Performance of cardiac output monitoring in the peri‐operative setting

Adequate cardiac output is essential to maintain oxygen and nutrient delivery to the tissues, but the optimal level of cardiac output around the time of major surgery or critical illness remains undefined. Many core interventions delivered in the peri-operative period, for example, fluids and vasoactive drugs, are given with the aim of increasing cardiac output, whether or not clinicians measure it directly or use it as a therapeutic end-point. Inadequate cardiac output (CO) and oxygen delivery may be encountered in high-risk patients undergoing major surgery or in critical illness, and have been associated with a higher incidence of complications, organ failure or death 1. Despite this, the utility of CO monitoring in the setting of surgery or critical care remains the subject of ongoing, and sometimes intense, debate. For many years, cardiac output estimation using bolus thermodilution via pulmonary artery catheter (PAC) has been considered the standard for accuracy. However, PAC insertion and management is invasive, time consuming, requires a high level of expertise and is associated with risk of complications. Thus, over the last three decades, minimally or non-invasive techniques of continuous CO estimation have become popular alternatives. In this edition of Anaesthesia, Truijen and colleagues report a study examining variability between CO measurements derived using thermodilution with a PAC, pulse contour analysis and non-invasive finger arterial pressure-derived continuous CO measurements in 27 patients undergoing cardiac surgery 2. The aims of this study were to assess the effect of haemodynamic variability on the concordance of CO measurements using three commonly used monitoring devices: single measurement bolus thermodilution using PAC; and continuous pulse contour analysis with invasive and non-invasive arterial pressure monitoring. The authors examined haemodynamic variability under different physiologic conditions ranging from relatively stable conditions with minimal variation in heart rate and blood pressure, to conditions which included much more variable blood pressure, heart rate and the presence of arrhythmia. Measurements were repeated before and after extracorporeal circulation, when changes in peripheral vascular physiology might be more marked. The authors classified variability into four grades: minimal haemodynamic variability, that is, heart rate and arterial pressure variation of less than 5% with no arrhythmias; then all grades of haemodynamic variability with or without the presence of variability in vascular tone detected using the Physiocal interval from the non-invasive CO system. They theorised that variability in haemodynamics and vascular tone would alter observed CO comparisons between continuous vs. discontinuous and invasive vs. non-invasive techniques. Key findings of this study were that, under conditions of relative haemodynamic and physiological stability, cardiac output measurements were comparable between thermodilution and non-invasive measurements. However, under more unstable conditions, these errors became larger, approaching 40%. As the authors note, this may be a result of limitations in both monitoring modalities, with thermodilution CO measurement being affected by haemodynamic variability (arrhythmias, heart rate variability, mean arterial pressure changes) and arterial waveform CO being affected by variability in peripheral vascular physiology. However, the physiological conditions which the authors defined as unstable or as having a high degree of variability in this study are commonly encountered in the peri-operative setting: arrhythmia; mechanical ventilation; thoracotomy; extracorporeal circulation; and changes in vascular tone. The authors argue that unless this variability is quantified in experimental protocols, studies comparing different cardiac output methodologies in unstable clinical settings cannot be meaningfully interpreted. The accuracy of newer, minimally invasive techniques of cardiac output measurement has been studied over many years, usually as each new device or technique emerges. Bolus thermodilution is commonly used as the reference standard for such studies. The accuracy and precision are compared, often under conditions of relative haemodynamic stability in healthy patients and not during the range of physiological disturbance or patient positioning that might be reasonably expected in major surgery. Historically, a percentage of error of ≤ 30% has been described as a somewhat arbitrary acceptable limit of agreement of accuracy between techniques 3. This is calculated by dividing the 95%CI of the bias by the mean value of cardiac output from both methods. This figure remains in widespread use, even in more contemporary studies 4. A recent meta-analysis reviewed accuracy and precision of non-invasive CO monitoring in the peri-operative setting in a total of 37 studies found an overall percentage error of 47%, with a high level of heterogeneity between studies. Thus, accuracy and precision, certainly by conventional measures, would appear to be inadequate for most non-invasive technologies. However, as few non-invasive CO monitoring modalities can achieve this level of accuracy or precision, its relevance has been questioned 5. Truijen et al. rightly point out that in a dynamic clinical setting, the accepted 'gold standard' for cardiac output measurement, the PAC, itself has limitations which may contribute to the large errors seen in these studies. This study questions the approach of uncritically accepting PAC-based measurements as a gold standard when considering clinically useful cardiac output measurement; all of the modalities studied measure cardiac output by different methods and none of them measure it directly. The intense focus on comparing the accuracy and precision of newer CO modalities with existing technologies – typically in very controlled settings – may have distracted clinicians from two critical questions. Firstly, whether a device is suitable for the specific clinical setting in which it is being used. Secondly, whether the measurements from these devices can usefully guide therapies which may improve patient outcomes. Squara et al. have argued that comparing continuous CO measurements with bolus thermodilution is akin to comparing cinematography with still photography, and that the ability of newer technologies to track changes in cardiac output with precision, and in a timely manner, are more useful than the absolute value for CO provided 6. The advantage of modern continuous cardiac output technologies over traditional PAC-based measurements is twofold: firstly, the technology is less invasive and therefore associated with less complications; secondly, it rapidly tracks changes in cardiac output and gives the clinician an ability to respond in a more timely fashion. The PAC may have a response time of up to 10 min making it unsuitable in many peri-operative settings. In continuous cardiac output monitoring using pulse contour analysis, the haemodynamic variation caused by heart–lung interactions, or the cardiovascular response to a rapid fluid bolus, may actually be useful as indices of preload and fluid responsiveness, and have been incorporated in to the information provided by these monitors as a useful clinical end-point. Rapid tracking of trends in cardiac output, along with these so-called dynamic indices of preload, can allow interventions such as fluids, vaso-active drugs, patient positioning or other to be delivered in real time and against therapeutic end-points. The PAC remains useful to diagnose and characterise complex haemodynamic disturbances, although this role is being superseded by expertise in peri- or intra-operative echocardiography. Transoesophageal and transthoracic echo are being used with increasing frequency in the peri-operative period and can provide a rapid assessment of haemodynamic status including: indices of preload and afterload; left ventricular systolic and diastolic function; right heart function; and valvular or other structural abnormalities. The strength of continuous cardiac output technologies is in providing minimally invasive, real-time information on trends in haemodynamic status and to guide response to therapy. In this case, absolute accuracy may be of less importance than precision and response time. There is now a bewildering array of continuous CO monitoring technologies available, of seemingly variable accuracy, and the bigger question remains: does using them to guide therapy in the peri-operative period actually improve patient outcomes? Although thermodilution cardiac output measurement can boast a high level of accuracy and precision, trials of PAC-guided management have failed to demonstrate benefit to patients in either the peri-operative or critical care setting 7. Although there is no evidence that the use of any cardiac output monitor per se improves outcomes, studies of peri-operative treatment algorithms based on minimally invasive continuous cardiac output monitoring have been more encouraging, suggesting reduced complications and duration of hospitalisation, particularly in major elective gastro-intestinal or orthopaedic surgery 8. These studies typically involve repeated fluid challenges to assess fluid responsiveness, emphasising the importance of accurate trend tracking in monitors rather than a requirement to give an accurate absolute value of cardiac output. A recent large clinical trial of restrictive vs. liberal fluid administration has reported that a restrictive fluid regimen was not superior to a liberal one and that lower rates of acute kidney injury were seen in the liberal group 9. Although assessing the impact of cardiac output monitoring was not a primary goal of the trial, the lack of benefit from 'unguided' fluid restriction suggests that there is still a place for individualised fluid regimens in surgical patients at high risk of complications, perhaps guided by cardiac output monitoring. Although studies comparing different cardiac output modalities may help refine the ongoing iterations of cardiac output-based treatment algorithms, large, pragmatic, clinical effectiveness trials are also needed to determine whether the complex intervention of monitor and algorithm is beneficial to patients when applied in a real-world clinical setting 10. Two such studies are currently underway: OPTIMISE-2 (http://optimiseii.org/) will investigate the use of a minimally invasive continuous cardiac output-guided algorithm in patients undergoing major elective gastro-intestinal surgery, and FLO-ELA (http://www.floela.org/) will test a similar intervention in the emergency laparotomy setting. Until these studies report, and demonstrate tangible patient benefit, the indication for use of this technology remains uncertain. Truijen et al. have underlined that all physiological monitors have limitations, and that the degree of interference from physiological variability is heavily dependent on the interaction between the underlying technology and the clinical context. In most commonly encountered peri-operative settings, we are unlikely to be able to achieve perfection in cardiac output measurement, but this may not be necessary. The existing evidence base suggests that a variety of cardiac output measurement techniques may be suitable to guide robust, usable algorithms, and a key focus must now be on confirming this effect in key patient groups by completing definitive large multi-centre trials. Patient benefit must be the ultimate goal of any technological development; there is no evidence that peri-operative clinicians simply having access to more – and more accurate – haemodynamic measurements results in meaningful improvements in patient care. ME has received an honorarium for speaking for Edwards Lifesciences. No other competing interests declared.