Introduction Monitoring has become a progressively more important part of the management of the critically ill patient over the last decade. However, the wide selection of available techniques offers a range of applicability and invasiveness, and the need to contain costs has contributed a significant extra economic dimension to the purely medical problems of choosing a monitoring system. Since its development, the pulmonary artery catheter has been widely used as a diagnostic and monitoring tool [1]. If current recommendations are followed, the risk-benefit ratio remains very low although, even in the most experienced hands, mishaps and failures can occur [2]. However, owing to alterations in compliance of the left ventricle and the risk of incorrectly estimating left heart filling from measured pressures, less indirect assessments of flows and volumes appear to allow better evaluation of preload [3–5]. More recently, automated technology has simplified haemodynamic assessment and it is relatively non-invasive. Echocardiography was first introduced in cardiological practice for determining pressure gradients in patients with mitral stenosis, but its use as a diagnostic tool has grown over the years. In particular, the introduction of echocardiography during and immediately after cardiac anaesthesia and surgery has led to a revolution in bedside diagnosis and monitoring. However, its role in other settings has still not been fully established. Lack of widespread experience and advanced training partly accounts for the reluctance to use the technique more widely. This review aims at a comprehensive coverage of the applications and usefulness of transoesophageal echocardiography and colour Doppler (TOE) for diagnosis and assessment of cardiac function in critically ill patients. Anatomy and images Separate authors have described the anatomy and related echocardiographic images. Standard views consist of cross-sectional imaging in the transverse plane [6,7]; the three major views are the short axis of the left ventricle at the level of the mid-papillary muscles (Fig. 1), the four-chamber and the short-axis view through the superior mediastinum. In addition, as much as possible of the aorta should be evaluated in order to complete the whole TOE investigation. At each level, various flows can be measured by means of colour flow examination and Doppler assessment. Biplanar and, subsequently, multiplanar imaging have made visualization of cardiac structures possible from different views, sometimes inaccessible from the transverse plane [8,9].Fig. 1.: Standard view at the transgastric level (short axis) of both the left (right side) and the right (left side) ventricle. Besides evaluation of global and regional contractility, an immediate idea of LV filling can be obtained. Right: systole; left: diastole.Determination of ventricular function Short-axis imaging in the transverse plane allows optimal visualization of both the left (LV) and right ventricle [6,7] and is the first image to be brought into view. The view gives immediate and relevant information about global contractility, LV ejection fraction and a first estimation of LV filling. Myocardial contractility can be evaluated in relation to the first three major areas of coronary perfusion. A correlation coefficient of 0.96 has been shown between the fractional area contraction (reduction in cross-sectional area of ventricular cavity) and the ejection fraction obtained by radionucleide angiography [10]. A hyperkinetic LV with a small end-systolic area and unenlarged end-diastolic area is indicative of hypovolaemia [11,12], although increased inotropic activity of the LV (e.g. after administration of inotropes) should be excluded [12]. Global and segmental wall movement can be routinely monitored intra-operatively at the level of the mid-papillary muscle [13–15], and these measurements help to predict post-operative morbidity and mortality [13]. The means for more concise evaluation of segmental wall motion have become available with the introduction of multiplane probes [8,9]. The longitudinal plane allows visualization of the apex of the LV and enables thrombi to be excluded; the whole posterior wall can be inspected in order to detect hypokinesis or an aneurysmal region. The zone perfused by the main left anterior descending artery—just beneath the mitral valve—is another part of the myocardium that can only be assessed in the longitudinal plane. Some appreciation of right ventricular contractility may also be obtained from the short-axis transverse plane, but a four chamber view is needed for a proper assesment of the contractility of the right ventricular free wall, from which hypokinesis can be distinguished or excluded. Acute overload of the right heart may also be characterized by hypokinesis of the free wall and dilatation of the right chamber (Table 1).Table 1: Evaluation of hypotension in the critically ill patient using TOE Evaluation of the valvular apparatus Mitral valve Imaging of the mitral valve and the subsequent assessment of the function of the valvular apparatus is one of the most valuable adjuncts of TOE in critically ill patients. Morphological abnormalities (such as insufficient closure of one or both of the mitral valve leaflets, myxomatous degeneration and calcification) can be discerned. Abnormalities in the motility of one or both leaflets (bulging, prolapse or flail) can be concisely evaluated by analysis of multiplanar images [8,9]. Colour flow mapping provides insights into the width [16], length [17], direction of regurgitant jets and into the extent of mitral regurgitation into the left atrium. The next step in the evaluation of mitral valve function is Doppler assessment (Fig. 2). The Doppler pattern comprises an early filling wave and an atrial contraction wave [18]. Transmitral spectral Doppler analysis of the flow pattern offers insight into both early and late filling of the LV in various critical situations (Table 2). Doppler flow recordings are easy to obtain, but they need to be interpreted with due allowance for recognized modifying influences such as age [19], mechanical ventilation [20], the position of the sample volume [21] and the preload [18,22]. The important features of the early filling phase are mainly related to factors such as LV systolic shortening and end-systolic volume [22], which affect the magnitude of the early diastolic transmitral pressure gradient [21,22].Fig. 2.: Relation of the pulmonary vein and transmitral flow Doppler patterns to invasive presssures in the LV. The atrial contraction wave of both pulmonary vein (reverse) and at the level of the mitral valve (forward) concur with the p-wave of the ECG. The systolic wave (S) of the pulmonary vein occurs during the systolic phase of the ECG. The diastolic phase (D) of the pulmonary vein flow and early (E) and late (A) filling waves at the level of the mitral valve occur during diastole on the ECG.Table 2: Elementary steps in the primary evaluation of a critically ill patient by transoesophageal echocardiography Assessment of pulmonary venous flows and derived Doppler signals have now become standard clinical practice in the evaluation of the function of the mitral valve apparatus [23]. The spectral Doppler pattern is characterized by various, well-defined waves [21,24]. Owing to atrial contraction, a flow reversal is seen, which corresponds well with mean pulmonary capillary wedge pressure (r=0.81) [21]. This wave is followed by a first antegrade systolic flow, generated by the atrial relaxation and suction [25]. A second systolic wave follows, which is strongly related to the left ventricular contraction. Finally, a diastolic wave is observed with a direct correlation (r=0.61) with the early peak filling wave at the transmitral level, as shown in Fig. 2[26]. The systolic fraction of the pulmonary flow correlates significantly with mean left atrial pressure (r=0.88) [26]. Although several interfering factors have been identified, such as heart rate [23], age and respiration [24], and sample position [25], it is well accepted that routine evaluation of mitral insufficiency is most easily and rapidly performed by pulsed Doppler at the level of the pulmonary veins, the right vein in particular [27]. Aortic valve The long axis of the LV in both transverse and longitudinal planes, permits the evaluation of the morphology of the aortic valve at the level of the upper mediastinum [6–9]. However, Doppler assessment of the flow through the valve is difficult owing to the intercept angle between sample volume and direction of the flow. Recently, the flow across the aortic valve has been analysed [28] by using a transverse short-axis view with counterclockwise rotation of the probe (Fig. 3). In particular, colour flow mapping and Doppler assessment is possible, yielding optimal visualization of the regurgitant jet in the LV cavity and calculation of the pressure gradient. Cardiac output can be calculated as the product of the time-velocity integral, heart rate and the effective cross-sectional area of the LV outflow tract [28] or aortic valve [29].Fig. 3.: Long axis transverse view of the LV. Optimal image to evaluate aortic valve function: aortic valve insufficiency and Doppler flow can be evaluated.Tricuspid valve and pulmonary valve Morphology and function of the tricuspid valve is easily evaluated from the transverse plane in the four-chamber view [6,7], although the angle of intercept often interferes with correct determination of the severity of any regurgitation. Analysis of pulmonary artery flows can be used to assess right ventricular function [30]. However, the periods before and during ejection are influenced not only by right ventricular afterload, but depend also on right ventricular preload and contractility, heart rate and breathing [31]. The combination of the assessment of the severity of tricuspid regurgitation and the degree of right ventricular dilatation is useful in clinical practice for evaluating pulmonary arterial hypertension [32]. Imaging the pulmonary valve is difficult. Visualization is best under-taken in the longitudinal plane although no optimal transvalvular flows can be obtained [8,9]. Suspected aortic dissection There have been many reports on the high sensitivity and specificity of TOE in diagnosing and determining the nature of an aortic dissection [33–36]. There are strong indications for undertaking TOE in patients with thoracic trauma with fractures of the upper ribs and/or pleural effusion, particularly on the left side [37]. In patients with a vertebral fracture in the thoracic region, transverse dissection should also be excluded [38]. TOE appears to be the technique of choice because it can rapidly be made available at the bedside of haemodynamically unstable patients and it costs relatively little. Diagnosis by TOE can be sometimes difficult, for example when there is interference by tracheal air with the diagnosis of ascending aortic dissection and the occurrence of oscillations etc. Nienaber et al.[34] reported a specificity of only 0.77 (six false positive diagnoses in 26 patients). This is in contrast with the results of a European multicentre study which showed a specificity and sensitivity of 0.98 and 0.99 respectively [33,35]. These data emphasize the difficulties and need for cautious interpretation, in particular when first using the technique. Compared with computed tomography and angiography, TOE is less invasive and allows bedside identification of entry and reentry points and the presence of a thrombus in the false lumen. In addition, some important anatomical characteristics can be identified by TOE such as the presence of aortic regurgitation or pericardial effusion and the extent of the dissection into the coronary arteries. Other pathological conditions When used to exlude other pathological conditions, TOE is not a primary diagnostic tool, but might be helpful in diagnosing and confirming suspected findings on X-ray. Atelectasis is a frequent respiratory complication in the ICU. Pleural effusion is also diagnosed easily when it is located in front of the spine and the aorta. Clinical implications TOE is a cost-effective and practical addition to the available monitoring and diagnostic equipment, allowing and improving effective haemodynamic determination at the bedside. In combination with invasive pressure monitoring, TOE has been shown to provide highly sensitive bedside cardiac images, immediately relatable to morphology and function [39–42]. Owing to its unique ability for evaluating beat to beat cardiac function, TOE could be used to monitor the haemodynamic effects of various interventions in anaesthetized and critically ill patients [20,43]. The most pressing question for solution during initial resuscitative management or after failed empirical treatment of critically ill patients is whether the primary cause of haemodynamic derangement is cardiac or non-cardiac in origin. Traditionally, this crucial question has always been answered by indirect measures or invasive methods such as pulmonary artery catheterization and measurement of cardiac output. In the therapeutic management of critically ill patients, TOE opens another window on the heart [42,44,45]. It can be used to determine those patients who need more prolonged invasive pressure monitoring. The principal steps in the evaluation of an ICU patient with TOE are shown in Table 1. In contrast with earlier descriptions [7], we propose that shortaxis imaging of the LV and RV should be undertaken first. Besides direct visualization of systolic function and evaluation of abnormalities of regional wall motion, preliminary information about preload and LV filling [11,12,45] is obtained. Therefore, this image should be the standard view for every TOE investigation in every critically ill patient. In further stages, a four-chamber view should be obtained with evaluation of the morphology and function of the different values. More specific items should be then assessed with particular views and techniques as appropriate. Obliteration of the left ventricular cavity may be interpreted as hypovolaemia [46], but it is not a characteristic image [12]. Therefore, the combination of left ventricular short-axis measurements and both transmitral and pulmonary vein flow assessments are essential for the appropriate interpretation of the state of filling in critically ill patients (Table 2 and Fig. 2). Information from haemodynamic monitoring may be difficult to interpret in hypotensive patients who need continuous inotropic or vasopressor support in a critical care setting: changes in compliance of the LV are the basis for the non-linear relation between pressure and preload [4,5,11]. Moreover, when there is intrinsic positive end-expiratory pressure, the disparity between values will increase even more. For differential diagnosis, the combined use of two-dimensional echocardiography and Doppler leads straightforwardly to a diagnosis (Table 1). The finding of a small hyperdynamic LV with a E/A of less than 1 in conjunction with a normal pulmonary venous flow pattern is consistent with hypovolaemia. Cardiogenic shock is characterized by global hypokinesia up to regional akinesia or dyskinesia and a reduced fractional area contraction of the left ventricule. The pulmonary venous flow pattern demonstrates an inverse ratio in this setting (i.e. S/D<1). Septic shock patients often show a hypercontractile left ventricle with a large variety of both pulmonary venous and transmitral flow patterns [47]. Combined septic-cardiogenic shock is presented by a combination of the echocardiographic features of cardiogenic and septic shock (Table 1). In massive pulmonary embolism with acute right ventricular failure, TOE can offer a rapid bedside diagnosis for patients who cannot be transported because of haemodynamic instability: it can demonstrate thrombi and right atrial and ventricular dilatation in conjunction with reduced right ventricular function [48]. Finally, cardiac tamponade can generally be diagnosed by either transthoracic or transoesophageal echocardiography but, in certain situations (e.g. after cardiac surgery), the transoesophageal approach is the only practicable technique. In particular, evidence of right atrial or ventricular collapse may lead to a diagnosis, although left atrial compression by local thrombi has also been described [49]. Limitations, pitfalls and complications Although TOE has introduced a new era of clinical surveillance and diagnosis at the bedside of the critically ill patient, some unique considerations and potential limitations should be recognized. In non-intubated as well as in ventilated ICU patients, transthoracic echocardiography still remains the first choice. TOE plays a particular role in ventilated patients both in general and in specific situations such as during and after cardiothoracic surgery, in the exclusion of thrombi [50,51] or aortic dissection [34] and for the assessment of the function of valvular prostheses [52–54]. Patients following trauma, who are needing TOE investigation, should be examined cautiously, because of the potential presence of pathological conditions in the oesophagus or neck, and the probe should be introduced gently. All patients should be monitored to the required minimum standards for anaesthesia, including oxygen saturation, ECG and non-invasive blood pressure measurement. All patients need intravenous (i.v.) administration of sedatives, most often in conjunction with topical lignocaine spray. Pitfalls and potential erroneous diagnoses are most often because of misinterpretation of normal and pathological images [55] and are closely related to the experience of the operator. They include false messages (e.g. trabeculations), membranes and echo-free spaces. Furthermore, mitral regurgitation should be assessed in relation to normal systemic pressure [56]. Optimizing the blood pressure before intepreting either colour flow mapping or Doppler patterns of transmitral and pulmonary venous flow will greatly increase their value (Fig. 2). Decreasing health-care funding will require any new diagnostic and monitoring techniques to be evaluated increasingly carefully in the future. Benson et al. carried out a cost-benefit analysis of TOE in three different cardiac surgery settings: congenital heart disease, valvular heart disease and coronary artery disease [57]. They found most impressive financial benefits for patients having congenital heart surgery and valve repair surgery. The significance of TOE for both patient care and the cost-benefit of valve repair surgery had already been suggested in an earlier study by Sheikh et al.[58]. Finally, although they reported no comparison with invasive monitoring, Heidenreich et al. emphasized the clinically important contribution of TOE to the diagnosis and subsequent management of hypotension in critically ill patients [59], in allowing prediction of prognosis and mortality in critical care situations. The question whether echocardiography offers supplementary information over and above that provided by invasive pressure monitoring remains unanswered in critically ill patients, although Benson et al. stressed the advantages of less invasive monitoring techniques compared with pulmonary artery catheterization in mitral valve repair surgery [57]. The advantages of bedside TOE will only enhance the care of critically ill patients if the technique is performed by skilled and well-trained investigators [60,61]. For this reason, it is essential that some general familiarity with the basics of echocardiography should be formally included in educational and training requirements for both anaesthesia and critical care [62]. Future developments Although, as an intermittent technique, TOE still cannot replace the pulmonary artery catheter, it is conceivable that TOE can help in refining diagnosis and therapy. In the near future, TOE will complement invasive monitoring by offering on-line information on cardiac volumes [15]. Integrating systemic pressure monitoring and on-line volume data in a useable everyday package will promote estimation of systolic and diastolic function [63]. Smaller, more compact machines and probes should allow clinicians to use the technique in a more continuous manner, both intra-operatively and in the ICU: this will improve both surveillance and diagnosis. Combining on-line pressure and volume data, in conjunction with two-dimensional imaging and colour Doppler, will provide bedside diagnosis and facilitate appropriate therapy, thus pushing back the frontiers of critical care medicine.