In 2015, the Pediatric Acute Lung Injury Consensus Conference (PALICC) developed a paediatric-specific definition for ARDS in children. This initiative was fuelled by the lack of a paediatric context in both the original NAECC definition of ARDS and its revised definition, the Berlin definition. By purely focusing on lung injury in adults, differences between children and adults related to risk factors, aetiologies and pathophysiology most likely have been overlooked.

The new definition of PARDS (Table 19.2) mirrors the Berlin definition in two ways, being timing of onset and exclusion of patients with cardiac failure or fluid overload that may explain the respiratory failure. Importantly, paediatric patients with causes of acute hypoxaemia that are unique to the perinatal period such as prematurity-related lung disease, perinatal lung injury (e.g. meconium aspiration syndrome, pneumonia and sepsis acquired during delivery) or other congenital abnormalities (e.g. congenital diaphragmatic hernia or alveolar capillary dysplasia) were excluded from the definition. As in adults, in order to qualify for PARDS, the symptoms of hypoxaemia and radiographic changes have to occur within 7 days of a known clinical insult. Noticeably, children with left ventricular heart dysfunction or congenital heart disease that fulfil all other PARDS criteria have PARDS if the acute hypoxaemia and new chest imaging changes cannot be explained by acute left ventricular heart failure or fluid overload. This is a huge improvement of the new PARDS definition over other pre-existing definitions.

There are two major differences between the Berlin definition and the PARDS definition; these are related to chest radiograph appearances and the metrics for oxygenation. Unlike the Berlin definition, diagnosis of PARDS requires the need for at least one consolidation on chest radiograph rather than mandating the need for bilateral consolidations. The rationale for this is the low sensitivity of chest radiography to detect (subtle) changes in consolidation, atelectasis or oedema and the poor correlation between consolidations on chest CT scanning and chest radiography [20, 66, 93] [24, 27, 37, 55]. In addition, the presence of bilateral infiltrates provided minimal predictive value for outcome in acute respiratory failure in both children and adults [76, 97, 98].

Aside from this, paediatric critical care practitioners also critiqued both definitions for including the PaO₂/FiO₂ ratio; this metric of oxygenation does not take ventilator settings or mode of ventilation (e.g. HFOV) into account. In addition, there is an increased use of pulse oximetry in children as a substitute for children without an indwelling arterial line, thereby creating an additional problem to the criteria requiring arterial blood gas measurement. Hence, the OI is incorporated in the PARDS definition. Based on a derivation set of two data sets from the USA and Australia/New Zealand, the following cut-off points were derived: OI 4–8 (mild PARDS), 8–16 (moderate PARDS) and > 16 (severe PARDS) [78]. Data from six other centres was used as a validation set, showing increased mortality with increasing PARDS severity, thereby supporting the substitution of OI for PaO₂/FiO₂. For patients without indwelling arterial lines, the oxygen saturation index (OSI) is used as an alternative provided that the SpO₂ is 97% or less [78]. The OSI uses the SpO₂ rather than the PaO₂.

ARDS can originate either from a direct pulmonary insult (i.e. pulmonary ARDS) such as a viral pneumonia or from extra-pulmonary insults such as sepsis (i.e. extra-pulmonary ARDS). The pathobiology of ARDS is a complex process, characterised by increased alveolo-capillary permeability, lung epithelial and endothelial injury and the accumulation of protein-rich oedema fluid during the acute phase of disease [146]. Surfactant homeostasis and haemostatic balance are altered. Much of the understanding of the pathobiology has come from clinical studies in adults or experimental models mimicking the adult clinical situation. Some of these underlying pathophysiological mechanisms are probably shared between children and adults, but it should also be appreciated that there are areas of significant difference including lung growth and development during (early) childhood, immune response and surfactant homeostasis. For instance, the lung parenchyma undergoes substantial structural remodelling and growth during childhood. In the neonatal period, the peripheral airspaces are subdivided by short and blunt tissue ridges into an increasing number of very shallow alveoli. It is by the age of 18 months that the lungs have developed into an adultlike structure [23]. After this period, the number of alveoli continues to grow through adolescence.

There are important differences in the immune response between infants and adults that are likely relevant [153]. In general, the immune system of a child < 1 year of age is relatively immature, including broad deficits in both innate and adaptive immunity [151]. Neonatal PMNs and monocytes function to a lesser degree compared with adults, with decreased chemotactic responses for as long as 1–2 years [54, 91, 100, 110]. Adaptive immunity is also different between young children and adults. At birth, the immune system has a strong T helper 2, anti-inflammatory predisposition with little ability of peripheral mononuclear cells to produce tumour necrosis factor alpha or other pro-inflammatory mediators compared with adults [13, 90]. In addition, at birth, T cells are biased towards regulatory T cells, thereby potentially suppressing the innate immune response [57, 157]. This anti-inflammatory predisposition may persist during early childhood [150]. Only a few studies have examined the role of immune responses in PARDS. Associations between elevated serum levels and interleukin (IL)-6 and PARDS and soluble intercellular adhesion molecule 1 with increased risk of death and/or prolonged duration of mechanical ventilation have been demonstrated in two small paediatric studies [44, 53]. Another small study showed an association of a genetic variant in the IL-1 receptor antagonist gene with ARDS in children with community-acquired pneumonia [114]. More recently, an association between markers of endothelial (angiopoetin 2) and pulmonary epithelial (soluble receptor for advanced glycation end products) injury and mortality in PARDS has been shown, especially in direct lung injury [152].

Mechanical ventilation (MV) is clearly of benefit for children with ARDS although there are many questions that require study (Fig. 19.2) [81]. Both controlled and assisted modes of ventilation are used by paediatric critical care practitioners, irrespective of the size of the breath (pressure versus volume). However, superiority of one mode over the other has not been demonstrated [122]. This also applies to the use of ventilator modes that allow the patient to trigger the machine breath. Adult data has suggested that patient–ventilator asynchrony (i.e. no synchronisation between patient’s demand and the delivered breaths) is associated with increased morbidity, but paediatric data is lacking [40]. Neurally adjusted ventilatory assist (NAVA) uses the electrical activity signal of the diaphragm to trigger inspiration. So far, paediatric experience with this alternative mode of triggering is limited. Piastra and colleagues showed that infants recovering from severe ARDS had a shorter duration of respiratory support when they were weaned with NAVA compared with pressure support ventilation [118].

MV may also exacerbate or even initiate lung injury and has therefore been identified as a predictor for poor outcome [131]. The development of this so-called ventilator-induced lung injury (VILI) has led to the concept of lung-protective ventilation (LPV). In brief, LPV encompasses the delivery of small tidal volume (Vt) to avoid end-inspiratory overdistension (i.e. volutrauma) and the application of a certain level of positive end-expiratory pressure (PEEP) to prevent repetitive opening and closure of alveoli (atelectrauma). The issue of LPV in PARDS is the subject of scientific debate. A large randomised controlled trial (RCT) in adult patients with ARDS according to the NAECC criteria showed improved patient outcome with 6 mL/kg ideal body weight (IBW) compared with 12 mL/kg/IBW [108]. However, other adult studies could not demonstrate a positive effect on mortality when 7 mL/kg IBW was compared with 10 mL/kg IBW [117]. Subsequent meta-analyses concluded that a Vt ≤ 10 mL/kg per predicted body weight was associated with higher survival [21, 22]. To date, a paediatric counterpart of the ARDS Network trial has not been performed. Remarkably however is the data coming from observational studies, showing an inverse or no relationship between Vt and mortality in children [46, 76](20) (Table 19.3). A systematic review of the available paediatric data could also not identify a specific threshold when Vt could be considered safe, irrespective of the presence of ARDS [39]. Experimental work has shown that very young animals were less susceptible to VILI compared with adult animals when subjected to injurious ventilation with supra-physiologic Vt (i.e. 30–40 mL/kg body weight) [84]. All of this challenges the role of volutrauma in PARDS. Various explanations may be proposed, including the incorrect measurement of Vt. The ventilator overestimates the Vt delivered to the patient in young children, signifying the need for correct measurement near the Y-piece of the patient circuit in children below 10–15 kg [25]. In addition, in adults, the Vt is calculated by predicted body weight for age, height and gender. In paediatric practice Vt is usually calculated by actual body weight. A paediatric counterpart of the ARDS Network trial is eagerly awaited, although such a trial seems very unlikely for various reasons including the need for a large sample size (i.e. > 500 patients) [82]. Interestingly, paediatric critical care practitioners may have already found the solution in targeting the optimal Vt in PARDS. Unlike in adult critical care, there is a large tendency to use pressure-controlled (PC) ventilation instead of volume-controlled (VC) ventilation. In contrast with the paediatric data showing an inverse relationship between tidal volume and mortality, these same studies did confirm an association between inspiratory pressures and mortality (Table 19.3). This is in line with the assumption that the patient with a more severe lung injury characterised by small residual inflatable lung volume available for alveolar ventilation and therefore gas exchange (which is also known as the baby lung concept) should receive tidal volumes below physiologic values [56]. Patient subgroup analysis from the ARDS Network trial showed that only patients with poor respiratory system compliance at study entry had a benefit in terms of survival when randomised to the 6 mL/kg study arm [41]. This observation suggests strongly that physicians tend to use lower Vt resulting in higher inspiratory airway pressures in the sicker patients (i.e. patients with lower respiratory system compliance) at the onset of ARDS and mechanical ventilation. This latter observation is in alignment with the data from the other paediatric observational study showing that patients with higher initial lung injury scores were ventilated by smaller tidal volumes and showed worse outcome [76]. This goes along with the concept of keeping lung tissue strain (the ratio between inflated volume and functional residual capacity) low in order to protect the lung [94]. Reanalysis of adult data also showed the importance of limiting the driving pressure (i.e. the ratio of Vt of compliance at a zero-flow state) in ARDS [30]. However, given the absence of paediatric data, no recommendation can be made on the upper limit of the plateau pressure that is allowable. The same applies to the level of PEEP. Levels of PEEP should be set to prevent lung unit collapse at expiration and to avoid tidal recruitment at each breath cycle (collapse–opening–recollapse injury). Although adult data indicated that higher levels of PEEP may be associated with lower mortality, such issues remain to be resolved in paediatrics. Remarkably, paediatric physicians tend to tolerate a higher level of FiO₂ instead of turning up the PEEP because they worry about haemodynamic consequences.

From a theoretical perspective, high-frequency oscillatory ventilation (HFOV) is an ideal LPV mode [83]. With HFOV, a continuous distending pressure (CDP) is generated maintaining lung volume, with superimposed small oscillations in a frequency range of 5–15 Hz allowing for gas exchange. Paediatric critical care practitioners cherish HFOV despite the lack of scientific evidence. To date, there has been only one randomised controlled trial comparing high-frequency oscillatory ventilation to conventional mechanical ventilation in 70 children with diffuse alveolar disease and/or air leak syndrome [8]. This study showed that HFOV utilising an aggressive volume recruitment strategy resulted in a significant improvement in oxygenation and a decreased requirement for supplemental oxygen at 30 days. However, 30-day mortality was not changed in this particular trial. A meta-analysis of all six paediatric and adult clinical trials demonstrated improved mortality in patients randomised to HFOV [133]. However, two large randomised studies in adults with moderate-to-severe (early) ARDS troubled the discussion on HFOV in PARDS [49, 155]. Whereas in the OSCillation for ARDS (OSCAR) trial no difference in 30-day mortality was observed, and the OSCILLation for ARDS Treated Early (OSCILLATE) trial was prematurely stopped (after the 500 patient analyses) by the steering committee because of a significantly higher in-hospital (47% vs 35%) and 60-day mortality (47% vs 38%) in the HFOV group. In the absence of paediatric trials, a post hoc analysis of data of paediatric patients enrolled in a protocolised sedation trial showed similar mortality rates but prolonged duration of MV among patients managed with HFOV, calling for a paediatric RCT to examine the role of HFOV in PARDS [12].

Weaning from the ventilator should start as early as possible, although no paediatric data is available identifying a specific threshold when to start and how to do the weaning itself. Also, at present no paediatric criteria have been validated to discriminate extubation success and failure as well as commonly used adult parameters to monitor weaning including the rapid shallow breathing index (RSBI) [109]. The same applies to the application of spontaneous breathing trials (SBT) and extubation readiness test (ERT). It is unclear what the optimal SBT in PARDS is (i.e. using either continuous positive airway pressure (CPAP), a low level of pressure support ventilation or T-piece ventilation).

Non-invasive positive pressure ventilation (NPPV) is increasingly being used to treat respiratory failure from a variety of aetiologies in children, including paediatric ARDS [124]. Despite this growing, there is much uncertainty about the indications, strategies and effects on patient outcome for the use of NPPV, especially in paediatric ARDS [47, 128]. There are several paediatric prospective and retrospective cohort studies describing the use of NPPV for acute respiratory failure, but most include a heterogeneous population that includes mild-to-severe ARDS. Thus, it remains the subject of debate if NPPV actually prevents the need for endotracheal intubation. Observational studies indicated that the median intubation rate for children with severe PARDS is 57% [47]. In only one randomised controlled trial of 50 children with acute hypoxaemia respiratory failure, the intubation rate was significantly lower (28% vs 60%, p = 0.045) when NPPV was compared with standard care. However, this particular study included also patients with mild PARDS.

The pathophysiologic and clinical features have triggered many investigators to study numerous pharmacological approaches to ARDS in critically ill adults [127]. Few of these approaches have been explored in critically ill children. As a consequence, much of the routine treatment for PARDS is based on data from adults or paediatric anecdotal experiences (Table 19.4). Despite this lack of scientific evidence, many pulmonary-specific therapies including inhaled nitric oxide (iNO), surfactant or steroids are used in daily practice [124, 125].