Infection is a common problem for patients in cardiothoracic intensive care units (CT ICUs) and although it may be the primary reason for admission, many more infections are acquired on the ICU. Estimates of incidence rates vary but are at least 5%. Undoubtedly, infection increases morbidity, the length of ICU admission and mortality.
This chapter considers infected patients in general along with specific infections of particular importance to patients admitted to the cardiothoracic ICU. Antibiotic therapy is considered in Chapter 12. Patients undergoing thoracic transplantation (Chapter 40 and Chapter 41) are particularly prone to infections and susceptible to a wider range of pathogens.
Many patients develop systemic inflammation following major surgery and cardiopulmonary bypass. It can be difficult to distinguish those in whom infection is the aetiology from the signs and symptoms of inflammation.
The identification of patients with sepsis and septic shock used to be based on heart rate, temperature, respiratory rate and the leucocyte count in the context of suspected infection. These criteria have poor predictive value of organ dysfunction and outcome in patients following cardiac surgery, maybe in part because they can be modified by recent cardiopulmonary bypass or concomitant therapies such as mechanical ventilation, pacing and renal replacement therapy. Recently, the consensus definition of sepsis has been modified and examines changes in the sequential organ failure score (SOFA) from baseline (see Table 32.1). This new system has not been validated specifically in the cardiothoracic patient but may be more useful. However, since organ dysfunction can be the result of an inadequate cardiac output, patients may have elevations in their SOFA score for an alternative reason.
|No hypotension||MAP <70 mmHg||Dopamine <5 μg/kg/min|
|<1.2||1.2–1.9||2.0–3.4||3.5–4.9||>5.0 or <200|
Clinicians often use laboratory investigations to guide their assessment of the presence of infection. However, it is normal for the C-reactive protein (CRP) and white cell count to increase following cardiac surgery even in the absence of infection; whether the magnitudes of these changes are greater in the setting of concomitant infection is not clear. Procalcitonin (PCT) is a peptide biomarker that has greater sensitivity and specificity for bacterial sepsis, and its resolution, than other cytokines or CRP. It has been employed in general populations of critically ill patients to limit the initiation or duration of antibiotic therapy. PCT rises following cardiopulmonary bypass with levels exceeding thresholds used in other populations to define bacterial infection. Even higher levels of PCT are observed with infection following cardiac surgery, but whether modifying antimicrobial practices based on these higher thresholds impacts on patient outcomes has never been tested. Moreover, in one study over half of patients with mediastinal infections did not have increased PCT levels.
The nature of the infecting organisms may be revealed by standard microbiological investigations of samples from normally sterile sites. Frequently these investigations are negative in the context of prior initiation of antibiotic therapy. Molecular techniques such as 16S RNA sequencing of tissues and fluids can be invaluable in settings such as infective endocarditis where standard techniques fail due to the presence of antibiotics or fastidious organisms. Similarly, PCR based assays of respiratory samples can identify pathogenic viruses and Pneumocystis jirovecii rapidly. 1,3 beta-D-glucan assay, mannan and anti-mannan antibody assays may help diagnose invasive fungal infection. Close liaison with a clinical microbiologist or infectious disease specialist is important.
Therefore, ultimately the diagnosis of an infected patient is based on clinical judgement of the clinical setting, laboratory results cogniscent that they may change for other reasons, microbiological data available and other investigations such as the chest radiograph.
The Surviving Sepsis Campaign has done much to educate clinicians about sepsis. It has created evidence-based guidelines for the management of patients with sepsis which are wide ranging and cover many aspects of care including antibiotic therapy, processes of care, management of concomitant respiratory failure and fluid resuscitation. Protocolisation of aspects of care has been associated with improved outcomes. Some of the key recommendations are highlighted in Table 32.2 and these are equally applicable to patients with cardiothoracic illness.
|Recommendation||Grade of recommendation*|
|Give antibiotics early|
|Identify (e.g. take microbiological cultures) and control the source of infection early, for example drain an abscess||1C|
|Crystalloids are the first choice for intravascular fluid replacement||1B|
|Human albumin solution can be considered in patients requiring considerable amounts of crystalloid; avoid starch solutions||1C|
|Noradrenaline is the first choice vasopressor||1B|
|Protocolised blood sugar management should aim for a blood glucose <180 mg/dl (10 mM)||1A|
|Hydrocortisone (200 mg/day) should only be considered if fluids and vasopressors do not restore haemodynamic stability||2C|
* Grade 1 recommendations are strong; Grade 2 are weak. The quality of the evidence is indicated as high (A, based on randomised controlled trials), moderate (B, downgraded RCTs or upgraded observational studies), low (C, well conducted observational studies with control RCTs) or very low (D, downgraded controlled studies or expert opinion).
The guidelines published in 2012 have been modified recently to remove specific goals of initial fluid resuscitation based on the central venous pressure and venous oxygen saturations. Three well-conducted large multicentre randomised trials – ProMISe, ProCess and ARISE – did not demonstrate a benefit of targeting these parameters during initial fluid resuscitation. Moreover, it is likely that optimal fluid resuscitation strategies may differ in patients with intrinsic cardiac dysfunction.
The hallmark circulatory changes associated with severe infection are arterial and venous vasodilatation and increased vascular permeability resulting in tissue oedema. This causes systemic hypotension that is mitigated normally by the physiological response of an increased cardiac output. Healthy adults can increase their cardiac index to well in excess of 4–5 l/min/m2. This may not be possible in those with heart failure, valvular heart disease or with therapies such as beta-adrenoreceptor blockade. Typically, clinicians administer considerable volumes of intravascular fluid to combat the increased venous compliance and hypotension. However, consideration should be given to the effects of further fluid administration on stroke volume (the primary goal of fluid administration) and the balance between the need to increase cardiac output versus the need to increase vascular tone with drugs such as noradrenaline and vasopressin. Consideration of these factors is ever more important in those with cardiac disease. For example, excessive fluid administration may adversely affect right ventricular dysfunction, which may be exacerbated by the increase in pulmonary vascular resistance that can occur during sepsis and systemic inflammation. Hypotension can be exacerbated further by the myocardial dysfunction related to systemic sepsis per se. In some settings inotrope therapy may be indicated. No study demonstrates a specific inotrope to be superior, but anecdotally milrinone is often associated with worsening vasoplegia and many prefer dobutamine or small to moderate doses of adrenaline. A recent study of levosimendan, a myocardial calcium sensitiser, showed no advantage in preventing organ dysfunction or mortality in sepsis.
Corticosteroids have been demonstrated to reduce the duration of hypotension in a general critical care population whilst not impacting on mortality. Restoration of the shock state may beneficially enhance coronary perfusion in those with coronary artery disease or significant right heart failure due to pulmonary hypertension.
Management of Specific Infections
Infective endocarditis (IE) is a challenging condition with poor outcomes for patients. In hospital mortality rates are as high as 30%. Its incidence is increasing at least in part due to the increasing number of patients at risk, for example those with intracardiac prosthetic material.
The management of infective endocarditis is complex and best undertaken by teams. Critical care physicians are often involved either following valvular interventions or if patients present with complications such as heart failure, uncontrolled infection or embolisation. Right sided endocarditis is less common and more often associated with intravenous drug abuse, congenital heart disease or invasive vascular devices such as central lines or pacing systems. It is generally tolerated better haemodynamically than left sided disease. Right sided endocarditis presents typically with multiple pulmonary septic emboli which may be manifested by breathlessness, chest pain and haemoptysis.
Antimicrobial therapy is the cornerstone of therapy in IE and choice of antibiotic is influenced by whether the endocarditis is on a native valve or prosthetic valve, the organism and the minimum inhibitory concentration (MIC) for a particular organism/antibiotic combination. There is less emphasis on aminoglycosides in recent guidelines due to renal toxicity. Rifampicin should not be instituted immediately as it probably has an antagonistic effect against other antibiotics with respect to replicating bacteria.
Heart failure may occur due to severe valvular regurgitation, a fistula (e.g. an acquired Gerbode ventriculoatrial defect) or rarely valvular obstruction. Severe heart failure refractory to medical therapies is an indication for expedited surgery.
Uncontrolled infection must be considered when fever persists or there is progressive perivalvular extension of infection as manifest by complications such as an aortic root abscess, fistula or pseudoaneurysms. Typically blood cultures become negative in a few days. New prolongation of the PR interval should raise suspicion of aortic root abscesses.
Cerebral embolisation from left sided endocarditis is a concern and can be associated with haemorrhagic transformation too. The risks of embolisation fall progressively following institution of antimicrobial therapy.
Surgery during the active phase of endocarditis has high risks and in general it is deferred until 4–6 weeks. Nevertheless it may be expedited in the setting of severe heart failure, uncontrolled infection or prevention of emboli.
Infection of implanted pacing systems or cardioverter defibrillators is increasing as more devices are inserted. It can be difficult to manage and may be associated with mortality as high as 10–22%. Infection can be in one or more of the endocardium, leads, or device and its pocket. Estimates of the incidence range from 0.5% to 2.2%. Risk factors for infection include inexperienced operators, low frequency of air changes in the operating environment, diabetes mellitus and renal failure. Management strategies include antibiotics and device removal. Device removal may be complex in those who are dependent on a device and if leads have been present for some time when they may become adherent to vascular structures. Device and lead removal may be complicated by bleeding, tamponade, pulmonary embolism and death.
Infection is a frequent complication of long term durable ventricular assist devices. Driveline infection is the most common and usually occurs at the exit site but infections may occur anywhere along the driveline up to the device pocket, or in the device or cannulae. Typically there is local trauma at the exit site or an inadvertent tug on the line. This underpins the careful initial surgical placement of the driveline and dressings that stabilise the line. Management strategies include the simultaneous use of antibiotics, fastidious exit site cleaning and source control with debridement of infected tissue and drainage of collections. Rarely, device exchange or cardiac transplantation may be considered.