Adolph Fick described the first method of CO estimation in 1870 [7]. Fick described how to compute an animal’s CO from arterial and venous blood oxygen measurements. Fick’s original principle was later adapted in the development of Stewart’s indicator-dilution method in 1897 [8], and Fegler’s thermodilution method in 1954 [9]. The introduction of the PAC in 1970 and its subsequent use in performing thermodilution measurements in humans translated the ability to measure CO from the experimental physiology laboratory to multiple clinical settings [10]. The direct Fick method was the reference standard by which all other methods of determining CO were evaluated until the introduction of the PAC. Currently the PAC is considered the “Gold Standard” against which other devices are compared. Remarkably, the accuracy of the CO measurements as determined by the PAC has never been established. Electromagnetometry and ultrasound using aortic flow-probes most closely represent a true “Gold Standard” for determination of CO but can only be performed in instrumented animals [11–13]. Despite the ubiquitous use of the PAC remarkably few studies have investigated the accuracy of the CO measurements as determined by thermodilution. A number of studies have compared the thermodilution CO with that measured by the Fick technique. These studies have reported a percentage error of between 56 and 83 % (with <30 % being clinically acceptable) [14–16]. Philips at al compared thermodilution CO with surgically implanted ultrasonic flow probes in an ovine model [11]. The percentage bias and precision was −17 % and 47 % respectively; the PAC under-measured dobutamine-induced CO changes by 20 % (relative 66 %) compared with the flow probe. This study found that the PAC was an inaccurate measure of CO and was unreliable for detection of CO changes less than 30 %. Critchely et al. using a similar methodology in pigs reported a precision of 26 % [17]. These studies suggest that the true CO has to change by at least 25 % to be detected by the PAC. Furthermore, the required change may be as high as 100 % depending on the monitor being used [18]. It is likely that multiple factors interact to affect the accuracy of the thermodilution CO calculation [19].

Transpulmonary thermodilution (TPTD) similar to the PAC calculates the CO by the indicator dilution method using the modified Stewart-Hamilton equation. With this method a known quantity of cold injectate is delivered via a central venous catheter and mixing of the thermal indicator occurs as it passes through the right atrium and ventricle, pulmonary circulation, left atrium, ventricle and aorta. A thermistor-tipped arterial line quantifies the change in temperature over time in a large proximal artery (femoral artery). A mono-exponential transformation of the curve with extrapolation of a truncated descending limb back to baseline allows calculation of area under the curve for CO measurement. TPTD suffers from many of the errors and limitations associated with CO determined by the PAC. However, the reproducibility of the CO measurements by TPTD appears significantly better than that of the PAC with a precision of about 7 % (compared to 25 % for the PAC) [20].

The extravascular lung water (EVLW) is the amount of water that is contained in the lungs outside the pulmonary vasculature, that is, the sum of interstitial, alveolar, intracellular, and lymphatic fluids, except pleural effusion. EVLW can be calculated from the descending limb (indicator dissipation) of the TPTD curve and is an accurate method of quantifying the degree of pulmonary edema (hydrostatic and permeability) [21]. An increased value of EVLWI is thus the pathophysiological hallmark of hydrostatic as well as inflammatory lung edema. This technique has been shown to compare favorably with the double indicator dilution technique and the ex–vivo gravimetric method [22–24]. Furthermore, this technique can detect small (10–20 %) increases in lung water [25]. The “normal” value for EVLWI is reported to be 5–7 mL/kg with values as high as 30 mL/kg during severe pulmonary edema. EVLW should be indexed to IBW rather than actual body weight [26]. The best EVLWI cut-off value to discriminate between normal lungs and lungs with diffuse alveolar damage is around 10 mL/kg [27]. EVLW has been demonstrated to be an accurate indicator of the severity of lung injury and a reliable prognostic indicator in patients with sepsis-induced acute lung injury [28, 29]. EVLW and the pulmonary vascular permeability index (PVPI) measured by TPTD have been demonstrated to be independent risk factors of day-28 mortality in patients with ARDS [30]. The EVLW is a very useful parameter to guide fluid removal (or not) in patients with ARDS and ARF. In addition, it is likely that using EVLW to guide fluid therapy in hemodynamically unstable ICU patients may reduce positive fluid balance, duration of mechanical ventilation and ultimately patient outcome (see Chap. 9).

The concept of pulse contour analysis is based on the relation between blood pressure, stroke volume (SV), arterial compliance, and systemic vascular resistance (SVR) [31]. If arterial compliance remains unchanged the area under the systolic portion of the arterial waveform is proportional to the stroke volume. The SV or CO can be calculated from the arterial pressure waveform if the arterial compliance and SVR is known. Although the pulse contour systems which are commercially available use different pressure-volume conversion algorithms, they are based on this basic principle. These systems can be divided into three categories:

Clinical data suggests that only those pulse contour devices that are calibrated to an external method have acceptable clinical accuracy (vascular compliance is measured rather than calculated using predictive algorithms). Furthermore, these devices should be re-calibrated when vascular tone changes (e.g. use of a vasoconstrictor or vasodilator) [32]. It would appear than a number of these monitors are no better than random number generators and should be used with great caution [32].

The esophageal Doppler technique measures blood flow velocity in the descending aorta by means of a Doppler transducer placed at the tip of a flexible probe. The probe is introduced into the esophagus of sedated, mechanically ventilated patients and then rotated so that the transducer faces the descending aorta and a characteristic aortic velocity signal is obtained. The CO is calculated based on the diameter of the aorta (measured or estimated), the distribution of the CO to the descending aorta and the measured flow velocity of blood in the aorta. As esophageal Doppler probes are inserted blindly, the resulting waveform is highly dependent on correct positioning. The clinician must adjust the depth, rotate the probe and adjust the gain to obtain an optimal signal [33]. Poor positioning of the esophageal probe tends to under-estimate the true CO. There is a significant learning curve in obtaining adequate Doppler signals and the correlations are better in studies where the investigator is not blinded to the results of the CO obtained with a PAC [34]. The greatest utility of the esophageal Doppler appears to be in the peri-operative setting (see Chap. 11).

A completely non-invasive Doppler technology, the USCOM (Ultrasound CO monitor, USCOM, Sydney, Australia), utilizes transaortic or transpulmonary Doppler ultrasound flow tracings to calculate cardiac output as the product of stroke volume and heart rate. Stroke volume is calculated from a proprietary algorithm applying ultrasound principles of blood velocity–time integral (VTI) measurements in the ventricular aortic/pulmonary outflow tract. Studies comparing USCOM measurements of cardiac output to those obtained with the PAC, TPTD, and aortic flow probes have shown reasonable agreement [11, 35–37]. The use of Doppler ultrasound to determine cardiac index has several inherent technological limitations. Potential sources of variation exist in the estimation of aortic/pulmonary outflow tract area, the determination of velocity-time integral as well as the variability with operator dependent measurements. With USCOM, the aortic/pulmonary outflow tract area is not directly measured, but calculated from a proprietary anthropometric algorithm based on the subject’s body height. Stroke distance is simply the distance a red blood cell travels per systolic stroke. This is measured as VTI of the Doppler flow profile of each systolic stroke. Thus, the accuracy of the USCOM technology depends on obtaining accurate, reproducible VTI values. A precise VTI measurement requires a good flow signal and its correct interpretation, both of which are heavily dependent on the subject and the operator. An improper technique of poor Doppler ultrasound beam alignment with blood flow at the aortic/pulmonary outflow tract will lead to suboptimal VTI measurements. A further limitation of this technique is that it is not conducive to continuous monitoring.