EUROPEAN SOCIETY OF INTEN SIVE CARE MEDICINE

IntensiveCare Medicine (1991) 17:I-VIH

Expert Paneh The use of the pulmonary artery catheter Participants: D. B e n n e t t ( U K ) , J. B o l d t ( F R G ) , L. B r o c h a r d ( F r a n c e ) , P. C o r i a t ( F r a n c e ) , J.-F. D h a i n a u t ( F r a n c e ) , D. E d w a r d s ( U K ) , E F e i h l ( S w i t z e r l a n d ) , J. G r o e n e v e l d ( T h e N e t h e r l a n d s ) , M . L a m y ( B e l g i u m ) , D. L u n d b e r g ( S w e d e n ) , E L e m a i r e ( F r a n c e ) , D. P a y e n ( F r a n c e ) , C. P e r r e t ( S w i t z e r l a n d ) , S. Reiz ( S w e d e n ) , J.-J. R o u b y ( F r a n c e ) , D. S c h e i d e g g e r ( S w i t z e r l a n d ) , M . S i n g e r (UK), P. S u t e r ( S w i t z e r l a n d ) , L. T h i j s ( T h e N e t h e r l a n d s ) a n d J.-L. V i n c e n t ( B e l g i u m ) Pulmonary artery (PA) catheterisation has become a well-established haemodynamic monitoring technique since its introduction some 20 years ago [11. It~allows monitoring of PA pressures, measurements of PA balloon-occluded (or capillary wedge) pressure (PAOP) and estimation of cardiac output by the thermodilution technique. Access to measurements of oxygen content of the mixed venous blood allows calculations of oxygen transport and oxygen consumption. PA catheterisation provides the necessary information for the derivation of other parameters such as systemic and pulmonary vascular resistance, and left and right ventricular stroke work index. Recent advances include: 1) continuous measurement of mixed venous oxygen saturation (8902) via a fibre-optic channel; 2) calculation of the right ventricular ejection fraction (RVEF) and RV volumes using a fast response the,rmistor; 3) temporary pacing (ventricular and/or atrial) using either electrodes on the outside of the catheter or lumens through which a pacing wire is inserted. Current developments include continuous measurements of cardiac output, oxygen delivery (DO2) and oxygen uptake (902). The emphasis in usage has shifted from its original predominance in the coronary care patient to the ICU and operating theatre environments. This has been accelerated, due in no small part to the catheter itself, both by a greater appreciation of haemodynamics and by the clinical utilization of oxygen txansport parameters. Cynics have also been heard to mutter darkly about the pecuniary benefits of routine insertion. Latterly, its role in the prevention of problems has been extended to haemodynamic "supra-optimization" in an attempt to reduce tissue oxygen debt in high risk patients [2].

Measurements of PA pressures Conventional measurements of PA pressures using the PA catheter are based on

pressure transmission by a fluid column to an external transducer. Like arterial pressure, PA pressure has two components: PA systolic (PAPs) and diastolic (PAPd), from which the monitor can calculate the mean PA pressure (PAPm). PAPs is influenced by the right ventricular (RV) stroke volume ejected, the compliance of the large pulmonary vessels and the PAPd. In turn, PAPd is determined by the downstream (small pulmonary vessels) resistances, the diastolic time, the pulmonary blood volume and the back-pressure against pulmonary blood drainage, i.e. the left ventricular (LV) filling pressure. Because of the oscillating nature of the circulation, the same mean value can correspond to different amplitudes of the oscillations, so that evaluation of PAPs and PAPd are also important. Among the determinants of the PAPs, compliance of the large pulmonary vessels seems the most critical. A change in pressure results in a change in the diameter of the elastic blood vessels which, in turn, determines the pressure distribution. Even if the relationship between the pulmonary circulation and RV function is not yet fully understood, for any given situation PAPs is a good index of RV systolic function. For any heart rate, an increase in stroke volume will result in an increase in PAPs, which can be buffered by an increased compliance of the pulmonary vessels. PAPd should be near-identical to the left atrial pressure and the LV enddiastolic pressure, except when there is an obstacle to blood flow within the pulmonary vasculature. This obstruction to flow can be macro- or microembolism, active vasoconstriction or structural thickening of the vessel walls. Right atrial pressure (RAP) was used extensively in critically ill and surgical patients because it was thought that changes in pressure were indicative of changes in the overall filling of both the right and left heart. Management decisions based on changes in RAP required the assumptions that RAP is a

reasonable measure of RV filling and that right and left functions are in parallel. Neither assumption is true, in particular during critical illness [4]. The PAOP measurement is associated with a transient obstruction to blood flow through a small PA. In these circumstances, PAOP, left atrial pressure and LVEDP are supposedly equal. There is, however, a variety of conditions that may invalidate this pressure relationship (Table 1).

Effect of intrathoracic pressure on measurement of PAOP West et al. have shown that there is a marked variation of blood flow within the lung as a result of inter-relationships between alveolar and vascular pressure (Fig. 1). Schematically, two factors may influence the pressure values and PAOP in particular as an index of left sided filling (during spontaneous or mechanical ventilation): the positioning of the PA catheter and the changes in pressures during the respiratory cycle. Fortunately, the main part of the lungs in a supine patient falls in zone III, and flow-directed PA catheters usually enter zone III because most of the blood is entering this area. There are two methods to confirm a zone-IIt catheter position: (1) A sudden increase or decrease of PEEP level should not influence the PAOP [5]; (2) A lateral chest roentgenogram taken with the catheter wedged can ascertain that the catheter tip is correctly positioned at or below left atrium level [61. One should note that zone III can also be converted in zone I or II by loss of intravascular volume (hypov01emia, balloon inflation), airway obstruction, or application of PEEP. An adequate wedge position of the PA catheter can be confirmed by three criteria: 1. the phasic contour of the PA waveform should change during PA occlusion; 2. PAPm should fall during occlusion;

Expert Panek The use of the pulmonary artery catheter (ESICM)

@

''

9

[

....

..~I

3. PO 2 and SO 2 should be higher in blood withdrawn from the wedged catheter than in arterial blood [7]. Changes in alveolar pressure during the respiratory cycle can markedly influence PAOP. During spontaneous breathing, alveolar pressure shifts from negative (inspiration) to positive (expiration). During mechanical ventilation, alveolar pressure increases during lung expansion. In either case, intrathoracic (and alveolar) pressure is closest to atmospheric pressure at the end of exhalation. Therefore, intravascular pressures should always be measured at this point to minimize the influence of intrathoracic pressure. Finally the practice of altering respiratory conditions to measure vascular pressures should be condemned because of the potential dangers of hypoxaemia, sudden increases in venous return and misinterpretation of the clinical situation.

Table 1. Conditions resulting in discrepancy between PAOP and LVEDP Mean PAOP> LVEDP

Positive pressure ventilation (with or without PEEP) Increased intrathoracic pressure Non-Zone III catheter placement Chronic obstructive lung disease Tachycardia Increased pulmonary resistance Mitral valve obstruction Pulmonary venous compression (tumour, fibrosis) Mitral regurgitation Left-to-right intracardiac shunt Mean PAOP < LVEDP

Noncompliant left ventricle Reduced pulmonary arterial tree (pneumonectomy, pulmonary embolism) Aortic regurgitation

eral-fold: it facilitates appropriate positioning of the PA catheter (during insertion of repositioning); it improves the validation of a PAOP tracing and limits the problems of "partial" PAOP and \N\ \\ "overinflation" or "underinflation" of \ \ \ \ the balloon; it will also facilitate the interpretation of the successive waves on the PAOP tracing and measurement of the PAPd-PAOP gradient. When the catheter is not wedged, it readily detects a spontaI ! neous migration of the PA catheter into a I I / I wedge position; furthermore, in this posi/ I Fig. 1. Schematic rep/ / tion a divergence in the two pressure trac/ / resentation of the es readily indicates a problem of zeroing i / / / West's zones of the I f or calibration of one pressure trace. i lung. PA, alveolar pressure; Pap, pulmonary artery pressure; Measurements o f cardiac output ev, venous pressure A major advantage of the PA catheter is the availability of cardiac output determiNumerous reports have documented nation by the thermodilution technique. the effects of pleural pressure on the rela- Hence cardiac output should be measured tionship between left atrial pressure and routinely in all monitored patients. ExPAOE A number of factors may alter this perimental and clinical studies have relationship, including a non-zone III po- shown good agreement between cardiac sition, some transducer-related artefacts, output values obtained by thermodilution eccentric balloon-occlusion, the existence and other techniques [9, 10]. The thermodilution technique is inaccurate in the of a pulmonary venous occlusive disease, an altered ventricular compliance and mi- ]~resence of intracardiac shunts or severe tral regurgitation. Moreover, even if tricuspid regurgitation [11]. However, PAOP correlates well with LVEDP, these moderate tricuspid regurgitation does not pressures might correlate poorly with invalidate the measurement. Controversy persists about the temLVEDV in the presence of an altered venperature of the injectate. Some authors tricular compliance [8]. the sigCardiac filling pressure is influenced have demonstrated that by several factors. Firstly, it is supposed to nal-to-noise rati~o was still satisfactory reflect the end-diastolic volume of the when the injectate was at room temperaventricle, but the pressure/volume rela- ture, and this represents a more practical tionship is influenced by the compliance procedure [12]. In any case, the injectate characteristics of the heart. The end-dia- temperature should be monitored at the stolic volume is influenced by the in- point of entry into the catheter. A large travascular volume, the pressure/volume volume of fluid (10 ml of D5W in adults) characteristics of the venous system, and is recommended to increase the signalalso the end-systolic volume which, in to-noise ratio. Timing of the injection in relation to turn, is influenced primarily by the conthe respiratory cycle also represents a contractile state and the end-systolic pressure. In the presence of hypovolaemia, troversial issue. Cardiac output is known PAOP can overestimate LVEDP by 4 to fluctuate during the respiratory cycle so mechanisms. Firstly, the PA catheter can measurements at various phases of the remove from a zone III to a non-zone III spiratory cycle have been suggested [13]. position so that the pressure measured re- For clinical purposes, however, timing of flects alveolar pressure rather than left the injection to the respiratory cycle proatrial pressure. Secondly, the atrial con- vides a cardiac output value which is not traction can contribute to a larger extent significantly different but reduces the to the filling. Thirdly, premature closing variability [14, 15]. Three to 5 measurements should be of the mitral valve during tachycardia will increase the left atrio-ventricular pressure averaged to obtain a cardiac output value. gradient. Fourthly, tachycardia limits the A change of at least 10% must be obtime for equilibration between PAOP and served to indicate a true alteration in cardiac output. left atrial pressure. In the unstable patient at least, a carSome types of catheters have been modified to include an additional lumen diac output measurement should always allowing simultaneous measurements of be complemented by simultaneous deterPA pressures and PAOE Although the mination of arterial and mixed venous oxclinical interest remains to be demonstrat- ygen contents (S~'O2 could suffice), aled, the advantages of this system are sev- lowing assessment of the relationship be-

Expert Panek The use of the pulmonary artery catheter (ESICM) tween DO2 and VO2. This represents a serious limitation to the non-invasive measurement of cardiac output as interpretation of a given cardiac output value may be incomplete. If there are doubts about the oxygen extraction capabilities of the patient, determination of a blood lactate level can help to recognise the presence of tissue hypoxia.

RVEF measurements Provision of a fast response thermistor and the development of computer algorithms have allowed measurement of both RV ejection fraction (RVEF) and cardiac output by the thermodilution technique. The technique has been validated appropriately [16-18], permitting its correct use in the evolution of myocardial function in various disease states. Modern computerized algorithms do not recognize plateaus but are instead based on a model using a pulsatile chamber's response to a pulsed input bolus. This mathematical system uses a single exponential washout curve. With this system the variability is reduced to 7 - 8 % [17, 18]. The limitations of the technique are the presence of arrhythmias or severe tachycardia (the technique may be less accurate when heart rate is above 120 bpm), tricuspid regurgitation or intracardiac shunts. As RVEF changes throughout the respiratory cycle, randomized RVEF measurements can show a large variability so that it is preferable to time RVEF measurements with the respiratory cycle in mechanically ventilated patients. Derived calculations of RV end-diastolic volumes could be useful guides to fluid therapy. Measurements of RVEF can be of particular interest in patients with acute pulmonary hypertension. Septic patients represent a future group of interest as a simultaneous decrease in arterial pressure can threaten RV coronary perfusion. RVEF also represents a useful prognostic parameter in patients with septic shock [19]. RVEF measurements have also been found useful in patients aRer trauma [20], myocardial infarction [211 or CABG [221. Furthermore, it can be used to assess the effects on the RV of vasoactive agents [23] or mechanical ventilation with PEEP

[24]. Continuous SVO2 monitoring The use of optic fibres allows continuous measurement of $90 2 in vivo. The measurement is based on reflection spectrophotometry: the emitted light of 2 or 3 selected wavelengths is reflected by the circulating red blood cells to a photodetector. This returned light transmission depends on the relative concentrations of

oxyhaemoglobin and haemoglobin in the circulating blood. The measurement can be influenced by blood pH, blood flow velocity and erythrocyte shape. Vessel wall artefacts can also represent a problem. Infusion of lipid emulsions [251 or methylene blue [26] or the presence of methaemt~globinaemia can also interfere with the measurement of $90 2. S'702 monitoring can be used for continuous assessment of the relationship between oxygen uptake and supply. It might be useful as an effective alarm system at the bedside, however the clinical indications are not clearly defined. In particular, its effectiveness must be better delineated.

Recommendations for insertion The PA catheter should be inserted percutaneously. Prior to insertion the catheter should be checked for balloon integrity and thermistor function, and the lumens should be flushed. The entire procedure should be sterile. After careful skin desinfection, an introducer sheath should be placed. The PA catheter should be protected by a sleeve limiting the risk of infection during manipulations. Once the catheter tip has been advanced by about 20 cm from the internal jugular or subclavian puncture site, the balloon should be completely inflated (1.5 cc). Not more than 15 cm should be introduced into the RV before a PA trace is obtained. When the PAOP trace is obtained after balloon inflation with 1.5 cc (and not less), the catheter is in position. A chest roentgenogram should be obtained routinely after catheter placement to document the appropriate position. The tip of the PA catheter should remain within 2 cm from the cardiac silhouette.

Complications While regarded as safe, PA catheterisation is an invasive procedure and a number of related complications have been recognised [27-31]. These complications are attributed to central venous cannulation (insertion), catheter passage and catheter presence. In a study reviewing 6245 PA catheter insertions, an incidence of morbidity as low as 0.4% was reported [27]. Complications of venous cannulation are identical for central venous access and PA catheter monitoring. ArteriaI puncture is a common complication of venous cannulation, especially via the internal jugular vein (carotid artery). Provided small bore catheters are employed, arterial puncture is rarely a serious complication. Haemorrhage, haematoma formation with upper airway obstruction and embolisation of arteriosclerotic plaques are rare but occasionally reported compli-

III cations of arterial puncture. Pneumothorax is also a rare but well recognised complication of either subclavian or (less commonly) internal jugular vein cannulation. In a patient treated with positive pressure ventilation the pneumothorax may become a life-threatening complication. During passage of the catheter, arrhythmias are common. The reported incidences vary significantly according to the type of underlying disease and the method of arrhythmia detection. Arrhythmias are reduced by passing the catheter with the balloon inflated. When arrhythmias do occur the catheter should be rapidly withdrawn. The routine use of prophylactic lidocaine prior to catheter insertion can reduce the incidence of ventricular arrhythmias [29] but is seldom required. Knotting of the PA catheter can also occur, especially in patients with low flow states and with large cardiac cavities. It can be avoided by using no more than 15 cm of the catheter length between the appearance of the RV trace and the appearance of the PA trace. If the PA cannot be reached within this distance, the PA catheter should be withdrawn into the right atrium and the procedure repeated. Complications can also occur when the PA catheter is in place. As with central venous catheterisation, a common problem is venous thrombosis at the site of insertion. This complication is usually asymptomatic. Its incidence has also decreased with the use of heparin-coated catheters. The distal migration of the PA catheter tip or a prolonged balloon inflation can occlude distal blood flow into the PA. Pulmonary infarction can represent a serious complication. Catheter-related infection and right heart endocarditis are recognised causes of morbidity and even mortality [28, 30]. The incidence of catheter-related infection is influenced by several factors including the sterility of the skin puncture and the insertion, the number of catheter manipulations, the care taken in the manipulation of the system and the duration of catheterisation. In a postmortem study [30], 7% of patients with a PA catheter had signs of infective endocarditis. This problem is, however, seldom recognised clinically. The most traumatic and acutely catastrophic complication is PA rupture and haemorrhage. The factors increasing the risk of PA perforation include advanced age, pulmonary hypertension, hypothermia and deviations from standard insertion techniques [31]. This complication has a reported mortality of about 50%. To avoid this complication one should minimize the frequency of balloon inflation and ensure the catheter tip location is in the proximal PA on initial placement. It is also impor-

Expert Panel: The use of the pulmonary artery catheter (ESICM)

IV tant to inflate the balloon progressively with the pressure trace under continuous scrutiny and to stop further inflation as soon as a balloon-occluded trace is obtained. The additional pressure port described above might be useful in this setting. Thus, invasive monitoring has its risks which have been well defined. These risks have appeared to decrease over the years. In each ICU the risk can decrease with the experience of the ICU staff and supervision of trainees and, in general terms, the frequency of use of the catheter. Nevertheless, benefits from PA monitoring should always be weighed against the risks related to the procedure. Invasive versus non-invasive techniques Recent developments in non-invasive techniques must be taken into account to evaluate the indications and usefulness of invasive monitoring. Echocardiography and Doppler echocardiography have become widely used in the ICU. They allow the non-invasive estimation of cardiac output [32], PA pressure [33], LV ejection fraction and volumes. Echocardiographic measurement of LV end-diastolic volume may be a better indicator of preload than PAOE The value of the examination is largely dependent on the experience and expertise of the echocardiographer. Transthoracic echocardiography cannot be repeated as often as invasive measurements and good images are frequently difficult to obtain in ICU patients; both limitations are less with transoesophageal echocardiography. Transthoracic bioimpedance cannot at present be recommended for the non-invasive monitoring of cardiac output [34, 35] or LV ejection fraction [36]. There is no non-invasive substitute for the monitoring of mixed venous oxygen saturation which allows valuable insights into the peripheral oxygenation.

Clinical applications Heart failure As mentioned earlier, the availability of bedside non-invasive techniques, such as echo-Doppler, has decreased the need for invasive haemodynamic monitoring in the critically ill patient. However, PA catheterisation can still be required for some diagnostic and therapeutic considerations. In the presence of acute myocardial infarction there are 4 important indications for PA catheterisation. Firstly, the development of eardiogenie shock requires the combined measurements of cardiac output and PAOP for rational use of volume expansion, inotropic drugs and

intra-aortic balloon counterpulsation [371. Secondly, in severe, acute LV failure (Killip class III) PA catheterisation can contribute to the diagnosis of acute mitral valve regurgitation or septal perforation. It is essential to select the correct form of therapy with vasoactive agents and fluids. hnportantly, careful volume expansion might be indicated in patients with hypovolaemia related to prior diuretic administration [38]. Thirdly, signs of predominant RV failure associated with arterial hypotension also represent an indication for PA catheterisation to assess biventricular function and to guide fluid therapy in the presence of a potential alteration of LV function [39]. However, the risk of catheter-induced ventricular arrhythmias should be considered in the presence of RV myocardial infarction. Fourthly, a persistent, unexplained tachycardia may be due to hypovolaemia, ventricular dysfunction or a stress-induced hyperdynamic response. PA catheterisation has a decisive role in the differential diagnosis of these conditions with obvious therapeutic implications. In situations where acute mitral valve regurgitation or septal perforation is suspected in the absence of severe acute LV failure, PA catheterisation is indicated only when echo-Doppler techniques are not available or non-contributory. In patients with pulmonary oedema and associated cardiac disease, PA catheterisation is indicated when there is doubt concerning the nature (cardiogenic or not) of the pulmonary oedema [401. It becomes mandatory in patients with cardiogenic pulmonary oedema who do not respond to initial therapy in order to reassess and guide therapy. In patients with severe oliguric renal failure and underlying acute heart failure abrupt changes in volaemia may occur, especially when dialysis is performed. The resulting changes in preload can be appreciated and treated using a PA catheter.

haemodynamic support. However, PA catheterisation should not delay pulmonary angiography. When thrombolysis is performed, peripheral insertion of the PA catheter may be preferable. In the absence of shock, echocardiographic techniques are usually sufficient to monitor the effects of vasoactive as well as thrombolytic therapy on RV function. Cardiac surgery Measurements of PA pressures and cardiac output can be useful both during and after cardiac surgery. Invasive monitoring can help in the early diagnosis of myocardial ischaemia during and after coronary artery bypass grafting (CABG) [41]. Even experienced anaesthesiologists sometimes fail to diagnose significant haemodynamic alterations in the absence of invasive haemodynamic monitoring [42]. In CABG, PA catheterisation has been recommended in patients with altered LV function (LVEF < 40%, LVEDP >20 mmHg), severe stenosis ( > 90%) of the left main coronary artery, history of multiple infarctions, redo surgery or advanced age ( > 7 0 years) [43]. PA catheterisation is also recommended in patients with recent myocardial infarction, in most cases of aortic and/or mitral replacement, especially when combined with CABG, and in the presence of pulmonary hypertension or cardiomyopathy. Modified PA catheters can be also useful to monitor SvO2 or RVEF or to provide immediate access for pacing. The Paceport (multi-purpose) catheter can be used for atrial or ventricular pacing, A-V sequential pacing, overdrive suppression of atrial or ventricular arrhythmias and diagnosis of complex arrhythmias. The importance of selective RV function after cardiac surgery has been stressed [44]. Direct measurements of RV volumes by the thermodilution technique can represent a useful adjunct to invasive monitoring [22].

Cardiac tamponade When suspected clinically, cardiac tamponade should be diagnosed by echocardiography so that PA catheterisation is only useful when this technique is not available or to document the efficacy of the pericardial drainage. In catheterised patients, invasive monitoring may also help to detect early recurrence of the tamponade. Pulmonary embolism In pulmonary embolism associated with obstructive shock, PA catheterisation is useful in making the diagnosis, assessing the severity of the disease and guiding the

Non-cardiac surgery Controlled data are lacking on the influence of the PA catheter on morbidity and mortality in patients undergoing non-cardiac surgery. Limited studies on patients undergoing abdominal aortic surgery and other major abdominal surgery have suggested that the values obtained from a PA catheter during and after surgery would be useful to detect myocardial ischaemia and to guide fluid replacement [45, 46]. A retrospective study of patients with previous myocardial infarction has suggested that invasive monitoring and aggressive treatment of circulatory aberrations might almost entirely eliminate the risk of

Expert Pane# The use of the pulmonary artery catheter (ESICM) reinfarction and death [471. Another uncontrolled study in patients subjected to abdominal aortic reconstructive surgery has suggested that the use of the PA catheter could significantly reduce cardiac morbidity [48]. PA catheterisation is useful in patients with LV dysfunction or at risk of developing pulmonary oedema but its value is more limited in detecting hypovolaemia or myocardial ischaemia. Unfortunately, neither simple nor sophisticated tests used preoperatively to assist the anaesthesiologist in his pre-operative risk assessment are very helpful in predicting the risk of developing a low stroke volume or an elevated PAOP in the perioperative period. Prospective studies should be performed to define which group of patients with cardiac disease might benefit from PA catheterisation during and especially after non-cardiac surgery.

Particular surgical indications Patients undergoing thoracic surgery may experience various hazardous intraoperative conditions, such as the lateral decubitus position with either one- or two-lung ventilation, major lung resection, aortic cross-clamping, large volume shifts, obstruction of a vena cava, presence of sepsis... PA catheterisation is recognised as potentially beneficial in high risk operations, and especially in high risk patients (acute or chronic respiratory failure, pulmonary hypertens i o n . . . ) . Similarly, in critically ill patients for whom anaesthesia is kept to a minimum, invasive monitoring can help to maintain physiological variables as close to normal as possible. Serious questions can be raised, however, regarding the accuracy of the data obtained during some types of thoracic surgery, namely during those performed in the lateral decubitus position or wkh one-lung anaesthesia. It is also important to avoid positioning the PA catheter in a pulmonary area to be resected. Continuous SgO 2 monitoring during anaesthesia may also indicate an increased oxygen consumption (light anaesthesia or muscular relaxation), decreased DO 2 or altered hypoxic pulmonary vasoconstriction. Thermodilution RVEF monitoring can also be useful in the diagnosis and management of RV failure following non-cardiac surgery in high risk patients [22]. PA catheterisation can be helpful during surgical procedures with major fluid shifts or metabolic changes such as extensive vascular procedures or liver transplantation. Reduced operative and reduced postoperative renal failure have been demonstrated in abdominal aortic aneurysm resection [49, 50]. Maximal hydration during surgery may improve the

early function of human renal transplants. As liver transplantation can be associated with cardiovascular instability, arrhythmias, massive fluid shifts, electrolyte imbalance, coagulation abnormalities and embolism, PA catheterisation should be used routinely for this operation. PA catheterisation may also be recommended for detection and treatment of venous air embolism during seated neurosurgical procedures [51]. A special PA introducer sheath with multiple orifices positioned by intravascular electrocardiography can provide an effective means of simultaneously monitoring PA pressures and aspirating air, thus affording diagnosis of significant pulmonary embolism, estimation of its severity and documentation of its successful treatment [52]. In pregnant women the PA catheter can be useful in the management of preeclampsia with hypotension, pulmonary oedema or oliguria [53]. Withdrawal of mixed venous blood can also document the presence of amniotic fluid emboli [54].

Respiratory failure Indications for and benefit from PA catheterisation in respiratory failure depend on several factors including clinical course, prior assessment of the cause and severity of the cardiorespiratory impairment, and the experience of the medical team in evaluation and management of such patients.

Acute respiratory failure It is clear that not all patients with acute respiratory failure, even with adult respiratory distress syndrome (ARDS), require PA catheterisation, However, a number of indications are widely recognized. First indications are in those with associated left heart failure, as acute or chronic LV impairment can influence the course of the respiratory failure, the type of respiratory support, the fluids and vasoactive therapy used as well as the type of weaning process [55]. For this indication, the main interest focuses on PAOP which should be kept as low as possible to limit microvascular leakage of fluids but high enough to maintain a sufficient cardiac output. In thoracic trauma with pulmonary contusion, an associated cardiac contusion is more common than clinically suspected. As this diagnosis may not bring any therapeutic implication, the risks of PA catheterisation must be weighed against the potential benefits. Secondly, positive pressure ventilation may induce cardiovascular alterations and organ dysfunction. In this situation, measurements of cardiac filling pressures, car-

diac output and systemic oxygen transport can be helpful. Although not fully established, the maintenance of a hyperdynamic state could improve the prognosis of these patients [2]. Thirdly, large fluid shifts can occur during sepsis or multiple organ failure, for instance in the presence of an altered capillary permeability or during continuous haemofiltration. These patients can develop haemodynamic instability requiring fluids and vasoactive agents, and this therapy can be guided by invasive haemodynamic monitoring. Pulmonary hypertension is an ominous sign in ARDS and may lead to acute RV failure. A number of vasodilating agents can be considered in this condition but should be administered under close respiratory and haemodynamic monitoring (including PA pressure, PAOP and cardiac output) [56, 571. Fourthly, PA catheterisation can be recommended in ARDS when pulmonary function does not improve over 24 to 48 h despite the appropriate therapy of the underlying disease, and adequate cardiocirculatory and ventilatory support. It is indeed quite common to then discover an unsuspected cardiovascular abnormality requiring a different therapeutic management.

Exacerbated chronic pulmonary disease In COPD patients with associated (recognised or suspected) LV dysfunction, PA catheterisation can provide a better diagnostic and therapeutic approach. In problems of weaning failure, PA catheterisation is also helpful to recognise (or exclude) LV impairment [55] and to consider the appropriate therapeutic measures.

Trauma and sepsis The role of invasive haemodynamic monitoring with a PA catheter in the traumatized and septic patient with shock is not clearly defined. Haemodynamic monitoring with the PA catheter can help to identify and treat those haemodynamic defects that correlate with a dismal outcome. However, the value of haemodynamic and metabolic variables in reflecting the severity and prognosis of shock associated with trauma or sepsis is not beyond debate. Hence, the indications for and implications of PA catheterisation in these conditions are hard to delineate. Monitoring of the PAOP, a major determinant of microvascular filtration in the lung, is nevertheless helpful in preventing (aggravation of) pulmonary oedema during fluid treatment of traumatic and septic shock [40].

VI Trauma. Although PA catheterisation is usually not indicated during early resuscitation of trauma victims, except perhaps in patients with preexisting heart or lung disease, the variables obtained with the catheter may be helpful to guide resuscitation in complicated cases, i.e. when trauma is severe, extensive or both, when trauma involves heart and lungs and when artificial ventilation has been instituted. Unsufficient response to initial resuscitation may be another indication to insert a PA catheter. Indeed, the "classical" resuscitation of a trauma victim based on the assessment of gross parameters such as a heart rate, blood pressure, urine output, and sometimes CVP, may not be associated with an increase in oxygen transport which may be necessary to prevent tissue hypoxia and to improve outcome [58]. The inadequacy of the "classical" resuscitation might be revealed only later when multiple organ failure develops. There are also data that a rise in D O 2 transport can reduce morbidity and mortality in surgical patients [2]. Septic shock. PA catheterisation may be performed when clinical judgement fails to diagnose the syndrome or to institute successful treatment. Management of septic shock in a patient with myocardial dysfunction may require the early insertion of a PA catheter and monitoring of PAOP and cardiac output. As in shock associated with trauma, it is likely that a rise in D O 2 and VO2 increases the chances for survival [59]. During septic shock, O 2 uptake seems dependent on 02 transport over a much wider range than normal, especially when the blood lactate level, an indicator of 02 deficit, is increased [60, 61]. Hence, therapy aiming at reversal of tissue hypoxia and a fall in lactate, accomplished by a high DO2, might improve outcome. Hence the PA catheter can play a central role in judging the course of septic shock and in guiding therapy by allowing measurements of O 2 transport variables [62, 63]. By providing data on filling and function of the heart, it allows to clearly delineate the choice between different forms of therapy (fluids and vasoactive agents), when clinical signs fail to do so [64].

Are prospective studies needed? Considerable controversy reigns over the value of PA catheterisation. The role of the PA catheter is to give accurate haemodynamic information, to monitor the haemodynamic status of patients at risk of developing haemodynamic alterations, and, if they develop, to select and monitor the therapeutic interventions. It

Expert Panel." The use of the pulmonary artery catheter (ESICM) is thus a diagnostic and monitoring tool, expected to provide a guide for therapy in the acutely ill. The introduction of the technique in anaesthesia and intensive care has undoubtedly provided us with a large amount of new knowledge. However, the question of its benefit for patients in terms of outcome has now been raised. Several conditions have to be met. Firstly, the risk associated with the invasive procedure must be outweighed by the expected benefit. Secondly, data must be collected accurately and interpreted. Thirdly, the haemodynamic parameters obtained should influence therapy. Fourthly, this change in therapy should improve outcome. Recent reports have suggested that PA catheterisation in selected groups of patients did not improve outcome [43, 65], whereas other reports suggested a beneficial effect [2, 47, 63]. Other recent work suggests that the manipulation of the parameters obtained with the PA catheter, in particular D O 2 and 902, will improve the prognosis in critically ill surgical patients [2]. Several prospective studies have shown that the assessment of haemodynamic status by doctors is often inaccurate and that the insertion of a PA catheter may alter therapy in a substantial number of cases [46, 66-69]. However, it has been suggested that doctors did not often use the available information optimally [701. The technique must also be used properly; the "red cap syndrome" is a familiar entity in some centres where the electrical connection to the cardiac output computer remains in its packaged state. Indeed, in a recent British survey [71], nearly a quarter of ICU users did not even possess a cardiac output computer. Medical and paramedical staff have to be well trained, not only in proper usage but in identification of problems and avoidance of complications. Thus the increasing use of the PA catheters in the past few years may simply reflect underutilisation of clinical skills as well as overuse of the catheter [65]. Such overutilisation may expose patients to undue risks although the incidence and the severity of the complications due to bedside haemodynamic monitoring have declined over the years because of our increased awareness of the nature of these complications and their prevention. If PA catheterisation yields a better diagnostic accuracy and a substantial change in therapy, a decrease in morbidity and mortality could be reasonably expected. However, conflicting results can be found in the literature. Rao et al. [47] compared a prospectively-studied group of cardiac surgery patients with PA catheterisation with a retrospectively-studied group treated without a PA

catheter. Their results suggested that invasive monitoring, allowing optimisation of the patient's haemodynamic status, was associated with a reduction in cardiac morbidity and mortality. Shoemaker et al. [2] suggested that manipulation of oxygen-derived variables in surgical patients at risk could reduce postoperative mortality. Reynolds et al. [3] related the increased use of invasive monitoring to an improved outcome in patients with septic shock. In these studies, however, several factors other than PA catheterisation could have influenced outcome. In two large studies no relationship between the use of the PA catheter and outcome was evidenced [43, 72]. In a study by Knaus et al. [72], comparing patient outcome from 12 ICUs, the inter-hospital variations in the use of PA catheters had no significant influence on outcome. In a recent prospective study, Tuman et al. [43] observed that the use of the PA catheter did not influence outcome after CABG. Gore et al. [65] analysed retrospectively the impact of PA catheterisation on the outcome of 3263 patients after acute myocardial infarction and observed no impact in terms of mortality, length of hospital stay or long term prognosis. They even observed an excess mortality in catheterised patients. These results seem so disturbing that E. D. Robin in a provocative editorial asked for a moratorium on the use of the PA catheter, at least in patients with myocardial infarction [73]. Although these elements stimulate the organisation of prospective randomised controlled studies on the use of the PA catheter, several important questions should be raised: 1) Are the routinely measured haemodynamic parameters adequately collected and interpreted [74]? 2) Is there a consensus on the therapy applied to any haemodynamic alteration? 3) Is the therapy applied able to influence outcome? Ignoring these complex aspects would result in clinical trials unable to demonstrate any beneficial effect of the PA catheter. If prospective studies are performed, they must be focused on very specific groups of patients. Although an altered mortality rate is incontrovertible proof, effects on morbidity, hospital stay and cost should also be examined, Whether the newer (and more expensive) refinements such as S v O 2 and RVEF add anything to these factors should be also analysed. Despite all these misgivings and shortcomings, there is, at present, no superior bedside technique for haemodynamic monitoring. Perhaps a complementary role will be found whereby a non-invasive technique will signal an earlier need for invasive monitoring; not only could this prove both cost-effective and practical but

Expert Panel." The use of the pulmonary artery catheter (ESICM)

may also produce the sought-after impact on survival statistics. Thoughout the world, there is considerable variation in usage of PA catheterisation. In the United States, approximately, 1000000 catheters are inserted annually. The recent study of British usage q u o t e d earlier revealed t h a t only 6 0 0 0 - 8000 catheters are inserted per year. A d j u s t e d for p o p u l a t i o n , this insertion rate is some 40 times lower. Obviously e c o n o m i c considerations play a part. D o u b t s were expressed over the benefit to risk ratio, the cost a n d the clinical indications for insertion. I n any case, a reasonable voice argues for a n earlier utilization in anticipation o f a clinical catastrophe. W h e n the c o n d i t i o n has b e c o m e irreversible, little can be achieved by the a d d i t i o n o f intensive monitoring.

References i. Swan HJC, Ganz W, Forrester JMH, Diamond G, Chonette D (1970) Catheterization of the heart in man with use of a flowdirected balloon-tipped catheter. N Engl J Med 283:447-451 2. Shoemaker WC, Appel PL, Kram HB, Waxman K, Lee TS (1988) Prospective trial of supranormal values of survivors as therapeutic goals in high-risk surgical patients. Chest 94:1176-1186 3. Mitzner W, Wagner EH (1989) Pulmonary and bronchial vascular resistance. In: Sharf SM, Cassidy SS (eds) Heart-lung interactions in health and disease. Dekker, 42:45-75 4. Mangano DT (1980) Monitoring pulmonary arterial pressure in coronary artery disease. Anesthesiology 53:364-370 5. Teboul JL, Besbes M, Axler O, BrunBuisson C, Andrivet P, Rekik N, Lemaire F (1988) A bedside index for determination of zone III condition of pulmonary artery (PA) catheters tips during mechanical ventilation (MV). Am Rev Respir Dis 137 [Suppl]: 139 6. Shasby DM, Dauber IM, Pfister S et al. (1981) Swan-Ganz catheter location and left atrial pressure determine the accuracy of the wedge pressure when positive endexpiratory pressure is used. Chest 80:666- 670 7. Suter PM, Lindauer JM, Fairley HB, Schlobohm RM (1975) Errors in data derived from pulmonary artery blood gas values. Crit Care Med 3:175-181 8. Calvin JE, Driedger AA, Sibbald WJ (1981) Does the pulmonary capillary wedge pressure predict left ventricular preload in critically ill patients? Crit Care Med 9:437-443 9. Stetz CW, MiIler RG, Kelly GE, Raffin TA (1982) Reliability of the thermodilution method in the determination of cardiac output in clinical practice. Am Roy Respir Dis 126:1001 - 1004 10. Kubo SH, Bnrchenal JEB, Cody RJ (1987) Comparison of direct fick and thermodilu-

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

tion cardiac output techniques at high flow rates. Am J Cardiol 59:384-386 Cigarroa RG, Lange RA, Williams RH, Bedotto JB, Hillis LD (i989) Underestimation of cardiac output by thermodilution in patients with tricuspid regurgitation. Am J Med 86:417-420 Pearl RG, Rosenthal MH, Nielson L, Ashton JP, Brown BW Jr (1986) Effect of injectate volume and temperature on thermodilution cardiac output determination. Anesthesiology 64:798- 801 Jansen JR, Schreuder J J, Bogaard JM, van Rooyen W, Versprille A (1981) Thermodilution technique for measurement of cardiac output during artificial ventilation. J Appl Physiol 50:584-591 Stevens H J, Thomas AR, Frederick GM et al (1985) Thermodilution cardiac output measurement: effects of respiratory cycle in its reproducibility. JAMA 253:2240 - 2242 Daper A, Parquier JN, Preiser JC, Contempr6 B, Vincent JL (1986) Timing of cardiac output measurements during mechanical ventilation. Acute Care 12:113-1t6 Kay HR, Afshari M, Barash P, Webler W, Iskandrian A, Bemis C, Hakki H, Mundth ED (1983) Measurement of ejection fraction by thermo-dilution techniques. J Surg Res 34:337-346 Vincent JL, Thirion M, Brimioulle S, Lejeune P, Kahn RJ (1986) Thermodilution measurement of right ventricular ejection fraction with a modified pulmonary artery catheter. Intensive Care Med 12:33-38 Dhainaut JF, Brunet F, Monsallier JE Villemant D, Devaux JY, Konno M, De Gournay JM, Armaganidis A, Iotti G, Huyghebaert ME Lanore JJ (1987) Bedside evaluation of right ventricular performance using a rapid computerized thermodilution method. Crit Care Med 15:148-152 Vincent JL, Reuse C, Frank N, Contempr6 B, Kahn RJ (1989) Right ventricular dysfunction in septic shock: assessment by measurement of right ventricular ejection fraction using the thermodilution technique. Acta Anesth Scand 33:34-38 Eddy AC, Rise CL (1989) The right ventricle: an emerging concern in the multiple injured patient. J Crit Care 4:58-66 Dhainaut JF, Ghannad E, Villemant D, Brunet F, Devaux JY, Schremmer B, Squara P, Weber S, Monsallier JF (1990) Role of tricuspid and left ventricular myocardial infarction in the treatment of right ventricular infarction-induced low cardiac output syndrome. Am J Cardiol (in press) Boldt J, Kling D, Moosdorf R, Hempelman G (i989) Influence of acute volume loading on right ventricular function after cardiopulmonary bypass. Crit Care Med 17:518-522 Vincent JL, Reuse C, Kahn RJ (1988) Effects on right ventricular function of a change from dopamine to dobutamine in critically ill patients. Crit Care Meal 16:659- 662

VII 24. Schulman DS, Biondi JW, Matthay RA, Barash PO, Zaret DL, Soufer R (1988) Effect of positive end-expiratory pressure on right ventricular performance. Am J Med 84:57-67 25. Seghal LR, Sehgal HL, Rosen AL, Gould SA, Moss GS (1984) Effects of intralipid on measurements of total hemoglobin and oxyhemoglobin in whole blood. Crit Care Med 12:907-909 26. Varon A J, Anderson HB, Civetta JM (1989) Desaturafion noted by pulmonary artery catheter oximeter after methylene blue injection. Anesthesiology 71:791 - 794 27. Shah KB, Rao TK, Laughlin S et al (I984) A review of pulmonary artery catheterization in 6245 patients. Anesthesiology 61:271-275 28. Davies M J, Cronin KD, Domainque E (1982) Pulmonary artery catheterization. An assessment of risks and benefits in 220 surgical patients. Anesth Intensive Care 10:9-14 29. Sprung CL, Marical EH, Garcia AA, Sequeira RF, Pozen RG (I983) Prophylactic use of lidocaine to prevent advanced ventricular artery arrhythmias during pulmonary artery catheterization: prospective, double-blind study. Am J Med 75:906-910 30. Rowley KM, Clubb KS, Smith GJW, Cabin HS (1984) Right-sided infective endocarditis as a consequence of flow-directed pulmonary-artery catheterization. A clinical study of 55 autopsied patients. N Engl J Med 311:1152-1156 31. Barash PG, Nardi D, Hammond G et al (1981) Catheter-induced pulmonary artery perforation, mechanisms, management and modifications. J. Thorac Cardiovasc Surg 82:5 - 12 32. Singer M, Bennett D (1989) Optimisation of positive end expiratory pressure for maximal delivery of oxygen to tissues using ocsophageal Doppler ultrasonography. Br Med J 298:1350-1353 33. Stevenson JG (1989) Comparison of several non-invasive methods for estimation of pulmonary artery pressure. J Am Soc Echo 2:157-171 34. Gotshall RW, Wood VC, Miles DS (1989) Comparison of two impedance cardiographic techniques for measuring cardiac output in critically ill patients. Crit Care Med 17:806-811 35. Preiser JC, Daper A, Parquier JN, Contempr6 B, Vincent JL (1989) Transthoracic electrical hioimpedance versus thermodilulion technique for cardiac output measurement during mechanical ventilation. Intensive Care Med 15:221-223 36. Miles DS, Gotshall RW, Quinones JD, Wulfeck DW, Kreitzer RD (1990) Impedance cardiography fails to measure accurately left ventricular ejection fraction. Crit Care Med 18:221-228 37. Forrester JS, Diamond GA, Swarm HJC (1977) Correlative classification of clinical and hemodynamic function after acute myocardial infarction. Am J Cardiol 39:137-145

VIII 38. Figueras J, Well MH (1979) Hypovolemia and hypotension complicating management of acute cardiogenic pulmonary edema. Am J Cardiol 44:1349-1355 39. Isner JM (1988) Right ventricular myocardial infarction. JAMA 259:826-834 40. Fein AM, Goldberg SK, Walkenstein MD et al (1984) Is pulmonary artery catheterization necessary for the diagnosis of pulmonary edema? Am Rev Respir Dis 129:1006-1009 41. Kaplan JA, Wells PH (I981) Early diagnosis of myocardial ischemia using the pulmonary artery catheter. Anesth Analg 60:789- 793 42. Waller JL, Johnson SP, Kaplan JA (1982) Usefulness of the pulmonary artery catheters during aortocoronary bypass surgery. Anesth Analg 61:221-222 43. Tuman K J, McCarthy R J, Spiess BD et al (1989) Effect of pulmonary artery catheterization on outcome in patients undergoing coronary artery surgery. Anesthesiology 70:199-206 44. Rabinovitch MA, Elstein J, Chiu C J, Rose CP, Arzoumanian A, Burgess JH (1983) Selective right ventricular dysfunction after coronary bypass grafting. J Thorac Cardiovasc Surg 86:444-450 45. Atria RR, Murphy JD, Snider M e t al (1976) Myocardial ischemia due to infrarenal aortic cross clamping during aortic surgery in patients with severe coronary artery disease. Circulation 53:961-965 46. Silverstein RR, Caldera DC, Cullen DJ et al (1979) Avoiding the hemodynamic consequences of aortic cross clamping and unclamping. Anesthesiology 50:462-466 47. Rao TLK, Jacobs KH, E1-Etr AA (1983) Reinfarction following anesthesia in patients with myocardial infarction. Anesthesiology 59:499- 505 48. Welis PH, Kaplan JA (1981) Optimal management of patients with ischemic heart disease for non-cardiac surgery by complimentary anesthesiologist and cardiologist interaction. Am Heart J 102:1929-1937 49. Whittemore AD, Clowes AW, Hechtman HB, Mannick JA (1980) Reduced operative mortality associated with maintenance of optimal cardiac performance. Ann Surg 192:414-421 50. Hesdorffer CS, Milne JF, Meyers AM, Clinton C, Botha R (1987) The value of Swan-Ganz catheterization and volume loading in preventing renal failure in patients undergoing abdominal aneurysmectomy. Clin Nephrol 28:272-276 51. Marshall WK, Bedford RF (1980) Use of a pulmonary-artery catheter for detection and treatment of venous air embolism: a prospective study in man. Anesthesiology 52:131-134

Expert Panel." The use of the pulmonary artery catheter (ESICM) 52. Bowdle TA, Artu AA (1988) Treatment of air embolism with a special pulmonary artery catheter introducer sheath in sitting dogs. Anesthesiology 68:107- t i 0 53. Clark SL, Cotton DB (i[988) Clinical indications for pulmonary artery catheterization in the patient with severe preeclampsia. Am J Obstet Gynecol 158:453-458 54. Masson RG, Ruggieri J (1985) Pulmonary microvascular cytology: a new diagnostic application of the pulmonary artery catheter. Chest 88:908-914 55. Lemaire F, Teboul JL, Cinotti Li~et al (1988) Acute left ventricular dysfunction during unsuccessful weaning from mechanical ventilation. Anesthesiology 69:171 - 179 56. Sibbald W, Short AIK, Driedger AA, Wells GA (1985) The immediate effects of isosorbide dinitrate on right ventricular function in patients with acute hypoxemic respiratory failure. A combined invasive and radionuclide study. Am Rev Respir Dis 131:862-868 57. Radermacher P, Santak B, W~ist H J, Tarnow J, Falke KJ (1990) Prostacyclin for the treatment of pulmonary hypertension in the adult respiratory distress syndrome: effects on pulmonary capillary pressure and ventilation-perfusion distributions. Anesthesiology 72:238- 244 58. Edwards JD, Redmond AD, NightingaIe P, Wilkins RG (1988) Oxygen consumption following trauma: a reappraisal in severely injured patients requiring ventilation. Br J Surg 75:690-692 59. Abraham E, Bland RD, Cobo JC, Shoemaker WC (1984) Sequential cardiorespiratory patterns associated with outcome in septic shock. Chest 85:75-80 60. Gilbert EM, Haupt MT, Mandanas RY, Huaringa AJ, Carlson RW (1986) The effect of fluid loading, blood transfusion, and catecholamine infusion on oxygen delivery and consumption in patients with sepsis. Am Rev Respir Dis 134:873-878 61. Astiz ME, Rackow EC, Falk JL, Kaufman BS, Weil MH (1987) Oxygen delivery and consumption in patients with hyperdynamic septic shock. Crit Care Med 15:26-28 62. Groeneveld ABJ, Kester ADM, Nauta JJP, Thijs LG (1987) Relation of arterial blood lactate to oxygen delivery and hemodynamic variables in human shock states. Circ Shock 22:35-53 63. Reynolds NH, Haupt MT, Thill-Baharozian MC, Carlson RW (1988) Impact of critical care physician staffing on patients with septic shock in a university medical intensive care unit. JAMA 260:3446-3450 64. Schreuder WO, Schneider A J, Groeneveld ABJ, Thijs LG (1989) The effect of dopa-

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

mine versus norepinephrine on hemodynamics in septic shock, with emphasis on right ventricular performance. Chest 95:1282-1288 Gore JM, Goldberg R J, Spodick DM, Alpert JS, Dalen JE (1987) A communitywide assessment of the use of pulmonary artery catheters in patients with acute myocardial infarction. Chest 92:721-727 Shell WE, Dewood MA, Peter 2", Mickle D, Prause JA, Forrester JS, Swan HJC (1982) Comparison of clinical signs and hemodynamic state in the early hours of transmural myocardial infarction. Am Heart J 104:521-528 Connors AF, McCaffree DR, Gray BA (1983) Evaluation of right-heart catheterization in the critically ill patient without acute myocardial infarction. N Engl J Med 308:263 - 267 Eisenberg PR, Jaffer AS, Schuster DP (1984) Clinical evaluation compared to pulmonary artery catheterization in the hemodynamic assessment of critically ill patients. Crit Care Med 12:549-553 Rekik N, Brochard L, Rauss A, BrunBuisson C, Teboul JL, Andrivet P, Besbes M, Messadi AA, Lemaire F (1989) Prospective assessment of the benefit and risk of Swan-Ganz catheter in critically ill patients (abstract). Am Rev Respir Dis 139:A17 Connors AF, Dawson NV, McCaffree DR, Gay BA, Siciliano CJ (1987) Assessing hemodynamics status in critically ill patients: do physicians use clinical informations optimally? J Crit Care 2:174-180 Singer M, Bennett ED (1989) Haemodynamic monitoring in the United Kingdom. Enough or too little? Chest 95:623 -626 Knaus WA, Draper EA, Wagner DP, Zimmerman JE (1986) An evaluation of outcome from intensive care in major medical centers. Ann Intern Med 104:410-418 Robin ED (1987) Death by pulmonary artery flow-directed catheter (editorial). Time for a moratorium? Chest 4:727-731 Silverman H J, Eppler JM, Pitman AP, Patz D (1987) Pulmonary artery wedge pressure measurements in patients on assisted ventilation. J Crit Care 2:115-120

Prof. J.-L. Vincent Department of Intensive Care Medicine Erasme University Hospital Route de Lennik 808 B-1070 Brussels Belgium