To date, the esophageal Doppler (ED) has been one of the most rigorously studied noninvasive CO measurement modalities. Side et al. described ED in 1971 and it was later refined by Singer et al. [25,26]. This technique uses a Doppler probe placed in the esophagus to measure blood flow in the descending thoracic aorta. The ED uses the Doppler Shift principle, which implies that when a transmitted sound wave is impeded by a structure, the reflected sound wave will vary in a frequency dependent manner with the structure’s characteristics. In the case of a fluid filled tube, such as the aorta, the magnitude of Doppler shift will vary in direct proportion to the velocity of flow in the tube (Fig. 27.1). Thus, the reflected sound wave can be used to determine flow velocity in the descending aorta. Multiplying this flow velocity by the ejection time and the cross-sectional area of the aorta provides an estimate of the stroke volume (SV). As this measurement does not account for the component of total stroke volume that travels to the coronary, carotid, and subclavian arteries, a correction factor must be applied to estimate the total SV. CO is then calculated by multiplying corrected stroke volume by the heart rate. The original versions of the ED system provided only Doppler shift data; therefore, the cross-sectional area of the aorta was estimated from a nomogram based on a patient’s height, weight, and age. Subsequently, a combined Doppler and ultrasound probe has been introduced to provide estimates of both aortic flow velocity and cross-sectional area [27]. The descending aortic cross-sectional area measured by this model correlated very well with that measured by transesophageal echocardiography. In addition, aortic blood flow measured with this model was well correlated with CO as measured by thermodilution [27].

Beyond providing an estimate of CO, ED systems can provide information about the preload and the contractility of the heart. Singer et al. analyzed the flow-velocity waveform derived from an ED system and discovered that the corrected flow time (FT_c) correlated with preload [26,28] (Fig. 27.2). These studies further demonstrated that as preload was increased or
decreased, the corrected flow time increased or decreased, respectively [26,28]. It is not clear, however, if following trends in FT_c in response to volume loading is superior to following trends to SV [29]. Wallmeyer et al. described a correlation between the peak velocity measured by Doppler and contractility measured by electromagnetic catheter measured flow [30]. Singer et al. further substantiated this finding by demonstrating that dobutamine infusions increased peak flow velocities measured by an ED system in a dose-dependent fashion [31]. These observations suggest that an experienced operator may be able to extrapolate useful hemodynamic parameters beyond the CO, through careful data synthesis.

The clinical usefulness of the ED system is still being determined. The majority of recent studies that have compared this system to the “gold standard” of thermodilution have been performed in either intraoperative or postoperative settings and have revealed mixed results. One single-center study of 35 patients that compared ED measurements of CO to simultaneous measurements of CO by thermodilution during off-pump coronary artery bypass graft showed very poor correlation between the two techniques [32]. Other studies, including a meta-analysis of 11 trials, have shown that ED systems are better at following changes in CO in response to fluid challenges than they are at measuring the absolute CO [33,34,35]. The authors of the meta-analysis also made an important point when discussing the reliability of comparing ED systems to thermodilution. They argued that the poor reproducibility inherent in the thermodilution technique will likely affect the limits of agreement between ED systems and thermodilution even if ED systems were reliable [33]. This concept was described by Bland and Altman [36] and has important implications when comparing the accuracy of absolute CO measured by any system when compared to thermodilution.

While comparing ED systems to thermodilution, technical advantages and disadvantages deserve consideration (Table 27.1). One advantage of the ED system is that it is continuous. Unlike the traditional bolus thermodilution techniques, an ED system can continuously display CO, which allows earlier recognition of hemodynamic deterioration or improvement in system responsiveness to a therapeutic intervention. In addition, an ED probe can be placed in minutes and has been associated with a low incidence of major iatrogenic complications [37,38,39]. Some data suggest that once inserted, an esophageal probe can be left in situ safely for more than 2 weeks [40]. One study determined that the training required to become proficient in the use of ED consisted of no more than 12 patients [41]. Furthermore, as the esophagus is a nonsterile environment, it is logical to assume that the infectious risk of ED probe use is less than that of a PAC placed percutaneously.

There are also technical disadvantages to the ED system. One is the high up-front cost of the system itself as compared to the PAC apparatus. This cost may represent a very real limitation in the number of systems that a facility can purchase and maintain. This financial obstacle must be balanced with the likelihood that multiple patients would have need of this system simultaneously, which would necessitate multiple systems. Another disadvantage of this system is that it can only be used in the intubated patient. Although a large percentage of critically ill and/or surgical patients who would benefit from this system fit this criterion, the nonintubated patient would be more problematic. Additional concerns would include the likely need for repositioning or recalibration in the ICU patient. Though surgical patients are often immobile, ICU patients are often repositioned frequently to prevent skin breakdown or to facilitate improved oxygenation. Such movements will increase the chance of probe position changes that will require frequent calibration and repositioning. Finally, Roeck et al. suggested that there is significant interobserver variability when measuring changes in stroke volume in response to fluid challenges with ED [35]. Poor reproducibility may limit the utility of this system.

As the ED is used more widely, outcome data will be crucial. To date, the majority of research has focused on the technique’s validity and feasibility. One notable study which compared intraoperative ED use with conventional monitoring during femoral neck fracture repair found a faster recovery time and significantly shorter hospital stay in the ED group [42]. Similarly, Gan et al. demonstrated in a prospective randomized trial of patients undergoing major elective surgery that stroke volume optimization using ED shortened hospital length of stay and resumption of PO intake as compared to conventional intraoperative care [43]. This latter finding may be due to less gut hypoperfusion which has also been demonstrated with the use
of ED [44]. A recent meta-analysis of nine trials of perioperative ED use also found improvements in length of stay as well as time to resuming an oral diet [45].

Although the above-mentioned data suggest that perioperative outcomes are improved with the use of ED, there are no robust parallel data for nonoperative ICU patients. The ultimate use of ED will depend on further outcome data, availability of equipment and local experience and expertise.

Pulse Contour Analysis (PCA) is another modality for measuring CO noninvasively that has been extensively studied. This method relies on the theory, first described by Frank in the early part of the twentieth century, that SV and CO can be derived from the characteristics of an aortic pressure waveform [46]. Wesseling et al. eventually published in 1983 an algorithm to link mathematically SV and the pressure waveform [47]. This original version calculated SV continuously by dividing the area under the curve of the aortic pressure waveform by the aortic impedance. As aortic impedance varies between patients, it had to be measured using another modality to initially calibrate the PCA system. The calibration method usually employed was arterial thermodilution. Aortic impedance, however, is not a static property. It is based on the complex interaction of the resistive and compliant elements of each vascular bed, which are often dynamic, especially in hemodynamically unstable patients. Since the first PCA algorithm was introduced, several unique algorithms have been created to model accurately the properties of the human vascular system for use in PCA systems.

PCA involves the use of an arterial placed catheter with a pressure transducer, which can measure pressure tracings on a beat-to-beat basis. Such catheters are now routinely used in operating rooms and ICUs as they provide a continuous measurement of blood pressure that is superior to intermittent noninvasive measurements in hemodynamically unstable patients. These catheters are interfaced with a PCA system, which uses its unique algorithm as well as the initial aortic impedance calibration data from a thermodilution measurement of CO to provide a continuously displayed measurement of CO. Obviously, the reliability of a PCA system depends upon the accuracy of the algorithm that it employs. Because each algorithm is unique in the weight that it ascribes to each element of vascular conductivity, it is impossible to ensure that a system will be able to reproduce the results of another system under similar conditions [48]. Keeping this in mind, one cannot conclude that all systems are equally reliable.

PiCCO (Pulsion SG, Munich, Germany) is a PCA system that has received considerable attention in the literature. Numerous studies have demonstrated good correlation between this system and pulmonary thermodilution in both critically ill and surgical patients [49,50,51,52,53]. Notably, this system did not require recalibration during these study periods, which were performed under static ventricular loading conditions. The system involves the placement of a femoral arterial catheter that is passed into the abdominal aorta. In addition to a pressure transducer, the catheter also contains a thermistor for arterial thermodilution. The system is calibrated by injecting cold saline via a central venous catheter at the right atrium in a manner similar to pulmonary arterial thermodilution. Instead of using a thermistor in the pulmonary artery, however, the thermistor on the femoral arterial catheter allows calculation of CO. This initial value of CO is then used to calibrate the PCA system that is attached to the arterial catheter. Because the arterial catheter is used for calibration, a PAC is not necessary. When compared with pulmonary artery thermodilution, the arterial thermodilution method was found to be accurate, implying that it is an acceptable method for calibration of a PCA system [49,50,51].

More recently, a novel PCA system known as the Flotrac (Edwards Lifesciences, LLC, Irvine, CA) has been introduced. It is designed to “autocalibrate” on a continuous basis. It calculates stroke volume using a general equation: SV = K × pulsatility, where K is a constant including arterial compliance and vascular resistance [54]. This constant is initially derived by patient variables such as height, weight, sex, and age by using a method described by Langewouters et al. [55] and is subsequently adjusted once per minute using arterial waveform characteristics. Pulsatility is determined by analyzing the standard deviation of the arterial pressure waveform over preceding 20-second intervals. Thus, the variables used to calculate SV are updated at least once per minute. This algorithm offers the advantage of not needing an alternative method for calculating CO for calibration purposes. When compared to pulmonary artery catheter thermodilution in a postcardiac surgery setting, this system showed good correlation over a wide range of COs. In addition, it appears that a radial artery catheter is just as accurate as a femoral artery catheter in this setting, which is another advantage of this system [54].

As mentioned earlier, the initial trials studying PCA systems used data from static ventricular loading conditions. Both the critically ill and the intraoperative patient, however, often experience rapid changes in ventricular preload. The accuracy of the PiCCO system with dynamic changes in preload was addressed in a subsequent study, which used a modified algorithm. Felbinger et al. showed that changes in CO in response to preload could be accurately measured in a cardiac surgical ICU population when compared to pulmonary thermodilution [56].

Although being able to monitor changes in CO during volume loading is important, being able to predict a priori when a patient would benefit from volume loading is perhaps more useful. Pulse pressures commonly vary throughout the respiratory cycle. Pulse pressure variation (PPV) is defined as the result of the minimum pulse pressure subtracted from the maximum pulse pressure divided by the mean of these two pressures.

The magnitude of the PPV in a patient can predict preload responsiveness [57,58,59]. Analogous to PPV, an additional piece of data that PCA systems can provide is the stroke volume variation (SVV). The SVV represents the change in percentage of SV over a preceding time period as a result of changes in SV due to ventilation. So far, the ability to use SVV to determine preload responsiveness has yielded mixed results. Reuter et al. found that SVV reliably decreased as cardiac index increased in response to preloading with colloids in ventilated postoperative cardiac surgical patients [60]. This finding supports the argument that the magnitude of SVV may be used to predict preload responsiveness. It is important to note that the tidal volumes used in this study were supraphysiologic (15 mL per kg), which results in a larger SVV and a resultant increase in the accuracy of this approach. Subsequently, another study used a smaller tidal volume (10 mL per kg) in a similar patient population and could not demonstrate a reliable relationship between SVV and an increase in cardiac index in response to preloading [61]. This finding suggests that when using lower tidal volume ventilation strategies, which are optimal for acute respiratory
distress syndrome (ARDS), PCA-derived SVV should not be used to estimate preload responsiveness.

Overall, the pulse contour analysis system offers several advantages over the traditional “gold standard” of pulmonary artery thermodilution (Table 27.2). Depending on the system, only an arterial catheter (Flotrac) or an arterial catheter and a central venous catheter (PiCCO) are required, both of which are commonly in place in critically ill and surgical patients. Thus, PACs and their possible risks can be avoided when using these systems. The PCA systems also provide a continuous measurement of CO as opposed to the intermittent nature of traditional thermodilution systems.

As with any system, there are disadvantages to the PCA system as well. The ability to use this system to determine preload responsiveness is questionable in patients who are being managed with recommended ventilatory strategies. In addition, some data suggest that in patients who have marked changes in blood pressure, the algorithm is not able to model adequately the changes in vascular resistance and compliance and, therefore, the accuracy of the measured CO declines [62]. Furthermore, a similar breakdown in the accuracy of measured CO has been suggested during the administration of vasoconstrictors [63], which are common in the critically ill patient.

The clinical utility of the pulse contour analysis system is still being determined. Future studies that may help in defining the system’s clinical role could focus on several points. First, a better understanding of how changes in blood pressure and vasoconstrictor use affect the accuracy of a particular system’s algorithm will help to determine when a system needs to be recalibrated to maintain its accuracy. In addition, an analysis of how SVV predicts preload responsiveness at lower tidal volumes will provide more applicable information. Finally, a paucity of data regarding how PCA systems affect patient outcomes exists at present. Comparisons between the outcomes seen with this system and pulmonary artery thermodilution may provide convincing evidence about the real usefulness of PCA. In particular, the common question “will the patient respond to fluids?” may be replaced by the question “should the patient be given fluids?” once adequate outcome data are available.

The Fick equation for calculating CO has been known for over a 100 years. Its underlying principle states that for a gas (X) whose uptake in the lung is transferred completely to the blood, the ratio of that gas’s consumption (VX) to the difference between the arterial (C_aX) and venous (C_vX) contents of the gas will equal the CO. In its original form, Fick used the example of oxygen (O₂) and described the following equation:

For this equation to be accurate, several conditions must exist. The first is that blood flow through the pulmonary capillaries must be constant. In order for this to occur, the right and left ventricular outputs must be equal (i.e., steady state) and there must be no respiratory variation of pulmonary capillary flow. Another condition critical to this method’s accuracy is an absence of shunts. As this method is dependent upon using gas exchange to calculate CO, any blood that does not participate in gas exchange will result in underestimation of CO. Furthermore, oxygen uptake by the lung itself must be minimal to maintain the integrity of this equation.

Although possible, the accurate measurement of [V with dot above]O₂ is clinically challenging, especially in patients who require high FiO₂ [64]. This prompted investigators to focus on using carbon dioxide production ([V with dot above]CO₂) in place of [V with dot above]O₂ [65,66,67]. As [V with dot above]O₂ is equal to [V with dot above]CO₂ divided by the respiratory quotient, they determined that CO could be calculated by [V with dot above]CO₂ divided by the arteriovenous difference between O₂ concentrations as well as the respiratory quotient (R). To measure O₂ concentrations continuously, arterial and venous oximeters were used to measure oxygen saturation (SO₂) and concentration was determined based on measured hemoglobin (Hgb) levels. This technique, therefore, relied upon the assumption that both R and hemoglobin levels remained constant during the measurement period.

Using this method, one study found good correlation with CO determined by thermodilution [67]. The drawback to this approach, however, is the need for an invasive central venous catheter to measure accurately venous oxygen saturations as well as initially to calibrate the system and determine R. Subsequently, the partial carbon dioxide rebreathing method was introduced in an attempt to avoid the need for such catheters.

The partial CO₂ rebreathing method is based upon the Fick equation for CO₂ [68]: