Neurodevelopment in Children Born Small for Gestational Age: A Randomized Trial of Nutrient-Enriched Versus Standard Formula and Comparison With a Reference Breastfed Group
Objective. Many studies have shown that children born small for gestational age (SGA) are at a neurodevelopmental disadvantage. We have shown that nutrient enrichment of formula fed to term SGA infants improves their growth and hypothesized that it also would improve their neurodevelopmental outcome.
Design. A randomized, controlled trial of standard term-infant (n = 147) or nutrient-enriched (n = 152) formula for the first 9 months. A reference group of 175 breastfed SGA infants was also recruited.
Setting. Subjects were recruited in 5 maternity hospitals in Cambridge, Nottingham, and Leicester, all in the United Kingdom.
Participants. Healthy, term infants (gestation: ≥37 weeks) with birth weight <10th centile.
Outcome Measures. Bayley mental and psychomotor scores at 18 months (primary) and developmental scores from Knobloch, Pasamanick, and Sherrard’s developmental screening inventory at 9 months (secondary).
Results. There was no significant intergroup difference in Bayley Mental Development Index (MDI) or Psychomotor Development Index (PDI) scores at 18 months. However, at 9 months, children fed the enriched formula had a significantly lower developmental quotient (99.5 vs 102.0; 95% confidence interval [CI] for difference: −4.6, −0.4). A significant disadvantage was seen in girls (−5.1; 95% CI: −7.8, −2.4) but not in boys (0.9; 95% CI: −2.4, 4.2).
Breastfed infants had significantly higher MDI and PDI scores at 18 months than formula-fed infants. Confounding factors accounted for ∼34% of the observed association between breastfeeding and MDI score and none of the association between breastfeeding and PDI score.
Conclusions. The previously reported enhanced linear growth in SGA children fed enriched formula was not matched by a neurodevelopmental advantage. At 9 months, girls fed the enriched formula had a significant developmental disadvantage, although this was not seen at 18 months. Later follow-up will determine any long-term effects on health or development. Meanwhile, use of enriched formula for term SGA children should not be promoted.
It seems that breastfeeding may be especially beneficial for neurodevelopment in children born SGA.
Many studies have shown that mean psychometric or academic test scores of children born small for gestational age (SGA) are lower than those of children with “appropriate” birth weight for gestational age. Among 8 different studies during the first 6 years of life, the majority demonstrated a disadvantage for those born SGA, although most of the studies were relatively small.1 Some studies have confirmed a similar disadvantage in late childhood or adolescence, when scores are more predictive of adult cognitive performance,2–7 although others did not.8–10 In a large British birth cohort study, subjects born SGA had significantly lower scores on a number of different cognitive tests at ages 8, 11, and 15 years. At age 26 they had poorer reading comprehension11 and were less likely to have professional or managerial jobs12 than those born with appropriate birth weight for gestational age. These differences remained after adjustment for social, demographic, and fetal or neonatal factors.
A suboptimal supply of nutrients in utero (whether because of suboptimal maternal nutrition or placental transfer of nutrients) is a cause of reduced fetal somatic and brain growth.13,14 In humans, the so-called critical brain growth spurt is between the third trimester and 2 years postterm13; thus, as Dobbing pointed out,15 this period represents an “opportunity actively to promote the proper growth of the human brain, by providing the best possible environmental conditions.”
The strategy of providing extra protein with only sufficient extra energy for protein utilization has already been adopted in the design of special formulas to improve growth rate in preterm infants both in hospital and postdischarge,16,17 and we now extend the use of the same formula, with almost 30% more protein than most standard formulas, for the nutrition of term SGA infants.
In this randomized trial in children born at term and SGA, we tested 2 primary hypotheses. The first was that children fed the nutrient-enriched formula would have improved growth. This proved to be the case; infants receiving enriched formula showed greater gains in length and head circumference than those fed term formula, not only during the 9-month intervention period but also up to 18 months.18
The other primary hypothesis was that children randomized to a nutrient-enriched formula from birth to 9 months would have Bayley Mental Development Index (MDI) and Psychomotor Development Index (PDI) scores one third of a standard deviation (SD) higher, at 18 months, than those fed a standard formula. Our secondary hypothesis was that children fed a nutrient-enriched formula would have one third of a SD advantage in developmental quotient at 9 months (score derived from the developmental screening inventory by Knobloch, Pasamanick, and Sherrard). Our findings are presented here.
We also studied a breastfed reference group and report associations between feeding practice and developmental scores.
SUBJECTS AND METHODS
Infants born in 1993, 1994, and 1995 were recruited into the randomized trial from 5 hospitals in Cambridge, Nottingham, and Leicester (all in the United Kingdom); in each, the study was approved by the local ethics committee. All infants were born at ≥37 weeks’ gestation and had a birth weight <10th centile for gestation and gender. This definition is similar to that used by the World Health Organization,19 although we used centiles from current United Kingdom growth charts.20 Infants with congenital abnormalities known to affect growth or development were ineligible. Mothers were approached only after they had unequivocally decided not to breastfeed and had commenced formula feeding.
A reference group of breastfed infants meeting the same birth-weight and gestation criteria was recruited, but infants were withdrawn if their mother stopped breastfeeding within 2 weeks of delivery. The duration of exclusive and partial breastfeeding was recorded.
Formula-fed infants were randomized to a standard versus nutrient-enriched formula. They commenced the trial diet within the first week of life, and it was supplied in powdered form until they were 9 months of age. The randomization schedule was prepared by a staff member who was not involved in any subsequent aspects of the study. It was generated in random permuted blocks and stratified by race (white/Asian) and birth-weight centile (<5th centile vs ≥5th centile for gestation and gender). The assignments were sealed in consecutively numbered, opaque envelopes.
The formulas were color-coded; the code was held by Farley Health Products and was not revealed to the investigators until after preliminary data analysis.
The nutrient-enriched formula (PremCare, Farley Health Products, a division of H.J. Heinz Co Ltd, Stockley Park, Uxbridge, United Kingdom) contained 301 kJ of energy per 100 mL, with 1.85 g of protein, 7.24 g of carbohydrate, and 3.96 g of fat. The enriched formula contained nearly 20% more protein in relation to energy than the standard formula (OsterMilk, also from Farley Health Products). This standard formula contained 284 kJ energy per 100 mL, with 1.45 g of protein, 6.96 g of carbohydrate, and 3.82 g of fat. Calcium, phosphorus, trace elements, and vitamins were also increased in the nutrient-enriched formula to support the projected increased growth performance. A full description of formula compositions is published elsewhere.18
We collected extensive sociodemographic data, as well as obstetric and neonatal data. Social class was coded into 6 categories by using the United Kingdom Registrar General’s Classification of Occupations and based on the occupation of the main family income earner. The mother’s education was coded according to categories published previously.21 This coding is based on educational attainment and, because there is a common public examination system in England, is likely to be more closely related to maternal intelligence quotient (IQ) scores than educational measures in most other countries.
The primary outcome measures, at 18 months, were the mental and motor scales of the Bayley Scales of Infant Development II, from which were derived the MDI and PDI.22 The MDI evaluates memory, habituation, problem solving, and language. The PDI tests control of gross muscle groups including movements associated with standing, walking, running, and jumping and also tests fine motor manipulations involved in prehension, adaptive use of writing implements, and imitation of hand movements. All items on both scales are scored on the basis of behavior elicited and observed by the assessor and not on the basis of parental report. The secondary outcome measure, at 9 months, was the developmental screening inventory of Knobloch et al.23 Using this, a developmental age is recorded for each of 5 subscales (adaptive, gross motor, fine motor, language, and personal-social). These ages were converted to quotients ([developmental age/chronological age] × 100). The overall developmental quotient was calculated as the mean of the 5 subscale quotients. This test was scored largely on the basis of assessor-observed behavior, although parental report was used for some items that were difficult to elicit at this age (eg, early language development).
Infant growth was also an outcome measure, and the results (reported elsewhere18) included weight, length, and head circumference at 9 and 18 months.
The primary hypothesis in respect to development was that there would be a one-third SD difference in Bayley MDI and PDI scores at 18 months between the 2 randomized groups. We calculated that, with 144 infants per group, we had 80% power at 5% significance to detect this difference, and a similar one-third SD difference in developmental scores at 9 months.
Mean scores in the randomized groups were compared by using the 2-sided Student’s t test.
Differences between the formula-fed and breastfed reference groups were compared by t and χ2 tests. Linear regression models were used to adjust for potentially confounding factors. Factors were retained in the model if their inclusion altered the regression coefficient for breastfeeding by ≥10%.24
Altogether, 299 infants were recruited into the randomized trial; 147 were allocated the standard formula (42% boys) and 152 the nutrient-supplemented formula (51% boys). There were 175 breastfed infants (53% boys). The trial profile, with the number of subjects seen at each stage and information about withdrawals, is shown in Fig 1. Baseline characteristics of the study groups are shown in Table 1.
There were no evident differences in baseline characteristics between children randomized to standard versus enriched formula (Table 1), although among the former group there were less boys (41.5% vs 50.7%). Values for baseline characteristics of children who were assessed at 9 or 18 months differed little from those shown.
There were no significant differences in the proportion of infants withdrawn from each formula group nor the numbers withdrawn by their parents or clinicians, and 18 of 147 (12%) control infants versus 25 of 152 (16%) of enriched formula-fed infants did not complete 9 months on the trial diet. Only 5 of the noncompliant subjects agreed to follow-up; 3 were in the control group and 2 were in the enriched-formula group. We followed up 125 (85%) of the control group and 121 (80%) of the enriched-formula group at 9 months and 122 (83%) and 118 (78%), respectively, at 18 months. In 5 of the children (all allocated the enriched formula and 4 of them boys), we were unable to complete the developmental assessment because they were uncooperative. Mean age (SD) at 9- and 18-month follow-up was the same in both groups (40  vs 40  weeks at 9 months and 80  vs 80  weeks at 18 months).
Mean values for developmental scores at 9 and 18 months, by randomized diet group, are shown in Table 2. At 18 months, mean MDI and PDI scores did not differ significantly between the 2 diet groups.
Mean overall developmental quotient at 9 months was significantly lower (by 2.5 [−4.6, −0.4] points) in the group fed nutrient-enriched formula, as were quotients on the gross motor and language subscales.
Findings at both 9 and 18 months were unchanged after exclusion of the 5 children (3 controls and 2 on enriched formula) who stopped having the assigned diet before 9 months but agreed to follow-up.
We tested for interaction between gender and diet to determine whether the influence of diet on developmental scores differed significantly according to gender (Table 3). In regression models with developmental scores as dependent variables and including gender, dietary allocation, and an interaction term (gender × diet), there was no evidence of significant statistical interaction between diet and gender for either Bayley MDI or PDI scores at 18 months. However, the interaction term was significant for overall quotient (P = .005) and gross motor (P = .03) and language (P = .003) subscale scores at 9 months of age. Girls fed the enriched formula had a significant 5.1-point disadvantage (−7.8, −2.4) in overall developmental quotient at 9 months, as well as significant disadvantages in all of the 5 individual subscales. After exclusion of children who did not remain on the assigned diet until 9 months, these findings were largely unchanged, although the disadvantage for girls fed enriched formula in the personal-social scale at 9 months was no longer statistically significant (95% confidence interval [CI]: −7.2, 0.1).
Comparison of Formula-Fed Infants With Breastfed Reference Group
Originally recruited as a reference group were 175 SGA breastfed infants, and 139 (79%) were seen at 18 months (Fig 1). Of these, 137 were developmentally assessed; 2 children (a boy and a girl) were uncooperative.
As expected, baseline characteristics of breastfed infants and their families differed from those of formula-fed infants (Table 1). Mothers of breastfed infants were more likely to have a university degree, come from higher social-class groups, and be hypertensive during this pregnancy and were significantly older and less likely to smoke. Mothers of breastfed infants had a greater head circumference and were taller than those of formula-fed children.
Mean values for developmental scores at 9 and 18 months in the breastfed group are shown in Table 2. Children who were breastfed had higher MDI and PDI scores at 18 months than those fed either formula. However, as indicated above and in Table 1, there were major sociodemographic differences between the breastfed and formula-fed groups that could account for these findings. We therefore constructed regression models to determine whether the significant difference between scores of breastfed versus formula-fed children was independent of potentially confounding factors that we measured. In these models we combined both formula groups, because their scores differed little at 18 months. Results are shown in Table 4. Without adjustment, breastfed children had an 11.7-point advantage in MDI score and a 6.0-point advantage in PDI score. Independent factors that were tested in the models were child’s gender and birth order, maternal age and education score, social class, maternal head circumference and height, and whether the mother smoked during her pregnancy.
We used the method advocated by Greenland24 and used in the final model only those factors that (if included) changed the regression coefficient by ≥10%. In the case of Bayley MDI, factors retained in the model were social class and maternal education, and there was a significant 7.7-point advantage for the breastfed children. If we included all the above-mentioned factors in the model, the regression coefficient for this advantage was 8.2.
None of the potentially confounding factors mentioned above changed the regression coefficient for the difference in Bayley PDI scores by ≥10%. Inclusion of all the above-mentioned factors in the model reduced the regression coefficient from 6.0 to 5.8 points.
There was no evidence of interaction between gender and whether the child was breastfed (in respect to MDI or PDI scores) either before or after adjustment as described above.
Information on duration of exclusive and partial breastfeeding was available for 93 of 137 (68%) breastfed subjects assessed at 18 months. The median duration of exclusive breastfeeding was 12 weeks (8, 16 [25th and 75th percentiles, respectively]) and of partial breastfeeding was 28 weeks (15, 51). Mothers for whom duration of breastfeeding data were available did not differ significantly from those for whom these data were not available. We found no evidence of an association between duration of exclusive or partial breastfeeding and developmental scores at 18 months.
At 9 months, breastfed children performed significantly better than those fed the enriched formula for each subscore and for overall quotient. They also performed better than those on the control formula in all but the gross motor and language scores. For the overall quotient, there was a 4.3-point (95% CI: 2.4, 6.3) advantage for breastfed versus control formula-fed children before adjustment; the advantage reducing to 3.7 points (1.3, 6.0) after adjustment for maternal education and to 3.8 points (1.5, 6.2) with all the above-mentioned factors in the model. In the comparison between breastfed and enriched formula-fed children, the unadjusted advantage in overall 9-month quotient for breastfed children was 6.8 points (5.0, 8.6), reducing to 5.8 points (3.8, 7.9) after adjustment for maternal education, and 6.1 points (3.9, 8.3) after adjustment for all the above-mentioned factors. There was no evidence of a diet-gender interaction in either case.
Randomized Trial of Nutrient-Enriched Versus Standard Formula
Children who were born SGA and fed the nutrient-enriched formula had no significant advantage in terms of the primary outcome measures, the Bayley MDI and PDI scores at 18 months, over those fed the standard formula. However, they did have a significantly lower mean developmental quotient (by 2.5 points [∼0.3 observed SD]), at 9 months than children fed the standard formula. Nine-month scores were a secondary outcome measure, and the developmental screening inventory of Knobloch et al23 was not designed or standardized as a robust research tool. Nevertheless, when trained observers tested children using this screening instrument, we found significant intergroup differences. There was a modest but significant correlation between 9-month overall quotient and Bayley MDI (correlation coefficient: 0.45; P < .0001) and PDI (correlation coefficient: 0.37; P < .0001) scores.
Our subgroup analyses according to gender were planned because of our finding that feeding preterm infants nutrient-enriched rather than standard formula in the neonatal unit conferred a long-term advantage in males but not females.25 We found no significant interaction between diet and gender in respect of Bayley MDI or PDI scores at 18 months (Table 3), but there was a significant interaction between gender and diet for the overall developmental quotient at 9 months. Girls fed the enriched formula had significantly lower scores in all subscales and in overall developmental quotient.
Infants receiving enriched formula had a greater gain in occipitofrontal head circumference by 9 and 18 months than those fed standard formula, and the differences were larger in girls,18 suggesting better brain growth in children fed the enriched formula, especially in girls. It was surprising therefore to find that girls born SGA at term who were fed the enriched formula had lower developmental scores at 9 months.
We do not understand the reason for this finding, and because we did not use a robust, validated test at this age, we do not wish to overstate its importance. Nevertheless, our findings are not without precedent. A previous randomized trial suggested that increasing protein intake in the neonatal unit could increase risk of neurodevelopmental disadvantage in low birth weight infants, particularly in girls.26 Goldman et al27 randomized 304 infants to diets containing either 2% or 4% protein, and 234 of 293 survivors were followed up at age 3. Those on the 4% protein milk had a significantly higher prevalence of strabismus, and among those children with birth weights <10th centile, 47% on the higher-protein diet versus 23% on the lower-protein diet had Stanford Binet IQ scores <90 at 3 years of age (P = .05).
Interestingly, 46% of boys on the higher-versus 41% on the lower-protein diet had IQ scores <90, whereas, in girls, 37% on the higher-versus 20% on the lower-protein diet had scores <90. However, this latter difference was not statistically significant. We calculated the proportion of children with Bayley MDI scores <90 in our study. Among girls, 56% of those on the enriched formula had MDI scores <90% vs 40% on the standard formula, but this difference was not statistically significant (95% CI for difference: −0.2; 34.0%). Among boys, 67% on enriched milk versus 63% on standard milk had low scores.
However, the diets in the 2 studies are not strictly comparable. The difference in protein content in our study (1.85% vs 1.45%) was not nearly as extreme as in the study by Goldman et al,27 and indeed both formulas in our study had a total protein content below that of the low-protein formula in the study by Goldman et al. Furthermore, in the earlier study, protein was increased without providing additional energy so it could be used,27 whereas, in our study, fat and carbohydrate content was also higher in the enriched formula.
More importantly, and regardless of whether our findings at 9 months are reliable, we have shown that neither boys nor girls born SGA at term derived a neurodevelopmental advantage from receiving a nutrient-enriched milk for the first 9 months.
Future follow-up of this cohort will determine whether there are any longer-term neurodevelopmental effects of feeding enriched formula to SGA infants.
Breastfeeding Versus Formula Feeding Comparison
There is now a large body of evidence demonstrating that breastfed children have a cognitive or developmental advantage over formula-fed children, but this is only the second study to investigate this issue in children born SGA, a group at increased risk of neurodevelopmental disadvantage.
Anderson et al28 recently reviewed studies, including ours in preterm infants,29 that could adjust for at least 5 potentially confounding factors and found that the advantage generally remained after such adjustment. There have also been 2 published studies involving subjects born at a time when infants from the least-advantaged strata of society were more likely to be breastfed. Hoeffer and Hardy30 studied children born in the United States between 1915 and 1921 and showed that breastfed children performed better than those who were fed artificially, although children who were exclusively breastfed beyond 9 months did less well. A study of elderly subjects born in the United Kingdom in the 1930s showed that, despite lower socioeconomic status, those who were breastfed had higher unadjusted IQ scores.31 Additional evidence comes from a large study in Hong Kong, recruiting children born between 1967 and 1976.
As we and others have pointed out,28,29,33 the remaining advantage after adjustment could be at least partly due to residual confounding, and we almost certainly failed to identify and accurately measure all confounding factors. In this study, adjustment for confounding factors that we measured accounted for ∼34% of the observed association between breastfeeding and MDI score. Thus in this group of children born SGA, residual confounding would have to account for 66% of the association between breastfeeding and MDI score. This is not impossible, but we consider it unlikely.
The advantage in terms of Bayley PDI scores was not at all reduced by similar adjustment. PDI scores were not significantly related to factors associated with maternal choice to breastfeed, thus we can be reasonably confident that this finding is not because of residual confounding. The only other study to report a motor advantage for breastfed infants was that of Ounsted et al,34 which, like ours, was a study of infants born SGA. They assessed 137 infants born SGA at 12 months of age and found that breastfed infants had significantly higher motor scores, whereas there was no difference in social scores.
We have shown that infants receiving enriched versus standard formula for the first 9 months of life showed greater gains in length and head circumference to both 9 and 18 months.18 Despite this improved head-circumference gain, there was no significant intergroup difference in Bayley MDI or PDI scores at 18 months, and at 9 months children fed the enriched formula had a significantly lower mean developmental quotient than those fed standard formula. The difference was especially marked in girls.
Additional evidence is needed before we can recommend feeding nutrient-enriched formula to children born SGA. Our findings here suggest that the enhancement in linear growth using enriched formula might be offset by reduced developmental scores at 9, although not 18, months. Later follow-up is needed to determine whether such an intervention would have a long-term effect on health or development.
Breastfeeding conferred a neurodevelopmental advantage, as well as an advantage in terms of growth.18 In this study of SGA children we demonstrated a substantial 6-point advantage in the Bayley motor scale as well an advantage in the mental scale. Breastfeeding is recommended for all infants, and it seems that it may be especially beneficial in those born SGA.
Farley Health Products, a division of H.J. Heinz Company Ltd (Stockley Park, Uxbridge, United Kingdom), contributed funding and supplied the trial formulas.
We thank the Research Staff who collected data in the study (Corina Adams, Helena Clements, Helen Clough, Ann Humphries, Catherine Leeson-Payne, Geraldine McHugh, Julie Owen, Mary Quinn, Dawn Rodd, and Emma Sutton) and the parents who allowed their infants to participate.
- Received December 16, 2002.
- Accepted June 4, 2003.
- Reprint requests to (R.M.) University of Melbourne Department of Paediatrics, Royal Children’s Hospital, Flemington Rd, Parkville, Victoria 3052, Australia. E-mail:
All authors were involved in study design; Drs Morley and Fewtrell analyzed the data; and all authors contributed to the final manuscript.
All the authors have at some time benefited from funding from the nutrition industry for research. All the authors other than Drs Abbott and MacFadyen also have benefited from funding for giving invited lectures. Dr Lucas has benefited from consultancies to the nutrition industry.
Dr Morley is now supported by VicHealth (the Victorian Health Promotion Foundation).
- ↵Grantham-McGregor SM. Small for gestational age, term babies, in the first six years of life. Eur J Clin Nutr.1998;52(suppl 1) :S59– S64
- ↵Westwood M, Kramer MS, Munz D, Lovett JM, Watters GV. Growth and development of full-term nonasphyxiated small-for-gestational-age newborns: follow-up through adolescence. Pediatrics.1983;71 :376– 382
- ↵Douglas JW, Gear R. Children of low birthweight in the 1946 national cohort. Behaviour and educational achievement in adolescence. Arch Dis Child.1976;51 :820– 827
- ↵Richards M, Hardy R, Kuh D, Wadsworth ME. Birth weight and cognitive function in the British 1946 birth cohort: longitudinal population based study. BMJ.2001;322 :199– 203
- ↵Dobbing J. Nutritional growth restriction and the nervous system. In: Davison AN, Thompson RHS, eds. The Molecular Basis of Neuropathology. London: Edward Arnold, 1981:221–223
- ↵Dobbing J. Nutrition and the developing brain. Lancet.1973;1(7793) :48
- ↵Lucas A, Gore SMG, Cole TJ, et al. A multicentre trial on the feeding of low birthweight infants: effects of diet on early growth. Arch Dis Child.1984;59 :722– 730
- ↵Lucas A, Fewtrell MS, Morley R, et al. Randomized trial of nutrient-enriched formula versus standard formula for postdischarge preterm infants. Pediatrics.2001;108 :703– 711
- ↵Fewtrell MS, Morley R, Abbott RA, et al. Catch-up growth in small-for-gestational-age term infants: a randomized trial. Am J Clin Nutr.2001;74 :516– 523
- ↵de Onis M, Habicht JP. Anthropometric reference data for international use: recommendations from a World Health Organization Expert Committee. Am J Clin Nutr.1996;64 :650– 658
- ↵Morley R, Cole TJ, Powell R, Lucas A. Mother’s choice to provide breast milk and developmental outcome. Arch Dis Child.1988;63 :1382– 1385
- ↵Bayley N. Bayley Scales of Infant Development. 2nd ed. San Antonio, TX: The Psychological Corporation; 1993
- ↵A Lucas, R Morley, TJ Cole. Randomised trial of early diet in preterm babies and later intelligence quotient. BMJ.1998;317 :1481– 1487
- ↵Anderson JW, Johnstone BM, Remley DT. Breast-feeding and cognitive development: a meta-analysis. Am J Clin Nutr.1999;70 :525– 535
- Golding J, Rogers IS, Emmett PM. Association between breast feeding, child development and behaviour. Early Hum Dev.1997;49(suppl) :S175– S184
- Copyright © 2004 by the American Academy of Pediatrics