Nutritional Management of the Critically Ill Neonate: A Position Paper of the ESPGHAN Committee on Nutrition

Abstract

Objectives: 

The nutritional management of critically ill term neonates and preterm infants varies widely, and controversies exist in regard to when to initiate nutrition, mode of feeding, energy requirements, and composition of enteral and parenteral feeds. Recommendations for nutritional support in critical illness are needed.

Methods: 

The ESPGHAN Committee on Nutrition (ESPGHAN-CoN) conducted a systematic literature search on nutritional support in critically ill neonates, including studies on basic metabolism. The Medline database and the Cochrane Library were used in the search for relevant publications. The quality of evidence was reviewed and discussed before voting on recommendations, and a consensus of 90% or more was required for the final approval. Important research gaps were also identified.

Results: 

This position paper provides clinical recommendations on nutritional support during different phases of critical illness in preterm and term neonates based on available literature and expert opinion.

Conclusion: 

Basic research along with adequately powered trials are urgently needed to resolve key uncertainties on metabolism and nutrient requirements in this heterogeneous patient population.

See “Parenteral Nutrition for Critically Ill Term and Preterm Neonates: A Commentary on the 2021 European Society for Paediatric Gastroenterology, Hepatology and Nutrition Position Paper” by van Goudoever and van den Akker on page 137.

What Is Known/What Is New

What Is Known

• Cumulative energy and protein deficits are common in preterm and term infants treated for critical illness.
• Critical illness induces an acute stress response that correlates with the duration and the severity of the injury insult and alters carbohydrate, protein and fat metabolism.
• Nutritional management in neonatal intensive care units varies widely and there is a lack of consensus in regard to optimal timing of initiating and advancement of nutritional support.

What Is New?

• The European Society for Paediatric Gastroenterology, Hepatology and Nutrition Committee on Nutrition provides a comprehensive review of the literature on nutritional support in critically ill preterm infants and term neonates.
• Evidence-based recommendation for nutritional support during the different phases of illness are provided and urgent research gaps are highlighted.

The nutritional management of critically ill neonates varies widely, and controversies exist in regard to when to initiate nutrition, mode of feeding, energy requirements, and composition of enteral and parenteral feeds. Recommendations for nutritional support in critical illness are needed.

Critical illness may be defined as any life-threatening condition induced by sepsis, major surgery, or other insults associated with tissue injuries, such as severe trauma, hypoxia-ischemia, severe cardiorespiratory compromise, or any other acute illness requiring intensive care. Although mechanisms of injury vary, there is typically an acute metabolic stress response, characterized by the activation of several cytokines [eg, interleukin (IL)-1, IL-6, IL-8, IL-10, tumor necrosis factor (TNF)-α], lipid mediators, such as lipoxins, resolvins, protectins, and maresins, and neuroendocrine hormones including catecholamines, glucagon, insulin-like growth factor 1 (IGF₁) and vasopressin ^(1,2). The aim of this metabolic response is to repair tissue damage, regain tissue homeostasis and resolve inflammation. It involves mechanisms to maintain plasma volume, increase cardiac output and mobilize energy reserves ^(1–3). A persistent or disrupted host response may result in exacerbated tissue catabolism (proteolysis and lipolysis), dysregulated glucose metabolism, impaired immune function, organ failure, and death.

Multiorgan dysfunction and mortality rates are higher in neonates compared to older infants, children and adolescents ⁽⁴⁾ and highlight the need for early identification of critical illness to improve outcomes (Table 1) ⁽⁵⁾. Early diagnosis is challenging especially in preterm infants who may already need complex medical support for longer-term conditions (eg, chronic lung disease) without being acutely unwell. Different scoring systems, that mostly include a combination of physiologic variables and biomarkers, have been developed to help identify infants or neonates with high illness severity and mortality risk, but few of these have been tested in prospective trials ^(6–14).

TABLE 1 - Common clinical variables and biomarkers used for the diagnosis and assessment of critical illness

Organ specific physiologic variables and biomarkers used for the diagnosis and assessment of critical illness
Typical underlying conditions for critical illness in preterm and term neonates	Organ system	Physiologic variables	Biomarkers
• Congenital heart defects with hemodynamic instability• Preterm infants with high illness severity score (ie, CRIB∗ II-score > 11, SNAP†-II > 30)• Major abdominal surgery• Necrotizing enterocolitis (NEC)• Sepsis• Severe asphyxia (hypoxic ischemic encephalopathy)• Severe respiratory distress syndrome (RDS)	Cardiovascular	Skin color, capillary refill timeHear rateArterial blood pressureCentral venous pressurePulmonary arterial pressure	Lactate
	Endocrine	Fatigue, muscle weakness, fluid retention, polyuria, and more	Glucose, glucose variability, vasopressin, osmolarity (serum/urine)
	Hematological	Petechiae and ecchymosis, disseminated intravascular coagulopathy	Platelet and leucocyte count, fibrinogen
	Hepatic	Liver size and consistency	Aspartate transaminase (ASAT), bilirubin, C-reactive protein (CRP), procalcitonin, prothrombin time or INR, serum(s): protein, transthyretin, s-urea
	Neurological	Seizures, pupillary reaction, Glasgow Coma Scale, core temperature >38.5 or <36°C, temperature variability
	Renal	Urine output	s-creatinine, s-urea, s-protein
	Respiratory	Hypocapnia, hypoxemia, mechanical ventilation, respiratory rate	pH, paCO2, PaO2/FiO2 ratio
Age-specific SIRS criteria for term neonates (5)
	Newborn (0−7 days)		Neonate (8−28 days)
Heart rate (beats/min)	<100 or >180		<100 or >180
Respiratory rate (breaths/min)	>50		>40
Systolic blood pressure (mmHg)	<59		<79
Leucocyte count (103/mm3)	<34		>19.5 or <5

The systemic inflammatory response syndrome (SIRS) refers to any combination of temperature variability (hypothermia, fever), tachycardia, tachypnea, and change in leucocyte count and is referred to as sepsis when it is triggered by an infection. PaCO₂= partial pressure of carbon dioxide; PaO₂= partial pressure of oxygen; FiO₂= fraction of inspired oxygen.

^∗CRIB = clinical risk index for babies.

^†SNAP = score for neonatal acute physiology.

The acute metabolic response in critical illness was initially categorized by Cuthbertson into the “ebb” and “flow” phases ^(1,15) to describe the immediate period of depressed metabolism and the subsequent more prolonged period of increased metabolic activity, characterized by catabolism. The release of cytokines and neuroendocrine hormones during the early phase of critical illness (6–24 hours) tends to be proportionate to illness severity and duration and counteracts the anabolic effects of insulin and IGF-1 ^(1,16–22). In this phase, degradation of stored glycogen is activated to cover energy needs (glycogenolysis) ⁽²³⁾. Concurrently, protein and fat are mobilized; protein to provide specific amino acid precursors for gluconeogenesis and the synthesis of acute-phase reactants (such as C-reactive protein [CRP] and fibrinogen) in the liver, and fat for the provision of free fatty acids and glycerol. Activated hepatic glucose production and reduced uptake of glucose by skeletal muscle increase the risk of hyperglycemia ^(1,2,23). The mobilization of energy through muscle and fat catabolism normally persists for several days, resulting in a temporary cessation of growth ^(2,15,24,25). Hence, during critical illness, it is prudent to adapt nutritional care according to the different phases of the metabolic stress response and to recognize the hepatic transition to anabolic protein metabolism when growth reoccurs. This would help avoid inappropriate nutrition, especially overfeeding during the early (catabolic) phase of illness and underfeeding during recovery. Consequently, it may help to consider nutritional support in three different phases ^(25–28); early acute phase, late acute phase, and the recovery phase (Fig. 1) ^(25,29,30). Precisely defining these phases in clinical practice may be challenging, but the early acute phase usually ends when clinical symptoms stabilize and acute cardio-respiratory support can be reduced, whereas in the recovery phase, intensive cardiorespiratory care is typically no longer required.

FIGURE 1: Simplified overview of different phases of critical illness. Note that the timing durations may be extremely variable.

Ideally, the nutritional care of critically ill neonates should be based on true measurements of energy expenditure (EE) to avoid the complications associated with under- and overfeeding. The gold standard to measure EE in healthy individuals is the stable isotope technique with doubly labeled water ^(31,32); however, when estimating energy requirements in clinical practice it is more common to use predictive equations or indirect calorimetry (based on mathematical algorithms and measurements of oxygen consumption and carbon dioxide production), because they are practical and provide results bedside ^(33–35). Nevertheless, predictive equations are often inaccurate in critical illness, especially in preterm infants, and indirect calorimetry is not easily and routinely used in most neonatal intensive care units (NICUs) ^(36–39).

A large number of observational studies ^(40–51) and surveys ^(38,52–56) on nutritional practices in preterm and term infants treated for critical illness show that prescribed and actual delivered intakes are generally below recommendations, and that cumulative protein- and energy deficits may account for as much as 40–45% of the variation in weight for age z scores during hospitalization ^(40,49). Adequate nutrient supply is complicated by carbohydrate and lipid intolerance that frequently results in hyperglycemia or lipemia, and this is often exacerbated by fluid restriction, concomitant medical infusions, interruption or intolerance to enteral feeding or lack of adequate access for enteral and parenteral nutritional support ^{(50,51,57,58)}. More proactive nutritional approaches during the first week of life in extreme preterm infants and during the acute phase of critical illness overall have been instigated ^(59–65), and some observational studies show positive associations between improved nutrient intakes and certain short-term clinical outcomes ^(46–49,66); however, this may be explained by reversed causality and the effect of confounders, and several randomized trials have failed to show long-term clinical benefits from early, enhanced nutritional support ^(67,68). On the contrary, based on studies in adults, it has been suggested that active nutritional support during the acute phase of illness may be harmful and that permissive underfeeding or even starvation improves outcomes, possibly by avoiding the induction of “nutri-traumas” such as hyperglycemia, suppressed autophagy, mitochondrial dysfunction, and refeeding syndrome ^(69,70).

A large randomized trial, the Early versus Late Parenteral Nutrition in the Pediatric Intensive Care Unit (PEPaNIC) trial, showed that withholding parenteral nutrition (PN) during the first week of acute illness improved early outcomes as compared to PN initiated during the first 24 hours after admission in children ⁽⁷¹⁾. Effects were similar in the subgroup of 209 term-born neonates recruited to the trial ⁽⁷²⁾. Despite this finding, many clinicians appear reluctant to limit early nutritional support due to (1) concerns about possible harm by not providing adequate nutrients during the first week of critical illness, particularly in neonates and undernourished children ⁽⁷³⁾, and (2) the belief that exogenous dietary protein provision is essential during critical illness ⁽⁷³⁾.

The results of the PEPaNIC trial require confirmation in other settings, especially by conducting studies focusing on nutritional support in critically ill preterm and term neonates, before firm recommendations for nutritional practice can be made.

SCOPE OF THE MANUSCRIPT AND DEFINITION OF THE TARGET PATIENT POPULATIONS

The scope of this position paper was to conduct a search of published literature on nutritional support during the first week of illness in critically ill neonates to provide practical recommendations for nutrient supply. By considering how critical illness affects energy needs and the metabolic utilization of carbohydrates, protein and lipids, and by evaluating the evidence base for nutritional support in critically ill preterm infants and term neonates, we aimed to answer four key questions related to the topic:

• 1. How does critical illness affect energy needs and the metabolic utilization of carbohydrates, protein and fat?
• 2. What is the evidence for the mode of feeding (enteral/parenteral)?
• 3. What is the evidence for the timing of initiation and the advancement of nutritional support?
• 4. What recommendations can be made based on the current body of evidence?

In cases of limited evidence, we considered data from other paediatric or adult trials to see if the results could be extrapolated to critically ill neonates. Important research gaps were highlighted.

The target patient population for the nutritional management recommendations for this position paper includes (1) preterm infants and term neonates (<4 weeks of age) requiring major surgical care, for example, congenital heart disorders, gastrointestinal malformations, necrotizing enterocolitis (NEC) or spontaneous intestinal perforation (SIP), (2) term neonates undergoing therapeutic hypothermia due to hypoxic-ischemic encephalopathy, and (3) preterm infants and term neonates with critical medical illnesses (eg, sepsis and multiorgan failure). Since neonatal illness severity scores predict time-dependent mortality and short-term morbidities better than birth weight and gestational age (GA) in very preterm infants (birth weight < 1500 g or GA < 32 weeks) ^(7,13,14), we further define preterm infants with high illness severity scores as being critically ill, for example, clinical risk index for babies (CRIB II) >11 points or a score for neonatal acute physiology II (SNAP-II) >30.

In the context of this paper, nutritional support is regarded as the provision of either enteral nutrition (EN) through a gastric tube or PN and does not specifically address the amount and delivery of intravenous fluids or glucose infusions. This is keeping with the recently published guidelines for nutritional support in the pediatric critically ill patient ⁽⁷⁴⁾.

Of note, this position paper does not discuss nutritional support in neonates with chronic inflammatory diseases or conditions, such as bronchopulmonary dysplasia (BPD) or patent ductus arteriosus (PDA).

METHODS

For this systematic literature review, the Medline database and the Cochrane Library were searched for relevant publications in English up to July 2020. Due to the limited number of randomized controlled trials, we also included cohort studies, case studies and surveys addressing the topic. Medline search terms included “critical illness” [MeSH Terms], “intensive care unit/or intensive care units,” “pediatric/or intensive care units,” neonatal, exp critical care, OR “parenteral nutrition, total” [Mesh:noexp], “parenteral nutrition,” OR “enteral nutrition” OR “amino acids/or fat emulsions, intravenous.” These terms were combined with MeSH terms and the key words “therapeutic hypothermia,” “hypoxic-ischemic encephalopathy,” and “resting energy expenditure.” The searches were restricted to the “newborn infant (birth to 1 month),” but included “premature infant” or “premature infants” or “very low birth weight” or “very low birth weight infants” or preterm or “preterm infants” or “preterm infant”. The primary search retrieved 1113 publications. After screening titles and/or abstracts 115 publications were evaluated and additional studies from the reference list of relevant publications were selected to complete the search. In total, 152 studies were included in this position paper to help provide evidence-based answers about nutritional management of the critically ill neonate. Based on available data, recommendations were proposed and discussed before anonymous online voting. Any recommendation that did not reach a consensus of 90% or more was rephrased until all authors agreed for publication.

ENERGY NEEDS IN HEALTHY AND CRITICALLY ILL NEONATES

Energy Needs in Healthy Preterm and Term Neonates

The main objective of feeding is to enable infants to fulfill their genetic potential for growth to optimize lifelong health and wellbeing ^(75,76). Energy requirements should cover (1) the energy needed to maintain basal metabolic functions (basal metabolic rate; BMR), (2) energy expended for physical activity, which is usually minimal, (3) diet-induced thermogenesis, which includes postprandial nutrient metabolism and the energy required for synthesis and organization of new tissue, (4) energy stored in new tissue (tissue growth), and (5) energy lost in stool and urine ^(27,77). Because measurement of the basal metabolic rate requires a 12–18-hour fast, a thermoneutral environment, and the patient being asleep, basal energy requirements are usually estimated by measuring resting energy expenditure (REE), that is, the energy expended by a person at rest, and includes thermoregulation and resting muscular activity ⁽⁷⁸⁾. In neonates, measured REE (or resting metabolic rate; RMR) also includes diet-induced thermogenesis as they are never strictly in a fasting state (Fig. 2A and B).

FIGURE 2: Energy balance in healthy and critically ill neonates. BMR = basal metabolic rate; REE = resting energy expenditure.

Studies of REE during the first few weeks of life in healthy preterm and term infants show that REE increases with increasing energy supply during the first weeks of life and that REE is directly proportional to the growth rate ^(79,80). Measured values for REE in healthy preterm infants are around 35–55 kcal · kg⁻¹ · day⁻¹ during the first 2 weeks of life when nutrient intakes are low, and increase to about 70 kcal · kg⁻¹ · day⁻¹ at 1 month of age ^(80–82). In healthy term neonates, REE increases from 45 to 50 kcal · kg⁻¹ · day⁻¹ to about 60 kcal · kg⁻¹ · day⁻¹ during the same time interval ^(79,80). Since the average energy requirements for growth are 3–5 kcal · g⁻¹ · day⁻¹⁽⁷⁷⁾, depending on the ratio of fat and protein deposition, very premature infants require an additional 50–70 kcal · kg⁻¹ · day⁻¹ of metabolizable energy to approximate intrauterine growth rates ^(77,83), whereas the energy needed for growth is around 30–40 kcal · kg⁻¹ · day⁻¹ in term neonates. In older children and adults, the energy fraction needed for growth in relation to the total EE is negligible ^(19,84). This means that the change in energy needs during critical illness is much larger in preterm and term neonates, putting them at extra risk of both over- and underfeeding.

Energy Needs in Critically Ill Neonates

Our literature search identified several small studies of EE during critical illness in different neonatal settings ^(85–93), but we did not find any studies of EE in neonates with HIE and/or receiving therapeutic hypothermia. Most studies reported REE during the initial acute phase ^{(85,86,89–93)}, but few studies included exclusively preterm infants or neonates <28 days of age. Studies in neonates after uncomplicated surgery show a significant brief increase in REE ⁽⁸⁵⁾, followed by REE in the range of the BMR (40–50 kcal · kg⁻¹ · day⁻¹) for 4–7 days postoperatively ^(85,89). Major surgery, surgery associated with other inflammatory insults, and sepsis provoke a greater metabolic challenge with a higher average postoperative peak in REE (50–60 kcal · kg⁻¹ · day⁻¹) and a slower resolution of the injury response ^(86,90,91). Data indicate that the metabolic stress response after major surgery is shorter in neonates and children than in adults ⁽⁹⁴⁾, and that premature infants (mean GA 29 ± 2.9 weeks, n = 18) exhibit an earlier anabolic recovery after acute illness compared to infants born nearer to term (GA 38.2 ± 1.8 weeks, n = 55) ⁽⁹⁵⁾. Figure 2B and C illustrates the difference in EE between healthy growing and critically ill neonates.

The respiratory quotient (RQ) obtained from indirect calorimetry has been proposed as a tool to adapt nutritional care during critical illness in neonates and children because the ratio of CO₂ production to oxygen consumption is dependent on the relative contributions of carbohydrate, protein and fat to the EE ^{(19,78,96,97)}: The RQ for carbohydrate is 1.0, for protein ∼0.85, for fat ∼0.7, and for the conversion of carbohydrate to fat (de novo lipogenesis) >1.0. In other words, overfeeding or high glucose intakes resulting in high lipogenic activity is energy demanding and increases RQ, whereas the use of endogenous fat stores to meet energy requirements decreases RQ. An observational study in 98 neonates and children showed that despite low sensitivity, RQ values >0.85 excluded underfeeding and RQ-values >1.0 identified overfeeding or excess carbohydrate intakes ⁽⁹⁷⁾. Targeting 64–70 kcal · kg⁻¹ · day⁻¹ provided as 2.5 g · kg⁻¹ · day⁻¹ of protein, 1–2 g · kg⁻¹ · day⁻¹ of fat and 10 g · kg⁻¹ · day⁻¹ of carbohydrate in the early postoperative phase resulted in an RQ > 1, suggestive of some overfeeding in an observational study of 10 surgical neonates ⁽¹⁹⁾. Along with the resolution of the early acute phase, RQ values dropped and caloric support could be increased from the 4^th day onwards.

The finding of a change in energy needs during the first week of critical illness is relatively consistent with EE studies in critically ill children, which show that REE values gradually increase with the increasing length of ICU stay (<4, 4–7, and >7 days) ⁽⁹⁸⁾, but also that EE is affected by other factors such as weight, temperature, heart rate, diastolic blood pressure, minute ventilation, and drugs ^(98,99).

MACRONUTRIENT UTILIZATION IN HEALTHY AND CRITICALLY ILL NEONATES

Carbohydrate Utilization

Carbohydrate and fat are the main macronutrients for provision of energy ^(27,75,100). Metabolic studies on parenteral macronutrient utilization have shown that glucose is the primary determinant for both glucose utilization and the metabolism of fat ^(101,102). Minimal and maximal glucose recommendations are normally based on the endogenous glucose production rate (GPR) and the glucose oxidation capacity. The estimated GPRs and glucose oxidation rates in preterm infants are ∼6 mg · kg⁻¹ · min⁻¹ (8.6 g · kg⁻¹ · day⁻¹) and ∼8 mg · kg⁻¹ · min⁻¹ (11.5 g · kg⁻¹ · day⁻¹), respectively, compared to 5 mg · kg⁻¹ · min⁻¹ (7.2 g · kg⁻¹ · day⁻¹) and 12.5 mg · kg⁻¹ · min⁻¹ (18 g · kg⁻¹ · day⁻¹) in term infants ^(26,101,103). If glucose supply exceeds the maximum glucose oxidation rate, glucose is converted to fat and fat oxidation ceases ^{(101,102,104)}; however, if the carbohydrate-to-fat ratio is reduced, the opposite occurs and fat utilization increases ^{(101,102,104)}. Nevertheless, these rates are only estimates and both higher and lower glucose intakes may sometimes be indicated and provided as long as the infants remain euglycemic ^{(101,105,106)}.

The utilization of exogenous glucose supply is however influenced by other factors including stress-induced insulin resistance ⁽²⁾, persistent endogenous gluconeogenesis and electrolyte disturbances (eg, hypophosphatemia) ^(26,107), all of which often occur during critical illness and are commonly encountered in extremely preterm infants ^(108–112). Observational studies show that 50–80% of critically ill children and extremely preterm infants experience hyperglycemia (>8.3 mmol/L) ^(23,113,114). Stress-induced insulin resistance may develop very rapidly as a response to the increased levels of stress hormones (ie, adrenaline, cortisol, glucagon, growth hormone) and inflammatory cytokines (TNF-α, IL-1, IL-6, IL-8, and more), resulting in (1) reduced uptake of glucose by the cells and (2) reduced ability to inhibit hepatic gluconeogenesis despite high glucose levels ⁽²³⁾.

Preterm infants are particularly susceptible to hyperglycemia due to immature regulatory mechanisms and decreased insulin production by the pancreatic beta cells ^(17,115). Hyperglycemia is associated with increased mortality and a range of morbidities such as multiorgan failure, nosocomial infections, intraventricular hemorrhage, NEC, and impaired neurodevelopment ^(113,114), but studies confirming causal relationships are lacking.

Glycemic variability, a measure of blood glucose fluctuations over time, has been shown to be associated with mortality in critically ill adult patients, independent of mean glucose concentrations ^(116,117). Tian et al studied the effect of high (>70 kcal · kg⁻¹ · day⁻¹) versus low caloric intake (<70 kcal · kg⁻¹ · day⁻¹) on serum glucose levels and glycemic variability in 37 preterm infants less than 30 weeks GA ⁽³⁰⁾. The infants were grouped into high or low metabolic stress groups based on peak serum CRP concentration within 72 hours of the injury insult (e.g. surgery, sepsis, NEC). High metabolic stress was defined as a CRP concentration of >50 mg/L. In the present study, the high-stress group had higher glucose levels and increased glycemic variability compared to the low-stress group. A caloric intake >70 kcal · kg⁻¹ · day⁻¹ was also found to worsen the injury-related hyperglycemia. Of interest is also the interaction between glucose concentrations and lactate levels during critical illness (U-shaped curve) ⁽¹¹⁸⁾. A large retrospective study in critically ill adults showed that high lactate levels combined with low glucose concentrations were associated with the highest risk of organ dysfunction and hospital mortality ⁽¹¹⁸⁾. In children, lactate resolution has been shown to be associated with a decreased risk of persistent organ dysfunction during critical illness and is used as a marker to predict outcome ⁽¹¹⁹⁾.

The prevention and treatment of hyperglycemia in preterm infants is controversial in terms of choosing between reducing carbohydrate supply or starting insulin ^(120,121). Insulin has been shown to reduce mortality and morbidity in critically ill adults ⁽¹²²⁾, but follow-up RCTs on the effects of prophylactic insulin infusion with tight glycemic control in adults, children and preterm infants did not show any clear benefit. On the contrary, many demonstrated an increased incidence of hypoglycemia ^{(121,123–127)} and even a higher 28-day mortality risk in a large RCT of very-low-birth-weight infants ⁽¹²⁷⁾. Interestingly, insulin treatment of established hyperglycemia was associated with a significant lower 28- and 70-day mortality in the population-based EXPRESS (extremely preterm infants in Sweden) cohort study ⁽¹¹⁴⁾. Present PN guidelines recommend to start insulin therapy if neonates in the NICU experience repeated blood glucose levels >10 mmol/L (180 mg/dL) despite a reasonable adaptation of the glucose infusion rate ⁽²⁶⁾. The target glucose level, < 10 mmol/L (180 mg/dL), is the same as the target glucose level recommended for critically ill children and adults ^(26,119,128).

Protein Utilization

Protein supply should preferably cover protein turnover and tissue growth, because protein synthesis is energy demanding and unlike glucose and fatty acids, amino acids cannot be stored ^(27,75,100). Amino acids that are not incorporated into protein are irreversibly oxidized to CO₂ and ammonia before they are converted to urea in the uric acid cycle and excreted in the urine, providing approximately 4 kcal/g ^(129–131). Excessive amino acid intakes may result in toxicity, such as hyperammonemia, metabolic acidosis, seizures or coma ^{(130,132,133)}. In healthy infants, protein needs are based on the sum of amino acids needed to ensure adequate growth and the amino acids needed to cover obligatory nitrogen losses ⁽¹³¹⁾. Certain nonessential amino acids may be synthesized in sufficient quantities from other amino acids and glucose, whereas others are essential or conditionally essential and need to be provided through the diet ^(129,131). The rate of protein synthesis is dependent on sufficient availability of one or more of the essential or conditionally essential amino acids, whereas an energy supply of 30–40 kcal per 1 g protein is recommended for optimal protein utilization (>25 kcal/g amino acids when administered parenterally) ^(129,131).

During critical illness, protein degradation is increased ⁽²⁴⁾. The release of cytokines, cortisol, and growth hormone promotes proteolysis of skeletal muscle so that specific amino acids can be redirected and used as substrates for hepatic gluconeogenesis (particularly alanine and glutamine), tissue repair, wound healing and positive acute phase reactants, like CRP, haptoglobin, fibrinogen and procalcitonin ^(19,24,134). At the same time, the synthesis of constitutive proteins (albumin and transthyretin), IGF₁, retinol-binding protein, and transferrin is reduced ⁽¹³⁴⁾. The endogenous protein breakdown does not seem to be attenuated by exogenous protein or carbohydrate supply and may rapidly lead to protein-caloric malnutrition ^{(1,24,26,135)}. This may play an important role in growth-restricted and premature infants with limited protein and energy stores.

Several studies in preterm infants and sick neonates and children show that a minimum intake of 55–58 kcal · kg⁻¹ · day⁻¹ and 1.3–1.5 g protein · kg⁻¹ · day⁻¹ are needed to achieve a positive protein balance ^{(57,136–138)}, albeit higher intakes may be needed in parenterally fed patients with severe illness ⁽⁵⁷⁾; however, since endogenous protein breakdown is relatively unaffected by exogenous protein supply, it is not clear whether a positive protein balance is possible or even desirable during the early, critical illness phase. In fact, autophagy, the recycling of cellular components into amino acids and fatty acids for cellular fuel, is enhanced by fasting ^(69,70). Fasting also promotes the downregulation of mitochondrial activity to sustain cell life (theory of adaptive hibernation) ⁽⁷⁰⁾. Both mechanisms are considered to play important roles during the inflammatory response and it is hypothesized that high energy and protein intakes may lead to hepatic overload, attenuated autophagy and disrupted mitochondrial function ^(19,70). On the other hand, excessive autophagy and malnutrition may worsen mitochondrial dysfunction and trigger cell death ^(69,139), highlighting the challenge of finding the right balance between meeting energy and nutrient needs while avoiding overfeeding.

When the acute stress phase resolves, urinary output commonly increases and endothelial membrane integrity improves ⁽¹⁾. The accompanying reduction in protein catabolism is reflected by decreased urinary nitrogen excretion, decreased CRP and increased albumin and transthyretin levels ^(1,140). CRP, transferrin and transthyretin concentrations have been shown to correlate with the magnitude and duration of the injury response in neonates and infants, ^(96,140,141), and CRP and transthyretin concentrations are predictive for the length of hospital stay and 30-day mortality ^(140,142). Thus, it has been suggested that these markers could be used to establish the shift from catabolism to anabolism. Other biochemical markers commonly used to assess protein tolerance and adapt nutritional care are acid-base status and urea levels ^(143–145); however, an increase in urea levels is an unspecific marker, which more often reflects the catabolic state of acute inflammation, intravascular volume depletion, inadequate energy supply, suboptimal amino acid composition, or may simply be a sign of appropriate amino acid oxidation or higher protein intakes ⁽¹³¹⁾.

Fat Utilization

Lipids provide fatty acids, which are concentrated sources of energy, building blocks of cell membranes and precursors of bioactive eicosanoids; important substrates in the regulation of inflammation, platelet aggregation and tissue repair ⁽¹⁴⁶⁾. During the acute metabolic stress response, catecholamines and growth hormone induce lipolysis with mobilization of free fatty acids and glycerol ^(17,19). Studies in sick preterm neonates and older infants have shown that fatty acids are used as a primary fuel source and that lipid oxidation is proportional to stress severity ⁽¹⁴⁷⁾.

As mentioned earlier, fat utilization is dependent on the supply of glucose. If glucose is provided in amounts higher than the upper threshold for oxidation, excess glucose will be redirected to de novo lipogenesis and the oxidation of fatty acids ceases ^(101,104). The conversion of glucose to fat is an energy-demanding process characterized by an increase in CO₂ production relative to oxygen consumption, which may lead to increased respiratory rate and ventilator dependency. Reducing the glucose to lipid ratio during parenteral nutritional support has been shown to promote fat utilization and reduce lipid peroxidation ^(101,104). The optimal ratio of glucose to lipids is not defined, but some have suggested limiting the amount of calories provided as glucose to a maximum of 50% caloric intake in critically ill adults ⁽¹⁰⁴⁾. Similarly important, if energy supplies are consistently below energy requirements, endogenous fat stores are needed for fat oxidation. In a study of 26 children who received intravenous carbohydrate exclusively after cardiac surgery, approximately 80% of the macronutrient utilization was from the oxidation of endogenous fat ⁽¹⁴⁸⁾. Such an increased fat utilization may be detrimental in preterm and growth-restricted infants, who have very limited fat stores and are at risk of essential fatty acid deficiencies ⁽¹⁴⁹⁾.

Historical studies suggested that parenteral lipid emulsions may have negative effects on pulmonary function, partly by reducing pulmonary diffusion capacity ⁽¹⁵⁰⁾, and by increasing pulmonary blood pressure and vascular resistance ⁽¹⁵¹⁾. Moreover, high contents of soybean oil and phytosterols are thought to induce inflammation and contribute to PN induced cholestasis (PNALD) ^(152,153). Newer composite LE with or without fish oil have been developed, providing fatty acids (oleic acid and fish oil) that may be more immunological inert or even promote anti-inflammatory mechanisms ^{(149,152,154)} but the evidence for beneficial effects of these emulsions are low ^(155–157). Whether newer composite lipid emulsions with or without fish oil may be more beneficial in critically ill neonates has not been studied and is thus unknown ⁽¹⁵⁵⁾. Several components of PN, in addition to lipids, generate oxidants such as hydrogen peroxide, lipoperoxides and ascorbylperoxides, so it is recommended to shield PN from light to reduce the oxidative load of premature infants and critically ill neonates ^(149,158). A recent meta-analysis, which included 800 premature infants from four trials, showed that mortality in the light-protected group was half of that in the light-exposed group, but whether this was due to the reduction of peroxides or other factors, such as reduced degradation of antioxidant vitamins, remains unclear ⁽¹⁵⁸⁾.

Intravenous lipid emulsions do not seem to affect platelet count or function, but there are conflicting data about lipid clearance during sepsis ⁽¹⁴⁹⁾. Serum triglyceride levels have been shown to correlate with illness severity in critically ill children ⁽⁹⁶⁾, and more frequent monitoring of triglyceride concentrations may thus be helpful when providing nutritional care. At present, serum triglyceride concentrations <3.0 mmol/L (265 mg/dL) are generally considered acceptable ⁽¹⁴⁹⁾. During suspected/confirmed sepsis, disseminated intravascular coagulation, thrombocytopenia, impaired liver function, increased triglyceride concentrations or metabolic acidosis, it may be prudent to decrease parenteral lipid supply ^(49,52,58); however, both high glucose intake and overfeeding may impair lipid utilization as described earlier, and the need for a reduction in glucose load should be considered before parenteral lipids are reduced ⁽¹⁴⁹⁾.

Mode of Feeding

The recommended route to administer nutritional support during critical illness is by EN where this is possible ⁽¹⁵⁹⁾. In contrast to critically ill children and adults, in whom early initiation of EN (within 24–48 hours) is considered standard of care ^{(28,74,160,161)}, many critically ill preterm infants and neonates receive nutritional support by the parenteral route because the gastrointestinal tract is immature or full enteral feeding has not been established. Nevertheless, minimal enteral nutrition (MEN), usually defined as the supply of nutritionally insignificant milk volumes of 12–24 mL · kg⁻¹ · day^{−1(162–164)}, is advocated whenever possible to maintain gut integrity ⁽¹⁶⁵⁾

Studies show that very- and extremely preterm infants may reach full enteral feeds by 7–14 days ^(166–168); however, in a large randomized, controlled trial of 2804 preterm infants, a faster increment of enteral feeding volumes (target 30 mL · kg⁻¹ · day⁻¹ vs 18 mL · kg⁻¹ · day⁻¹) did not improve survival without moderate or severe neurodevelopmental disability at 24 months, nor did it affect the risk of late-onset septicemia or NEC ⁽¹⁶⁶⁾. Importantly, median (SD) age at randomization was 4 (3–6) days and the actual intake volumes were lower than intended, so that median days to reach full milk feeding volumes were 7 and 10 days in the faster and slower increment groups, respectively.

We did not identify any randomized controlled trials on the role of feeding mode in critically ill term neonates. A secondary analysis of the PEPaNIC trial showed that low mean EN intake was associated with newly acquired infections, hypoglycemia, duration of mechanical ventilation, length of PICU and hospital stay, but also that all these associations, except for hypoglycemia, disappeared after adjustment for confounders ⁽¹⁶⁹⁾. Two small retrospective observational studies ^(170,171) and one survey ⁽¹⁷²⁾ reported feeding practices in infants undergoing therapeutic hypothermia for HIE. Available data suggest that MEN is safe and feasible ^(170,171). Unfortunately, none of these studies reported details on the parenteral support given. An ongoing, large retrospective national cohort study aims two give some answers to the optimum enteral and PN strategy for term neonates during and after therapeutic hypothermia ⁽¹⁷³⁾, but the results have not yet been published.

In sum, there is insufficient data from studies in critically ill preterm infants or term neonates to determine whether EN is more beneficial than PN during the acute phase of illness, or whether higher EN intake is superior to lower EN intake. The use of EN as compared to PN has not been shown to have an effect on overall mortality in critically ill adult patients, but to decrease infectious complications and ICU length of stay, most likely caused by reduced macronutrient intake with EN ⁽¹⁷⁴⁾. Interestingly, data from two large RCTs do not support a preference for early enteral compared with early parenteral nutrition when nutritional support is targeted at similar calorie and protein intakes (n = 4810 critically ill adults) ^(175,176). On the contrary, these data suggest that early enteral nutrition may be of more harm; mainly by increasing the risk of digestive complications ^(175,176) and, as found in the PEPaNIC trial, the risk of hypoglycemia ⁽¹⁷⁶⁾.

TIME OF INITIATION OF NUTRITIONAL SUPPORT

Preterm Infants

Historically, it was believed that premature infants tolerated fluid and calorie restriction without negative long-term effects ⁽¹⁷⁷⁾, but in the late 1950s and early 1960s experimental studies showed that early malnutrition permanently affected the growth of organs, including the brain ^(177,178). Since then, a myriad of studies have investigated the safety and efficacy of the timing of nutrient supply after birth in very preterm infants ^{(143,144,167,179–196)}. Based on this body of evidence, current European guidelines for PN in children advocate early initiation of PN in preterm infants, including 1.5–2.5 g · kg⁻¹ · day⁻¹ of amino acids and 1–2 g · kg⁻¹ · day⁻¹ of lipids, with a gradual increase to target within a few days ^(129,149). Since most of the studies performed do not include critically ill preterm infants nor discriminate between healthier and sicker neonates at inclusion, we do not know whether the results can be generalized to the critically ill preterm infant. In a large exploratory, secondary study of a randomized trial including 1366 extremely low birth weight infants ⁽⁴⁸⁾, total daily energy intake during the first week of life was found to mediate the effect of critical illness on later outcomes ⁽⁴⁸⁾; however, recent reviews and meta-analyses on the effect of initial higher vs lower intakes of parenteral amino acids supply on clinical outcomes in very preterm and low birth weight infants, did not find clear evidence for long-term benefits with early higher intakes when summarizing data from ∼1500 infants ^(67,68,131). Apart from modest effects on weight gain, protein balance and glucose control, early and higher AA intakes increased the risk of raised urea levels and metabolic acidosis.

Term Neonates

We did not identify any RCT on the clinical effect of the timing of nutritional support in critically ill neonates admitted to a NICU ^(197,198), but two RCTs and one secondary analysis of an RCT have studied the effect of early vs delayed nutritional support in neonates and children admitted to a pediatric intensive care unit (PICU) ^(71,199,200).

The multicenter PEPaNIC trial (n = 1440) ⁽⁷¹⁾, included 209 neonates (mostly admitted due to cardiac and abdominal surgery), and studied the effect of withholding PN for 1 week compared to providing full nutrition up to caloric targets by initiating PN within 24 hours after admission, if enteral nutrition was insufficient to cover 80% of the caloric target. To match fluid intakes between the groups, the late PN group received a mixture of glucose 50 mg/mL and saline (9 mg/mL). All infants received intravenous trace elements, minerals, and vitamins, and glucose concentrations were controlled with insulin according to local target ranges ⁽⁷¹⁾. Even though the median total macronutrient administration up to day 4 were only 50–70% of target in the early PN group, mean total energy intake exceeded 55 kcal · kg⁻¹ · day⁻¹ on day 2, 70 kcal · kg⁻¹ · day⁻¹ on day 3 and 80 kcal · kg⁻¹ · day⁻¹ on day 4 in the infants with a weight <10 kg ^(71,201). In contrast, the mean total energy intake in the late PN group reached 25 kcal · kg⁻¹ · day⁻¹ on day 2 and remained between 40 and 45 kcal · kg⁻¹ · day⁻¹ on days 3 and 4, which is more in line with both estimated energy requirements and current energy recommendations for the acute phase of critical illness (40–50 kcal · kg⁻¹ · day⁻¹) ^(25,27). The primary study showed that permissive underfeeding reduced the incidence of new infections, shortened the duration of intensive care dependency, and reduced length of hospital stay ⁽⁷¹⁾. The strongest effect was observed in neonates and undernourished children ^(72,202). The PEPaNIC findings are in line with the results of the adult EPaNIC study, which showed that early combined parenteral/enteral nutrition delayed recovery irrespective of critical illness severity ⁽²⁰³⁾.

Secondary observational analyses of the PEPaNIC trial indicate that early provision of parenteral amino acids could explain the worse clinical outcomes in the intervention group ⁽²⁰¹⁾. The risk of all studied outcomes gradually increased up to a median daily amino acid dose of 1.15 g/kg [interquartile range (IQR) 1.10–1.22], representing 40–50% of reference doses for age and weight; however, higher doses did not further increase the risk of harm. The authors discuss whether this negative effect of parenteral amino acids is caused by the suppression of autophagy, an amino acid load above the metabolic capacity of the liver and kidneys or a suboptimal composition of the amino acid formulations used ⁽²⁰¹⁾. Interestingly, higher average doses of lipids were associated with a greater likelihood of earlier live discharge and with a greater likelihood of earlier live weaning from mechanical ventilation in the neonates. These findings support the notion that lipids play an important role during critical illness. Higher plasma 3-hydroxybutyrate (3HB) and lower blood glucose concentrations were also associated with earlier weaning from mechanical ventilator support and live discharge ⁽²⁰⁴⁾. The ketogenic fasting response in the late PN group was however not associated with new infections. Other adjusted analyses revealed that a higher average blood glucose concentration was an independent risk factor for infections, whereas a higher daily glucose dose was protective ⁽²⁰¹⁾. These findings are in line with other studies indicating that factors other than glucose intakes have an impact on glucose concentrations ^(114,188). The low average glucose intake in the late PN group (2–3 vs 5–6 mg · kg⁻¹ · min⁻¹) resulted in a significantly higher risk of hypoglycemia ⁽⁷²⁾. Recent data from follow-ups at 2 and 4 years after PICU admission show that delayed PN did not negatively affect survival, anthropometrics, health status, or neurocognitive development, but reduced emotional and behavioral problems ^(205,206). Interestingly, critically ill term neonates appeared less vulnerable to the developmental harm caused by early-PN compared to infants >29 days, toddlers and children up to 5 years of age ⁽²⁰⁷⁾.

The second study on the effect of timing of nutritional support, was a small trial on the effect of enteral nutrition given within 24 hours compared to after 48 hours in children with burns (n = 77, age range 3.1–18.4 years) ⁽¹⁹⁹⁾. This comprehensive study included blood and urine measurements of several endocrine markers of inflammation throughout the study period, and biweekly indirect calorimetry. The incidence of reportable adverse events, including bowel necrosis, multisystem organ and renal failure, was higher in the early feeding group, but this was not significant. Early enteral feeding reduced caloric deficits, stimulated early insulin secretion and first-week nitrogen balance, but there was no difference between the groups in regard to endocrine status, morbidity, mortality, hypermetabolism or length of hospital stay.

The third study investigated the effects of different initiation times of PN (within 48 hours, after 48 hours to 72 hours, after 72 hours to 7 days, and after 7 days) on outcomes in children with severe traumatic brain injury (n = 77 + 13) ⁽²⁰⁰⁾. In this observational part of an RCT (Cool Kids), initiating nutritional support before 72 hours was associated with favorable outcomes, including improved survival with earlier PN initiation (within 48 hours). Of note, most of the children in this trial were between 3 and 15 years.

It is challenging to make a clear recommendation in regard to the best timing of nutritional support in critically ill term neonates based on existing evidence. Strengths and limitations of the PEPaNIC trial have been extensively debated previously ^(208–212). Although the trial showed no benefit of early PN in any specific age group and a particular risk of harm in children aged between 29 days and 11 months ⁽²⁰⁷⁾, it is important to stress that PN was started within 24 hours after admission if enteral intakes were not > 80% of target. This is not in line with the nutritional guidelines at that time, which only recommended commencing PN after a few days if patients were not able to reach 60% of caloric intake with EN ^(213,75). As presented in this literature review, the risk of overfeeding during the acute phase of the metabolic stress response is high and current evidence support a cautious supply of nutrients at or just below basic energy needs until the resolution of the acute phase. In the weight group 0-10 kg of the PEPaNIC trial, energy supplies in the early PN group exceeded 150% of theoretical basal energy requirements and protein supplies more than 60-70% of the reference dose for age and weight during the early acute phase ⁽²⁰¹⁾. Since growing infants are at higher risk of overfeeding compared to older children and adults ⁽⁷¹⁾, this may partly explain the larger negative effects found in this group. Moreover, the nutritional interventions tested in the PEPaNIC trial represent two extremes of nutrient intakes, and do thus not answer whether delaying PN for 7 days improves outcomes as compared to providing PN at or just below REE during the early acute phase of critical illness. In addition, the PEPaNIC trial did not include preterm infants and since they have very different nutrient requirements, the results from term neonates and children cannot be generalized to preterm infants.

The latest Cochrane review on early versus late PN for critically ill term and late preterm infants ⁽¹⁹⁸⁾, which only identified data from the PEPaNIC trial ⁽⁷²⁾, concluded that there is insufficient evidence from RCTs for recommendations for or against nutritional support during the first week of critical illness. Current available guideline recommendations for nutritional support in critically ill children ⁽⁷⁴⁾, including term infants ⁽¹⁶⁰⁾, advocate the initiation of EN within 24–48 hours after admission. A minimal enteral protein intake of 1.5 g · kg⁻¹ · day⁻¹ may be considered to avoid a negative protein balance, but energy targets should not exceed REE during the early acute phase of illness ⁽¹⁶⁰⁾. Moreover, recommendations aim to achieve at least two thirds of the prescribed energy target by the end of the first week ⁽⁷⁴⁾. This is based on observational cohort studies, which show positive associations between energy intakes greater than 67–80% of estimated energy targets and improved outcomes ^(214,215), and that energy intakes above 55 kcal · kg⁻¹ · day⁻¹ are needed to maintain energy equilibrium in young children and infants >1 month of age ^(57,136). Due to the potential harm of early commencement of PN, recommendations regarding initiation of parenteral support differ; however, in general, it is recommended that initiation of PN should be considered on an individual basis and that PN support can be delayed until day 8 in term infants and children with normal nutritional state and low risk of nutritional deterioration ^(27,74,129).

SUMMARY

The ESPGHAN-CoN reviewed relevant studies on nutritional support in critically ill preterm and term neonates, including studies on basic metabolism. A body of evidence support that measured EE during critical illness is substantially lower than predicted EE of healthy growing neonates; in the range of 40–50 kcal · kg⁻¹ · day⁻¹ with minimal metabolic stress and 50–60 kcal · kg⁻¹ · day⁻¹ with severe inflammatory stress (sepsis, preexisting inflammation, etc). Nonetheless, there is insufficient data to make firm recommendations on the optimal composition and timing of nutritional support.

In preterm infants, available evidence does not support any significant changes to current guidelines, which recommend that critically ill preterm infants should receive nutritional support started at (or reduced to) the minimal amount needed to cover basal metabolic rate and basic macronutrient needs during the early acute phase ^{(26,27,129,149)}. For many preterm infants, this means that they will need PN.

In critically ill term neonates, initiation of PN within 24 hours is not routinely recommended; however, considering the limitations of the PEPaNIC trial and the observed low risk of long-term harm from early PN in critically ill neonates, the ESPGHAN-CoN does not support a change towards withholding parenteral nutritional support for 7 days as standard nutritional care. This position paper suggests considering careful initiation of nutritional support, including micronutrients, just below or at predicted REE after 48–72 hours. Even though recent RCT data in adults implicate that permissive underfeeding with PN is safe ^(175,176), the ESPGHAN-CON only recommends commencing PN to prevent nutritional deficiencies when adequate enteral nutrition is not feasible.

In both critically ill premature infants and term neonates, the phases of clinical illness should be assessed daily for adjustments of nutritional care. Although only indicative, biomarkers such as CRP, serum-glucose, glycemic variability, lactate, serum triglycerides, transthyretin and urea may be used along with clinical parameters to help recognize the return of growth anabolism. We estimate that the early acute phase is likely to last between 2–4 days and the late acute phase about 3–6 days. This means that full nutrient intakes (target nutrition) may not be appropriate until between 5–10 days after the acute insult, depending on illness severity and duration. Of note, premature infants may exhibit an earlier return to anabolism after acute illness compared to infants born nearer to term. During the recovery phase, nutritional supply in the upper range of recommendations should be considered to help cover cumulative deficits and to promote tissue repair, and catch-up growth.

CONCLUSION AND RECOMMENDATIONS

This literature review identifies that there are insufficient data to determine the optimal composition and timing of nutritional support in preterm infants and term neonates who are critically ill. Based on the reviewed literature and a thorough evaluation and interpretation of present parenteral and enteral macronutrient guidelines ^{(26,27,76,129,149)}, the ESPGHAN-CoN makes the following cautious recommendations for nutritional management during different phases of critical illness:

General Recommendations

• If critical illness is suspected, establish the diagnosis by assessing clinical and biological markers (Table 1).
• Theoretical energy and macronutrient needs during different phases of critical illness are given in Table 2.
• Minimal enteral nutrition (MEN) should be initiated within 48 hours if feasible.
• Gradually advance nutrient intakes (∼1.3–1.5 times REE) when the clinical state and the inflammatory response are resolving. The transition from the catabolic to the anabolic phase seems to start between 3 and 7 days after the insult, but may occur earlier (24–48 hours) in preterm infants or neonates with less illness severity or be delayed in neonates with severe injury insults.
• During the recovery phase, consider to increase target above estimated needs to cover cumulative deficits and promote catch-up growth.

TABLE 2 - Theoretical energy and macronutrient needs during different phases of critical illness in the neonate

	Preterm infants			Term neonates <28 days
	Early acute	Late acute	Recovery	Early acute	Late acute	Recovery
Energy (kcal · kg−1 · day−1)
Enteral	40–55	70–95	110–160	35–50	55–80	90–120
Parenteral∗	40–55	60–80	90–120	15–40	45–70	75–85
Glucose (g · kg−1 · day−1)†
Enteral	5–8	7–11	11–15 (18)	4–6	6–10	9–15
Parenteral∗	5–8 (10)	7–10 (12)	11–14 (17)	4–7 (10)	6–10	8–14
Glucose (∼mg · kg−1 · min−1)
Enteral	3.5–5.5	5–7.5	7.5–10.5 (12.5)	3–5	4–7	6–10.5
Parenteral∗	3.5–5.5 (7.0)	5–7 (8.5)	7.5–10 (12)	3–5 (10)	4–7	5.5–10
Protein (g · kg−1 · day−1)
Enteral	1.0–2.0	2.0–3.0	3.5–4.5	<1.5	1.5–2.5	2.0–3.5
Parenteral∗	1.0–2.0	2.0–3.0	2.5–3.5	0 (−1.0)	1.5–2.5	2.0–3.0
Lipids (g · kg−1 · day−1)
Enteral	2.0–3.0	3.0–6.0	5.0–8.0	< 3.0	3.0–4.5	4.0–6.0
Parenteral∗ , ‡	1.0–2.0	2.0–3.0	3.0–4.0	0 (–1.5)	1.5–2.5	3.0–4.0

^∗When supplementing parenteral nutrition, enteral intakes need to be considered (subtracted from estimated total needs) to optimize nutrient supply and reduce the risk of overfeeding. Note that parenteral energy needs are lower than enteral requirements, and that the maximum ranges of protein (amino acids) and lipids are lower than when given enterally.

^†The glucose supply should be guided by plasma glucose measurements to avoid hypo- and hyperglycemia.

^‡Lipids should be an integral part of PN (30–50% of nonprotein calories) and the nonprotein energy to protein ratio >25 kcal/g protein to facilitate protein utilization.

Monitoring

• Assess the phase of clinical illness every 24 hours. Biomarkers such as CRP, glucose, glycemic variability, lactate, transthyretin and urea may be used along with clinical parameters to recognize the transition from the acute to the stable phase when anabolic protein metabolism reoccurs.
• In case of hyperglycemia, start insulin infusion if glucose levels remain > 10 mmol/L (180 mg/dL) despite reasonable adaptation of the glucose infusion rate, i.e, ∼4 mg · kg⁻¹ · min⁻¹ (6 g · kg⁻¹ · day⁻¹) in preterm infants and ∼ 3 mg · kg⁻¹ · min⁻¹ (4 g · kg⁻¹ · day⁻¹) in term neonates.
• Extremely preterm infants and small for GA infants are at risk of refeeding syndrome, particularly hypophosphatemia. Monitor and replace phosphate and potassium if values are low.

The different nutritional objectives between term neonates and preterm infants are highlighted in Figure 3.

FIGURE 3: Recommendation for the nutritional management of the critically ill neonate.

Research Gaps

This review calls attention to the fact that basic research and adequately powered trials are urgently needed to resolve key uncertainties on metabolism and nutrient requirements in critically ill neonates.

The ESPGHAN-CoN recommends that future research include:

• RCTs on the timing and amount of PN in critically ill preterm infants and term neonates.
• Studies on permissive underfeeding during the early phase of acute illness followed by a gradual increase to nutritional target compared to delayed PN in critically ill neonates.
• The clinical effects of enteral vs parenteral nutritional support at isocaloric intakes in critically ill neonates.
• Role of specific amino acids and fatty acids during critical illness.
• Studies to determine whether nutritional requirements are sex-specific in critically ill neonates.

These trials need to consider different levels of illness severity, include flexible study designs to account for heterogeneity in the populations studied, and be large enough to determine meaningful differences in functional outcomes, both in the short and long term.