Statistics from Altmetric.com
Over 100 years ago Karl Ludwig stated that:
The fundamental problems in the circulation derive from the fact that the supply of adequate amounts of blood to the organs of the body is the main purpose of the circulation while the pressures that are necessary to achieve it are of secondary importance; but the measurement of flow is difficult while that of pressure is easy so that our knowledge of flow is usually derivatory.
Are we, in 1998, any nearer to the routine monitoring of flow? The more salient question may be whether flow measurement is actually useful in terms of clinical management or patient outcome. The wherewithal to monitor flow exists. Techniques for cardiac output measurement, albeit of varying accuracy, invasiveness and complexity, have been available commercially for the past 25 years, although their use in the UK is relatively sparse, both in intensive and coronary care units1 and operating theatres.2 The national confidential enquiry into perioperative deaths2 covering the years 1993–94 reviewed 1802 deaths occurring within 30 days of an operation. Three quarters of these patients were deemed moderate to very high risk, yet only 4.2% had a Swan-Ganz (pulmonary artery) catheter in situ during surgery.
The Swan-Ganz catheter: friend or foe?
Several studies have shown outcome benefit from flow directed haemodynamic manipulation in the high risk surgical patient using either invasive3 4 or non-invasive5 6monitoring techniques. Only sporadic studies have demonstrated an advantage for the critically ill ICU patient.7 However, a recent retrospective study by Connors and colleagues8suggested that patients receiving a Swan-Ganz catheter on day 1 of their ICU admission were 39% more likely to die compared with patients matched for disease and illness severity by complex statistical manipulations who did not receive the catheter. The accompanying editorial9 advocated an immediate moratorium on catheter use, a situation10 parallelling an earlier study by Goreet al, 11 who found an increase in mortality in patients with acute myocardial infarction, the only perceived change in management over historical controls being the increased use of Swan-Ganz catheters.
The study by Connors et al has been heavily criticised; nevertheless, it served as the catalyst for a recent US consensus meeting,12 which acknowledged the lack of conclusive data showing benefit or harm, yet reaffirmed its likely worth. A prospective, randomised trialrather than an immediate ban was recommended, a view supported by the US Food and Drug Administration.
It is more likely that fault, if any, lies with how the catheter is used rather than from any intrinsic damage caused by the catheter itself. To contemplate a study comparing management using the Swan-Ganz catheter against bedside clinical acumen alone undermines Ludwig’s maxim and is, in my opinion, badly misdirected. A number of studies in both intensive care7 13 14 and coronary care unit15 16 populations has shown the unreliability of clinical and radiological evaluation of haemodynamic status with experienced specialists faring no better than junior staff and medical students.13 A first priority in evaluating the Swan-Ganz catheter should be to ensure adequate training. A recent European questionnaire17 examining proper use and interpretation of data mirrored an earlier US study18 showing that competency was highly variable, with only 70% of questions being answered correctly. Second, how the catheter is used must be addressed. At the time of Connors et al’s study,8 many ICUs had adopted the supranormalised oxygen delivery philosophy advocated by Shoemaker whereby the circulation was driven with fluid loading followed by increasing doses of dobutamine to obtain predetermined raised values in cardiac index, oxygen delivery, and consumption.3 This approach proved highly successful in high risk non-cardiac surgical patients3 4 and was subsequently adopted by many ICUs. However, it took several years before—at best—no benefit19 and—at worst—harm20 was demonstrated in such critically ill patients in prospective, randomised, controlled studies.
Several non-invasive techniques exist for flow monitoring but none has yet penetrated clinical practice to any great extent. The major issues are reliability and familiarity. Unfortunately, many commercial devices have been launched without proper validation and often with obvious design flaws. Furthermore, thermodilution, an imperfect gold standard,21 22 which is often used imperfectly,17 18 is often used as the comparator technique. These problems have served to undermine confidence in the various technologies, which upgraded equipment and newer models have yet to fully overcome.
The two best described techniques are Doppler ultrasound and thoracic bioimpedance. Doppler technology has enjoyed a renaissance since the advent of combined Doppler echocardiography, in particular, from the transoesophageal approach, which has provided new possibilities for both diagnosis and monitoring.23-25 Some authorities advocate its superiority over invasive techniques26; however, expense, complexity, and the necessary expertise have restricted its widespread application in the UK. A simpler technique uses standalone Doppler ultrasound that interrogates the blood vessel—usually ascending or thoracic descending aorta—from suprasternal27 28 or oesophageal29 30approaches, respectively. This enables monitoring of blood flow velocity from which a reasonable estimate of cardiac output can be made. The shape of the flow velocity waveform also readily permits assessment of left ventricular contractility and filling, and systemic vascular resistance.30 31 A Starling-type curve can be constructed from the stroke volume response to 200 ml fluid challenges. To date, oesophageal Doppler ultrasound is the only non-invasive technology used successfully to direct intraoperative fluid loading, resulting in outcome benefit and significant reductions in hospital stay in prospective, randomised, controlled trials of cardiac5 and orthopaedic6 surgery. A relevant finding from these studies was the greater sensitivity of stroke volume over cardiac output, as compensatory tachycardia tended to maintain cardiac output in the face of mild to moderate hypovolaemia. Thus, overall, the control groups showed no change in output but had significant falls in stroke volume.
An editorial32 accompanying the hip fracture repair study6 spelled out the potential benefits of this non-invasive flow monitoring technique in high risk patients undergoing surgery, but emphasised the need for multicentre confirmation before its widespread adoption. It is worth reiterating that outcome was improved by acting on the information provided to achieve the set goal of optimal filling. Which haemodynamic goals are applicable to different patient populations (high risk surgery, severe heart failure, sepsis), and whether non-invasive techniques are equal, or even superior, to invasive monitoring techniques requires extensive study, but encouragement can be drawn from these initial investigations.
Thoracic bioimpedance has had a more chequered course. In 1966 Kubicek described the thorax as a cylinder evenly perfused with blood of specific resistivity. In 1983 Sramek demonstrated that the thorax behaves electrically more like a truncated cone. Pulsatile thoracic aortic blood flow causes negative impedance changes with a maximum rate of change during systole between pairs of electrodes placed on the neck and upper abdomen. Various new correction factors and equations have been advanced (table 1) and incorporated into commercial systems using pairs of electrodes placed around neck, thorax, and upper abdomen. Unfortunately, validation studies have produced highly inconsistent results. Generally, the technique appears to be reasonably reliable in healthy volunteers but performs unpredictably in the critically ill.33-37 Thus, the paper by Spiering et alin this issue38 is rather surprising in view of the major discrepancy between bioimpedance and dye dilution in normal subjects. Despite the authors’ assurances, I wonder if an unrecognised methodological problem affected the study outcome. If bioimpedance is to develop as a clinical tool it must address these reliability issues, alert the operator to erroneous readings, and regain user confidence.
Other techniques are also available for either intermittent or continuous monitoring of flow. Echocardiographic measurement of ventricular volumes at end systole and end diastole enables stroke volume to be measured with reasonable accuracy23-25; however, at least six different methods have been described and user expertise is a significant factor. Recently developed technology allows three dimensional reconstruction of images of the left ventricle, although ventricular volumes are still generally calculated from cross sectional and M mode images, accepting assumptions about left ventricular shape.
Analysis of the contour of the aortic pulse wave allows monitoring of cardiac output, as the rate blood flows from arteries to veins is proportional to the rate of fall of arterial pressure. This can be performed non-invasively from a finger probe39 or from an indwelling pressure transduced arterial cannula.40-42 There is not an extensive literature assessing its reliability in different conditions; furthermore, it cannot be used for measurement of cardiac output unless first calibrated against another technique.
Thermal dilution techniques can be performed without the need for pulmonary artery catheterisation,43 although validation data are scanty. A commercially available device using this technology (COLD; Pulsion, Munich, Germany also claims to measure global end diastolic volume (the sum of the end diastolic volumes of both left and right atria and ventricles), intrathoracic blood volume (a possible indicator of cardiac preload), and extravascular lung water.43 44
The Fick principle states that oxygen consumption (Vo 2) equals cardiac output multiplied by the arteriovenous oxygen concentration difference. For cardiac output measurement the arterial oxygen concentration is measured from a peripheral arterial blood sample, the venous oxygen concentration from pulmonary arterial blood, and Vo 2 is derived from minute ventilation and inspired and expired gas analysis.45-48 In a variation of this method,co 2 production can be used instead of oxygen consumption although large discrepancies have been found in comparative studies in the critically ill.48 In theory, the Fick technique is the gold standard for cardiac output measurement but it is invasive and methodological error is not uncommon—for example, when high inspired oxygen concentrations are being administered. Using the indirect Fick approach, the carbon dioxide rebreathing technique is a popular technique among sports physiologists as it is totally non-invasive and more reliable at exercise than at rest.47 49 50 The rebreathing manoeuvre is used to estimate mixed venous Pco 2(PVco 2), which, combined with end tidal Pco 2 and co 2production, gives a non-invasive Fick estimate of the cardiac output. Reproducibility is not as high as other techniques48 49and it has yet to be applied at the bedside of the sick patient.
Finally, the dye dilution technique may enjoy renewed interest as a bedside cardiac output monitoring technique through online means of measuring plasma concentrations of indocyanine green, the standard indicator in use today.44 45 51 Alternatively, a new approach using lithium chloride as the indicator52 53with plasma concentrations being measured by a lithium selective electrode in a flow through cell connected by a three way tap to a standard arterial cannula, may also merit consideration.
There is sufficient evidence to show that flow monitoring leading to directed haemodynamic management is beneficial, at least in certain patient subsets. Advantage is generally gained by those at risk of, rather than having, established organ failure, such as the high risk surgical patient. Few studies have demonstrated outcome benefit in critically ill patients in whom metabolic, inflammatory, and cellular processes are often too far advanced for benefit to be gained from haemodynamic manipulations that prevent or reverse tissue hypoxia, a potent stimulus of various inflammatory pathways. Current use of flow monitoring generally precludes pre-emptive circulatory optimisation to prevent, or at least, minimise tissue hypoxia. This is equally applicable to the trauma patient or the patient in acute heart failure, where hypovolaemia often passes unrecognised and untreated until hypotension and organ failure have supervened. Invasive measurement of cardiac output is coming under increasing scrutiny and the place of non-invasive technology merits re-appraisal in the light of newer and more reliable equipment and techniques. This should provide impetus for further investigation, particularly in patients in the early stages of illness.