Validation of Mean Arterial Pressure as an Indicator of Acute
Changes in Cardiac Output
JASON E. PRASSO,* GEORGE BERBERIAN,† SANTOS E. CABRERIZA,† T. ALEXANDER QUINN,‡ LAUREN J. CURTIS,† DAVID G. RABKIN,†
ALAN D. WEINBERG,§ AND HENRY M. SPOTNITZ†
Changes in mean arterial pressure (MAP) are often assumed
to reflect changes in cardiac output (CO). A linear relation-
ship is postulated to exist between these two quantities based
upon the circuit model for systemic circulation. Previous
studies have correlated changes in CO and MAP. However, to
our knowledge, no studies have tested the relationship be-
tween CO and MAP in vivo without changes in systemic
vascular resistance. Research on baroreceptor stimulation
and vasomotor response has shown that vasomotor tone
changes 15 to 60 seconds after an acute change in CO.
Maximal activation of vasomotor response occurs after ap-
proximately 30 seconds. Thus MAP should correlate directly
with CO during acute changes (<15 seconds). To test this, we
examined the relationship between CO and MAP during 10
second occlusions of the inferior vena cava in anesthetized
pigs. A linear relationship existed between CO and MAP in
seven pigs (%MAP ? 0.60[%CO] ? 0.41, p ? 0.0001). This
study validates the use of MAP as an indicator of acute
changes in CO. Fluctuations in MAP correlate well with acute
changes in CO in the absence of changes in vascular tone.
ASAIO Journal 2005; 51:22–25.
Important hemodynamic variables of the systemic circulation
are cardiac output (CO), systemic vascular resistance (SVR),
and arteriovenous pressure gradient (between the left ventricle
and right atrium).1–3Substituting mean arterial pressure (MAP)
for the pressure gradient, these are related by an adapted
version of Ohm’s Law:1,2
MAP ? CO ? SVR
This assumes that right atrial pressure is 0. SVR is thus defined
as the ratio of MAP to CO. Because MAP represents the mean
hydrostatic pressure gradient within the entire vascular tree,3it
should vary linearly with CO if SVR remains constant.4In the
clinical setting, fluctuations in MAP are often assumed to
directly reflect changes in CO, which is usually measured by
invasive means. This relationship, however, has not been ver-
ified with intact reflexes and constant SVR.
Previous laboratory studies have assessed the importance of
SVR in the MAP/CO relation and have postulated that it is curvi-
linear. In a study that varied MAP while applying constant posi-
tive pressure to the carotid sinuses, CO and MAP were related by
a curve convex to the pressure axis.5A second study using pump
controlled flow of 60–140 ml/min kg?1with intact baroreflex
response demonstrated a relation that was also convex to the
pressure axis.6However, both studies showed that after carotid
sinus denervation and bilateral vagotomy, the pressure flow re-
and Fick based cardiac output measurements took approximately
30 seconds for each reading.
Both of these studies, however, fail to minimize changes in
SVR while testing the CO/MAP relation. Stimulation of the
baroreceptors and the carotid sinus receptors has been shown
to be the primary effector of changes in MAP.7One study
reported that under steady state conditions, changes in CO
account for less than 10% of changes in blood pressure; the
remaining 90% are attributed to baroreceptor activation.8,9A
study of autoregulation of the total systemic circulation in
decapitated dogs demonstrated that during sustained acute
changes in MAP of between 25 and 50 mm Hg, changes in
SVR facilitated a gradual return to steady state control values of
CO, oxygen consumption, and arteriovenous oxygen differ-
ence within approximately 35 minutes.10A second study
showed that before denervation, the vascular bed responded to
changes in CO using pressure regulation, whereas after dener-
vation, flow regulation was observed instead.11Because SVR
can have such drastic effects upon MAP and CO, in determin-
ing the true relationship between these variables, changes in
vasomotor tone must be minimized while SVR and the nervous
reflexes remain intact.
Changes in CO resulting from physical exercise do not affect
SVR for at least 15–60 seconds after the onset of muscular
stimulation.12Studies of hemorrhage and shock have shown
that maximal activation of the vasomotor response occurs
approximately 30 seconds after a detectable change in CO or
MAP.13Thus changes in SVR should not be present during the
first 15 seconds after a change in CO or MAP.
In this experiment, both MAP and CO were measured during
acute decreases in CO in pigs. Trials lasted 10 seconds to
avoid any effects of changes in SVR. We sought to determine
the relationship between CO and MAP and to verify whether
current clinical use of Equation 1 to predict changes in CO
based upon changes in MAP is indeed valid. Such a validation
could be of clinical value.
From *Columbia College, New York; †Surgery Columbia University,
New York; ‡Department of Biomedical Engineering, Columbia Col-
lege, New York; and the §Department of Biostatistics, Columbia Col-
lege, New York.
Submitted for consideration June 2004; accepted for publication in
revised form October 2004.
Correspondence: Dr. Henry M. Spotnitz, 622 West 168thStreet PH
14–103, New York, NY 10032.
ASAIO Journal 2005
Materials and Methods
All animals received humane care in compliance with the
Principles of Laboratory Animal Care developed by the Insti-
tute of Laboratory Animal Resources and the Guide for the
Care and Use of Laboratory Animals written by the Institute of
Laboratory Animal Resources and published by the National
Institutes of Health (NIH Publication number 85–23, revised
Seven domestic pigs (35–65 kg) were anesthetized using
ketamine hydrochloride (20 mg/kg intramuscular), xylazine
hydrochloride (0.5 mg/kg intramuscular), and atropine sulfate
(2 mg/kg intramuscular). Pigs were intubated and mechani-
cally ventilated. Anesthesia was maintained with inhalation
isoflourane (1.75–2.25%) mixed with 100% oxygen to avoid
cardiovascular effects of xylazine. Electrocardiogram leads
were attached to the limbs, a 0.9% saline infusion was started,
and the left femoral artery was instrumented with an 18 guage
angiocatheter attached to a pressure transducer to measure
MAP. Midline sternotomy and longitudinal pericardiotomy
were performed, and an ultrasonic flow probe (Transonic Sys-
tems Inc., Ithaca, NY), validated for measurement of CO by
Dean et al.,14was filled with acoustic coupling gel and placed
around the ascending aorta. The inferior vena cava (IVC) was
isolated and an umbilical tape snare placed around it proximal
to the diaphragm.
Inferior Vena Cava Occlusion
Immediately preceding occlusion of the IVC, all instrumen-
tation was calibrated and checked for accuracy. Mechanical
ventilation was stopped to prevent respiratory cycle associated
fluctuations in CO and MAP. The snare around the IVC was
then tightened, reducing venous return to the heart and caus-
ing an acute drop in CO. CO and MAP were measured during
the occlusion for a 10 second period (Figure 1). The snare was
then released, and mechanical ventilation was resumed 5
seconds after occlusion. A minimum of two occlusions were
performed on each animal. Following experimentation, all
pigs were humanely killed.
Analog data for electrocardiogram, mean arterial pressure,
and aortic flow velocity were digitized at 200 Hz using an
analog to digital converter (Chart v3.6.3/s software, Powerlab
16SP; ADInstruments Inc., Milford, MA) and recorded on a
digital computer (Power Macintosh 7100/66; Apple Computer,
Cupertino, CA) (Figure 1).
The aortic flow velocity profile was integrated between
R-wave peaks to obtain a measurement of CO, and the arterial
pressure waveform was integrated as well to determine the
corresponding MAP measurements for each heartbeat during
the first 10 seconds after occlusion of the IVC. Baseline MAP
and CO data were recorded during the 5–10 seconds imme-
diately preceding each occlusion of the IVC to later calculate
percent change in CO and MAP.
Percentage changes in CO versus percentage changes in
MAP were modeled via the PROC MIXED procedure in SAS
(SAS Institute Inc., Cary, NC). Because of the fact that repeated
measurements within animals may be correlated, this proce-
dure allows one to model this “correlation structure” as a
covariance pattern. This accurate estimate will allow for im-
proved estimates of the standard errors of measurement and
therefore more powerful tests. A likelihood ratio test or a
procedure known as Akaike’s information criterion15is used to
discern which covariance pattern allows for the best fit. We
therefore chose the “compound symmetry” structure, for cor-
relations that are constant for any two points in time. Regres-
sion equations were then generated with adjusted standard
Representative tracings from an IVC occlusion are presented
in Figure 1. Section A illustrates baseline values of CO and
MAP. At arrow 1, both arterial pressure and aortic flow begin
to decrease because of vena caval occlusion. At arrow 2, IVC
occlusion stops, establishing interval B, which corresponds to
the section of data used for analysis. Interval C represents the
recovery period after the vena caval snare was released. Pe-
ripheral MAP peaks occur after aortic flow peaks as the pres-
sure catheter is downstream of the aortic flow probe. There is
no visible change in heart rate, further supporting the assump-
tion that no systemic reflex responses occurred in the 10
second interval tested. A total of 17 cardiac cycles are present
in interval B in Figure 1. Neither MAP nor CO is restored to
baseline values immediately after release of the IVC snare
Figure 1. Representative data tracings from Chart 3.6.3/s. Base-
line CO and MAP measurements were obtained from interval A.
Data collection begins at arrow 1 and ends at arrow 2. Interval B
contains the 17 heartbeats analyzed in the experiment. Interval C
represents the recovery period. CO, cardiac output; MAP, mean
MAP AND ACUTE CHANGES IN CARDIAC OUTPUT
because blood must first travel through the pulmonary circuit
before CO and MAP return to normal physiologic values.
Figure 2 illustrates the relation between CO and MAP for
each of the 17 heartbeats illustrated in Figure 1. The solid
trendline represents unaltered, nonnormalized data, found to
exist in the relation MAP ? 26.14 (CO) ? 38.45 (r ? 0.99). A
linear relation was found in all other experiments as well. The
broken trendline in Figure 2 shows CO and MAP as a percent-
age of baseline values. These normalized data exist in the
relation %MAP ? 0.64 (%CO) ? 1.86 (r ? 0.99). Similar data
were recorded for all IVC occlusions in all experiments and
underwent the same analysis.
Analyses were performed for all IVC occlusions in all exper-
iments (n ? 7) as well as for IVC occlusions 1, 2, and 3
individually. Table 1 illustrates the corresponding relations
that were obtained upon statistical analysis as well as the
corresponding error measurements. Figure 3 illustrates the
universally obtained direct relationship found between per-
centage change in CO and percentage change in MAP. No
statistically significant difference was noted between IVC oc-
clusions 1, 2, and 3 in all experiments (p ? 0.0001).
The results of this experiment demonstrate that in the ab-
sence of changes in SVR, a clear linear trend exists between
CO and MAP during acute decreases in CO in pigs. Statistical
analysis reveals that percentage changes in CO are related to
percentage changes in MAP by the equation %MAP ? 0.60
(%CO) ? 0.41. Overall, these data support MAP as a valid
indicator of changes in CO when SVR is constant. During the
first 10 seconds of fluctuations in CO, MAP should vary with
CO. After 15 seconds or more, however, a baroreceptor re-
sponse is likely to alter this linear relationship.8,12,13
Previous studies have suggested that the CO/MAP relation is
not linear while baroreflex responses are intact. However, the
duration of data collection in these experiments was so long
(several minutes) that changes in SVR assuredly occurred. This
might explain the curvilinear relationship between CO and
MAP seen by Sagawa and Eisner5and Levy et al.6during their
Studies have also suggested that in the presence of vascular
reflex responses, changes in CO account for less than 10% of
changes in the arterial blood pressure, with the remaining 90%
attributed to changes in vascular tone.8,9Experiments in which
the baroreceptors are directly stimulated or in which arterial
pressure is the controlled variable are thus subject to substan-
tial error because of dynamic changes in SVR.3,13,16Our study,
however, varied CO and tested the relation before any reflex
changes could occur. Additionally, we noted no statistically
significant difference in the CO/MAP relation between IVC
occlusions in each pig despite the findings of a previous study
that indicated that subsequent measurements of the CO and
MAP relationship in the same animal over the course of hours
can yield curves with significantly different slopes.4
Some authors maintain that occlusion of the IVC artificially
changes SVR with secondary effects upon CO and MAP. Guy-
ton et al.16found that the effect of changes in venous resistance
upon CO and MAP is eight times less than equivalent changes
in arterial resistance. Venous capacitance has been shown to
be 18 to 30 times greater than arterial capacitance.17Thus
occlusion of the IVC during 10 second testing intervals allows
for storage of blood in the venous tree without producing
noticeable changes in afterload, which would drastically
change CO and MAP.1,17,18
The data obtained in our experiment imply that changes in
MAP within 15 seconds of an intervention can be used to
Figure 2. CO/MAP relation from Figure 1 for 17 beat vena caval
occlusion is represented by solid black trendline. CO/MAP data are
reexpressed as percentage of baseline value on the axes at the top
and right of the graph and are represented by the broken trendline.
Linear regression equations and correlation coefficients are pro-
vided. CO, cardiac output; MAP, mean arterial pressure.
Table 1. Linear Regression Equations for Percentage of
%MAP ? 0.60 (%CO) ? 0.25
%MAP ? 0.63 (%CO) ? 1.62
%MAP ? 0.58 (%CO) ? 0.91
%MAP ? 0.60 (%CO) ? 0.41
n ? number of animals. MAP, mean arterial pressure; CO, cardiac
Figure 3. Overall CO/MAP relation for 19 IVC occlusions in seven
experiments expressed as a percentage of baseline values. Linear
regression equation and p values are provided. CO, cardiac output;
MAP, mean arterial pressure; IVC, inferior vena cava.
PRASSO ET AL.
estimate changes in CO. Examples of such interventions in- Download full-text
clude clinical protocols in which biventricular pacing may be
optimized. A variety of atrioventricular and ventricular-ven-
tricular delays can be tested within 15 second time intervals.
Acute changes in MAP can be considered to reflect any
changes in CO caused by these pacemaker protocol changes.
Initial attempts to apply this principle in clinical data on biven-
tricular pacing have shown low correlation coefficients, re-
flecting scatter in the data and small absolute changes in CO.
These studies are continuing.
Clinically, SVR is quite variable. Many patients undergoing
cardiac surgery exhibit signs of extremely low SVR immedi-
ately after separation from cardiopulmonary bypass that can
only be corrected through use of vasoactive drugs.19Both
patients with diabetes and patients with preoperative ejection
fractions of less than 40% often show higher SVR levels post-
operatively than patients with less compromised circulatory
physiology because of increased preoperative SVR.19,20This
suggests that acute compensatory fluctuations in SVR are lim-
ited in magnitude and duration and that an adaptive mecha-
nism is required to alter SVR for prolonged periods in patients
with initially lower SVR levels. Because SVR cannot be
changed sufficiently by CO and MAP alone to maintain patient
stability in these cases, pharmacologic support is required to
keep patients with a need for large changes in vasomotor tone
viable.19,21Using fluctuations in CO and MAP to predict
changes in SVR is therefore not a reliable method because
there are confounding factors affecting the body’s baroreflex
Future studies would be useful in clarifying whether the
CO-MAP relationship is maintained under other physiologic
circumstances. These studies most significantly might include
an evaluation of the relationship in chronic heart failure to
determine whether it is affected by alterations in the renin-
angiotensin axis and other feedback mechanisms. A series of
IVC occlusions performed in severely hypertensive pigs might
also be useful to determine whether increased vascular filling
pressures produce an altered relationship in extreme values of
CO. Testing at extreme pressures when vessels are distended
might reveal a different relationship between the CO-MAP
values beyond the values for MAP tested in this experiment. In
addition, studies involving acute increases in CO might be
valuable in demonstrating that this equation works in the
opposite manner in which it was tested in this experiment.
The use of MAP for assessing changes in CO has been
validated in this experiment for acute changes (duration ? 10
seconds) from baseline vital signs. We feel confident that in the
first 10 seconds of fluctuations in CO, MAP will accurately
reflect these changes without the interference of SVR.
Supported in part by the National Heart, Lung and Blood Institute of
the National Institute of Health (RO1 HL 48109 to Dr. Spotnitz) and in
part by the Department of Surgery, Columbia University College of
Physicians and Surgeons.
1. Guyton AC: Circulatory Physiology: Cardiac Output and its Reg-
ulation. Philadelphia: Saunders, 1963.
2. Chandran KB: Cardiovascular Mechanics. New York: New York
University Press, 1992.
3. Shepherd JT, Abboud FM: The cardiovascular system, in Geiger SR
(ed) Handbook of Physiology: A Critical, Comprehensive Pre-
sentation of Physiological Knowledge and Concepts. Bethesda,
MD: American Physiological Society, 1983, pp. 483–487.
4. Folkow B: A study of the factors influencing the tone of denervated
blood vessels perfused at various pressures. Acta Physiol Scand
27: 99–117, 1952.
5. Sagawa K, Eisner A: Static pressure-flow relation in the total
systemic vascular bed of the dog and its modification by the
baroreceptor reflex. Circ Res 36: 406–413, 1975.
6. Levy MN, Brind SH, Brandlin FR, Phillips FA: The relationship
between pressure and flow in the systemic circulation of the
dog. Circ Res 11: 372–380, 1954.
7. Mukkamala R, Toska K, Cohen RJ: Noninvasive identification of
the total peripheral resistance baroreflex. Am J Physiol Heart
Circ Physiol 284: H947–H959, 2003.
8. Liu HK, Guild SJ, Ringwood JV: Dynamic baroreflex control of
blood pressure: Influence of the heart vs. peripheral resistance.
Am J Physiol Regulatory Integrative Com Physiol 283: R533–
9. Bevegard BS, Shepherd JT: Circulatory effects of stimulating the
carotid arterial stretch receptors in man at rest during exercise.
J Clin Inv 45: 132–142, 1966.
10. Granger HJ, Guyton AC: Autoregulation of the total systemic
circulation following destruction of the central nervous system
in the dog. Circulation Research 25: 379–388, 1969.
11. Liedtke AJ, Urschel CW, Kirk ES: Total systemic autoregulation in
the dog and its inhibition by baroreceptor reflexes. Circ Res 32:
12. Guyton AC, Douglas BH, Langston JB, Richardson TQ: Instanta-
neous increase in mean circulatory pressure and cardiac output
at onset of muscular activity. Circ Res 11: 431–441, 1962.
13. Guyton AC, Hall JE: Textbook of Medical Physiology. Philadel-
phia: Saunders, 1996.
14. Dean DA, Jia CX, Cabreriza SE: Validation study of a new transit
time ultrasonic flow probe for continuous great vessel measure-
ments. ASAIO J 42: M671–M676, 1996.
15. Akaike H: A new look at the statistical model identification. IEEE
Trans Automatic Control 19: 716–723, 1974.
16. Guyton AC, Abernathy B, Langston JB, Kaufmann BN, Fairchild
HM: Relative importance of venous and arterial resistances in
controlling venous return and cardiac output. Am J Physiol 196:
17. Guyton AC, Armstrong GG, Chipley PL: Pressure-volume curves
of the arterial and venous systems in live dogs. Am J Physiol
184: 253–258, 1956.
18. Allison JL, Sagawa K, Kumada M: An open-loop analysis of the
aortic arch barostatic reflex. Am J Physiol 217: 1576–1584,
19. Salenger R, Gammie JS, Vander S: Postoperative care of cardiac
surgical patients, in Cohn LH, Edmunds LH (ed), Cardiac Sur-
gery in the Adult, 2nd ed., New York: McGraw-Hill, 2003, pp.
20. Christakis GT, Fremes SE, Koch JP, et al: Determinants of low
systemic vascular resistance during cardiopulmonary bypass.
Ann Thorac Surg 58: 1040–1049, 1994.
21. Schwinn DA, Mc Intyre RW, Hawkins ED, Kates RA, Reves JG:
?-1-Adrenergic responsiveness during coronary artery bypass
surgery: Effect of preoperative ejection fraction. Anesthesiology
69: 206–217, 1988.
MAP AND ACUTE CHANGES IN CARDIAC OUTPUT