fluid administration was performed using
colloid solutions (albumin, fresh frozen plasma, or
hydroxyethylstarch) in 253 instances, and crystalloid
solutions (serum saline solution or Ringer’s lactate)
in 153 instances (Table 1). In nine studies, the volume
infused was predetermined and ranged from 250 to
500 mL for colloids and from 300 to 500 mL for
crystalloids (Table 1). In two studies,5,10 volume infu-
sion was performed until a rise in RAP ⱖ 2 mm Hg
was obtained; hence, the volume of serum saline
solution infused varied from 100 to 950 mL. In
another study,8 fluid was administered until a rise
in PAOP ⱖ 3 mm Hg was obtained. In this case,
the volume infused was 938 ⫾ 480 mL for serum
saline solution and 574 ⫾ 187 mL for 5% albumin
or fresh frozen plasma. The speeds of fluid infusion are
reported in Table 1. In all studies but one,9 hemo-
dynamic measurements were performed just before
and immediately at the end of fluid infusion.
The hemodynamic response to volume expansion
was defined by an increase in stroke volume in five
studies and in cardiac output in seven studies (Table
1). The values of stroke volume or cardiac output
increase used to define responder and nonresponder
patients are presented in Table 1.

Results
There were 406 fluid challenges in 334 patients
(Table 1). Most of the patients were septic (55%) and
receiving mechanical ventilation (84%). The decision
RAP
Before volume expansion, RAP was not signifi-
cantly lower in responders than in nonresponders in
three of five studies2,4,12 (Fig 1). The two remaining
studies3,8 reported a lower value of baseline RAP in

Table 2—Criteria Used to Decide Volume Expansion*

Source Criteria

Calvin et al2 Cardiac index ⬍ 3.5 L/min/m and PAOP ⬍ 12 mm Hg in septic and trauma patients
2

Cardiac index ⬍ 2.5 L/min/m2 and PAOP ⬍ 20 mm Hg in acutely ill patients with a defined cardiac cause
Schneider et al3 Systematic infusion in patients with septic shock
Reuse et al4 Systolic BP ⬍ 90 mm Hg or cardiac index ⬍ 2.5 L/min/m2 or heart rate ⬎ 120/min or decreased urine
output (⬍ 25 mL/h)
Magder et al5 Clinical impression that the cardiac output was inadequate for tissue needs and would respond to volume
loading
Diebel et al6 Oliguria (urine output ⬍ 30 mL/h) or hypotension or in an attempt to optimize oxygen delivery
Diebel et al7 In an attempt to increase oxygen delivery to ⬎ 600 mL/min/m2 or to reach a plateau in the oxygen
consumption-delivery relationship
Wagner and Leatherman8 One or more clinical conditions that suggested the possibility of inadequate preload: hypotension (⬍ 100
mm Hg), oliguria (⬍ 30 mL/h), tachycardia (⬎ 100/min), lactic acidosis, cool extremities, vasopressor
wean, azotemia
Tavernier et al9 Sepsis-induced hypotension (systolic BP ⬍ 90 mm Hg or its reduction by ⱖ 40 mm Hg from usual values)
Magder and Lagonidis10 As part of routine testing to assess cardiac filling status if PAOP ⱕ 18 mm Hg
Tousignant et al11 PAOP ⬍ 20 mm Hg or inotropic support or low urine output and adequate gas exchange
Michard et al12 Systolic BP ⬍ 90 mm Hg or the need of vasoactive drugs (dopamine ⬎ 5 ␮g/kg/min or norepinephrine)
and PAOP ⬍ 18 mm Hg and Pao2/Fio2 ⬎ 100 mm Hg
Feissel et al13 Systematic infusion in septic shock patients with preserved left ventricular systolic function and Pao2/Fio2
⬎ 100 mm Hg
*Fio2 ⫽ fraction of inspired oxygen.

2002 Critical Care Reviews

 Figure 1. Mean RAP before volume expansion in responders and nonresponders.

responders than in nonresponders (Fig 1), and a sponders (Table 3). In the first study,6 the mean
significant relationship between the baseline RAP value of PAOP was significantly higher in responder
(r2 ⫽ 0.20), and the increase in stroke volume in patients (14 ⫾ 7 mm Hg vs 7 ⫾ 2 mm Hg, p ⬍ 0,01).
response to volume expansion was reported by Wag- In contrast, the two other studies8,11 reported a
ner and Leatherman.8 However, the marked overlap significantly lower value of PAOP at baseline in
of individual RAP values did not allow the identifi- responders than in nonresponders (Table 3), and a
cation of a RAP threshold value discriminating re- significant relationship between the baseline PAOP
sponders and nonresponders before fluid was admin- (r2 ⫽ 0.33) and the increase in stroke volume in
istered. response to volume expansion was reported by Wag-
ner and Leatherman.8 However, in none of these
PAOP studies, a PAOP cutoff value was proposed to predict
the hemodynamic response to volume expansion
Before volume expansion, PAOP was not signifi- before fluid was administered.
cantly lower in responders than in nonresponders in
seven of nine studies2– 4,6,7,9,12 (Table 3). Three stud- RVEDV
ies6,8,11 reported a significant difference between the
baseline value of PAOP in responders and nonre- Before volume expansion, RVEDV index was not
significantly lower in responders than in nonre-
sponders in four of six studies2– 4,8 (Fig 2). In the two
Table 3—PAOP Before Volume Expansion in
remaining studies of Diebel et al,6,7 RVEDV index
Responders and Nonresponders* was significantly lower at baseline in responders than
in nonresponders (Fig 2), RVEDV index ⬍ 90
PAOP, mm Hg mL/m2 was associated with a high rate of response
Source Responders Nonresponders (100% and 64%, respectively), and RVEDV index
Calvin et al2 8⫾1 7⫾2
⬎ 138 mL/m2 was associated with the lack of re-
Schneider et al3 10 ⫾ 1 10 ⫾ 1 sponse to volume expansion. However, when the
Reuse et al4 10 ⫾ 4 10 ⫾ 3 RVEDV index ranged from 90 to 138 mL/m2, no
Diebel et al6 14 ⫾ 7 7 ⫾ 2† threshold value was proposed to discriminate re-
Diebel et al7 16 ⫾ 6 15 ⫾ 5 sponder and nonresponder patients before volume
Wagner and Leatherman8 10 ⫾ 3 14 ⫾ 4†
Tavernier et al9 10 ⫾ 4 12 ⫾ 3
expansion. Moreover, another study8 reported a
Tousignant et al11 12 ⫾ 3 16 ⫾ 3† positive response to volume expansion in four of nine
Michard et al12 10 ⫾ 3 11 ⫾ 2 patients with a RVEDV index ⬎ 138 mL/m2, a lack
*Values are expressed as mean ⫾ SD, except for the study of of response in three of nine patients despite a
Schneider et al3 (mean ⫾ SEM). RVEDV ⬍ 90 mL/m2, and a significant but weak
†p ⬍ 0.05 responders vs nonresponders. relationship between the baseline RVEDV index

www.chestjournal.org CHEST / 121 / 6 / JUNE, 2002 2003

 Figure 2. Mean RVEDV index before volume expansion in responders and nonresponders.
(r2 ⫽ 0.19) and the increase in stroke volume in
response to volume expansion.
LVEDA
In two studies,9,11 the LVEDA before volume
expansion was significantly lower in responders than
in nonresponders (Table 4). In the study of Tav-
ernier et al,9 a significant and negative relationship
(r2 ⫽ 0.4, p ⫽ 0.01) was also reported between the
baseline value of LVEDA index and the percentage
of increase in stroke volume in response to volume
expansion. However, using receiver operating char-
acteristic curve analysis, Tavernier et al9 demon-
strated minimal value of LVEDA index to discrimi-
nate responder and nonresponder patients. In the
study of Tousignant et al,11 a marked overlap of
baseline individual LVEDA values was observed so
that a given value of LVEDA could not be used to
predict the hemodynamic response to fluid infusion.
Table 4 —LVEDA Before Volume Expansion in
Responders and Nonresponders*
LVEDA, cm2/m2
Source Responders Nonresponders
Tavernier et al9 9⫾3 12 ⫾ 4†
Tousignant et al11 15 ⫾ 5‡ 20 ⫾ 5†‡
Feissel et al13 10 ⫾ 4 10 ⫾ 2
*Data are presented as mean ⫾ SD.
†p ⬍ 0.05 responders vs nonresponders.
‡Area expressed in centimeters squared.
2004
Moreover, in another study,13 responder and nonre-
sponder patients were not different with regard to
the baseline value of LVEDA index (10 ⫾ 4 cm2/m2
vs 10 ⫾ 2 cm2/m2), and no significant relationship
(r2 ⫽ 0.11, p ⫽ 0.17) was observed between the
baseline value of LVEDA index and the percentage
of increase in cardiac index in response to volume
expansion.
⌬RAP
In patients with spontaneous breathing activity,
two studies from Magder et al5,10 demonstrated that
an inspiratory decrease in RAP ⱖ 1 mm Hg pre-
dicted a positive response to volume expansion, with
positive predictive values of 77% and 84% and
negative predictive values of 81% and 93% (Table 5).
⌬down
In sedated patients receiving mechanical ventila-
tion with sepsis-induced hypotension, one study9
demonstrated that the ⌬down was significantly
greater (11 ⫾ 4 mm Hg vs 4 ⫾ 2 mm Hg,
p ⫽ 0.0001) in responders than in nonresponders,
and that the ⌬down threshold value of 5 mm Hg was
able to discriminate responders and nonresponders
with a positive predictive value of 95% and a nega-
tive predictive value of 93% (Table 5). Moreover,
this study9 reported a positive and good relationship
(r2 ⫽ 0.58, p ⫽ 0.001) between the baseline value of
⌬down and the percentage of increase in stroke
volume in response to volume expansion.
Critical Care Reviews
 Table 5—Positive and Negative Predictive Values of Dynamic Parameters
Patients, Parameters
Source No. Tested
Magder et al5 33 ⌬RAP
Tavernier et al9 35 ⌬down
Magder and Lagonidis10 29 ⌬RAP
Michard et al12 40 ⌬PP
Feissel et al13 19 ⌬Vpeak
⌬PP
In sedated patients receiving mechanical ventila-
tion with acute circulatory failure related to sepsis,
one study12 demonstrated that ⌬PP was significantly
greater (24 ⫾ 9% vs 7 ⫾ 3%, p ⬍ 0.001) in respond-
ers than in nonresponders, and that a ⌬PP threshold
value of 13% allowed discrimination between re-
sponder and nonresponder patients with a positive
predictive value of 94% and a negative predictive
value of 96% (Table 5). Moreover, in this study,12 the
value of ⌬PP before fluid administration was signif-
icantly and closely correlated (r2 ⫽ 0.85, p ⬍ 0.001)
with the volume expansion-induced changes in car-
diac output, such that the higher ⌬PP at baseline, the
greater was the increase in cardiac output in re-
sponse to fluid infusion.
⌬Vpeak
In sedated patients receiving mechanical ventila-
tion with septic shock, one study13 demonstrated that
⌬Vpeak was significantly greater (20 ⫾ 6% vs
10 ⫾ 3%, p ⬍ 0.01) in responder patients than in
nonresponder patients, and that a ⌬Vpeak threshold
value of 12% allowed discrimination between re-
sponder and nonresponder patients with a positive
predictive value of 91% and a negative predictive
value of 100% (Table 5). Moreover, a positive and
tight linear correlation (r2 ⫽ 0.83, p ⬍ 0.001) was
found between the ⌬Vpeak before volume expansion
and the volume expansion-induced changes in car-
diac output.
Discussion
The present analysis emphasizes the minimal clin-
ical value of ventricular preload indicators and the
higher value of dynamic parameters (testing the
cardiovascular response to respiratory changes in
pleural pressure) in predicting fluid responsiveness
in critically ill patients. It has been suggested that a
beneficial hemodynamic effect of volume expansion
cannot be expected in critically ill patients with a
www.chestjournal.org
Positive Negative
Best Threshold Predictive Predictive
Value Value, % Value, %
1 mm Hg 84 93
5 mm Hg 95 93
1 mm Hg 77 81
13% 94 96
12% 91 100
RAP ⬎ 12 mm Hg17 and/or a PAOP ⬎ 12 mm Hg or
⬎ 15 mm Hg.15,18 In this regard, RAP and PAOP
have been reported to be lower in responders than in
nonresponder patients in two studies (Fig 1, Table
3). Moreover, a significant relationship between the
increase in stroke volume in response to volume
expansion and the baseline RAP (r2 ⫽ 0.20) or the
baseline PAOP (r2 ⫽ 0.33) was reported by Wagner
and Leatherman,8 suggesting that the lower RAP or
PAOP before volume expansion, the greater the
increase in stroke volume in response to fluid infu-
sion. However, although statistically significant,
these relationships were weak because a given value
of RAP or of PAOP could not be used to discriminate
responders and nonresponders before fluid was ad-
ministered. Moreover, in all other clinical studies
(Fig 1, Table 3), no difference between responder
and nonresponder patients was observed with regard
to the baseline value of RAP and of PAOP, and no
relationship was reported between cardiac filling
pressures before volume expansion and the hemo-
dynamic response to volume expansion. Finally, it
must be noted that fluid infusion has been shown
to significantly increase cardiac output in some
critically ill patients with central venous pressures
⬎ 15 mm Hg.19
Two studies of Diebel et al6,7 reported a lower
value of RVEDV index in responder than in non-
responder patients, and suggested that a beneficial
hemodynamic effect of volume expansion was likely
(rate of response 100% and 64%) when the RVEDV
index was below 90 mL/m2 and very unlikely (rate
of response of 0%) when the RVEDV index was
⬎ 138 mL/m2. However, when the RVEDV index
ranged from 90 to 138 mL/m2, no cutoff value could
be proposed to discriminate responder and nonre-
sponder patients. Moreover, Wagner and Leatherman8
reported positive responses to volume expansion in
patients with a RVEDV index ⬎ 138 mL/m2, and
the lack of response in patients with a RVEDV index
⬍ 90 mL/m2. Finally, in four of six studies investigating
whether RVEDV could predict fluid responsiveness,
no significant difference was observed between re-
CHEST / 121 / 6 / JUNE, 2002 2005
sponders and nonresponders with regard to the base-
line value of RVEDV index (Fig 2).
The echocardiographic measurement of LVEDA
has been shown to reflect more accurately the left
ventricular preload when compared with PAOP,20
and to improve the ability to detect changes in left
ventricular function caused by acute blood loss.21 In
nine anesthetized mongrel dogs, Swenson et al22
reported a significant relationship between baseline
LVEDA and changes in cardiac output induced by
IV fluid therapy, suggesting that LVEDA could be an
indicator of fluid responsiveness. In this regard,
LVEDA was found to be significantly lower in
responders than in nonresponders in two clinical
studies,9,11 and a significant relationship between the
baseline LVEDA index and the changes in stroke
volume induced by volume expansion has also been
reported.9 However, using receiver operating char-
acteristic curve analysis, Tavernier et al9 demon-
strated in patients with sepsis-induced hypotension
the minimal value of a given LVEDA index value to
discriminate responders and nonresponders before
fluid was administered. Moreover, in the study of
Tousignant et al,11 including medical-surgical ICU
patients, considerable overlap of baseline individual
values of LVEDA was observed between responders
and nonresponders, supporting the interpretation
that a specific LVEDA value cannot reliably predict
fluid responsiveness in an individual patient. Re-
cently, in patients with septic shock, Feissel et al13
did not observe any difference between the mean
baseline value of LVEDA index in responders and
nonresponders, neither any relationship between the
baseline value of LVEDA index and the percentage
of change in cardiac index in response to volume
expansion.
Therefore, all clinical studies have emphasized the
lack of value of ventricular preload indicators as
predictors of fluid responsiveness in critically ill
patients. Methodologic and physiologic reasons
could be advanced to explain these findings. First,
RAP, PAOP, RVEDV, and LVEDA are not always
accurate indicators of ventricular preload. Indeed,
RAP and PAOP have been shown to overestimate
transmural pressures in patients with external23 or
intrinsic24 PEEP. The PAOP is highly dependent on
left ventricular compliance,25 which is frequently
decreased in ICU patients (sepsis, ischemic, or hy-
pertrophic cardiopathy). Because it is the transmural
pressures and not intracavitary pressures such as
RAP and PAOP that are related to end-diastolic
volumes via the chamber compliance, it is not sur-
prising that those surrogates bear little relationship
to fluid responsiveness. The evaluation of RVEDV
by thermodilution has been shown influenced by
tricuspid regurgitation,26 which is frequently en-
2006
countered in patients with pulmonary hypertension
(ARDS, mechanical ventilation with PEEP). The
estimation of the LVEDA by echocardiography does
not always accurately reflect left ventricular end-
diastolic volume27 and hence LV preload. Second, in
case of right ventricular dysfunction, a beneficial
hemodynamic effect of volume expansion cannot be
expected, even in the case of low left ventricular
preload.28 Third, knowing the preinfusion end-
diastolic volume tells little about the diastolic cham-
ber compliance. In this regard, hypovolemia can be
associated with a normal or high LVEDA value in
patients with dilated cardiopathy. Finally, two mat-
ters must be stressed: (1) the increase in end-
diastolic volume as a result of fluid therapy depends
on the partitioning of the fluid into the different
cardiovascular compliances organized in series, and
(2) the rise in stroke volume as a result of end-
diastolic volume increase depends on ventricular
function since a decrease in ventricular contractility
decreases the slope of the relationship between
end-diastolic volume and stroke volume.1 Therefore,
a patient can be nonresponder to a fluid challenge
because of high venous compliance, low ventricular
compliance and/or ventricular dysfunction. In this
regard, it is not so surprising that bedside indicators
of cardiac chambers dimensions are not accurate
predictors of fluid responsiveness in ICU patients in
whom venous capacitance, ventricular compliance,
and contractility are frequently altered.
Assuming that respiratory changes in pleural pres-
sure induce greater changes in RAP when the right
ventricle is highly compliant than when it is poorly
compliant, Magder et al investigated whether the
inspiratory decrease in RAP could be used to predict
fluid responsiveness.5,10 Two studies5,10 demon-
strated that a positive response to volume expansion
was very likely in patients with an inspiratory de-
crease in RAP ⱖ 1 mm Hg, while it was unlikely if
the inspiratory decrease in RAP was ⬍ 1 mm Hg.
Unfortunately, most of ICU patients with acute
circulatory failure are sedated and receiving mechan-
ical ventilation, thus are unable to produce an in-
spiratory decrease in pleural pressure sufficient to
decrease the RAP.10 In this condition, analysis of the
respiratory changes in left ventricular stroke volume
has been proposed to predict fluid responsiveness.
Indeed, by decreasing the venous return pressure
gradient, mechanical insufflation may decrease the
right ventricular filling,29 and consequently the right
ventricular output if the right ventricle is sensitive to
changes in preload. In this condition, the following
decrease in left ventricular filling may also induce a
significant decrease in left ventricular output if the
left ventricle is sensitive to changes in preload.
Therefore, the magnitude of the respiratory changes
Critical Care Reviews
in left ventricular stroke volume, which reflects the
sensitivity of the heart to changes in preload induced
by mechanical insufflation, has been proposed as a
predictor of fluid responsiveness. Because the arte-
rial pulse pressure (systolic minus diastolic pressure)
is directly proportional to left ventricular stroke
volume,30 the respiratory changes in left ventricular
stroke volume have been shown reflected by changes
in pulse pressure.31 Accordingly, the respiratory
changes in pulse pressure have been shown to
accurately predict fluid responsiveness in patients
receiving mechanical ventilation with acute circula-
tory failure related to sepsis.12 The analysis of the
respiratory changes in systolic pressure has also been
proposed to assess fluid responsiveness. However,
the systolic pressure variation induced by mechanical
ventilation results not only from changes in aortic
transmural pressure (mainly related to changes in
left ventricular stroke volume), but also from
changes in extramural pressure (ie, from changes in
pleural pressure).32,33 Therefore, the systolic pres-
sure variation is a less specific indicator of changes in
left ventricular stroke volume and hence a less
accurate predictor of fluid responsiveness than the
pulse pressure variation.12 In this regard, it has been
proposed to discriminate the inspiratory increase in
systolic pressure (not necessarily due to a change in
left ventricular stroke volume) from the ⌬down,
which in contrast necessarily reflects a change in left
ventricular stroke volume.34 Experimental and clin-
ical studies34,35 have emphasized the influence of
volume status on ⌬down (hemorrhage increases
⌬down, while volume expansion decreases ⌬down),
and Tavernier et al9 demonstrated that ⌬down is an
accurate predictor of fluid responsiveness in septic
patients with hypotension.
The analysis of the arterial pressure waveform is
not possible in patients with cardiac arrhythmias.36
Indeed, in this condition, the changes in arterial
pressure do not reflect the effects of mechanical
insufflation on left ventricular stroke volume. It must
be emphasized that the evaluation of ⌬down and of
⌬PP requires invasive arterial pressure catheteriza-
tion. However, in shock states, estimation of BP
using a cuff is commonly inaccurate, and use of an
arterial cannula provides a more appropriate and
reproducible measurement of arterial pressure.15
Interestingly, Feissel et al13 have recently demon-
strated that Doppler echocardiographic imaging of
aortic blood velocity could be used to assess nonin-
vasively the respiratory changes in aortic blood ve-
locity and to predict fluid responsiveness in patients
with septic shock. It must be noted that ⌬down,
⌬PP, and ⌬Vpeak have been shown to be accurate
predictors of fluid responsiveness in sedated patients
receiving mechanical ventilation with sepsis.
www.chestjournal.org
Whether they also predict fluid responsiveness in
nonsedated, spontaneously breathing patients with-
out sepsis remains to be determined.
It must be emphasized that various types and
volumes of fluid, speeds of fluid infusion, and defi-
nitions of responders to volume expansion have been
used in the studies analyzed (Table 1). This may have
a significant influence on the results and conclusions
of the studies. Indeed, the hemodynamic effects of
an hypertonic colloid infusion are expected to be
more dramatic than those of an equal volume of
isotonic crystalloid infusion. Because of intravascu-
lar-extravascular equilibration, the speed of volume
infusion should also greatly influence the hemody-
namic response, particularly in septic patients with
systemic capillary leakness. Moreover, because of
different definitions of responders from one study to
another, some patients considered as responders in
some studies, would have been considered as non-
responders in other studies. Unfortunately, because
individual data were not available in all but one
study, a comparison of the predictive value of each
parameter using the same definition of responders
was not possible. Finally, the predictive value of
dynamic parameters has been tested by only few
studies. Therefore, further studies are required to
confirm the high value of dynamic parameters in
discriminating responder and nonresponder patients
before fluid infusion. However, our analysis empha-
sizes the minimal value of static ventricular preload
parameters as predictors of fluid responsiveness and
strongly supports the use of the dynamic parameters
in the decision-making process concerning volume
expansion in critically ill patients.
ACKNOWLEDGMENT: The authors thank Dr. Denis Chemla
for helpful discussion.
References
1 Braunwald E, Sonnenblick EH, Ross J. Mechanisms of
cardiac contraction and relaxation. In: Braunwald E, ed.
Heart disease. Philadelphia, PA: W.B. Saunders, 1988; 389 –
425
2 Calvin JE, Driedger AA, Sibbald WJ. The hemodynamic
effect of rapid fluid infusion in critically ill patients. Surgery
1981; 90:61–76
3 Schneider AJ, Teule GJJ, Groeneveld ABJ, et al. Biventricular
performance during volume loading in patients with early
septic shock, with emphasis on the right ventricle: a combined
hemodynamic and radionuclide study. Am Heart J 1988;
116:103–112
4 Reuse C, Vincent JL, Pinsky MR. Measurements of right
ventricular volumes during fluid challenge. Chest 1990; 98:
1450 –1454
5 Magder S, Georgiadis G, Cheong T. Respiratory variations in
right atrial pressure predict the response to fluid challenge. J
Crit Care 1992; 7:76 – 85
6 Diebel LN, Wilson RF, Tagett MG, et al. End-diastolic
CHEST / 121 / 6 / JUNE, 2002 2007
 volume: a better indicator of preload in the critically ill. Arch
Surg 1992; 127:817– 822
7 Diebel L, Wilson RF, Heins J, et al. End-diastolic volume vs
pulmonary artery wedge pressure in evaluating cardiac pre-
load in trauma patients. J Trauma 1994; 37:950 –955
8 Wagner JG, Leatherman JW. Right ventricular end-diastolic
volume as a predictor of the hemodynamic response to a fluid
challenge. Chest 1998; 113:1048 –1054
9 Tavernier B, Makhotine O, Lebuffe G, et al. Systolic pressure
variation as a guide to fluid therapy in patients with sepsis-
induced hypotension. Anesthesiology 1998; 89:1313–1321
10 Magder S, Lagonidis D. Effectiveness of albumin vs normal
saline as a test of volume responsiveness in post-cardiac
surgery patients. J Crit Care 1999; 14:164 –171
11 Tousignant CP, Walsh F, Mazer CD. The use of transesoph-
ageal echocardiography for preload assessment in critically ill
patients. Anesth Analg 2000; 90:351–355
12 Michard F, Boussat S, Chemla D, et al. Relation between
respiratory changes in arterial pulse pressure and fluid re-
sponsiveness in septic patients with acute circulatory failure.
Am J Respir Crit Care Med 2000; 162:134 –138
13 Feissel M, Michard F, Mangin I, et al. Respiratory changes in
aortic blood velocity as an indicator of fluid responsiveness in
ventilated patients with septic shock. Chest 2001; 119:867–
873
14 Boldt J, Lenz M, Kumle B, et al. Volume replacement
strategies on ICUs: results from a postal survey. Intensive
Care Med 1998; 24:147–151
15 Task Force of the American College of Critical Care Medi-
cine, Society of Critical Care Medicine. Practice parameters
for hemodynamic support of sepsis in adult patients in sepsis.
Crit Care Med 1999; 27:639 – 660
16 Vincent JL. Hemodynamic support in septic shock. Intensive
Care Med 2001; 27:S80 –S92
17 Magder S. More respect for the CVP. Intensive Care Med
1998; 24:651– 653
18 Packman MI, Rackow EC. Optimum left heart filling pres-
sure during fluid resuscitation of patients with hypovolemic
and septic shock. Crit Care Med 1983; 11:165–169
19 Baek S-E, Makabali GG, Bryan-Brown CW, et al. Plasma
expansion in surgical patients with high central venous pres-
sure (CVP): the relationship of blood volume to hematocrit,
CVP, pulmonary wedge pressure, and cardiorespiratory
changes. Surgery 1975; 78:304 –315
20 Thys DM, Hillel Z, Goldman ME, et al. A comparison of
hemodynamic indexes derived by invasive monitoring and
two-dimensional echocardiography. Anesthesiology 1987; 67:
630 – 634
21 Cheung AT, Savino JS, Weiss SJ, et al. Echocardiographic and
hemodynamic indexes of left ventricular preload in patients
with normal and abnormal ventricular function. Anesthesiol-
ogy 1994; 81:376 –387
2008
22 Swenson JD, Harkin C, Pace NL, et al. Transesophageal
echocardiography: an objective tool in defining maximum
ventricular response to IV fluid therapy. Anesth Analg 1996;
83:1149 –1153
23 Pinsky M, Vincent JL, De Smet JM. Estimating left ventric-
ular filling pressure during positive end-expiratory pressure in
humans. Am Rev Respir Dis 1991; 143:25–31
24 Teboul JL, Pinsky MR, Mercat A, et al. Estimating cardiac
filling pressure in mechanically ventilated patients with hy-
perinflation. Crit Care Med 2000; 28:3631–3636
25 Raper R, Sibbald WJ. Misled by the wedge? The Swan-Ganz
catheter and left ventricular preload. Chest 1986; 89:427– 434
26 Spinale FG, Mukherjee R, Tanaka R, et al. The effects of
valvular regurgitation on thermodilution ejection fraction
measurements. Chest 1992; 101:723–731
27 Urbanowicz JH, Shaaban MJ, Cohen NH, et al. Comparison
of transesophageal echocardiographic and scintigraphic esti-
mates of left ventricular end-diastolic volume index and
ejection fraction in patients following coronary artery bypass
grafting. Anesthesiology 1990; 72:607– 612
28 Magder S. The cardiovascular management of the critically ill
patients. In: Pinsky MR, ed. Applied cardiovascular physiol-
ogy. Berlin, Germany: Springer, 1997; 28 –35
29 Morgan BC, Martin WE, Hornbein TF, et al. Hemodynamic
effects of intermittent positive pressure ventilation. Anesthe-
siology 1966; 27:584 –590
30 Chemla D, Hébert JL, Coirault C, et al. Total arterial
compliance estimated by the stroke volume-to-aortic pulse
pressure ratio in humans. Am J Physiol 1998; 274:H500 –
H505
31 Jardin F, Farcot JC, Gueret P, et al. Cyclic changes in arterial
pulse during respiratory support. Circulation 1983; 68:266 –
274
32 Robotham JL, Cherry D, Mitzner W, et al. A re-evaluation of
the hemodynamic consequences of intermittent positive pres-
sure ventilation. Crit Care Med 1983; 11:783–793
33 Denault YD, Gasior TA, Gorcsan III J, et al. Determinants of
aortic pressure variation during positive-pressure ventilation
in man. Chest 1999; 116:176 –186
34 Perel A, Pizov R, Cotev S. Systolic BP: variation is a sensitive
indicator of hypovolemia in ventilated dogs subjected to
graded hemorrhage. Anesthesiology 1987; 67:498 –502
35 Coriat P, Vrillon M, Perel A, et al. A comparison of systolic
BP variations and echocardiographic estimates of end-
diastolic left ventricular size in patients after aortic surgery.
Anesth Analg 1994; 78:46 –53
36 Michard F, Teboul JL. Respiratory changes in arterial pres-
sure in mechanically ventilated patients. In: Vincent JL, ed.
Yearbook of intensive care and emergency medicine. Berlin,
Germany: Springer, 2000; 696 –704
Critical Care Reviews