Strategies for assisting ventilation include endotracheal intubation and positive pressure ventilation and recently several other modes including noninvasive techniques.

Ventilatory support using high-flow humidified nasal oxygen has been introduced into neonatal practice for respiratory support. There are limited reports of its use in older children with respiratory distress (Spentzas et al. 2009; McGinley et al. 2009). There are a number of commercial systems available. There is considerable variability in the pressures and flows delivered to infants with these systems (Dani et al. 2009; Lampland et al. 2009), and it is suggested that pressure limitation devices should be incorporated (Lampland et al. 2009). More research is required before these can be recommended for respiratory support in children (Randolph 2009) with acute respiratory distress.

Noninvasive ventilation has been attempted in children with acute respiratory failure due to a wide variety of conditions (Teague 2005) including neuromuscular conditions (Katz et al. 2004; Reddy et al. 2004; Hartmann et al. 1994), asthma (Haggenmacher et al. 2005; Ram et al. 2005) as well as acute respiratory failure due to infections (Prado et al. 2005; Piastra et al. 2004; Fortenberry et al. 1995).

After an initial report of nCPAP in an infant with apnoea related to RSV infection (McNamara and Sullivan 1997), Thia et al. (2008) demonstrated that nCPAP administered to infants with bronchiolitis and hypercapnia was associated with significant improvement.

A report on 14 patients from Chile (Prado et al. 2005) showed that nCPAP or BiPAP administered via nasal mask in acute respiratory failure due to infection was associated with significant improvement and that most children did not need invasive ventilation.

A recent Cochrane review of negative pressure ventilation in acute respiratory failure (Shah et al. 2005) concluded that there was not enough evidence available to make a recommendation as to its clinical applicability.

Thus it may be possible to use conventional or noninvasive (including nCPAP, BiPAP and mask) ventilation in children with acute respiratory failure related to infection. Unfortunately firm recommendations cannot be made until challenges in terms of masks (including fit), indications and techniques are elucidated (Teague 2005; Loh et al. 2007) and proposed predictors of success (Bernet et al. 2005) confirmed.

Anaesthesia for endotracheal intubation for conventional ventilation is essential, but there is considerable controversy regarding the optimal agents for anaesthesia in acute sepsis (Zelicof-Paul et al. 2005). Selection of appropriate agents will depend on experience and training.

Etomidate has been used in a number of studies, but there remains considerably concern regarding possible adrenal suppression following its use in sepsis (Zed et al. 2006; Zuckerbraun et al. 2006). It does however create favourable conditions for intubation (Zelicof-Paul et al. 2005) and is less likely to cause hypotension than agents such as midazolam (Choi et al. 2004).

Ketamine has been used extensively in many settings with a very low complication rate (Melendez and Bachur 2009; Jankiewicz and Nowakowski 1991) and is an agent of choice in many emergencies, while propofol and thiopentone have been used with some success.

Mechanical ventilation in respiratory failure has been one of the core functions of intensive care since the earliest days of the speciality (Grenvik and Pinsky 2009). The goals of mechanical ventilation have primarily been focused on the maintenance of acceptable blood gases in patients with failure of the respiratory system, but there is also a considerable body of literature related to cardiopulmonary interactions and the effect of mechanical ventilation on cardiovascular function (Bronicki and Anas 2009).

Essentially the goals of ventilation in sepsis and organ failure could be summarized as:

The priorities of ventilation in sepsis, from the immediate resuscitation at the time of presentation through the recovery phases of the illness, may differ as the physiology changes. Thus acceptable goals will depend on the stage of the illness.

Haemodynamic instability is a common feature of severe sepsis in children (Goldstein et al. 2005). While adults are likely to have normal to high cardiac output with low systemic vascular resistance in septic shock, children are more likely to have low cardiac output with high systemic vascular resistance (Ceneviva et al. 1998; Carcillo et al. 2002). The impact of mechanical ventilation on haemodynamics depends on the interaction of many factors (Luecke and Pelosi 2005) including lung and chest wall mechanics (O’Quin et al. 1985) and characteristics of ventricular function (whether there is right, left or bi-ventricular failure) (Bronicki and Anas 2009). Cardiopulmonary interactions are generally complex (Bronicki and Anas 2009), more so in the critically ill (Pinsky 1985, 1990, 1994, 1997, 2005, 2007; Jellinek et al. 2000), and may be affected by a wide variety of issues in sepsis.

Ventilation may alter venous return in patient with sepsis. For instance, in sepsis cardiac output may be decreased because of venodilatation and reduced venous return as well as high right atrial pressures due to positive pressure ventilation (Bronicki and Anas 2009). The relationship between ventilator pressures and right atrial pressure depends on the interaction between respiratory efforts and ventilator pressures, lung resistance and compliance, chest and abdominal wall compliance and the characteristics of the pericardium and cardiac muscle (Kingma et al. 1987). Generally lung compliance in children is relatively low, while chest wall compliance is high, with the result that ventilator pressures are poorly transmitted to the right atrium. This may however not be true in sepsis, where the chest wall may become very non-compliant as a result of oedema, inflammation and raised intra-abdominal pressure.

Ventilation may also affect both right and left ventricular outputs. Right ventricular output varies depending on pulmonary vascular tone and resistance which is highest at low and high lung volumes and lowest at functional residual capacity (Bronicki and Anas 2009; Jardin 1997; Jardin and Vieillard-Baron 2003). Thus positive pressure ventilation in sepsis may reduce pulmonary vascular resistance (and impedance) and improve right ventricular function by optimizing FiO₂, pH and lung volumes. However, over-distension of the alveoli may be associated with deterioration of right ventricular function.

Left ventricular afterload is also increased if negative intrathoracic pressures are generated during respiration (Bronicki and Anas 2009), and thus positive pressure ventilation may substantially reduce left ventricular afterload of patients in respiratory distress who have been generating very low intrathoracic pressures (as may be seen in sepsis). This may in turn improve cardiac output.

However, the left and right ventricles do not function in isolation and ventricular interactions that may determine the effect of ventilation on cardiac output (Tyberg et al. 2000; Biondi et al. 1988).

For instance, if respiratory distress and suboptimal cardiac output coexist, the respiratory muscles may consume a significant proportion of cardiac output. Mechanical ventilation may thus contribute significantly to an improvement in organ perfusion by reducing respiratory muscle activity (Aubier et al. 1981; Viires et al. 1983).

If left ventricular dysfunction is present in the septic patient, positive pressure ventilation may be associated with significant improvements in cardiac output (Bronicki and Anas 2009), while positive pressure ventilation may substantially reduce cardiac output if the patient is hypovolaemic.

In patients with acute lung injury recruitment, manoeuvres have been used in an attempt to improve respiratory function. These manoeuvres have been associated with significant decreases in cardiac output in both animal and human adult studies (Lim et al. 2004; Toth et al. 2007).

Cardiac function in severe sepsis can therefore be profoundly affected adversely or positively by positive pressure ventilation. The exact effects will depend on the specific situation and may have to be established by careful consideration of the physiological state and sometimes a process of trial and error.

The optimal ventilatory pattern to minimize damage to the lungs will depend both on the disease and on the specific lung pathology.

Although the interplay is complex, the presence of predisposing factors (such as haemorrhagic shock) may exacerbate lung injury from inadequate ventilation (Bouadma et al. 2007), while inadequate ventilation may also exacerbate the inflammatory response to shock.

The presence of infection (or inflammation) also increases the lung and global inflammatory response to ventilation with high tidal volumes and low PEEP.

Nahum et al. (1997) demonstrated in a dog model that injurious patterns of ventilation (low PEEP and high tidal volume) significantly increased the translocation of bacteria from the lung into the circulation and the inflammatory responses within the lung.

Altemeier et al. showed in rabbits that ventilation with 10 ml/kg tidal volume had a synergistic effect, increasing lung injury, with the systemic administration of LPS (Altemeier et al. 2004). Bregeon et al. using a rabbit model showed that when E. coli lipopolysaccharide was administered intravenously, ventilation with zero PEEP and 10 ml/kg of tidal volume was associated with significant deterioration in lung mechanics, hypoxaemia and histological damage. Both authors demonstrated that mechanical ventilation alone did not induce lung injury, but mechanical ventilation together with aspiration of LPS was associated with an augmented inflammatory response and the development of ARDS changes (Altemeier et al. 2005).

Similar findings have been reported by others (Dhanireddy et al. 2006; O’Mahony et al. 2006; Bem et al. 2009) using a variety of models of ventilation during sepsis.

Thus a consistent theme in much experimental work in sepsis, mechanical ventilation (particularly with tidal volumes of 10 ml/kg or more) and high FiO₂ is associated with exacerbation of inflammatory responses both in the lung and throughout the body. There is little experimental data available on children, but it would seem reasonable to use “lung protective” strategies of ventilation in the setting of sepsis. There may also be a place for cautious use of “recruitment” followed by reduced pressure ventilation to minimize lung damage (Rimensberger et al. 1999).

Another consideration is the very common presence of fever in patients with sepsis. In an experimental study of ventilator-induced lung injury in rabbits, Suzuki et al. (2004) demonstrated that lung damage using potentially injurious forms of ventilation was exacerbated by hyperthermia relative to hypothermia.