Article Text

Download PDFPDF

Improving head growth in very preterm infants – a randomised controlled trial I: neonatal outcomes
  1. M J Tan,
  2. R W Cooke
  1. School of Reproductive and Developmental Medicine, University of Liverpool, Liverpool, UK
  1. Professor R W I Cooke, Neonatal Unit, Liverpool Women’s Hospital, Crown Street, Liverpool L8 7SS, UK; mc19{at}


Background: Infants born very preterm are at an increased risk of poor growth in the post-natal period. Poor brain growth in this critical period may result from inadequate nutrition, and has long-term effects on neurodevelopmental outcome.

Aims: To examine the feasibility of providing macronutrients at amounts above current recommendations (hyperalimentation) to improve nutrition and head growth in preterm infants.

Methods: 142 infants <29 weeks' gestation were randomised to hyperalimented or standard parenteral and enteral nutrition. Growth was monitored from birth to 36 weeks' postmenstrual age (PMA). The primary outcome measure was occipitofrontal circumference (OFC) at 36 weeks' PMA.

Results: 55 infants in the intervention group and 59 infants in the control group survived to 36 weeks' PMA. 11 (16%) infants in the intervention group and 13 (18%) infants in the control group were small for gestational age (SGA). There was no statistically significant difference between the two groups in the primary outcome measure or other growth variables. Babies in the intervention group received significantly more energy and protein, but 80% were still in a cumulative protein/energy deficit at the end of 4 weeks. 20 (24%, p = 0.008) of those in deficit at 4 weeks had an OFC of more than 2 SD below the mean at 36 weeks' PMA, as opposed to none of those not in deficit.

Conclusion: Cumulative energy/protein deficit is predictive of poor head growth, but the delivery of adequate intakes remains a challenge in the preterm.

Statistics from

Survival of very preterm infants has improved over the last few decades.14 Rates of cerebral palsy and neurosensory impairment decreased after 2000, although cognitive outcomes have not improved.57

Growth restriction in later life is common in preterm survivors.810 At 36 weeks' postmenstrual age (PMA), up to 89% to 97% of infants have weight, length and occipitofrontal circumference (OFC) below the 10th centile.11 12 Most remain so at 5 years.13 At 7 years their size remains significantly less than term controls.14 Small OFC was associated with poorer cognitive and motor performance. OFC correlates closely with brain volume.15 Preterms with small OFC at 8 months have poorer cognitive outcome later.16 Growth failure in this population may relate to inadequate early nutrition. Infants with higher energy intake in the first few weeks of life had better developmental outcome.11 17 18

As most very preterm infants are intolerant of full enteral feeding in the immediate post-natal period, they rely mainly on parenteral nutrition (PN). As the gut matures and feed tolerance improves, increasing amounts of enteral feeds are given. This may take several weeks. Most clinicians adopt a cautious approach to advancement of enteral feeds because of the risk of developing necrotising enterocolitis (NEC).19 Early enteral feeds may result in better weight gain.9 20 21 Infants at the greatest risk of poor growth are those who received PN for prolonged periods.9 22 23 Fluid restriction and multiple drug infusions may prevent adequate PN in the early post-natal period.

The European Society of Paediatric Gastroenterology and Nutrition (ESPGAN) and the American Association of Pediatrics (AAP) recommend an intake of 120 kcal/kg/day of energy and 3 g/kg/day of protein for preterm infants.24 25 This does not take into account the additional requirements needed in chronic lung disease (CLD) or required for “catch-up” growth.26 27 Embleton et al showed mean cumulative energy deficit of 813 kcal/kg and protein deficit of 23 g/kg by the fifth week of life.

Hyperalimentation is defined as providing parenteral or enteral nutrition with macronutrients at amounts above current recommendation. We hypothesised that improving nutrition in the early post-natal period would improve head growth. We conducted a randomised controlled trial of hyperalimentation in very preterm infants.


All infants born before 29 weeks' gestation were eligible. Triplets and infants of higher multiplicity, those admitted after 7 days of age and infants with major congenital abnormalities were excluded. This single-centre study was conducted between January 2004 and January 2007.

Infants were recruited within 7 days of age. Variable-length block randomisation was used. The randomisation codes were kept in sequentially numbered, opaque and sealed envelopes. Following written informed consent from the parents, the infants were randomised to receive either a hyperalimented or standard feeding regimen until 34 weeks' PMA.

The hyperalimented regimen comprised PN that contained 20% more energy (117 kcal/kg/day) with proportional increase in dextrose (16.3 g/kg/day), protein (4 g/kg/day) and fat (4 g/kg/day). The standard regimen comprised PN that contained 93 kcal/kg/day, and followed the recommendations by ESPGAN 25 for dextrose (13.5 g/kg/day), protein (3 g/kg/day) and fat (3 g/kg/day). The micronutrients within the two PN were the same and as recommended. PN began within the first 24 h after birth when possible. PN was increased stepwise from 1 g/kg/day protein and lipid to 4 g/kg/day protein and lipid over 7 days for the intervention group, and to 3 g/kg/day protein and lipid over 5 days for the control group. The carbohydrate intake was dependant upon the total fluid allowance of each infant, which was increased from 60 and 90 ml/kg/day to 150 and 165 ml/kg/day in the first 5 days. Hyperglycaemia (blood glucose >12 mmol/l) was managed with insulin infusions.

Infants started milk within 48 h or when clinically stable. The decision to start milk was made by the clinician responsible for the infant. Expressed breast milk (EBM) was the first choice. The infants of mothers who chose not to provide EBM, received preterm formula: Nutriprem (Cow & Gate) for the intervention group and Osterprem (Farleys) for the control group. The target energy and protein intakes based on estimated composition of EBM were: 133 to 150 kcal/kg/day and 4 g/kg/day for the intervention group and 133 kcal/kg/day and 3.3 g/kg/day for the control group.28

The volume of milk given was increased by 6–12 ml every 24 h, as tolerated, until the target volume of 165 ml/kg/day was met. Milk was withheld in infants who had signs suggestive of NEC. EBM was fortified with Nutricia (Cow & Gate) breast milk fortifier once the infants tolerated volumes >75 ml/kg/day. PN was discontinued once infants received >50% of their total fluid as milk. Ten per cent dextrose with electrolytes was given intravenously until full enteral feeding was achieved. Milk was increased by clinicians to 180 ml/kg/day, if weight gain was unsatisfactory.

OFC, lower leg length (LLL) and weight were recorded weekly. OFC was measured using a standard OFC non-stretchable lasso tape (Child Growth Foundation, London, UK), LLL with a knemometre (Force institute, Copenhagen, Denmark), and weight using digital scales (Seca 757 class III). Actual intakes of parenteral and enteral feeds for each infant were recorded. Daily serum electrolytes and creatinine in the first 2 weeks of life and then weekly thereafter, serum urea on the eighth day of PN, weekly serum triglyceride for infants who remained on PN and weekly bilirubin and liver transaminases were measured in all infants. Energy and protein intakes were calculated as intravenous intakes plus 85% of enteral intakes, to give metabolisable energy and protein intakes (assuming 85% of enteral intakes are absorbed in the gut).29 30 Energy and protein balance were estimated by subtracting actual cumulative energy and protein intakes from recommended intake (120 kcal/kg and 3 g/kg for enteral feeds or the equivalent in PN).

The primary outcome measure was the OFC at 36 weeks' PMA. The secondary outcomes were LLL, mid-arm circumference (MAC), total body length and weight at 36 weeks' PMA. OFC, length and weight measurements were converted into standard deviation scores (SDS).31 Total body length was measured using a standard infant measuring mat (Child Growth Foundation) and MAC using a non-stretchable disposable measuring tape. Apart from the primary outcome measure, all measurements were performed by the author (MJT). Trained observers blind to assignment measured the OFC. Only OFC and weight measurements performed by local staff were available for infants transferred back early to their units.

Outcomes were defined as follows: patent ductus arteriosus (PDA) the presence of a haemodynamically significant PDA; sepsis by a positive blood, urine or CSF culture, or raised C reactive protein with clinical signs of infection; antenatal steroids as any betamethasone at least 12 h before delivery; NEC as all cases surgically confirmed or in which there was a strong clinical suspicion leading to medical treatment; cholestasis as raised conjugated bilirubin above 30 μmol/l, and raised liver transaminases; severe intraventricular haemorrhage (IVH) as IVH of Grade III and above; ventilatory support as mechanical ventilation; CLD as oxygen requirement at 36 weeks' PMA. Clinical risk index for babies (CRIB II) scores were also calculated.32

Sample size and statistical analysis

Sample size was based on a previous unit audit in 2003 which showed that among infants <1500 g, half had an OFC more than 2 SD below the population mean (based on Gairdner–Pearson growth reference) at discharge.33 To reduce the numbers with small heads by 50%, with alpha at 0.05 and power at 80%, 110 infants would be required. Allowing for deaths and loss to follow-up of around 20%, 140 infants were required.

Data analysis was performed by intention to treat using SPSS 12. Student t test, chi-squared test, Mann–Whitney U test and analysis of variance (ANOVA) were used where appropriate. Bivariate correlation analyses were used to calculate correlation coefficients and effect sizes for both cumulative energy and protein deficits. Regression analyses of clinical factors most likely to affect head growth were performed. Post hoc analyses of primary outcome measure by sex and by SGA (small for gestational age) or AGA (appropriate for gestational age) were performed.


Between January 2004 and January 2006, 176 eligible infants were admitted to the unit (fig 1). Twenty-six parents refused consent for their infant to be recruited and eight were missed. One hundred and forty-two infants were randomised, 91 to the intervention group and 81 to the control group. Three infants were subsequently found to have major malformations; one in the control group had Trisomy 21 and died later, two in the intervention group had an atrioventricular septal defect and congenital Cytomegalovirus infection, respectively, but both were included in the analyses. Thirteen intervention and 15 control infants died prior to 36 weeks' PMA. Fifty-five intervention and 59 control infants are included in the analyses, including three who died subsequently (two intervention, one control). Characteristics at entry for all infants are presented in table 1. SGA is defined as birth weight below the 10th percentile. At 36 weeks' PMA, survivors in the intervention group had lower gestation (19 vs 15 below 26 weeks) and birth weight (median 900 g vs 965 g).

Figure 1 Schematic diagram showing outcomes of all eligible babies and number of babies who survived to 36 weeks' PMA.
Table 1 Entry characteristics of all babies

Babies in the intervention group received significantly more energy (mean difference 144 kcal/kg 95% CI 56 to 232 kcal/kg, p<0.01) and protein (mean difference 9 g/kg 95% CI 7 to 12 g/kg, p<0.001) than controls by the end of 4 weeks. Ten (22%) from the intervention group compared to 3 (6.4%) from the control group achieved positive energy balance (not statistically significant), while 23 (51%) compared to 1 (2%) achieved positive protein balance at 4 weeks (p<0.001).

OFC and weight at 36 weeks' PMA were obtained in 112 of the survivors. Two babies were not measured; one had hydrocephalus and the other had been withdrawn from the study at parental request. Both were in the control group. Full measurements including LLL, total body length and MAC at 36 weeks' PMA were measured in 68 infants. The rest of the infants had returned to their referring units and only OFC and weight measurements were available. When two measurements were obtained for a parameter, the measurement from the left limb was used. Clinical, nutritional and growth outcomes are shown in table 2. All growth outcomes at 36 weeks' PMA were not statistically different between the two groups. Using ANOVA, there were no differences in the weekly growth rate of OFC, LLL and weight between the two groups in the first 7 weeks of life. Infants in the intervention group regained their birth weights significantly earlier. Linear growth measured using LLL gain in the first 2 weeks of life was greater in the intervention group (0.28 mm/day vs 0.20 mm/day, p = 0.05). Subsequent LLL gain did not differ significantly between the two groups. Post hoc analysis of OFC SDS at 36 weeks' PMA by subgroups (sex and SGA or AGA), showed no statistical difference between the intervention and control groups.

Table 2 Clinical, nutritional and growth outcomes of all survivors

At the end of 4 weeks, 80% of babies in the intervention group compared to 97.8% in the control group were still in overall energy/protein deficit (p = 0.02). When growth was compared for babies in both groups who were or were not in positive energy and protein balance at 4 weeks, mean SD score of OFC (−0.11 vs −1.17, p = 0.006), length (−1.78 vs −2.65, p = 0.046), weight (−0.77 vs −1.52, p = 0.008) and actual LLL (10.8 vs 10.2, p = 0.03), at 36 weeks' PMA were significantly greater in those that did not have a deficit. None of the babies who were in overall positive balance had an OFC more than 2 SD below the mean, compared to 20 (24.4%, p = 0.008) of those who had an overall deficit at 4 weeks.

The relationships between head growth and cumulative energy and protein balance are shown in figs 2 and 3. The correlation coefficients were statistically significant (R = 0.44 and R = 0.34). The relationship was even more pronounced in babies below 27 weeks (correlation coefficients of 0.54 and 0.46).

Figure 2 Scatter plot and graph showing correlation between estimated cumulative energy balance at 28 days and z score of OFC at 36 weeks' PMA. OFC, occipitofrontal circumference; PMA, postmenstrual age.
Figure 3 Scatter plot and graph showing correlation between estimated cumulative protein balance at 28 days and z score of OFC at 36 weeks' PMA. HC, head circumference; OFC, occipitofrontal circumference; PMA, postmenstrual age.

Babies in the intervention group received mechanical ventilation and PN for longer. When “days on ventilation” was adjusted by birth weight, singleton, chorioamnionitis and antenatal steroid, the effect of randomisation became insignificant (p = 0.095). There was no statistical difference in total hospital stay or number of babies who required oxygen at 36 weeks' PMA.

Using logistic regression, factors significantly associated with OFC more than 2 SD below the mean were explored. The best model showed that factors most predictive of OFC more than 2 SD below the mean at 36 weeks' PMA were days to full enteral feeding (p = 0.024), CRIB II score (p = 0.036), SD score of first recorded OFC (p = 0.006) and cumulative protein and energy deficit at 4 weeks (p = 0.029).


We have shown that there is a relationship between OFC at 36 weeks' PMA and nutritional intake. Cumulative energy and protein deficits occur commonly among extremely preterm infants in the first few weeks of life. Attention to nutritional intake may reduce these deficits and improve overall growth. Increased energy and protein intakes improve head growth in the first few weeks of life, particularly among infants below 27 weeks' gestation.

Babies in the intervention group appear to have been on PN and respiratory support for longer. This may be explained by more survivors with lower gestation in the intervention group. Early linear growth (lower leg length in the first 14 days) was better among babies in the intervention group and there were no significant differences in the length of hospital stay and rate of CLD between the groups.

We examined head growth as a proxy for brain growth and development. The correlations between head growth and energy as well as protein intakes are significant. This is consistent with other studies that have investigated the relationship between nutrient intake and growth.17 22 34 Georgieff et al showed a significant correlation between suboptimal head growth and duration in which infants received an energy intake of less than 85 kcal/kg/day.22 Embleton et al showed that there is a correlation between nutritional deficit and weight gain.34 Although the babies in their study were more mature, they found that 45% of the change in weight was accounted for by cumulative energy deficit. Cumulative protein deficit had no significant effect. Using multivariate analysis, Berry et al showed that energy intake correlated positively with growth in the first 56 days and protein intake with growth in the first 14 days.17 In the later period of growth (15–56 days), weight increased by 1 g per 7.8 kcal intake. Energy intake was a predictor of growth in the earlier period when protein intake was not included in the analysis.

Wilson et al randomised 125 ill preterm babies who weighed <1500 g to aggressive or standard nutritional.35 They increased dextrose, amino acid and lipid in PN to 12.5%, 3.5 g/kg/day and 3.5 g/kg/day, respectively. Infants in the intervention group received the same enteral feeds but were started on feeds earlier, regardless of clinical state. They found that early growth was better in the intervention group while late growth was better in the control group. The number of babies with weight and length below the 10th centile were significantly reduced and the numbers of babies with OFC below the 10th centile were halved. The infants in their control group only received a maximum of 2.5 g/kg/day amino acids and 2 g/kg/day lipids in the PN. This is less than the babies in the control group of our study and also less than the current recommended values.

Delivering higher amounts of energy and protein to ill extremely preterm infants is fraught with difficulties. These include problems with maintaining normoglycaemia, intravenous access, fluid restriction, protein restriction due to poor renal function, feed intolerance and concerns regarding NEC. In addition, the infants in both groups who were given preterm formula would have received similar energy intakes. There were difficulties in increasing energy intake in preterm formula as breast milk fortifiers are not licensed for addition to formula.

There appeared to be an excess of pulmonary morbidity in the intervention group compared to the control group. The reason for this is unclear. It is possible that as more infants with lower gestation survived in the intervention group, this increased the overall days on respiratory support. When other factors were adjusted for, including chorioamnionitis, antenatal steroid and birth weight, the effect of randomisation on the days on respiratory support became insignificant. There have been concerns regarding possible harmful effects of intravenous lipids on pulmonary function.36 However, only infants above 28 weeks' gestation were included in the above study. The use of lipid emulsion at 4 g/kg/day appears to be well tolerated in preterm infants.37 There is insufficient evidence to suggest that higher lipid intake increases the risk of pulmonary morbidity or CLD. Studies have focused mainly on the early initiation of parenteral lipid and have not shown any detrimental effects.38 39

What is already known on this topic

  • Inadequate early nutrition often leads to poor post-natal growth in preterm infants.

  • Later cognitive ability is related to height and head circumference in children born preterm.

  • Post-natal growth failure often leads to small stature and head size at school age.

What this study adds

  • If adequate early nutrition can be achieved, post-natal growth failure is prevented.

  • Hyperalimentation was not associated with significant adverse effects, although there was a suggestion of longer mechanical ventilation.

The limitations of this study include its lack of blinding and that it is underpowered to show a significant difference in the OFC of our study population. It was not practical to blind this study as individual components of the PN had to be prescribed based on daily blood biochemistry, renal function and glucose homeostasis. We anticipated that more babies in the intervention group would require treatment with insulin. Breast milk when fortified has a shelf-life of less than 24 h.

Our sample size was calculated based on a previous audit which showed that 50% of infants <1500 g had OFC more than 2 SD below the mean at discharge. Only 15% of infants in our control group had an OFC more than 2 SD below the mean. Compared to the feeding regimen in 2003, energy and protein intakes were greater in our control group. Babies started PN on day 1 instead of day 3 of life. Current feeding regimen also included the use of minimal enteral nutrition (MEN) in the first few days of life. MEN has been shown to improve growth.40 This may explain the lack of statistical difference when comparing OFC outcomes in both groups.

We have shown a correlation between improved energy and protein intakes and OFC in extremely preterm infants. We found difficulties in increasing energy and protein intakes beyond that currently delivered. Cooperation with the infant feeding industry may be necessary to develop preterm formulae with higher energy and protein content that are safe for extremely preterm infants.



  • Competing interests: None declared.

  • Ethics approval: This study received approval from the local research ethics committee, Liverpool, UK.

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.

Linked Articles

  • Fantoms
    Martin Ward Platt