gilbert - cardiac output measurement

Learning objectives

After reading this article, you should be able to:

C understand the relationship between cardiac output, stroke

volume, heart rate and arterial pressure

CLINICAL MEASUREMENT

Cardiac outputmeasurementMichael Gilbert

C describe how estimates of cardiac output are derived from

other measured variables

C outline additional information derived from each measurement

technique

AbstractCardiac output measurement is used to guide fluid and inotropic drugtherapy. Techniques employ modelling of the circulation to derive es-timates of cardiac output from readily measured variables, includingthermodilution, analysis of arterial pressure waveforms, Doppler mea-surements of blood flow velocity, and electrical bioimpedance.

Keywords Bioimpedance; cardiac output; oesophageal Doppler;pulse contour analysis; thermodilution

Royal College of Anaesthetists CPD Matrix: 1A03, 2A04

Cardiac output is the product of ventricular stroke volume and

heart rate, and estimates of it are used to guide fluid and

inotropic therapy in intraoperative and critical care settings, in an

attempt to improve clinical outcomes.

The pressureeflow relationship

When the stroke volume is ejected from either the left or right

ventricle, pressure is generated in the aorta and pulmonary ar-

teries by a combination volume change and the propagation and

reflection of waves generated by the energy of ejection. Arterial

pressure is the sum of these effects (Figure 1).

Volume change: the great vessels are compliant, so during sys-

tole, more blood is ejected into them than actually leaves. In

diastole, they passively empty into smaller arteries, and return to

their original calibre, at a rate determined by arterial compliance

and vascular resistance.1 The volume of blood entering a vessel

or cardiac chamber must equal the volume of blood leaving,

during each cardiac cycle, or distension would occur. Stroke

volume therefore consists of systolic and diastolic components,

preventing systolic pressure overshoot during rapid ejection, and

allowing for it to be delivered to the arteries throughout the

cardiac cycle.2 Aortic pressure peaks during systole and declines

exponentially during diastole, determined by a time constant ðtÞ.

Wave generation: the energy of stroke volume ejection gener-

ates compression and decompression waves in the arterial sys-

tem, which rapidly propagate and are reflected at points of

branching and calibre change. Waves propagating through the

arterial system are the result of successive ventricular ejections,

rather than harmonic oscillation. This is demonstrated by the

Michael Gilbert MB ChB FRCA is a Consultant CardiothoracicAnaesthetist at Morriston Hospital, Swansea, UK. Conflicts ofinterest: none.

ANAESTHESIA AND INTENSIVE CARE MEDICINE --:- 1

Please cite this article in press as: Gilbert M, Cardiac output measuremen10.1016/j.mpaic.2015.11.005

exponential pressure decline to an asymptote, during the pro-

longed diastole after a premature ventricular complex.3

Measurement techniques

The circulation is a complex system of branching vessels with

variable flow velocity, calibre and compliance, so pressure and

flow vary across measurement sites. Blood flow measurement

creates challenges, because of difficulty siting instruments in or

near blood vessels of interest. The relationships between arterial

pressure, blood flow, vascular resistance and vessel dimensions

are determined by Ohm’s Law and the Hagen-Poiseuille Law of

laminar flow. Haemodynamic monitors measure a spectrum of

variables, including arterial pressure, blood velocity, indicator

dilution, or electrical impedance to estimate cardiac output, and

other values reflecting haemodynamic status.

Measurement of indicator dilution

Indicator dilution employs the principle that a known volume

(V0) and concentration (C0) of an indicator is injected into the

circulation, diluting the indicator within a volume of distribu-

tion (V1). The indicator concentration (C1) is detected at a

remote site, where it initially increases and then decreases in a

non-stepwise manner. If V1 ¼ C0 � V0/C1 and cardiac output (Q)

¼ V1/t, then cardiac output is inversely related to the area under

the concentrationetime curve (Figure 2a).

Thermodilution uses the concept of a thermal indicator, where

cold crystalloid solution injected into the circulation produces a

blood temperature change at the measurement site.4,5 Cardiac

output is inversely proportional to the area under the blood

temperature changeetime curve.

The pulmonary artery catheter was developed as the route by

which boluses of cold crystalloid solution could be injected into

the right atrium, with proximal and distal thermistors to detect

injectate temperature and blood temperature in the pulmonary

artery. The catheter is inserted via a central vein and is floated

into the pulmonary artery using a small inflatable balloon on the

catheter tip. As the temperature change is measured across the

right ventricle, this technique estimates right ventricular stroke

volume.

Continuous cardiac output (CCO) monitors use an element in

the proximal part of the catheter to intermittently heat the sur-

rounding blood, as it passes through the right atrium. The tem-

perature change is much smaller than with cold crystalloid

� 2016 Published by Elsevier Ltd.

t, Anaesthesia and intensive care medicine (2016), http://dx.doi.org/

http://dx.doi.org/10.1016/j.mpaic.2015.11.005

ECG, aortic and radial artery pressure andbioimpedance waveforms

Note systolic pressure augmentation and secondary, non-dicrotic peaks caused by wave effects in the radial artery waveform. Ventricular activation, Q, aortic valve opening, A, aortic valve closing, C, pulmo-nary valve opening, P, and mitral valve opening, M. LVET, left ventricular ejection time; PEP, pre-ejection period. Bioimpedance, Z, bioimpedance change, ΔZ.

Figure 1


injection, making the measurement more prone to error from

‘thermal noise’.

Transpulmonary thermodilution (TPTD) employs temperature

measurement in a peripheral systemic artery (e.g. femoral artery)

following cold crystalloid injection into a central vein. This has

the advantage of not requiring catheterization of the pulmonary

artery. The indicator is diluted to a greater extent, producing a

smaller thermodilution curve. This technique estimates left

ventricular stroke volume.

Lithium dilution cardiac output (LiDCO) monitors employ the

injection of lithium into a central vein, and measurement of



lithium concentration with an ion-sensitive electrode attached to

a peripheral arterial line. Lithium recirculation limits the ability

to repeat measurements at short time intervals.

Additional clinical data derived from indicator dilutionmeasurement

Global end-diastolic volume (GEDV) is a reflection of the ade-

quacy of preload. It represents end-diastolic volume of all cardiac

chambers, and is calculated as the difference between ITTV and

PTV (Figure 2b). The ‘normal’ range is 680e800 ml m�2, which

is greater than the actual end-diastolic volume of the cardiac

chambers.

Extravascular lung water (EVLW) is a reflection of the severity

of pulmonary oedema. It is theoretically represented by the

volume of indicator sequestered in the lung during transit, and is

the difference between pulmonary thermal volume (PTV) and

pulmonary blood volume (PBV). Experimentally derived

formulae are used to calculate EVLW, which estimate the dif-

ference between ITTV and the sum of GEDV and PBV [Reuter,

Bendjelid]. One example is: EVLW ¼ (CO � MTTT) � (1.25 �GEDV).

There is controversy about EVLW and GEDV calculations.

Firstly, CO, MTTT and EDTT are all derived from the same

thermodilution curve, so any error in the detection of the indi-

cator is multiplied. Secondly, there is conflict between the as-

sumptions that the calculation of cardiac output should occur

with no loss of indicator while at the same time the calculation of

EVLW assumes that there is. Thirdly, GEDV is calculated from

cardiac output, assuming that they are invariably related, which

they are not.6

Pulse contour analysis

Early investigators discovered an empirical correlation between

central pulse pressure and stroke volume. Analysis of the pe-

ripheral arterial pressure waveform to derive estimates of stroke

volume and cardiac output necessitates accurate reproduction of

the central aortic pressure waveform and an estimation of arterial

resistance.

The time constant ðtÞ of an exponential decay curve fitted to

the diastolic part of the central aortic pressure waveform pro-

vides a combined estimate of large vessel compliance (C) and

total arterial resistance (R) (Figure 1). To derive a value for

arterial resistance, compliance must be estimated. Current tech-

nology employs two main models:

Windkessel model: so-called because the term was used by

Starling to describe the compliance ‘chamber’ represented by the

distensible aorta (Figure 3a). This model attempts to fit the area

under the curve (AUC) of arterial waveforms measured over

several cardiac cycles to physiological models which predict the

central aortic pressure waveform. These approaches require

calibration against another method of CO measurement, such as

thermodilution or indicator dilution, to calculate a constant

which represents large vessel compliance. Derived values for

compliance and vascular resistance are then entered in a calcu-

lation of cardiac output, and subsequent changes in measured




ln ΔT

THERMODILUTION CURVE

MEAN TRANSIT TIME (MTTT)

EXTRAPOLATED DECLINE

EXP DECAY TIME (EDTT)

ΔT

a

b

RA RV PBV

EVLW

LA LV

INJECTION OF COLD INDICATOR

DETECTION OF INDICATOR IN PULMONARY ARTERY

TRANSPULMONARY DETECTION OF INDICATOR

Thermodilution curves

(a) Upper graph: Temperature change plotted against time. Recirculation of the indicator causes the slope of the curve to flatten out, diverging from the exponential decline. The dashed line shows extrapolation of the curve to exclude the effect of recirculation.Lower graph: Semi-logarithmic plot of temperature change against time, where extrapolation of the line of decline allows accurate calculation of the duration of ‘indicator’ detection. (b) Volume indices are calculations based on estimated cardiac output and specific time intervals derived from the thermodilution curve: Intrathoracic thermal volume (ITTV) = CO × MTTT. Pulmonary thermal volume (PTV) = CO × EDTT.

Figure 2


values are used to estimate cardiac output from the point of

calibration.7

Empirical approach: this technique does not attempt to fit AUC

data to a physiological model, but works from the assumption

that stroke volume is related to the standard deviation of

measured arterial pressure data (sAP), which is proportional to

pulse pressure. This echoes the earliest techniques of estimating

stroke volume, but inaccuracies of the original assumption are

now corrected for by the application of a conversion factor. A

multivariate polynomial equation is used to calculate the con-

version factor, including variables such as skewness and kurtosis



of the data (Figure 3b). Stroke volume is given as the product of

sAP and the correction factor.8

There is a significant factor which confounds calculation of

stroke volume using the peripheral arterial waveform.

Forward-propagating and reflected waves distort the waveform

and, in particular, the exponential diastolic pressure decline.

These waves are also augmented by vasoconstriction and

blunted by vasodilatation, further distorting the waveform in

hypovolaemia and sepsis, precisely when the measurements are

most crucial.

Recent work on wave intensity analysis, has focussed on

modelling which separates ‘reservoir’ (Windkessel) pressure, the




Figure 3 (a) Hydrodynamic and electrical models of the systemic circulation, showing the Windkessel (WK) representing the compliant large ar-teries and the parallel resistances of the branching peripheral arterial system (R), represented in the electrical models as capacitance and resis-tance respectively. The two-, three- and four-component models are refinements which better describe flow and pressure changes in theascending aorta and the peripheral arteries. Reproduced under Creative Commons Licence from Westerhof N, Lankhaar J-W, Westerhof BE.7 (b)Effect of vascular resistance: increased leftward skewness of pressure data is related to increased vascular resistance or vasoconstriction. (c)Effect of large vessel compliance: increased kurtosis (broadness) indicates decreased aortic compliance.


ANAESTHESIA AND INTENSIVE CARE MEDICINE --:- 4 � 2016 Published by Elsevier Ltd.

Please cite this article in press as: Gilbert M, Cardiac output measurement, Anaesthesia and intensive care medicine (2016), http://dx.doi.org/10.1016/j.mpaic.2015.11.005


Oesophageal Doppler waveform

The figure shows the measurement variables, peak blood flow velocity, velocity time integral, ejection time, interpeaktime.

Figure 4


component of arterial pressure related to volume change in the

aorta, from ‘wave pressure’, the component resulting from

augmentation and attenuation by these waves.3,9 The accurate

reproduction of the central aortic pressure waveform would

allow for more accurate calculation of systolic and pulse pres-

sures, and the time constant of the exponential diastolic pressure

decline.

Doppler measurement of velocity

Descending aortic blood velocity is measured by a probe in the

oesophagus, which emits a 4e5 Hz pulsed or continuous Doppler

signal at a shallow angle to the aorta. Ultrasound travels at a

speed of 1540 m s�1 in blood, and is reflected by cellular com-

ponents moving away from the ultrasound emitter. Rarefaction

of the reflected wave increases wavelength, lowering its fre-

quency, which is measured by the probe (Figure 4).

Blood flow velocity is calculated using the Doppler equation,

which is proportional to the difference between the frequencies

of emitted and reflected waves. Two Doppler shifts occur, first

when ultrasound travels between the source and the blood, and a

second when the ultrasound is reflected back to the source. The

velocity calculation should only take account of the Doppler shift

of the reflected wave, so the total frequency change is halved.

The velocity calculation includes a correction factor for the angle



of insonation, which is the angle between the axis of the Doppler

beam and the axis of blood flow. At an angle of <20� the Dopplershift is only reduced by 6% and can be corrected for, but at

angles beyond 20�, the calculation of velocity becomes increas-

ingly inaccurate. At 90� (i.e. with the beam perpendicular to the

axis of blood flow) no Doppler shift is detected.

Cross-sectional area of the aorta is either measured using M-

mode ultrasound, or is estimated using a nomogram.

Stroke volume (SV) is calculated by multiplying the cross-

sectional area by the integral of the blood velocity e ejection

time curve. Cardiac output is calculated as the product of SV and

heart rate, which is calculated using the time between adjacent

velocity peaks.

Additional clinical data derived from Dopplermeasurement

Mean acceleration (MA) is the mean slope of the ascent of the

velocityetime curve. It is an indicator of contractility. Unlike

peak velocity, it is independent of afterload, so it is reproducible

independent of vascular tone.

Corrected flow time (CFT) is an indicator of preload. It is

calculated by indexing the duration of ejection against ‘normal’




THORACIC LENGTH, L

AORTIC CROSS SECTIONAL AREA, AATHORACIC CROSS SECTIONAL AREA, AT

ρB ρT

A bioimpedance model of the thorax

The blood within aorta and great vessels are represented by the inner cylinder, with impedance (ρB)and cross-sectional area (AA). The thorax is represented by the by the outer cylinder, with impedance (ρT) and cross-sectional area (AT).

Figure 5


heart rate of 60 beats min�1. CFT <450 ms indicates that the

ventricular ejection ends prematurely, a marker of

hypovolaemia.

Bioimpedance

The human body is able to conduct electrical current because of

the presence of charged ions within blood and interstitial fluid.

Impedance (Z), a measure of resistance in alternating current

circuits, cannot be directly measured. In the presence of constant

current, bioimpedance is calculated by measurement of voltage,

using Ohm’s law. Current passes differentially through the body

along high impedance and low impedance pathways. The lowest

impedances are in blood (150 U cm�1) and plasma (63 U cm�1),

while the highest are in air (1275 U cm�1) and cardiac muscle

(750 U cm�1).

Thoracic bioimpedance systems apply low-amperage alter-

nating current of 1.4e1.8 mA, at a frequency of 30e75 Hz be-

tween electrodes applied to the base of the neck (thoracic inlet)

and the costal margin (thoracic outlet). Thoracic baseline

impedance (Z0) is inversely proportional to the total current-

conducting fluid content of the thorax (TFC). The individual

contributions to conductivity (the inverse of impedance) of the

intravascular, interstitial and alveolar fluids cannot be separated,

so the thorax is modelled as two concentric cylinders. The inner

cylinder represents the low impedance of blood and the outer

cylinder represents the high impedance components of the rest of

the thorax (Figure 5).

The blood volume of the thorax increases transiently during

systole, increasing the volume of the low impedance conduit,

relative to the rest of the thorax. The volume within venous

capacitance vessels and the pulmonary microcirculation varies

only slightly with respiration, so this change is mainly due to the

increase in the aorta and pulmonary arteries. The change in



bioimpedance (DZ) during ventricular ejection (LVET) is pro-

portional to left ventricular stroke volume.10 This is represented

by the DZ peak between aortic valve opening and closing

(Figure 1).

The calculation of stroke volume, requires estimation of the

contribution to conductivity from the different blood compo-

nents, as these exhibit different impedances. Haematocrit, elec-

trolyte concentrations, age, sex and weight are used as correction

factors in this calculation.

Additional clinical data derived from bioimpedancemeasurement

Systolic time ratio is the ratio between the durations of electrical

and mechanical systole. Pre-ejection period (PEP) represents the

time between the beginning of electrical activation of the

ventricle (q-wave on ECG) and the beginning of ventricular

ejection, or mechanical systole (LVET). This ratio increases in

cardiac failure.

Velocity index is an indicator of the peak velocity of blood in the

aorta. This is affected by both myocardial contractility and

afterload, and can remain in the normal range even as myocar-

dial contractility decreases.

Acceleration index represents the acceleration of blood into the

aorta in the first 10e20 ms of ventricular ejection, and is calcu-

lated from the steep ascent on the DZ-time waveform. It is an

indicator of myocardial contractility, with higher values indi-

cating increased contractility. A

REFERENCES1 G€odje O, H€oke K, Goetz AE. Reliability of a new algorithm for

continuous cardiac output determination by pulse-contour



http://refhub.elsevier.com/S1472-0299(15)00271-4/sref1






analysis during hemodynamic instability. Crit Care Med 2002; 30:52e8.

2 Thiele RH, Durieux ME. Arterial waveform analysis for the anes-thesiologist: past, present, and future concepts. Anesth Analg2011; 113: 766e76.

3 Parker KH. Introduction to wave intensity analysis. http://www.bg.ic.ac.uk/research/k.parker/wave_intensity_web/welcome.htmlMost recently (accessed 9 Jun 2015).

4 Reuter DA, Huang C, Edrich T, et al. Cardiac output monitoringusing indicator-dilution techniques: basics, limits, and perspec-tives. Anesth Analg 2010; 110: 799e811.

5 Fegler G. Measurement of cardiac output in anæsthetized ani-mals by a thermo-dilution method. Q J Exp Physiol 1954; 39:153e64.



6 Bigatello LM, Kistler EB, Noto A. Limitations of volumetric indicesobtained by transthoracic thermodilution. Minerva Anaesthesiol2010; 76: 945e9.

7 Westerhof N, Lankhaar J-W, Westerhof BE. The arterial wind-kessel. Med Biol Eng Comput 2009; 47: 131e41.

8 Edwards Lifesciences. Getting ml/beat from mmHg. http://ht.edwards.com/resourcegallery/products/mininvasive/pdfs/flotrac_algorithm.pdf. (accessed 24 Jun 2015).

9 Mukkamala R, Xu D. Continuous and less invasive central he-modynamic monitoring by blood pressure waveform analysis. AmJ Physiol Heart Circ Physiol 2010; 299: H584e99.

10 Stevanovi�c P, �S�cepanovi�c R, Radovanovi�c D. Thoracic electricalbioimpedance theory and clinical possibilities in perioperativemedicine. Signae Vitae 2008; 3(suppl 1): S22e7.










http://www.bg.ic.ac.uk/research/k.parker/wave_intensity_web/welcome.html

http://www.bg.ic.ac.uk/research/k.parker/wave_intensity_web/welcome.html
















http://ht.edwards.com/resourcegallery/products/mininvasive/pdfs/flotrac_algorithm.pdf
















gilbert - cardiac output measurement

Documents