You are on page 1of 6

Arterial pressure allows monitoring the changes in cardiac output induced by volume expansion but not by norepinephrine*

Xavier Monnet, MD, PhD; Alexia Letierce, PhD; Olfa Hamzaoui, MD; Denis Chemla, MD, PhD; Nadia Anguel, MD; David Osman, MD; Christian Richard, MD; Jean-Louis Teboul, MD, PhD

Objective: To evaluate to which extent the systemic arterial pulse pressure could be used as a surrogate of cardiac output for assessing the effects of a fluid challenge and of norepinephrine. Design: Observational study. Setting: Medical intensive care unit. Patients: Patients with an acute circulatory failure who received

a fluid challenge (228 patients, group 1) or in whom norepinephrine was introduced or increased (145 patients, group 2). Interventions: We measured the systolic, diastolic, and mean arterial pressure, pulse pressure, and the transpulmonary ther- modilution cardiac output before and after the therapeutic inter- ventions. Main Results: In group 1, the fluid challenge significantly increased cardiac output by 24% 25%. It significantly increased cardiac output by > 15% ( 35% 27%) in 142 patients (“re- sponders”). The fluid-induced changes in cardiac output were correlated with the changes in pulse pressure (r .56, p < .0001), systolic arterial pressure (r .55, p < .0001), diastolic arterial pressure (r .37, p < .0001), and mean arterial pressure

( r .52, p < .0001). At multivariate analysis, changes in pulse

pressure were significantly related to changes in stroke volume (multiple r .52) and to age (r .12). A fluid-induced increase

in pulse pressure of >17% allowed detecting a fluid-induced increase in cardiac output of > 15% with a sensitivity of 65[56 – 72]% and a specificity of 85[76 –92]%. The area under the receiver operating characteristic curves for the fluid-induced changes in mean arterial pressure and in diastolic arterial pressure was significantly lower than for pulse pressure. In group 2, the intro- duction/increase of norepinephrine significantly increased car- diac output by 14% 18%. The changes in cardiac output induced by the introduction/increase in the dose of norepineph- rine were correlated with the changes in pulse pressure and systolic arterial pressure (r .21 and .29, respectively, p .001) but to a significantly lesser extent than in group 1. Conclusions: Pulse pressure and systolic arterial pressure could be used for detecting the fluid-induced changes in cardiac output, in spite of a significant proportion of false-negative cases. By contrast, the changes in pulse pressure and systolic arterial pressure were unable to detect the changes in cardiac output induced by norepinephrine. (Crit Care Med 2011; 39:000 –000) KEY WORDS : fluid challenge; volume expansion; norepinephrine; vasopressors; arterial pressure; cardiac output; arterial compli- ance; pulse wave amplification

I n patients with an acute circula- tory failure, the question whether cardiac output should be moni- tored or not is still a matter of

debate. Recent guidelines recommend to

*See also p. 000. From the Service de Re´ animation me´ dicale (XM, OH, NA, DO, CR, J-LT) and Unite´ de Recherche clinique (AL), Hoˆ pital de Biceˆtre, AP-HP, Le Kremlin-Biceˆtre; Faculte´ de Me´ decine Paris-Sud (XM, OH, DC, NA, DO, CR, J-LT), Univ Paris-Sud, Le Kremlin-Biceˆtre; and Service de Physiologie (DC), Hoˆ pital Antoine Be´ cle`re, AP-HP, Clamart, France. Supplemental digital content is available for this ar- ticle. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (http://www.ccmjournal.com). Dr. Monnet and Dr. Teboul consulted for Pulsion. The remaining authors have not disclosed any poten- tial conflicts of interest. For information regarding this article, E-mail:

xavier.monnet@bct.aphp.fr Copyright © 2011 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins

DOI: 10.1097/CCM.0b013e31820edcf0

measure cardiac output in patients with ventricular failure or persistent shock de- spite adequate fluid resuscitation (1). Al- though these recommendations are not supported by a high level of evidence, they suggest that a “basic” hemodynamic monitoring based on the sole arterial pressure might be sufficient for monitor- ing initial fluid resuscitation but insuffi- cient for monitoring later resuscitation with fluid and vasopressors (1). Since it tends to maintain arterial pres- sure stable while cardiac output varies, the sympathetic regulation might preclude us- ing arterial pressure for monitoring the treatment-induced changes of cardiac out- put. However, among the different values of arterial pressure, the pulse pressure (PP, i.e., the difference between systolic [SAP] and diastolic [DAP] arterial pressure) is physiologically related to stroke volume (2). It might thus be able to reflect the changes in cardiac output induced by dif- ferent treatments. Nevertheless, the rela-

tionship between PP and stroke volume is affected by two physiologic phenomena: the arterial compliance and, if PP is measured at the peripheral level, the pulse wave am- plification phenomenon (3). Since vaso- pressors might alter both arterial compli- ance and pulse wave amplification to a larger extent than volume expansion, the ability of PP to reflect the changes in stroke volume (and cardiac output) might be bet- ter for fluid therapy than for vasopressors, but this has never been demonstrated. Thus, in the present study, we inves- tigated to what extent PP could be used for indicating absolute values and treat- ment-induced changes of cardiac output in a large population of patients with an acute circulatory failure treated with fluid and norepinephrine.

METHODS

Patients. This study was approved by our institutional review board. A deferred consent was asked from the patient’s surrogate as soon

review board. A deferred consent was asked from the patient’s surrogate as soon Crit Care Med

Table 1. Patients characteristics at baseline

Characteristic

Patients Receiving Volume Expansion (n 228)

Patients With an Introduction/ Increase of NE (n 145)

Gender (no of patients, F/M)

95/133

58/87

Age (mean SD , years) Simplified acute physiology score II (mean SD )

63 13 52 11

63 12 43 18

Type of shock (no of patients, %) Septic Hypovolemic Post cardiac arrest Drug poisoning Other Patients receiving NE at baseline (no. of patients, %) NE dose at baseline (median 25%–75%

193 (85) 14 (6) 8 (4) 7 (3) 6 (2) 212 (93) 0.5 0.3–1.1

145 (100) 0 (0) 0 (0) 0 (0) 0 (0) 86 (59) 0.5 0.2–0.7

interquartile , g/kg/min) Acute respiratory distress syndrome (no. of patients, %)

137 (61)

73 (50)

There was no statistical difference between groups. NE, norepinephrine.

as possible. As he/she recovered conscious- ness, a deferred consent was asked from the patient. If the patient or his/her next of kin refused to consent, patient’s data were not entered into analysis. We included patients (1) if they presented an acute circulatory failure defined by the

presence of at least one of the following crite- ria: 1) SAP 90 mm Hg (or fall of SAP 50

mm Hg in patients previously known as hy-

pertensive), 2) urinary flow 0.5 mL/kg/min

for more than 2 hrs, 3) tachycardia 100 beats/min, or 4) presence of skin mottling and (2) if a fluid challenge was administered (group 1, n 228 patients) or if norepineph-

rine was introduced or its dose was increased

(group 2, n 145 patients, different from patients of group 1), as ordered by the attend- ing physician. The decision to administer fluid

and to introduce/increase the dose of norepi- nephrine was based on standard care criteria (e.g., positive fluid responsiveness indicators for fluid, low arterial pressure for vasopres- sors). Patients’ characteristics at baseline are listed in Table 1. Twenty-two patients were not ventilated. Among the overall 353 patients

who were intubated, 70 were ventilated in the

pressure support mode. Among the 283 re-

maining patients, all of whom were ventilated

in the assist control mode, 223 exhibited some

spontaneous triggering of the ventilator. In

patients ventilated in the assist-control mode,

tidal volume was 6.5% 2.1% mL/kg.

Measurements. All patients had an internal

jugular vein catheter and a thermistor-tipped arterial catheter (PV2015L20N, Pulsion Medi- cal Systems, Munich, Germany) in the femoral artery that was connected to a PiCCO-Plus or

a PiCCO-2 device (Pulsion Medical Systems,

Munich, Germany) for measuring cardiac out- put through transpulmonary thermodilution. The femoral arterial catheter was connected to the pressure sensor PV8215 by a PV8215 mon-

itoring kit (Pulsion Medical Systems, Munich, Germany) and invasive arterial pressure was measured by the PiCCO-Plus or PiCCO-2 de- vice. Before all pressure measurements, the arterial line was carefully flushed and zeroed to atmospheric pressure. For the measure- ment of cardiac output, the values of three thermodilution measurements were averaged. The PiCCO-Plus and the PiCCO-2 devices both measure cardiac output and arterial pressure in a similar way. Study Design. Before fluid infusion in group 1 and before the introduction/increase in the dose of norepinephrine in group 2, a transpulmonary thermodilution was per- formed and the values of cardiac output, stroke volume, global end-diastolic volume, and extravascular lung water (all obtained from transpulmonary thermodilution), the values of SAP, DAP, mean arterial pressure (MAP), and PP were recorded. In group 1, a 500-mL saline bolus was infused over 20 min. In this group, norepi- nephrine was administered at baseline (in 228 patients) and its dose was unchanged during fluid infusion. In group 2, norepinephrine was introduced (in 59 patients) at a dose of 0.24 (0.13– 0.48) g/kg/min or its dose was in- creased (in 86 patients) from 0.48 (0.24 – 0.71) g/kg/min to 0.62 (0.43–1.07) g/kg/min. The values of cardiac output, stroke vol- ume, global end-diastolic volume, and ex- travascular lung water, the values of SAP, DAP, MAP, and PP were also obtained after the therapeutic intervention, i.e., immediately af- ter fluid infusion in group 1 and 5 min after the stabilization of MAP in group 2. Since predicting fluid responsiveness is an impor- tant issue in critically ill patients, we separated patients in group 1 between “fluid responders” and “fluid nonresponders.” A positive fluid re- sponse was defined by a treatment-induced increase in cardiac output of 15% (4 –7). We

also used some other definitions of the fluid responsiveness: an increase in cardiac output of 10%, an increase in stroke volume of 15%, and an increase in stroke volume of 10%. Statistical Analysis. All continuous vari- ables except the dose of norepinephrine were normally distributed at Kolmogorov-Smirnov test. Results are expressed as mean SD , as median (25%–75% interquartile range) or as mean (95% confidence interval), as appropri- ate. Comparisons of variables between before vs. after fluid administration were assessed by using a paired Student’s t test. Comparisons between responders vs. nonresponders were assessed by using a two-sample Student’s t test or a Mann-Whitney U test, as appropriate. Cor- relations were assessed by the Pearson coeffi- cient and correlation coefficients were com- pared between group 1 and group 2 using the Fisher transformation (8). The correlation analysis was also performed in the four quar- tiles of age of the total population. A stepwise regression analysis was performed to look for independent variables related to the changes in PP (in %). A p value of .20 was necessary for a variable to enter regression analysis. Elsewhere, a p value of .05 was considered statistically significant. In group 1, receiver operating characteristic (ROC) curves (with 95% confidence intervals) were constructed for testing the ability of the fluid-induced changes in SAP, DAP, MAP, and PP to predict fluid responsiveness. The areas under the ROC curves were compared using a Hanley-McNeil test (9). Since the two groups of patients could have different arterial pressures at baseline, we also analyzed the data in a subpopulation of each groups with matched baseline MAP (for detailed methods, see Supplemental Digital Content 1, http://links.lww.com/CCM/A228). The statistical analysis was performed by using the Statview5.0 software (Abacus Concepts, Berkeley, CA) and the MedCalc8.1.0.0 software (Mariakerke, Belgium).

RESULTS

Effects of the Therapeutic Interven- tions on Hemodynamic Variables. In group 1 (n 228 patients), volume ex- pansion increased cardiac output by 24% 25% ( p .05). It significantly in- creased cardiac output of 15% ( 35% 27%) in 142 patients (“responders”) (Table 2). In the 86 remaining patients of group 1 (“nonresponders”), volume ex- pansion significantly increased cardiac output by 7% 5% (Table 2). In the whole group 1, volume expansion in- creased SAP, DAP, MAP, and PP by 15% 19%, 9% 16%, 13% 17%, and 21% 29%, respectively (Table 2). The magnitude of these changes was signifi- cantly larger in responder than in nonre- sponder patients (Table 2).

was signifi- cantly larger in responder than in nonre- sponder patients (Table 2). 2 Crit Care

Table 2. Hemodynamic variables in patients receiving volume expansion (n 228)

 

Before Volume

After Volume

Variable

Category

Expansion

Expansion

Heart rate (mean SD , beats/min)

Nonresponders

92 23 98 22 107 24 109 21 52 13 53 13 71 16 71 14 56 19 56 18 703 185 649 203 b 11 5

90 22 a 96 21 a,b 115 26 a 129 23 a,b 55 13 a 59 13 a,b 75 16 a 82 16 a,b 60 20 a 70 18 a,b 808 280 a 766 264 a,b 11 5

Responders

Systolic arterial pressure (mean SD , mm Hg)

Nonresponders

Responders Diastolic arterial pressure (mean SD , mm Hg) Nonresponders Responders

Mean arterial pressure (mean SD , mm Hg)

Nonresponders

Responders

Pulse arterial pressure (mean SD , mm Hg)

Nonresponders

Responders

Global end-diastolic volume (mean SD , mL/m 2 ) Nonresponders Responders

Extravascular lung water (mean SD , mL/kg predicted body weight)

Nonresponders

Responders

10 6 3.2 1.0 2.7 0.9 b

10 5 3.4 1.1 a 3.5 1.1 b

Cardiac index (mean SD , L/min/m 2 )

Nonresponders

Responders

“Responders” refers to patients in whom volume expansion increased cardiac index by 15% (n 86). “Nonresponders” refers to the other patients (n 142). a p .05 vs. before volume expansion (comparisons in rows); b p .05 vs. nonresponders (comparisons in columns).

Table 3. Hemodynamic variables in patients with an introduction/increase of norepinephrine (n 145)

Before Introduction/

After Introduction/

 

Increase of

Increase of

Variable

Norepinephrine

Norepinephrine

Heart rate (mean SD , beats/min) Systolic arterial pressure (mean SD , mm Hg) Diastolic arterial pressure (mean SD , mm Hg) Mean arterial pressure (mean SD , mm Hg) Pulse arterial pressure (mean SD , mm Hg) Global end-diastolic volume (mean SD , mL/m 2 ) Extravascular lung water (mean SD , mL/kg predicted body weight) Cardiac index (mean SD , L/min/m 2 )

97 22 83 14 39 7 54 8 45 14 692 137 10 4

96 21 120 21 a 52 9 a 75 10 a 67 22 a 735 154 a 10 5

3.1 1.0

3.5 1.1 a

a p .05 vs. before introduction/increase in NE (comparisons in rows).

In group 2 (n 145 patients), the introduction/increase in dose of norepi- nephrine increased cardiac output by 14% 18% ( p .05) (Table 3). In the whole group 2, the introduction/ increase in the dose of norepinephrine significantly increased SAP, DAP, MAP, and PP by 46% 30%, 9% 16%, 38% 24%, and 41% 23%, respec- tively (Table 3). Relationship Between the Changes in Arterial Pressure and the Changes

in Cardiac Output and Stroke Volume. In

group 1, the fluid-induced changes (in

%) in cardiac output were correlated

with the changes (in %) in PP ( r .56,

p .0001) (Fig. 1), SAP ( r .55, p

.0001), DAP ( r .37, p .0001), and MAP ( r .52, p .0001). The fluid- induced changes (in %) in stroke vol-

ume were correlated with the changes (in %) in PP ( r .49, p .0001), SAP ( r .44, p .0001), DAP ( r .25, p .0002), and MAP ( r .39, p .0001). There was not statistical difference be- tween the r coefficients found for the correlation between changes in cardiac output and change in PP on the one side and for the correlation between changes in cardiac output and changes in SAP on the other side. Multivariate analysis indicated that the changes in PP (in %) were significantly related to the changes in stroke volume ( r .52) and to age ( r .12). When the popula- tion was divided in 4 quartiles of age, the changes (in %) in PP were related to the changes (in %) in stroke volume for all quartiles with different coefficients of correlation: r .40 ( p .002) for the

first quartile of age (from 36 to 53 yrs,

n 59), r .48 ( p .0001) for the

second quartile of age (from 54 to 62 yrs, n 50), r .54 ( p .0001) for the

third quartile of age (from 63 to 74 yrs,

n 70), r .68 ( p .0001) for the

fourth quartile of age (from 75 to 88

yrs, n 49). In group 2, the changes (in %) in cardiac output induced by the introduc- tion/increase in dose of norepinephrine were correlated with the changes (in %)

in PP (r .21, p .001, Fig. 2), SAP ( r

.29, p .004), and MAP ( r .21, p .01) but not in DAP ( r .13, p .12). All these correlations were significantly lower than in group 1 ( p .0001, .003, and .0007 for the changes in PP, SAP, and MAP, respectively). The changes (in %) in stroke volume induced by the introduction/increase in dose of norepi- nephrine were correlated with the changes (in %) in PP ( r .17, p .04), SAP ( r .18, p .003), and MAP ( r .17, p .04) but not in DAP ( r .01,

p .92). All these correlations were

significantly lower than in group 1 ( p .0007, .007, and .02 for the changes in PP, SAP, and MAP, respectively). Multi-

variate analysis indicated that the changes in PP (in %) were related to the changes (in %) in stroke volume ( r .17) but not to age. Ability of the Changes in Arterial Pressure to Detect Fluid Responsiveness

in Group 1. The ability of the changes in

the different values of arterial pressure to detect a fluid-induced increase in cardiac

output of 15% in group 1 is described in Table 4 and Figure 2. The fluid- induced changes in PP and in SAP exhi- bited the best diagnostic values, with similar areas under the ROC curves. A fluid-induced increase in PP of 17% allowed detecting a fluid-induced in- crease in cardiac output of 15% with a sensitivity of 65% (95% confidence inter-

val: 56%–72%) and a specificity of 85% (95% confidence interval: 76%–92%), i.e., with 6% and 22% of false-positive and false-negative cases, respectively. The area under the ROC curves for the fluid- induced changes in MAP and for the flu- id-induced changes in DAP was signifi- cantly lower than for the fluid-induced changes in PP (Table 4). The areas under the ROC curve describing the ability of the fluid-induced changes in PP to detect a fluid-induced increase in cardiac output

of 15% were not different among pa-

tients belonging to the first, the second, the third, and the fourth age quartiles

tients belonging to the first, the second, the third, and the fourth age quartiles Crit Care

A

300 250 200 150 100 50 0 -50 -50 0 50 100 150 200 250
300
250
200
150
100
50
0
-50
-50
0
50
100 150 200 250 300
Changes in PP (in %) induced
by the fluid challenge

Changes in CO (in %) induced by the fluid challenge

B * r = 0.56 r = 0.21 300 n = 228 n = 145
B
*
r
= 0.56
r
= 0.21
300
n
= 228
n
= 145
250
200
150
100
50
0
-50
-50
0
50
100 150 200 250 300
Changes in PP (in %) induced
by the introduction/increase in NE

Changes in CO (in %) induced by the introduction/increase in NE

Figure 1. Correlation between the changes in arterial pulse pressure (PP , in % change from baseline) and cardiac output (CO , in % change from baseline) induced by a fluid challenge (A , n 228 pairs of measurements) and by the introduction or increase in dose of norepinephrine. (B , n 145 pairs of measurements). *p .0001 vs. r in A .

100 80 60 40 Diagnostic ability of the fluid-induced changes (in %) in: 20 0
100
80
60
40
Diagnostic ability of
the fluid-induced changes (in %) in:
20
0
0
20
40
60
80
100
Sensitivity (%)

100-Specificity (%)

PP

SAP

MAP

DAP

Figure 2. Receiving Operating Characteristics curves describing the ability of the fluid-induced changes (in %) in arterial pulse (PP) and systolic (SAP), diastolic (DAP), and mean (MAP) arterial pressure to detect a fluid-induced increase in cardiac index of 15%.

(0.81 0.05, 0.78 0.07, 0.78 0.05, and 0.80 0.06, respectively). Similar results were obtained when using dif- ferent definitions of fluid responsive- ness (see Supplemental Digital Content 1, http://links.lww.com/CCM/A228).

Influence of Baseline Arterial Pres- sure. The analysis of the subsets of groups 1 and 2 matched for baseline MAP showed similar results than in the whole popula- tion (see Supplemental Digital Content,

http://links.lww.com/CCM/A228).

DISCUSSION

This study suggests that PP and SAP

could be used for detecting the changes

in cardiac output induced by a fluid chal-

lenge, in spite of a significant proportion of false-negative cases in which PP and

SAP did not change while cardiac output

increased. By contrast, the changes in PP

and SAP were unable to detect the

changes in cardiac output induced by the norepinephrine. When taking care of patients with cir- culatory shock, the question is frequently pending whether cardiac output must be monitored or if a basic monitoring with the sole arterial pressure will be suffi- cient. International guidelines recom- mend monitoring cardiac output in pa- tients with shock refractory to initial fluid therapy (1), but this is not sup- ported by a substantial scientific back- ground. The goal of the present study was to provide some physiologic basis to this guideline. In the first place, our study indicates that the different values of arterial pres- sure are not equivalent for monitoring the treatment-induced changes in stroke volume or cardiac output. Physiologi- cally, the changes in MAP are dissociated from the changes in cardiac output due to the sympathetic modulation of the ar- terial tone, which tends to maintain MAP constant while cardiac output varies (10). According to this physiologic paradigm, we found that the changes in MAP were unable to reflect the changes in cardiac output induced either by fluid infusion or by the introduction or increase in norepi- nephrine. In particular, the introduction/ increase in norepinephrine augmented MAP to a large extent, but this was re- lated to arterial vasoconstriction and not correlated with a simultaneous change in stroke volume or cardiac output. Another physiologic paradigm is that, in contrast with MAP, PP is directly re-

Table 4. Diagnostic ability of the fluid-induced changes in arterial pressures values to detect a fluid-induced increase in cardiac index of 15% in patients receiving fluid infusion (group 1)

 

Area Under the Receiver Operating Characteristic Curve

 

Positive

Negative

p vs.

Best Cutoff

Predictive

Predictive

Youden

 

Variable

.500

Value

Sensitivity

Specificity

Value

Value

Index

Changes

in

arterial

pulse pressure

0.784 0.03 0.757 0.03 0.692 0.04 a 0.598 0.04 a

.0001

17%

65 56–72 74 66–81

85 76–92 67 57–77

88 80–93 59 50–68

0.50

Changes

in

systolic

arterial pressure

.0001

8%

79 71–85

61 51–71

0.41

Changes

in

mean arterial pressure diastolic arterial pressure

.001

13%

46 38–55 84 74–91 67 29–45 83 73–90

82 72–90 49 40–57 78 66–87 44 37–52

0.30

Changes

in

.01

11%

0.50

37–52 0.30 Changes in .01 11% 0.50 n 228, mean SD or mean 95% confidence interval

lated to stroke volume (2). In theory, PP might thus be a better candidate for mon- itoring the changes in cardiac output at the bedside (3). However, PP is also in- versely correlated with arterial compli- ance, which might differ among patients and might change over time in a same patient (11). Furthermore, the propor-

tionality between PP and stroke volume is physiologically expected at the aortic level but not at the peripheral arterial due to the pulse wave amplification phenom- enon (2). In the present study, we quan- tified the extent to which these two phys- iologic issues preclude to use the changes in PP for detecting the changes in stroke volume and cardiac output. In group 1, the fluid-induced changes in PP were significantly correlated with the fluid-induced changes in cardiac out- put. These results are in line with those

of previous small human studies in which

a decrease in PP paralleled the decrease

in stroke volume induced by graded low body negative pressure (12) or compres- sion of the right atrium (13). In our study, the correlation with the fluid- induced changes in PP was not better for the fluid-induced changes in stroke vol- ume than for the fluid-induced changes in cardiac output, since heart rate was not decreased to a large extent by volume expansion in our study population. In turn, the changes in PP exhibited a good diagnostic accuracy for detecting fluid re- sponsiveness. This indicates that PP was an acceptable surrogate of cardiac output for assessing the effects of fluid therapy. Since DAP was only slightly modified by volume expansion, SAP (i.e., the sum of DAP and PP) exhibited a similar diagnos- tic accuracy. Although the fluid-induced changes in PP and SAP and in cardiac output were significantly correlated, these correla- tions were not excellent (r .56 and .55, respectively). This might be related to the physiologic issues of arterial compliance and pulse wave amplification phenome- non (10). According to this hypothesis, we found that the coefficient of correla- tion between changes in PP and in stroke volume induced by the fluid challenge was lower in the youngest patients, in whom the arterial compliance was as- sumed to be higher and the pulse wave amplification lower than in the oldest. In group 2, it is not surprising that norepinephrine increased DAP and MAP, which are related to the arterial tone (3). By contrast, an interesting finding was that the relationship between PP and car-

diac output was poor. This indicates that changes of mechanical properties of the arterial system induced by norepineph- rine, including arterial compliance and pulse wave amplification, were so marked that PP could not be used for tracking trends in cardiac output. Nevertheless, an important limitation of our study was that we did not precisely investigate arte- rial compliance and the pulse wave am- plification phenomenon. As a clinical application, the present study clarifies in which manner clinicians should use arterial pressure for monitor- ing therapy during circulatory failure, with very different attitudes regarding fluid and vasopressors. The results sug- gest that when PP increases 17% with fluid administration, the clinician can be reasonably confident that a patient is fluid responsive (6% of false-positive cases only). By contrast, if PP does not increase 17%, the high number of false-negative cases (22%) precludes drawing any reasonable conclusion con- cerning fluid responsiveness. This issue might be particularly relevant when per- forming repetitive fluid challenges, a pro- cedure that has been recently revisited (14). Underestimating the response of cardiac output to fluid boluses might clearly lead to fluid under-resuscitation. Concerning the monitoring of norepi- nephrine administration, PP cannot be used for assessing the effects that norepi- nephrine can exert on stroke volume (15, 16). However, despite our results might change routine practice, their clinical relevance is highly limited by the fact that this study was not designed for evi- dencing any different impact of arterial pressure vs. cardiac output monitoring on outcome. This issue should be inves- tigated in further studies. Some monitoring devices use the physiologic relationship between the am- plitude of the arterial curve and stroke volume for estimating cardiac output from an arterial line. For this purpose, these devices estimate arterial compl- iance and vasomotor tone from a complex analysis of the arterial pressure curve and, for some of these systems, with help of an external calibration (17). In accor- dance with our results, the fact that changes in arterial compliance and/or va- somotor tone are more marked with nor- epinephrine than with fluid therapy might explain why some systems have greater difficulty to reliably estimate the changes in cardiac output when induced

by norepinephrine than by volume expan- sion (17). To summarize, the changes in PP or SAP but not in MAP could help for de- tecting the hemodynamic effects of a fluid challenge. If PP increased 17% during the fluid challenge, one may sup- pose with acceptable confidence that car- diac output increased 15%. By contrast, if PP did not increase 17% during the fluid challenge, one could not ascertain that cardiac output did not improve. The changes in PP or SAP could not be used for monitoring the effects of norepineph- rine on cardiac output.

ACKNOWLEDGMENTS

We are greatly indebted to Prof. Lau- rence Meyer, from the Department of Epidemiology and Public Health of Biceˆtre Hospital, for her help in statis- tical analysis.

REFERENCES

1. Antonelli M, Levy M, Andrews PJ, et al: He- modynamic monitoring in shock and impli- cations for management. International Con- sensus Conference, Paris, France, 27–28 April 2006. Intensive Care Med 2007; 33:

575–590

2. Chemla D, He´bert JL, Coirault C, et al:

Total arterial compliance estimated by stroke volume-to-aortic pulse pressure ra- tio in humans. Am J Physiol 1998; 274:

H500 –H505

3. Lamia B, Chemla D, Richard C, et al: Clinical review: Interpretation of arterial pressure wave in shock states. Crit Care 2005; 9:601– 606

4. Michard F, Boussat S, Chemla D, et al: Rela- tion between respiratory changes in arterial pulse pressure and fluid responsiveness in septic patients with acute circulatory failure. Am J Respir Crit Care Med 2000; 162:

134 –138

5. Monnet X, Osman D, Ridel C, et al: Predict- ing volume responsiveness by using the end- expiratory occlusion in mechanically venti- lated intensive care unit patients. Crit Care Med 2009; 37:951–956

6. Monnet X, Rienzo M, Osman D, et al: Esoph- ageal Doppler monitoring predicts fluid re- sponsiveness in critically ill ventilated pa- tients. Intensive Care Med 2005; 31:

1195–1201

7. Monnet X, Rienzo M, Osman D, et al: Passive leg raising predicts fluid responsiveness in the critically ill. Crit Care Med 2006; 34:

1402–1407

8. Fisher RA: Statistical Methods for Research Workers. Fourteenth Edition. Edinburgh, London, UK, Oliver & Boyd, 1970

9. Hanley JA, McNeil BJ: A method of compar- ing the areas under receiver operating char-

JA, McNeil BJ: A method of compar- ing the areas under receiver operating char- Crit Care

acteristic curves derived from the same cases. Radiology 1983; 148:839 – 843

10. Nichols W, O’Rourke M: Contours of pressure and flow waves in arteries. In: Blood Flow in Arteries. Nichols W, O’Rourke MS (Eds). London, UK, Oxford University Press; 1998, pp 170–200

11. O’Rourke B: Evidence for mitochondrial K channels and their role in cardioprotection. Circ Res 2004; 94:420 – 432

12. Convertino VA, Cooke WH, Holcomb JB: Ar- terial pulse pressure and its association with reduced stroke volume during progressive

central hypovolemia. J Trauma 2006; 61:

629 – 634

13. Marquez J, McCurry K, Severyn DA, et al:

Ability of pulse power, esophageal Doppler, and arterial pulse pressure to estimate rapid changes in stroke volume in humans. Crit Care Med 2008; 36:3001–3007

14. Vincent JL, Weil MH: Fluid challenge re- visited. Crit Care Med 2006; 34:1333–1337

15. Hamzaoui O, Georger JF, Monnet X, et al:

Early administration of norepinephrine in- creases cardiac preload and cardiac output in

septic patients with life-threatening hypoten- sion. Crit Care 2010; 14:R142

16. Monnet X, Jabot J, Maizel J, et al: Norepi- nephrine increases cardiac preload and re- duces preload dependency assessed by pas- sive leg raising in septic shock patients Crit Care Med 2011; In Press

17. Monnet X, Anguel N, Naudin B, et al: Arterial pressure-based cardiac output in septic pa- tients: Different accuracy of pulse contour and uncalibrated pressure waveform devices. Crit Care 2010; 14:R109

pulse contour and uncalibrated pressure waveform devices. Crit Care 2010; 14:R109 6 Crit Care Med 2011