A review of feeding intolerance in critically ill children

Enteral feeding is delivered either through the gastric route (into the stomach) or into the intestine in the critically ill child. During critical illness, the child (aged 0–17 years) has to adapt and respond to both the disease process causing the critical illness (and does this in a physiological age-dependant manner) and the medications and therapies applied to the child in the intensive care unit. Gastric dysmotility, a problem in the critically ill child, is defined by Martinez et al. as “the functional capacity of the stomach to move the contents forward by abnormally slow and/or uncoordinated activity of the gastric or antroduodenal musculature” [20]. This results in the phenomena of delayed gastric emptying which is estimated to occur in around half of critically ill children [21]. The causes for delayed gastric emptying are multifactorial; alterations in hormonal responses (increased levels of ghrelin, motilin, cholecystokinin, glucagon-like-peptide-1, peptide-YY, amylin) and alterations of vagal tonus, vasoactive intestinal peptide, and nitric oxide secretion are combined with the side effects of opiates, sedatives, neuromuscular blocking agents, and catecholamines alter GI motility and slow transit time, usually in a dose-dependent manner [20]. As a result, stomach motor discoordination, antroduodenal discoordination, and gut contraction/relaxation discoordination will occur. An altered gastrointestinal motility, which is multifactorial, is the primary mechanism underlying many of the signs and symptoms of feeding intolerance in the critically ill children such as large gastric residual volumes, constipation, and abdominal distention [18]. Other mechanisms include gut tissue alteration, due to gut inflammation or hypoperfusion, which are responsible for malabsorption and bowel movement alterations. Furthermore, pressure changes induced by positive pressure ventilation may affect the renin-angiotensin system and consequently reduce splanchnic perfusion. Many children admitted to the PICU also have a degree of circulatory shock, and this has a direct effect on reducing gut perfusion as may direct compression, either from trauma or from abdominal compartment syndrome. The critically ill child is also lying (usually supine) and immobile, a non-physiological state. Administering enteral nutrition may then place additional oxygen demands on the gut, which can result in ischemia with negative consequences [18, 20].

Surveys of PICU healthcare professionals (nurses, physicians, and dieticians) reveal a number of commonly used signs and symptoms to define feeding intolerance [16, 40, 43, 44]. These include gastric residual volumes (GRV), vomiting, diarrhea, abdominal distention and pain/discomfort, and the use of serum lactate [41, 46]. However, evidence to support each of these as an indicator of feeding intolerance is weak or absent, and in critically ill children, a multitude of non-feed-related factors can produce these signs and symptoms. Nutrition research in critically ill children also frequently uses “feeding intolerance” as an outcome measure and a plethora of different definitions has been used to define this [5, 7, 9, 10, 19‐22, 28, 33‐35, 37, 45, 48, 50] [Table 1]. In pediatrics, these include the use of gastric residual volume alone, possibly including a change in the aspirate colour or in combination with abdominal signs and symptoms. Most [87%, 13/15] studies used GRV either as the main parameter to determine feeding intolerance or in combination with other parameters in 54% (8/15) (Table 1). The GRV critical threshold used varied markedly from a threshold of 2 ml/kg body weight to more than 50% of the volume of the feed given over the previous 4 h [Table 1]. Two studies defined feeding intolerance as the number of stools per 24 h period and/or the presence of vomiting or abdominal pain and/or tenderness [10, 19]. Similarly, an adult systematic review of 72 intensive care studies [2] found that 43 different definitions of feeding intolerance were used, with GRV used in 88% of these alone or in combination with other signs.

First, signs of feeding tolerance/intolerance (i.e., feeding intolerance markers) have to be distinguished from signs or conditions that imply that the patient is not ready to be fed (i.e., feeding intolerance predictors). Feeding intolerance predictors, encountered prior to feed infusion, are a presumption that feeds will not be tolerated. They relate to various anatomical and pathophysiological conditions that need to be treated prior to feeding initiation. Conversely, feeding intolerance markers are signs/symptoms observed during EN administration, presumed to be indicative of feeding intolerance. Healthcare professionals then have to establish their causality, distinguishing feeding issues and non-feed-related causes, in order to manage them appropriately to ensure future feeding tolerance. Markers and predictors include increased gastric residual volume, upper gastrointestinal (GI) signs and symptoms, bowel sounds and frequency of bowel movements, and lastly abdominal pain and/or distention. Each of these will now be discussed.

Gastric residual volume used as a surrogate marker of gastric emptying is used extensively in a majority of intensive care units (neonatal, pediatric, and adult) [6, 13, 40, 44, 46]. However, the evidence for it as a valid marker of gastric emptying is not supported [21]. An increased GRV can occur in up to 50% critically ill children [21, 46]. GRV comprises both the enteral feed administered, in addition to gastric secretions produced, and many clinicians fail to consider the large impact of the latter on this volume. Furthermore, there is an inaccurate assumption that the relationship between feed volume infused and GRV is linear [46]. However, a prospective study in critically ill children showed higher GRVs did not correlate with delayed gastric emptying [21]. The volume of GRV obtained is known to be affected by multiple factors: patient position, feeding tube tip position and diameter of feeding tube being aspirated, the type of tube (rigid aspiration tube or soft silicone or polyurethane nutrition tube), the syringe size used to aspirate, the nurses’ technique in aspiration, and the type of feeding delivery method used (bolus versus continuous) [1, 46]. If aspirate is obtained, there is a belief that this volume is entirely accurate and decisions are made based on this volume, with no evidence to underpin any GRV threshold level to indicate “intolerance.” Clinicians’ fears around increased GRV relate to the perceived risk of vomiting and pulmonary aspiration with increased GRV, but this is not supported in children or adults [27, 30, 42]. For children, an arbitrary cutoff volume is not useful given the wide weight and age variations. If a GRV threshold had to be set to define intolerance, this should rather be set in milliliter per kilogram or milliliter per body surface area but should also account for volumes administered. However, no research has ever shown that a critical amount is significant, or predicts a greater risk of aspiration.

Vomiting in critically ill children can occur in up to 31% patients [18] and is perceived to be dangerous due to its risk of aspiration and pneumonia. When the vomit contains the administered feed, it is considered a feeding intolerance sign. However, the cause of vomiting may not be related to enteral feeding. Indeed, the amount or type of feed is not necessarily responsible for any upper GI signs and symptoms, and as a consequence, treating the cause may be more appropriate than withholding feeds. Gastroesophageal reflux and vomiting may reflect bowel obstruction or paralytic ileus, but can also be induced by a number of other factors. Vomiting in the critically ill child may also be caused by the wakening child and irritation of the endotracheal tube, by oral and endotracheal suctioning and coughing with secretions as well as by incorrect gastric tube position or inadequate gastric decompression. In infants, gastroesophageal reflux is common due to the immaturity of the lower esophageal sphincter. Furthermore, a number of common PICU medications (opiates, sedatives, catecholamines) or less common (chemotherapeutic agents) can cause nausea and vomiting. Iatrogenic withdrawal syndrome, which is also frequent in the pediatric setting, can also result in both vomiting and diarrhea [11]. Supine positioning may also increase the likelihood of regurgitation and vomiting in the critically ill population. Vomiting, can, however be prevented using antiemetic medications, adapting feed delivery (continuous feeding or post pyloric feeding) before withholding feeds.

The presence and frequency of bowel sounds may be a helpful indicator of bowel function and may indicate paralytic ileus or bowel obstruction. However, these can be difficult to assess with certain therapies (such as high-frequency oscillatory ventilation). Additionally, the correlation between bowel sounds and delayed gastric emptying in the ICU is poor [38]. In the past decade, early rehabilitation after surgery concepts in children has become well established; these include early post-operative feeding [31, 32] and they found return of bowel sounds was a poor marker of gut motility. Recommended practice is therefore that bowel sounds should not be used to decide whether to initiate or withhold feeding. There is a clear relationship between enteral nutrition (EN) and bowel movements in the critically ill. Patients require some enteral input to generate any bowel movements. If this is delayed or only delivered in small amounts, stool output will consequently be affected. Surprisingly, constipation, which is frequent in critically ill children, is less commonly considered as a sign of feeding intolerance in PICU. A study of 150 children [17] found constipation occurred in 47% of children who spent more than 3 days in the PICU. A number of factors influenced the development of constipation (defined as 3 days without a bowel movement) including immobility, opiates and other commonly used PICU drugs, delayed enteral feed administration, splanchnic hypoperfusion, inflammation, sepsis, and electrolyte disturbances (most notably hypercalcemia, hypokalemia, hypomagnesemia, and hypophosphatemia). They found a significantly higher incidence in surgical patients and in older higher weight children, in addition to those who received little enteral nutrition which was started later [17]. Diarrhea (still inadequately defined in neonates and small infants) can also occur in critically ill children, although the reported incidence is lower than that reported in critically ill adults of 15–38% [36, 47]. Diarrhea is commonly defined in adults as more than three to five liquid stools per day [36], but in infants, it is normal to pass a semi-liquid stool after each feed, making the definition more difficult. The Bristol stool chart is validated in adults and older children [14], but no validated tool is available in infants and toddlers. Diarrhea can be caused by various non-feed-related factors such as infections (Clostridium difficile) or other microbiome-acquired imbalance due to gut failure, antibiotic use, or fasting/feeding regimen changes and shock states and hypoalbunemia [36, 47]. In addition, formula type and the osmolality of feed administered and delivery method (continuous versus bolus) can also produce diarrhea and impact on this [36]. Finally, the first stool in critically ill children is loose most of the time [36], with no link to type or amounts of feeds. This may be more likely to be related to gut inflammation, constipation, and gut microbiome alteration. In other words, both constipation and diarrhea are multifactorial and commonly non-feed related, in the PICU setting.

Abdominal distention is another subjective sign that may indicate feed intolerance [3] resulting from altered gut motility, but its specificity and sensitivity are once again limited. It can be mimicked or concealed by generalized edema, ascites, a full bladder, peritoneal dialysis, and swallowing air with hand ventilation or non-invasive ventilation, if there is inadequate gastric decompression. Nevertheless, abdominal girth can be objectively measured and assessed [39], although there is no validated threshold of distention that is significant (in absolute value or percentage). Like many other PICU assessment parameters, the absolute value (abdominal girth if measured in a consistent way and place) can be used as a trend over time. Intra-abdominal pressure, estimated by urinary bladder pressure monitoring through the urinary catheter [39], is not a sign of feeding intolerance but rather a sign of “risk to feed” if elevated. Indeed, high intra-abdominal pressure is the consequence of severe ascites or intra-abdominal compression. Abdominal pain or discomfort is not only very subjective, but in many critically ill children difficult to assess, and caused by a number of factors including trauma, surgery, wounds, trapped gas, constipation, gut dysmotility, medications, lines, and surgical drains, with no link to feeding tolerance potentially. In children who can self-report, it is an important sign, but it is problematic for many critically ill children.

Reduced splanchnic perfusion can alter the tolerance to EN and has been reported to induce both functional and structural changes in the GI tract [20]. EN increases oxygen and metabolic demands of the gut, which may lead to oxygen supply and delivery imbalances if the gut is underperfused. Vasoactive drugs may worsen this inadequate intestinal perfusion further. Serum lactate is a commonly used factor to determine inadequate tissue perfusion [43], but other methods of tissue perfusion assessment, such as splanchnic near-infrared spectroscopy, are possible, though used rarely [12, 43]. However, due to fears around suboptimal tissue perfusion, serum lactate in particular is frequently used as an indicator to delay or withhold enteral feeding. Although increased lactate indicates inadequate tissue perfusion, using it as a maker to stop or withhold feeds is problematic and subjective, as variable lactate thresholds are used [43]. In the PICU context, lactate is likely to be used as a feeding intolerance predictor, rather than a feeding intolerance sign. No research has examined the use of splanchnic NIRS in relation to enteral feeding tolerance in critically ill children, but one study in infants after congenital heart surgery has shown splanchnic NIRS to be a valid and easy non-invasive monitoring tool to assess splanchnic oxygen demand and supply [12].

Table 2 summarizes the signs and symptoms used to assess feeding intolerance.

In essence, the complexity of critically ill children means that numerous factors can produce these, often very subjective, signs and symptoms. Healthcare staff often interpret these as feeding intolerance when they are not because of their insufficient specificity and sensitivity. Improved awareness by healthcare professionals of the limitations of these signs and symptoms is needed, as was called for in the first review paper of feeding intolerance in children in 2004 [46]. Feeding intolerance markers also need to be clearly distinguished from feeding intolerance predictors. The healthcare professional, firstly a nurse at the bedside, has to try to consider these factors and assess feeding tolerance, often seeking guidance from physicians, dieticians, and/or a unit protocol. However, identifying the causes of the sign/symptoms is a complex process. Often, a fear of the perceived risk and severity of feeding intolerance (such as vomiting and pulmonary aspiration) causes enteral nutrition cessation as a first action. However, identification and treatment of the cause of signs/symptoms would be better in the PICU setting, to avoid compromising nutrition goals. Most units utilize one or a combination of the simplistic markers listed above as a threshold for delaying, withholding, or initiating EN or additional therapies such as prokinetics. A more holistic approach should supplement local guidelines, but this requires better training to improve the knowledge of healthcare professionals.

Common action taken to reduced feed intolerance in critically ill children include changing feed delivery method from intermittent bolus to continuous feeding or from gastric to transpyloric feeding, changing feed formulation (from polymeric to semi-elemental), or administering prokinetic agents. However, few of these have been comparatively studied. The two studies to examine continuous versus intermittent feeding [7, 10] found neither to be superior. Similarly, the two trials [24, 37] comparing gastric with transpyloric feeding found no difference, but the outcome used was ventilator-associated pneumonia, not feed tolerance. There are no studies on optimal feed formulation and feed tolerance/intolerance. Only one study on intravenous erythromycin [8] (used at a prokinetic dosage to facilitate transpyloric tube placement, not to improve feed tolerance) found the drug safe, but did not examine its impact on tolerance. Therefore, there is no evidence to suggest which, if any, interventions are effective in improving feed intolerance.

There is little or no development or progress in defining feeding intolerance made since a review paper on feeding intolerance in children was published in 2004 [46]. The international PICU community now urgently needs to agree a pragmatic consensus definition for feeding intolerance in the PICU, taking into account all the items described previously (feed related and non-feed related). Given the paucity of evidence around the components currently used to define feeding intolerance, this will not be evidence based, but would at least promote some consistency, both for research and for practice. This could then be used in future research, where feeding intolerance is used as an outcome.