Cardiac output monitoring: an integrative perspective

Cardiac output monitoring in the critically ill patient is standard practice in order to ensure tissue oxygenation [1] and has been traditionally accomplished using the pulmonary artery catheter (PAC). In recent years, however, the value of PAC has been questioned with some suggesting that its use might not only be unnecessary but also potentially harmful [1]. This notion, together with the availability of new less invasive cardiac output measuring devices, has markedly decreased the widespread use of the PAC [2]. Today, various devices are available to measure or estimate cardiac output using different methods. Some of these less invasive devices track stroke volume (SV) continuously and provide dynamic indices of fluid responsiveness, others allow assessment of volumetric preload variables, and some also provide continuous measurement of central venous saturation via the use of proprietary catheters that are attached to the same monitor. All these variables — together with cardiac output — may result in an improved hemodynamic assessment of the critically ill patient. However, it is important to appreciate that each device has its inherent limitations and that no cardiac output monitoring device can change patient outcome unless its use is coupled with an intervention that by itself has been associated with improved patient outcomes. Therefore, the concept of hemodynamic optimization is increasingly recognized as a cornerstone in the management of critically ill patients and has been shown to be associated with improved outcome in the perioperative [3] and in the intensive care unit (ICU) [4] setting.


Introduction
Cardiac output monitoring in the critically ill patient is standard practice in order to ensure tissue oxygenation [1] and has been traditionally accomplished using the pulmonary artery catheter (PAC). In recent years, however, the value of PAC has been questioned with some suggesting that its use might not only be unnecessary but also potentially harmful [1]. Th is notion, together with the availability of new less invasive cardiac output measuring devices, has markedly decreased the widespread use of the PAC [2]. Today, various devices are available to measure or estimate cardiac output using diff erent methods. Some of these less invasive devices track stroke volume (SV) continuously and provide dynamic indices of fl uid responsiveness, others allow assessment of volumetric preload variables, and some also provide continuous measurement of central venous saturation via the use of proprietary catheters that are attached to the same monitor. All these variablestogether with cardiac output -may result in an improved hemodynamic assessment of the critically ill patient. However, it is important to appreciate that each device has its inherent limitations and that no cardiac output monitoring device can change patient outcome unless its use is coupled with an intervention that by itself has been associated with improved patient outcomes. Th erefore, the concept of hemodynamic optimization is increasingly recognized as a cornerstone in the management of critically ill patients and has been shown to be associated with improved outcome in the perioperative [3] and in the intensive care unit (ICU) [4] setting.
Th e aim of this article is to provide a systematic update of the currently available and most commonly used cardiac output monitoring devices. In addition, an integrated approach for the use of these diff erent devices in critically ill patients will be presented taking into considera tion the devices' technical characteristics, their perfor mance and typical limitations, and also any additional hemodynamic variables they may off er.

Overview of cardiac output monitoring devices
When selecting a cardiac output monitoring device for clinical use, diff erent factors play a role (Table 1): Institutional factors may largely limit the choice of the available devices. On the other hand important device-related factors, e.g., invasiveness (Fig. 1), may restrict the area of application. Moreover, patient specifi c conditions may dictate the use of an invasive or a particular minimallyor non-invasive device.

Invasive cardiac output monitoring
Th e PAC was the clinical standard for cardiac output monitoring for more than 20 years and the technique has been extensively investigated. Its complications are well known and despite developments in recent years, the PAC has a distinct role in patient care. An in-depth review is beyond the scope of this article, but some technical aspects and limitations need to be noted: Cardiac output measurement by intermittent pulmonary artery thermodilution, which is based on the Stewart-Hamilton principle, is considered to be the `reference cardiac output monitoring standard' against which all new cardiac output measuring devices are compared. How ever, operator dependence, various patient conditions (e.g., mitral or tricuspid valve insuffi ciency, shunt) or misplacement of the PAC may infl uence reliable cardiac output assessment [6]. In contrast, continuous cardiac output assessment may overcome some of these limitations. Intermittent thermal fi lament heating induces pulmonary artery temperature changes that are measured via a distal thermistor and matched with the input signal. Based on the cross correlation of in-and output signals intermittent cardiac output values are produced from a thermodilution wash-out curve. Th ese values are then averaged for the display of continuous cardiac output readings, which results in a delayed response time of several minutes after induction of cardiac output changes (e.g., for Opti-Q TM , Abbott, Abbott Park, IL and Vigilance TM catheters, Edwards LifeSiences, Irvine, CA) [7]. A so-called fast response continuous cardiac output catheter (truCCOMS TM , Omega Critical Care, East Klibride, GB) allows a more synchronized continuous cardiac output monitoring [8]. Th e additional hemodynamic variables that can be assessed via PAC and are most often used are conventional fi lling pressures, pulmonary artery pressures, and mixed venous oxygen saturation (SvO 2 ). Th erefore, the PAC is still indicated when additional monitoring of pulmonary artery pressures and SvO 2 is desirable. It is also indicated in situations where less invasive techniques are contraindicated or fail to provide accurate cardiac output values.

Minimally invasive cardiac output monitoring
Minimally invasive cardiac output monitoring devices use one of four main principles to measure cardiac output: Pulse contour analysis, pulsed Doppler tech nology, applied Fick principle, and bioimpedance/bioreactance. Furthermore, devices that use pulse contour analysis can also be classifi ed into calibrated and uncalibrated systems.

Pulse pressure analysis
Pulse pressure analysis is based on the principle that SV can be continuously estimated by analyzing the arterial pressure waveform obtained from an arterial line. Th e characteristics of the arterial pressure waveform are aff ected by the interaction between SV and individual vascular compliance, aortic impedance and peripheral arterial resistance. For reliable cardiac output measurement using all devices that employ pulse pressure analysis technology, optimal arterial waveform signal (i.e., eliminating damping or increased tubing resonance) is a prerequisite. Moreover, it cannot be overemphasized that severe arrhythmias may reduce the accuracy of cardiac output measurement, and that the use of an intra-aortic balloon pump precludes adequate performance of the device. Furthermore, pulse pressure analysis may be of limited accuracy during periods of hemodynamic instability, i.e., rapid changes in vascular resistance. Th is may especially be a problem for uncalibrated pulse pressure analysis. In contrast, calibrated pulse pressure analysis may require frequent re-calibration for accurate cardiac output estimation in these situations. A growing number of calibrated and uncalibrated devices that measure the cardiac output based on the pulse pressure analysis method are available.
PiCCOplus TM system (Pulsion Medical Systems, Munich, Germany): Th e PiCCO TM system uses a dedicated thermistor-tipped catheter, which is typically placed in the femoral artery, in order to assess SV on a beat-to-beat basis. Alternatively a radial or brachial catheter may be employed, but these catheters have to be longer than the femoral one for the adequate assessment of the aortic arterial pressure wave signal. Cardiac output calibration via transpulmonary thermodilution requires the insertion of a central venous line. Th e calibration process is also used for the adjustment of individual aortic impedance and needs to be repeated every eight hours in hemo dynamically stable patients. However, during situations of hemodynamic instability, calibration needs to be done more frequently (eventually every hour) [9]. Nevertheless, a variety of studies have successfully validated the PiCCOplus TM system in diff erent patient populations [10,11].
Th e launch of an uncalibrated device from Pulsion Medical Systems, the PulsioFlex TM system, can be expected in 2011. Th e system will require a specifi c additional sensor, which can be connected to a regular invasive arterial pressure monitoring set.
LiDCO TM plus and LiDCO TM rapid system: Th e LiDCO TM plus and LiDCO TM rapid systems (LiDCO Ltd, London, UK) use the same pulse pressure algorithm (PulseCO TM ) to track continuous changes in SV. Th is algorithm is based on the assumption that the net power change in the system in a heartbeat is the diff erence between the amount of blood entering the system (SV) and the amount of blood fl owing out peripherally. It uses the principle of conservation of mass (power) and assumes that following correction for compliance there is a linear relationship between netpower and netfl ow. Th erefore, the LiDCO systems should be considered as pulse power analysis techniques. Th e LiDCO TM plus requires calibration using the transpulmonary lithium indicator dilution technique, which can be performed via a peripheral venous line [12]. In contrast, the LiDCO TM rapid uses nomograms for cardiac output estimation. Clinical studies have demonstrated reliable estimation of cardiac output using PulseCO as long as no major hemodynamic changes are observed [13]. Regarding the LiDCO TM plus, the reliability of the lithium calibra tion system may be negatively aff ected by high peak doses of muscle relaxants, which cross-react with the lithium sensor. Th is can be tackled if the lithium calibration is performed before or 30 minutes after the administration of a muscle relaxant. Th e LiDCO TM plus system, in combination with a hemodynamic treatment protocol (targeting an oxygen delivery > 600 ml/min/m 2 , was shown to be associated with reduced complications and length of hospital stay in patients after major general surgery [14]. Th e primary indication for the uncalibrated LiDCO TM rapid is its perioperative use for SV optimization. Th erefore, the LiDCOrapid trend analysis is more important than absolute cardiac output values (which may diff er when compared with cardiac output assessed by PAC).
FloTrac TM /Vigileo TM system: Th e FloTrac TM /Vigileo TM system (Edwards LifeSciences, Irvine, USA) requires a proprietary transducer, the FloTrac TM , which is attached to a standard non-proprietary radial or femoral arterial catheter and is connected to the Vigileo TM monitor. Th e FloTrac TM /Vigileo TM system does not require calibration. To estimate cardiac output, the standard deviation of pulse pressure sampled during a time window of 20 seconds is correlated with `normal' SV based on the patient's demographic data (age, sex, height, and weight) and a built-in database containing information about cardiac output assessed by the PAC in a variety of clinical scenarios. Impedance is also derived from these data, whereas vascular compliance and resistance are determined using arterial waveform analysis. After confl icting results of early validation studies, the cardiac output algorithm has been repeatedly modifi ed in the last 5 years. Th is has resulted in an improved performance primarily in perioperative setting [15,16]. Further software modifi cations addressed the issue of limited accuracy during hyperdynamic situations and preliminary data showed improved cardiac output measure ments under these specifi c conditions. However, accuracy of the device during rapid hemodynamic changes remains a major concern [17]. Nevertheless, a study using the Flotrac TM /Vigileo TM system for intraoperative hemodynamic optimization recently demonstrated a decreased complication rate and a reduced length of hospital stay [18].
A new cardiac output monitoring device based on pulse pressure analysis, which is calibrated by transpulmonary thermodilution -the EV 1000 TM /VolumeView TM system from Edwards Lifesciences -is currently being tested and will soon be released for its use in daily practice.
Pressure recording analytical method (PRAM): Another method to estimate SV continuously without calibration is the PRAM -MostCare® (Vytech, Padova, Italy), which is based on mathematical assessment of the pressure signal obtained from an arterial line without calibration.
PRAM has been validated so far in a porcine model under various hemodynamic states [19] and in humans undergoing cardiac surgery [20]. Similar to other devices that use pulse contour analysis, the accuracy of PRAMderived cardiac output is aff ected by the quality of the pressure signal and by factors that interfere with the ability to detect a pressure signal.
Nexfi n TM : Th e Nexfi n TM HD (BMEYE B.V, Amsterdam, Netherlands) is a completely non-invasive pulse pressure analysis device that assesses pulse pressure using photoelectric plethysmography in combination with a volume-clamp technique (infl atable fi nger cuff ). Cardiac output is derived using the so-called Modelfl ow method (simulation of a three-element Windkessel model).
Regarding validation of the device, only limited published data are available [21].

Doppler cardiac output monitoring devices
Cardiac output can be estimated non-invasively using esophageal or transthoracic Doppler probes. Esophageal Doppler devices measure blood fl ow in the descending aorta and estimate cardiac output by multiplying the cross sectional area of the aorta by blood fl ow velocity. Th e aortic diameter is obtained from a built-in nomogram or by direct measurement using M-mode echocardiography. Several esophageal Doppler probes are available commercially: ODM II TM (Abbott, Maiden head, UK), CardioQ TM (Deltex Medical Ltd, Chichester, Sussex, UK), and HemoSonic100 TM (Arrow, Reading, PA, USA). Th e latter device is a combination of a Doppler and an Mmode probe, the production of which has been stopped recently. Th ere are several limitations for the use of esophageal Doppler devices. First, the device measures blood fl ow in the descending aorta and makes an assumption of a fi xed partition between fl ow to the cephalic vessels and to the descending aorta. Although this may be valid in healthy volunteers, this relationship may change in patients with co-morbidities and under conditions of hemodynamic instability. Second, Doppler probes are smaller than conventional transesophageal echocardiography probes and position may change unintentionally, thus limiting continuous cardiac output assessment. Since probe position is crucial to obtaining an accurate measurement of aortic blood fl ow, this device is operator-dependent and studies have shown that 10-12 insertions are required to obtain accurate measurements [22] with an intra-and inter-observer variability of 8-12% [23]. Moreover, aortic cross-sectional area is not constant but rather dynamic in any individual patient. Th us, the use of a nomogram may result in less accurate cardiac output estimation. Despite some limitations of esophageal Doppler devices, their utility appears to be confi rmed by several perioperative hemodynamic optimi zation studies that have consistently demonstrated a reduction in complication rates and hospital length of stay [24].
Alternatively to the esophageal route, the transthoracic approach may be used to assess cardiac output, albeit intermittently. Th e USCOM TM device (USCOM, Sidney, Australia) targets the pulmonary and aortic valves accessed via the parasternal and suprasternal windows in order to assess cardiac output completely non-invasively. Validation studies have revealed confl icting results, which could be explained primarily by the inherent problem of variable signal detection [25,26].

Applied Fick principle
Partial CO 2 rebreathing: Th e NICO TM system (Novametrix Medical Systems, Wallingford, USA) applies Fick principle to carbon dioxide (CO 2 ) in order to obtain cardiac output measurement in intubated, sedated, and mechanically ventilated patients using a proprietary disposable re-breathing loop that is attached to the ventilator circuit. Th e NICO TM system consists of a mainstream infrared sensor to measure CO 2 , a disposable airfl ow sensor, and a pulse oximeter. CO 2 production is calculated as the product of CO 2 concentration and airfl ow during a breathing cycle, whereas arterial CO 2 content is derived from end-tidal CO 2 and its corresponding dissociation curve. Every three minutes, a partial re-breathing state is generated using the attached rebreathing loop, which results in an increased end-tidal CO 2 and reduced CO 2 elimination. Assuming that cardiac output does not change signifi cantly between normal and re-breathing states, the diff erence between normal and re-breathing ratios are used to calculate cardiac output. Th ere are several limitations to this device including the need for intubation and mechanical ventilation with fi xed ventilator settings and minimal gas exchange abnor malities [27]. Variations in ventilator settings, mechanicallyassisted spontaneous breathing, the presence of increased pulmonary shunt fraction, and hemodynamic instability have been associated with decreased accuracy [28]. Th us, this technique may be applied in a precisely defi ned clinical setting to mechanically ventilated patients only.
Pulsed dye densitometry: Th e DDG-330® analyzer (Nihon Kohden, Tokyo, Japan) allows intermittent cardiac output measurement based on transpulmonary dye dilution with transcutaneous signal detection adapted from pulse oximetry (pulsed dye densitometry): Th e concentration of indocyanine green (ICG) is estimated in the arterial blood fl ow by optical absorbance measurements after its venous injection. Cardiac output is calculated from the dye dilution curve according to the Stewart-Hamilton principle. Unfortunately a variety of factors, e.g., vasoconstriction, interstitial edema, movement or ambient light artefacts, may limit reliable intermittent cardiac output assessment [29].

Bioimpedance and bioreactance
Electrical bioimpedance uses electric current stimulation for identifi cation of thoracic or body impedance variations induced by cyclic changes in blood fl ow caused by the heart beating. Cardiac output is continuously estimated using skin electrodes (BioZ®, CardioDynamics, San Diego, USA) or electrodes mounted on an endotracheal tube (ECOM TM , Conmed Corp, Utica, USA) by analyzing the occurring signal variation with diff erent mathematical models. Despite many adjustments of the mathematical algorithms, clinical validation studies continue to show confl icting results [30,31].
Recently, however, Bioreactance® (NICOM®, Cheetah Medical Ltd, Maidenhead, Berkshire, UK) a modifi cation of thoracic bioimpedance, has been introduced [32]. In contrast to bioimpedance, which is based on the analysis of transthoracic voltage amplitude changes in response to high frequency current, the Bioreactance® technique analyzes the frequency spectra variations of the delivered oscillating current. Th is approach is supposed to result in a higher signal-to-noise ratio and thus in an improved performance of the device. In fact, initial validation studies reveal promising results [32,33].

Additional hemodynamic variables
Apart from SV and cardiac output, hemodynamic monitoring devices provide various additional hemodynamic variables (Table 2); namely, static preload variables, functional hemodynamic variables, and continuous central venous oxygen saturation (ScvO 2 ).

Static preload variables
Various cardiac output monitoring devices require a central venous line for calibration of the system. Th us, central venous pressure (CVP) is briefl y reviewed here. CVP is traditionally assessed as an estimate of cardiac preload since true preload, which is defi ned as enddiastolic myocardial fi ber tension, cannot be measured at the bedside. Several factors, however, aff ect CVP readings including impaired right ventricular (RV) func tion, and severe pulmonary or valvular heart disease. Although the majority of physicians use CVP in order to guide fl uid therapy [34], several studies have shown lack of correlation between CVP and SV [35,36]. Moreover, absolute CVP cannot be used to assess preload responsive ness. Th erefore, the utility of CVP is limited and changes in trend over time and cyclic changes induced by mechanical ventilation are more important than absolute numbers. In contrast to the pressure preload variables, the so-called volumetric preload variables are considered to be superior indicators of preload. Global end-diastolic volume (GEDV) and extravascular lung water (EVLW) are static volumetric parameters that are assessed by transpulmonary thermodilution, which is required for the calibration of the PiCCOplus device and the upcoming EV1000/VolumeView device. Diff erent studies have shown a better correlation between GEDV and SV than between the latter and static pressure preload [35]. GEDV could thus be used to better guide perioperative fl uid therapy than pressure preload parameters [37]. EVLW on the other hand can be used to diff erentiate between cardiac versus non-cardiac pulmonary edema, and has been identifi ed as an independent predictor of survival in critically ill patients [38]. It may, therefore, be of value in tailoring therapy in patients with acute respiratory distress syndrome (ARDS).

Functional hemodynamic variables
Pulse pressure analysis devices provide an automated quantifi cation of SV variation (SVV) and some also allow the determination of pulse pressure variation (PPV). Th e basis of these functional variables is cyclic changes in intrathoracic pressure during positive pressure ventilation which induce changes in SV and pulse pressure as a result of a reduction in preload. Th e diff erent functional hemodynamic variables have been shown to be able to predict fl uid responsiveness in various studies [39], whereas static preload variables have not [40]. Nonetheless, it has to be emphasized that cardiovascular and ventilatory limitations, such as arrhythmias, right heart failure, spontaneous breathing activity, and low tidal volume (< 8 ml/kg body weight) aff ect the reliability of these dynamic indices of fl uid responsiveness. Under these circumstances, `passive leg raising' could be employed to assess fl uid responsiveness as it results in an internal fl uid shift from the legs to the central compartment caused by the modifi ed Trendelenburg position.
Th is technique has been demonstrated to reliably determine fl uid responsiveness in critically ill patients [41].

Central venous oxygen saturation
ScvO 2 is used as a global marker of the balance between systemic oxygen supply and demand [42]. It can be easily measured by obtaining a blood sample drawn from a central venous catheter, compared with SvO 2 , which requires placement of a PAC and the withdrawal of blood from the distal port of the catheter. In addition to intermittent measurements using a blood sample and a blood gas analyzer, both ScvO 2 and SvO 2 can be measured continuously using proprietary central venous and pulmonary artery catheters, respectively. Th ere are no outcome studies that compare intermittent versus continuous measurements of ScvO 2 or SvO 2 ; however, the only study that showed a survival benefi t using ScvO 2 as a resuscitation endpoint employed continuous measure ment [43]. Using proprietary catheters, continuous measurements of ScvO 2 can be obtained from both the Vigileo TM and the PiCCO TM systems. As far as its clinical utility is concerned, ScvO 2 has been used as a resuscitation endpoint in patients with severe sepsis and septic shock [43]. It is important to realize that absolute ScvO 2 and SvO 2 values may diff er considerably in diff erent clinical situations; however, a strong correlation of their trends over time has been demonstrated [44].

Integrative concept
Considering the technical features and the typical limitations of the diff erent cardiac output monitoring techniques it is obvious that no single device can comply with all clinical requirements. Th erefore, diff erent devices may be used in an integrative concept along a typical clinical patient pathway (Fig. 2) based on the invasiveness of the devices and the available additional hemodynamic variables ( Table 2). Bioreactance may be used on the ward or in the emergency department to assess cardiac output initially in order to confi rm a preliminary diagnosis. Its use may be expanded in the perioperative and ICU setting. Partial CO 2 -rebreathing requires an intubated and mechanically ventilated patient for cardiac output estimation. Th us, this technique may be primarily used during an operation. Uncalibrated pulse pressure analysis devices may be the primary choice in a perioperative setting as they provide functional hemodynamic variables and thus allow comprehensive hemody namic management. In contrast, calibrated systems may be required when postoperative complications or hemodynamicinstability occur and increased device accuracy or volumetric variables are needed for improved patient management. In the presence of factors that aff ect the accuracy of all minimally invasive cardiac output monitoring devices, or when pulmonary artery pressure monitoring or right heart failure treatment is required, PAC insertion may be required for patient specifi c therapy.

Conclusion
Various devices that allow continuous cardiac output measurement in the critically ill patient are commercially available today. Th eir presence does not completely preclude but does increasingly limit the use of the PAC. A variety of factors (institutional, device related, and patient specifi c) infl uence the selection of a cardiac output monitoring device and clinicians need to understand the underlying principles and the inherent limitations of these devices. A selection of these techniques may be used in an integrative approach along a patient pathway. In combination with ScvO 2 measurements, the assessment of volumetric preload variables and the functional hemodynamic variables that they provide may obviate the need for a PAC in the treatment of critically ill patients.