Acute heart failure is usually defined as the rapid onset of, or change in, signs and/or symptoms of heart failure resulting in the need for urgent treatment.1 It can occur as the first manifestation of a failing heart (acute de novo heart failure) or it can occur in a patient with a chronic history of heart failure, in which case the term ‘acute decompensation’ is often applied.
The acute heart failure syndrome can present clinically in a number of ways. The current European Society of Cardiology (ESC) guidance1 lists six clinical ‘types’ (see Figure 1):
- worsening, or decompensated, chronic heart failure, usually with peripheral oedema and systemic and pulmonary congestion;
- pulmonary oedema, with respiratory distress, tachypnoea and systemic oxygen desaturation;
- hypertensive heart failure, with signs and symptoms of heart failure but also hypertension and usually relatively well-preserved left ventricular (LV) systolic function. Patients are often tachycardic, with signs of vasoconstriction and pulmonary congestion, but without much in the way of peripheral oedema;
- cardiogenic shock, with symptoms and signs of poor tissue perfusion due to low cardiac output and low blood pressure;
- isolated right heart failure, with a low output syndrome without pulmonary congestion but raised jugular venous pressure; and
- acute coronary syndrome and heart failure.
Another method of classification is to use the history and physical examination to classify patients into one of four haemodynamic profiles based on congestion and perfusion (see Figure 2).2 Patients can be described as ‘dry and warm’, ‘wet and warm’, ‘dry and cold’ or ‘wet and cold’. Mortality is twice as high in ‘wet and warm’ patients, and 2.5-fold higher in ‘wet and cold’ patients than in those who are ‘dry and warm’, after adjustment for other prognostic factors.2 It has been suggested that these simple bedside profiles may be used to guide and monitor therapy non-invasively in patients with decompensated heart failure.3 In some patients, more invasive methods of monitoring are required.
Acute heart failure poses a significant burden on the health services across Europe. In the UK, heart failure is the cause of approximately 5 % of emergency medical admissions, and heart failure patients account for 10 % of bed-days due to the relatively long duration of hospital admission.4 It is, therefore, not surprising that hospitalisation is responsible for 60–70 % of the total direct costs of heart failure to the healthcare system.5
The prevalence of heart failure in developed countries is likely to be around 2 %, increasing steeply with age.4 The mean age of patients hospitalised in the UK National heart failure audit was 77.3 years. Men outnumber women in younger age groups, but the sex ratio is reversed in very old age groups, where women heavily outnumber men.
The median length of stay was nine days in Europe, with an in-hospital mortality of 6–10 % rising to 15 % by three months6 and to 30 % by 12 months.7 Mortality is particularly high for patients with cardiogenic shock, with an in-hospital mortality of around 50 %. In general, readmission rates are high, with typical values of around 20–25 % at six months6 and 30 % at 12 months.8 Much effort has been expended to improve the post-discharge care of such patients, with many now entering chronic disease management programmes.
Characteristics of Patients
Surveys and registries in Europe and North America report a consistent picture of the characteristics of patients admitted with acute heart failure.6 The typical patient is elderly (median age in the early 70s) and there is a slight predominance of men. Almost 70 % will present with acute decompensation of chronic heart failure and the remainder with acute de novo heart failure, most patients having a history of both coronary artery disease and hypertension. Co-morbidity is very common, atrial fibrillation, type 1 diabetes, renal dysfunction and chronic lung disease being often present. Preserved systolic function is found in perhaps one-third of patients, particularly in elderly women. Acute coronary syndrome is frequently present in de novo cases (up to 40 %), but is less common in decompensation of chronic heart failure. Arrhythmia and infection are also frequently seen, as is non-compliance with medication for those with a history of heart failure. Progressive valvular heart disease such as progressive mitral regurgitation (which may accompany adverse LV remodelling) may also be an important cause of acute decompensation. Critical aortic stenosis may present as de novo acute heart failure, especially in the elderly not under medical follow-up.
There is a clear consensus on what constitutes good practice for evaluating patients with acute heart failure,1,9 although the evidence base for this consensus is sparse.
First, a thorough history taking and physical examination to identify cardiac and non-cardiac disorders and behaviours that might cause or accelerate the development or progression of heart failure should be undertaken. The physical examination should include assessment of the patient’s heart rate, blood pressure (taking note that hypotension is usually pathological, whereas normotension does not always ensure adequate organ perfusion), volume status, peripheral perfusion, skin temperature and venous filling pressure. Cardiac auscultation for S3 or S4 and valvular abnormality, and auscultation for lung crackles and pleural effusions should be performed.
Second, the following investigations should be considered:
- electrocardiogram (ECG) to provide information on heart rate, rhythm and conduction abnormalities, and perhaps on likely aetiology, such as ST elevation myocardial infarction;
- chest radiograph to assess the degree of pulmonary congestion, cardiomegaly, pleural effusion and lung pathology – bearing in mind the limitations of a supine chest radiograph;
- arterial blood gases to assess oxygenation, pCO2 and acid–base balance, particularly in patients with respiratory distress or signs of low organ perfusion. A rising lactate is usually a very late feature of severe cardiogenic shock. Beware of the unreliability of data from non-invasive pulse oximetry in patients with poor peripheral perfusion;
- laboratory investigations: full blood count, urea and electrolytes, serum creatinine, plasma glucose, liver function tests, international normalised ratio and troponin if appropriate – noting that mild elevation is not uncommon in acute heart failure, even in the absence of a clinical history of acute coronary syndrome;
- natriurietc peptides: their measurement may be helpful in excluding heart failure in the emergency setting, although it may take time for the peptide concentrations to rise in very acute cases. The evidence base is much stronger for the value in prognostication, based on both baseline values and on how much the peptide concentration falls prior to discharge. There is little evidence of benefit of serial monitoring in patients during admission (see discussion below); and
- 2D and Doppler echocardiography, which can provide important information on the underlying cardiac dysfunction. Regional and global LV function as well as right ventricular (RV) systolic function can be assessed, along with LV diastolic function, ventricular dyssynchrony, valvular structure and function (allowing, for example, to exclude critical aortic stenosis in patients presenting with de novo acute heart failure), pericardial abnormality (allowing to exclude pericardial collection and tamponade, which may present with features of cardiogenic shock) and mechanical complications of acute myocardial infarction (such as ventricular septal defect or severe mitral regurgitation). 2D and Doppler echocardiography can be useful in serial monitoring of cardiac status and haemodynamics (see discussion below).
It is important to realise that clinical examination is helpful but that no clinical sign is completely reliable in detecting haemodynamic abnormalities or in classifying patients according to the type of heart failure as per the schema shown in Figure 2.10 Most clinical signs are relatively specific, but are not sensitive for haemodynamic abnormalities: they are useful when present but can be absent even when haemodynamics are perturbed (see Table 1). Early 2D and Doppler echocardiography are paramount in the assessment of patients presenting with acute decompensated heart failure.
Monitoring should be started as soon as possible after arrival in hospital. In all critically ill patients, ‘routine’ observations, such as temperature, respiratory rate, heart rate, blood pressure (with a greater focus on mean arterial blood pressure), oxygenation, urine output and ECG should be considered essential. If oxygenation is a problem, then this should be monitored, and certainly if the patient requires a fraction of inhaled oxygen greater than that of air.
Arterial line insertion may be necessary if there is severe haemodynamic instability or if frequent arterial blood samples are required. Central venous lines may be useful for allowing the delivery of fluids and drugs (for example, inotropic agents) and monitoring of central venous pressure (CVP) and central venous oxygen saturation (ScvO2). A raised CVP is usually suggestive of disease, but it is not diagnostic of the underlying process. The measurement of ScvO2 may be useful; for example, a low ScvO2 (e.g., <50 %) is associated with tissue ischaemia, and is best used to monitor resuscitation attempts. However, the variability of ScvO2 means it does not reflect the mixed venous oxygen saturations that are measured from the pulmonary artery (PA) (e.g., measured when calculating the Fick cardiac output) tightly enough and is therefore not recommended as their surrogate.
Pulmonary Artery Catheterisation
Most patients do not require PA catheterisation. A clear indication for the insertion of such an indwelling catheter should be present.11
A recent meta-analysis of 13 published clinical trials demonstrated no benefit of a PA catheter on morbidity and mortality in a number of disorders, including heart failure.11 One of the largest studies on heart failure, the Evaluation study of congestive heart failure and pulmonary artery catheterization effectiveness (ESCAPE), showed no benefit from the routine use of such catheterisation.12
PA pressure measurement may prove useful in distinguishing cardiac and non-cardiac causes of shock in patients with complex cardiorespiratory disease or co-morbidity. Knowing the PA pressure may also be helpful in patients who are not responding to conventional therapy as one would expect, or where it is difficult to assess the volume status and perfusion of the patient. If hypotension or worsening renal function develops during therapy, data from the PA catheter may help distinguish excessive diuresis from inadequate perfusion. Patients with cardiogenic shock, or who are being considered for mechanical support or transplantation, will require a full haemodynamic assessment.
It should be remembered that the pulmonary capillary wedge pressure (PCWP) may be an unreliable estimate of LV end-diastolic pressure in the presence of mitral stenosis, aortic regurgitation, pulmonary venous occlusive disease, high airway pressure, RV dysfunction or other causes of ventricular interdependence, a poorly compliant left ventricle or invasive ventilation. Severe tricuspid regurgitation makes the thermodilution method of calculating cardiac output unreliable.
Serial 2D Echocardiography and Doppler Examination
Serial non-invasive semi-quantitative assessment, by Doppler echocardiography, of RV and LV filling pressure, stroke volume and PA pressure may obviate the need for invasive evaluation or monitoring.
In advanced heart failure, assessing cardiac output by the LV outflow method using Doppler echocardiography13 is more reliable than using thermodilution, and is highly correlated with the Fick method of estimation (with a systematic bias to slight underestimation by the Doppler method).14
The LV outflow tract (LVOT) diameter (d) is measured in systole from the parasternal long-axis view just below the insertion of the aortic cusps, and the LVOT cross-sectional area is calculated according to the formula of πr2 or 0.785d2 (the mean of three measurements). The LVOT flow velocity is measured by pulsed-wave Doppler from the apical 5-chamber view, with the sample volume in the middle of the outflow tract immediately below the aortic cusps and the time velocity integrals calculated (averaged over five consecutive cycles for sinus rhythm or averaged over 10 cycles for atrial fibrillation). Cardiac output is then calculated as follows: the LVOT time–velocity integral multiplied by the LVOT cross-sectional area multiplied by the heart rate.13
Even simpler methods include ‘oesophageal Doppler’ monitoring (ODM), which allows monitoring of the mean velocity of blood travelling through the descending thoracic aorta during ventricular systole. This can be translated into cardiac output by using a nomogram, or by using an M-mode echocardiographic estimate of aortic diameter. In a systematic review, ODM has been shown to have high validity for monitoring changes in cardiac output.13
LV filling pressures may be assessed by measuring the ‘transmitral flow in early diastole on pulsed Doppler’ divided by the ‘diastolic mitral annular velocity on tissue Doppler’ (E:Ea ratio). In patients in the intensive care setting, an E:Ea ratio >15 identified a PCWP >15 mmHg with a sensitivity of 86 % and a specificity of 88 %.15 However, more recently, a study in patients presenting with decompensated heart failure showed a lower sensitivity and specificity of the E:Ea ratio of >15 in predicting elevated PCWP (66 % and 50 %, respectively) particularly in patients with large LV volumes and those receiving cardiac resynchronisation therapy.16
Other Methods of Monitoring Haemodynamics
Other non-invasive or minimally invasive methods of monitoring cardiac output include pulse-contour analysis, pulse-power analysis and impedance cardiography.17
Pulse-contour analysis uses data from a peripheral artery (invasively or non-invasively) or fingertip (non-invasively) and models the stroke volume from the systolic, diastolic or both portions of the pressure waveform (this is based on the concept that the contour of the arterial pressure waveform is proportional to stroke volume). Such systems need to be calibrated against transpulmonary thermodilution (for example, cold normal saline is injected into a central vein and is detected by a thermistor within a peripheral arterial line, for example a PiCCO™ system), but thereafter pulse-contour analysis can provide serial or continuous monitoring of cardiac output. However, this technique may give inaccurate cardiac output measurements in low cardiac output states, in rhythm disturbances and in the presence of regurgitant valve lesions, shunts, pulmonary embolism and pneumonectomy. Furthermore, it is highly dependent on arterial compliance (for example, changes in vascular resistance, i.e., vasoconstriction) and therefore requires frequent calibrations (every one to four hours).
Pulse-power analysis uses data from the arterial pulse pressure wave to calculate changes in stroke volume (rather than absolute values), although calibration against transpulmonary thermodilution can provide reasonably accurate estimates of cardiac output.
In an attempt to determine cardiac output accurately from pulsecontour analysis or pulse-power analysis without the need for calibration against transpulmonary thermodilution, various algorithms have been developed (e.g., FloTrac™ and LiDCOrapid™), but further work is necessary to determine the validity of such simpler systems. Impedance cardiography continuously measures electrical resistance changes throughout the thorax as aortic blood volume increases and decreases during the cardiac cycle, thus deriving changes in stroke volume, cardiac output, myocardial contractility, total thoracic fluid volume and systemic vascular resistance. Four pairs of cutaneous electrodes are required, each pair comprising a transmitting and sensing electrode. This method has been validated against other techniques in a number of clinical conditions, including heart failure,18 but requires further study before it can replace current approaches.
Serial Natriuretic Peptide Measurement
Including measurements of plasma brain natriuretic peptide (BNP) or N-terminal prohormone of BNP (NT-proBNP) in the initial assessment of a patient presenting to the accident and emergency department with symptoms that might be due to heart failure can help to confirm or refute the diagnosis, and lead to a more rapid and effective use of healthcare resources.19–21
Failure of plasma concentration to fall during an admission is a poor prognostic sign.22 In chronic heart failure, adjusting therapy to drive down natriuretic peptide concentrations to a target level may improve outcome, particularly in younger patients, but has yet to become part of routine practice.23 There is no evidence that the serial measurement of natriuretic peptide concentration is a useful addition to monitoring acute heart failure in the intensive care setting, Doppler measurements being more useful.15
The treatment of acute heart failure is beyond the scope of this article, but the goals of therapy are to improve symptoms, stabilise the haemodynamic condition rapidly, resolve congestion and improve prognosis. A treatment strategy should be clearly identified, based upon the patient’s particular circumstances. The majority of patients will require long-term therapy of what will become chronic heart failure – with enrolment in a chronic heart-failure management programme likely to optimise outcomes. However, in selected patients with cardiogenic shock who do not respond to standard therapies, mechanical circulatory support with a ventricular assist device could be considered. This support should be instituted early, as it is contraindicated in the presence of multiorgan dysfunction. The Interagency Registry for Mechanically Assisted Circulatory Support has published referral criteria for patients who should be considered for mechanical support.24 For example, patients with cardiogenic shock despite escalating support are recommended to be referred within hours, and those with progressive decline in clinical status with inotrope-dependent severe acute heart failure should be referred within days, provided multiorgan failure has not supervened.24
Earlier Detection of Decompensation
Often patients with acute heart failure present in extremis, with clear evidence from their history of decompensation occurring over days to weeks beforehand. Much effort has focused on how to detect such deterioration early, in the hope that simple measures (improved compliance with medication, dietary measures, a small increase in oral diuretic dose) might correct the situation before it gets to crisis point.
Patient monitoring, particularly of daily weight, is recommended in most guidelines,1,9 but it is likely to identify only a proportion of patients with cardiac heart failure who are deteriorating. A weight gain of greater than 2 kg over 48–72 hours has good specificity (97 %) but poor sensitivity (9 %) for predicting clinical deterioration.25
Remote monitoring of weight and symptoms has been shown to reduce the risk of mortality and heart-failure hospitalisation in many small studies,26 but a recent large randomised controlled trial suggests that this may not be the case if patients have only mild symptoms, are on optimal drug therapy and have had heart failure for a number of years.27
An increasing number of patients with heart failure have a therapeutic device implanted, such as a cardiac resynchronisation therapy device or a defibrillator. In addition to delivering therapy, such devices can record and transmit information on a large number of physiological variables, such as intrathoracic impedance, heart rate variability, patient activity and frequency of atrial and ventricular arrhythmia.
Initially, it was hoped that one variable could be identified that would provide a clinically useful marker of impending decompensation, but this has proved not to be the case. More recently, algorithms that combine trends in several such ‘diagnostics’ have been developed that help identify patients who may be at increased risk of decompensation. The value of this should not be overstated: although such algorithms can identify patients at a five-fold increased risk of decompensation in the following month, the absolute risk is still only of the order of 5 % or so.28
It is vital that the monitoring personnel do not over-react to streams of ‘diagnostic’ data, as this may only serve to increase patient anxiety and unnecessary medication changes, and may lead to an increase in hospitalisation. This effect was seen in a randomised study of adding an ‘alert’ to clinical management when a trend in decreased intrathoracic impedance was seen on remote monitoring.29 Further large randomised trials of a strategy of remote monitoring of patients with implanted devices are currently under way.
Other completely implantable monitoring-only technologies have been developed to measure RV, PA or left atrial pressure reliably.30–32 Although not yet at the stage where they can be implemented in routine practice, early data suggest considerable promise. For example, in a randomised trial of more than 500 patients with New York Heart Association functional class III heart failure and an implanted battery-free pressure transducer in the PA, daily interrogation of the device by the patient at home enabled the treating physician to drive down PA pressure by modifying drug therapy. This led to a substantial reduction in the need for rehospitalisation because of heart failure.31
Chronic left atrial pressure monitoring is also technically feasible and has been shown to give the patient and treating physician a reliable haemodynamic target for therapy.32 Considerable validity is given to such an approach from the correlation between such measurements and patients’ clinical status.
Figure 3 depicts the time trends in RV and estimated PA pressure in patients before and after a decompensation of chronic heart failure.33 Further work is necessary to identify which patients are most likely to benefit from such a ‘high-tech’ approach to monitoring, and how best to integrate the information flow from these devices into a chronic disease management programme.
Acute heart failure is a serious clinical syndrome, associated with a high mortality and a high risk of subsequent rehospitalisation. Most patients are elderly and have considerable co-morbidity. A thorough history, physical examination and 2D and Doppler echocardiography are vital parts of the initial evaluation, and will determine the monitoring and treatment strategy. Monitoring will include frequent reassessment of vital signs, such as blood pressure, heart rate, body temperature, peripheral perfusion, urine output and pulmonary and systemic congestion. PA catheterisation should only be performed if there is a good rationale for it – usually in patients for whom an assessment of congestion and perfusion is difficult, when the therapeutic response is not adequate or when mechanical intervention is being considered. Minimally invasive or non-invasive monitoring with a range of techniques, including ODM, is routine in intensive care practice. There is no evidence that the serial measurement of plasma natriuretic peptides is useful in the acute setting, but it can be useful for prognostication and tailoring therapy in the chronic disease management phase. Increasingly, patients with heart failure have implanted devices that allow remote monitoring of a range of physiological variables, but the utility and cost-effectiveness of this approach to allow earlier detection of decompensation requires further work.