Article Text

Download PDFPDF

Balancing the risks and benefits of parenteral nutrition for preterm infants: can we define the optimal composition?
  1. Nicholas D Embleton1,2,
  2. Colin Morgan3,
  3. Caroline King4
  1. 1Newcastle Neonatal Service, Newcastle Hospitals NHS Foundation Trust, Newcastle, UK
  2. 2Institute of Health and Society, Newcastle University, Newcastle, UK
  3. 3Neonatal Intensive Care Unit, Liverpool Women's Hospital, Liverpool, UK
  4. 4Neonatal Units, Imperial College Healthcare NHS Trust, London, UK
  1. Correspondence to Dr Nicholas D Embleton, Newcastle Neonatal Service, Newcastle Hospitals NHS Foundation Trust, Newcastle NE1 4LP, UK; Nicholas.embleton{at}


Nutrient intakes in preterm infants are frequently inadequate and are associated with worse neuro-developmental outcome. Preterm infants take time to establish enteral intakes, and parenteral nutrition (PN) is now an integral component of care. Despite this, the evidence base for PN intakes is extremely limited. There remains uncertainty over safe initial and maximum amounts of macronutrients, and the optimal amino acid and lipid composition. Studies have tended to focus on short-term growth measures and there are few studies with long-term follow-up. There may be a tradeoff between improving cognitive outcomes while minimising metabolic harm that means determining the optimal regimen will require long-term follow-up. Given the importance of appropriate nutrition for long-term metabolic and cognitive health, and the associated healthcare costs, optimising the composition of PN deserves to be seen as a research priority in neonatal medicine.

  • Neonatology
  • Nutrition

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Parenteral nutrition (PN) is established as standard-of-care for preterm infants, but despite its widespread use, the evidence base for the optimal composition is extremely limited.1 A recent National Confidential Enquiry into Patient Outcome and Death enquiry into neonatal PN highlighted a diverse range of practices, many of which were considered to be substandard across the UK.2 This may, in part, reflect uncertainty around intake recommendations in the early postnatal period. Preterm infants have limited nutrient stores, with strong evidence that nutrient intakes in early postnatal life relate to cognitive outcomes.3 The commonest use of neonatal PN in preterm infants is as ‘bridging nutrition’ while enteral nutrition is established, but it is also an essential part of management where enteral feeding is not possible, for example, in infants with necrotising enterocolitis (NEC). The wide range in requirements between, for example, a 24-week infant establishing feeds, and a growth-restricted 32-week infant following NEC make defining a single optimal composition for ‘preterm infants’ impossible. Nevertheless, the aim of this article is to describe the clinical challenges of determining optimal PN intakes, particularly macronutrients, such as protein, lipid and carbohydrate, in otherwise stable, preterm infants.

The evolution of neonatal PN in the late 1960s4 required several technical challenges to be overcome.5 Since then, dramatic advances in neonatal care have meant that infants weighing 500–750 g frequently survive, and PN administration is now a daily occurrence in most neonatal intensive care units (NICU). Unfortunately, improvements in the evidence base for PN intakes have failed to match the success of increased survival. Although the nutrient requirements to enable growth to approximate that of the in utero fetus are well described,6 there remain substantial challenges in terms of assessing growth in clinical practice (which is more complex than simply weight gain) and defining the optimal rate of growth ex utero which may not be the same as that when in utero.7 Systematic reviews of PN have shown a benefit on time to regain birth weight and early measures of growth,8 but there are no current controlled trial data to show a long-term benefit on growth, metabolic or cognitive outcomes.

While many observational studies suggest that higher rates of weight gain9 or nutrient intakes3 are associated with improved neurodevelopmental outcomes, these studies may lack adequate adjustment for confounders, and are at risk from reverse causation. While there are some limited, randomised, controlled trial (RCT) data to suggest that higher enteral nutrient intakes improve cognition in adolescence,10 there are no equivalent data for parenteral intakes. Nevertheless, these strong links between early nutrient intakes and later outcome, suggest that suboptimal provision of parenteral nutrients in early life is likely to have life-long adverse cognitive impacts. A recent RCT has shown a benefit of higher parenteral nutrient intakes on head growth,11 but longer-term follow-up will be needed to determine if this results in functional benefit. While growth (or proxy measures such as weight gain) is an important measure of health status for preterm infants, it has many shortfalls as an outcome measure in itself. The optimal pattern of growth needs to balance the potentially competing concerns of cognitive benefits versus the risks of adverse metabolic programming.

Most extremely preterm infants take at least 10–14 days to tolerate full enteral nutrition, so the composition of a typical neonatal PN regimen must be designed to complement any enteral intake, and when combined, aim to provide all essential nutrients during this period. Because infants receive nutrients via the parenteral and enteral route in the early postnatal period, trial designs aiming to determine the effect of early parenteral or enteral nutrition are necessarily complex. The postnatal age and speed of increase of PN that results in the optimal balance of risks and benefits such as line-associated sepsis from prolonged PN, and increased feed intolerance or NEC from faster enteral feeds is not clear, but is currently being examined in RCTs, such as the Speed of Increasing milk Feeds Trial (SIFT ( Additionally, given the multitude of nutrients that need to be provided by PN, progress would be painstakingly slow if a pharmaceutical-type RCT approach to each and every nutrient was adopted. Even then, nutrient interdependence will mean that the first limiting nutrient will set the ceiling for outcomes: a null finding for single nutrient enhancement may be obtained because of inadequate provision of other nutrients or cofactors. Currently used neurodevelopmental tools (such as Bayley Scale of Infant Development) provide important data on global outcome, but may lack the precision to detect important differences in the more specific cognitive domains likely to be affected by individual nutrients.12

Protein and amino acids

At 24 weeks gestation, a fetus weighing 500 g is composed of ∼90% water with only ∼50 g of ‘dry’ tissue.13 Extremely preterm infants have no energy stores as such, and must catabolise body protein to meet energy requirements if these are not met by the diet. The rapidity with which malnutrition occurs is dramatic, and birth of an extremely preterm infant must be seen as a nutritional emergency. Inevitable nitrogen losses (in urine, faeces, skin cells and secretions) are equivalent to ∼1 g/kg/day of protein in a preterm infant, but may be 50% higher in those born <28 weeks gestation.14 The predicted daily protein accretion of a fetus at 28 weeks gestation is ∼2 g/kg,13 meaning that an intake of at least 3–3.5 g/kg (protein or amino acid (AA) equivalent) is needed to promote protein accretion and allow for obligatory losses, if lean mass accretion is to proceed at a rate approximately the same as the in utero fetus. Higher intakes may be needed in the smallest infants and/or following a catabolic episode, or if catch-up growth is appropriate. Considering that obligatory protein losses in an extremely preterm infant are <1.5 g/kg/day, and that ∼2 g/kg/day protein are accreted in utero, it is possible to estimate that a preterm infant's lean mass will be less than 90% of the equivalent in utero fetus after just 48 h, unless exogenous AA are administered.

A series of studies in the 1980s and 1990s focused on nitrogen balance in the first week of life at differing levels of macronutrient intakes and broadly support AA intakes of 3–3.5 g/kg/day once full PN is established, if nitrogen retention similar to the in utero fetus is the primary objective.15 More recent studies confirmed these data,16 ,17 but there are now some data to support even higher intakes (3.5–4 g/kg/day) within the first week.11 ,18–20 Many of the earlier studies used metabolisable energy intakes lower (30–70 kcal/kg/day) than are commonly used now, although the optimal protein-energy ratio has yet to be well defined. While high AA infusions commencing immediately after birth (>3 g/kg/day) may mimic the observed high AA oxidation rate seen in utero,21 preterm babies must function without the help of the placenta to remove potentially toxic metabolites. A recent RCT data showed that 3.6 g/kg/day gave no additional advantage to 2.4 g/kg/day in the first 48 h with respect to nitrogen accretion, but was associated with more metabolic imbalances,18 while another showed improved head growth on ∼2 g/kg/day during the first 2 days increasing to 3.8 g by day 5.11

Splanchnic metabolism is important, but AA provided in PN avoid hepatic and splanchnic ‘first pass’ metabolism.22 Individual AA requirements are poorly defined, with evidence that some ‘non-essential’ AA are ‘conditionally essential’ in preterm infants including arginine, glutamine, glycine, proline, taurine, and tyrosine.22 ,23 Without knowledge of individual AA requirements, it may not be possible to define the optimal AA composition that best meet needs.24

There are additional practical challenges in AA delivery, such as solubility and precipitation problems associated with certain AA such as tyrosine and cysteine that will influence optimal PN composition.25 Current commercially available AA solutions appear to result in acceptable levels of nitrogen retention and plasma AA profiles, although the plasma concentrations of many conditionally essential AA are frequently lower than in utero references.26 ,27 However, there is substantial debate around the appropriate plasma AA reference for a preterm infant receiving PN in terms of safety and efficacy: are in utero or cord blood levels appropriate, or is the plasma AA profile of a healthy breastfed infant a more appropriate reference? One small recent study suggested an association between raised concentrations of AA and worse developmental outcome.28 Although the focus of this article is on PN nutrient composition, several studies indicate that achieving AA intakes that improve growth are not just dependent on PN AA composition, but also influenced by the use of clear nutritional strategies, standardised and/or concentrated solutions and PN service organisation.


Although difficult to precisely measure, it can be extrapolated from published data that resting energy expenditure is likely to be around 50–60 Kcal/kg/day in most stable growing preterm infants, although it will be higher in those with additional demands from coexisting morbidities such as sepsis.29 To enable growth, at least 100 kcal/kg/day will be needed for exclusively PN-fed babies although requirements are likely to increase as the proportion of enteral nutrition increases. The energy cost of growth depends on the precise tissue accreted and cannot be directly determined in preterm infants. Stable isotope studies have demonstrated the highly dynamic nature of protein turnover in preterm infants and show that ∼5 g protein is synthesised and catabolised for every 1 g protein that is finally accreted.30–32 This high level of protein synthesis and catabolism demands high protein and energy intakes.

Intravenous lipid

Intravenous lipid provides a concentrated source of energy, a delivery mechanism for fat-soluble vitamins, and also provides essential fatty acids (EFA). Although the amount required to prevent EFA deficiency is quite low (<0.5 g/kg/day), the requirements for optimal neurodevelopmental outcomes are likely to be considerably higher. Concerns regarding lipid deposition in the lungs, and the vasoactive effects of hyperperoxides and proinflammatory cytokines on pulmonary vascular function led to more cautious introduction in the past. While many authorities recommend commencement at 1–2 g/kg/day on the first day, increasing to 3–4 g/kg/day where PN is the sole source of nutrition, there are no data demonstrating clear long-term benefit.1 ,33 Increased lipid supply may improve early nitrogen retention,34 but there are limited data at the upper end of the intake range (3–4 g/kg), and no reliable data for intakes >4 g/kg/day. Systematic reviews suggest there is no disadvantage to commencement before day 5, but few studies compared very early initiation immediately after birth to 48 h of age, that is now commonly practiced.34 ,35 There is no consensus about whether there is a need to monitor for the presence of lipaemia, or raised levels of either cholesterol or triglycerides, or the acceptable upper limit that should be tolerated in parenterally fed infants.

Currently available intravenous lipid solutions were not designed with the objective of meeting the needs of extremely preterm infants. The most widely used products are produced from soybean oil and have a long track record of short-term safety, but do not provide the optimal blend of fatty acids. Newer lipids using olive oil, or a combination of soybean oil, medium-chain triglycerides, olive oil and fish oil are now increasingly used.34 ,36 ,37 Studies in older children suggest that these newer formulations might be advantageous, with those containing fish oil appearing to result in a lower incidence of liver inflammation and more appropriate fatty acids profiles.36 ,38 However an excess of eicosapentaenoic acid compared to arachidonic acid, as found in some of these new fish oil-based emulsions, has been associated with reduced growth when fed enterally to preterm infants.39 Currently, there are no long-term outcome data on preterm infants receiving newer lipid formulations. These products arose because of a desire to reduce the adverse proinflammatory hepatic effects seen in adults (and children), so while there may be a potential cognitive advantage to preterm infants of these lipids, this was not the driver for innovation. Like many areas of ‘medicines’ research, neonatal practice appears to lag behind.

Monitoring of nutrient intakes

There are no neonatal studies that help determine the optimal monitoring regimen for PN, but frequent electrolyte and glucose monitoring is important. Hyperglycaemia is common, but there appears to be no benefit in routine basal insulin administration.40 Additionally, the glucose threshold at which insulin is started (that balances the risks of insulin with any benefits) has not been defined, nor have the relative benefits of insulin been compared with alternate strategies such as decreasing carbohydrate intake. There is increasing recognition of a link between early weight gain, levels of IGF-1, and retinopathy of prematurity that suggest increased AA and glucose intakes may be of benefit to brain growth.41 ,42 Whether insulin has a role in modulating these processes remains to be determined.

Measurement of phosphate over the first few days of PN is not common but is likely to be important: phosphate is an important substrate for muscle function, and a component of lean tissue. It seems possible that some babies may need ventilatory support in the first few days simply because of inadequate phosphate levels. A recent study suggested that higher AA intakes without adequate phosphate supply, especially in infants who were growth restricted in utero, might cause a metabolic derangement similar to the ‘re-feeding’ syndrome seen in adults following starvation.43 It is not possible to assess the adequacy of PN AA intakes (ie, ‘protein’ supply) in individual patients: plasma AA measurement is expensive and there is no agreed reference, total serum protein concentration does not reflect anabolism, albumin has a half-life of several weeks, and urea is not a sensitive indicator of nitrogen intake over the first few days. Proteins with shorter half-lives (eg, retinol-binding protein) are not measured in clinical practice but might be useful indicators of short-term protein status. Acidosis or hyperammonaemia are rarely observed using modern PN AA formulations.

Balancing the risks

Determining the optimal nutrient composition of PN requires clinicians to balance multiple competing risks for several constituents in a situation where the evidence base is extremely poor. PN composition may be associated with manufacturing risks: solutions can become contaminated with toxins such as aluminium leached from glass vials containing calcium gluconate resulting in worse neurodevelopmental44 and bone outcomes in adolescents.45 There are multiple other potential adverse effects as a consequence of PN composition: for example, one study has shown an association between raised cholesterol levels in early postnatal life and aortic stiffness in later life.46 While there is general agreement that most preterm infants <1500 g birth weight are likely to benefit, the precise cutoffs at which the risks outweigh the benefits have yet to be determined (see table 1).

Table 1

Examples of risk-benefit ‘tradeoffs’ associated with parenteral nutrition composition

Because of the problems associated with observational studies, many of these uncertainties can only be resolved with large RCTs and long-term follow-up of metabolic, growth and cognitive outcomes that attempt to quantify the competing risks and benefits. In the absence of RCT data, standardised PN regimens with robust audit, and use of large-scale databases that include long-term developmental outcome may be used to refine practice. There are several areas of uncertainty, some of which we have listed in table 2. The wide variability in current practice reflects a lack of evidence on which to base robust guidelines, and the fact that most current recommendations are based on ‘expert opinion’. Given that PN is a critical determinant of survival and long-term outcome, carries with it substantial healthcare costs, and is administered to tens of thousands of preterm infants worldwide every year, this uncertainty reflects a woeful lack of well-designed research studies.

Table 2

Examples of current uncertainties and potential ways forward



  • Contributors All three authors are responsible for the paper, and have participated in the concept and design, drafting or revising, and have approved this manuscript as submitted.

  • Competing interests None.

  • Provenance and peer review Commissioned; externally peer reviewed.

Linked Articles

  • Fantoms
    Martin Ward Platt