点击显示 收起
1 From MRC Childhood Nutrition Research Centre, Institute of Child Health, London; the Menzies Centre for Population Health Research, Hobart, Tasmania, Australia; and the Clinical Epidemiology and Biostatistics Unit, University of Melbourne Department of Paediatrics.
2 Supported by Farley Health Products (a division of HJ Heinz Company Ltd), which supplied the trial formulas and provided financial assistance. 3 Reprints not available. Address correspondence to R Morley, Department of Paediatrics, Royal Children's Hospital, Flemington Road, Parkville, Victoria 3052, Australia. E-mail: morleyr{at}cryptic.rch.unimelb.edu.au.
ABSTRACT
Background: Preterm children are at high risk of poor growth performance. In 2 randomized trials, preterm infants fed preterm formula grew better in the neonatal period than those fed banked donor breast milk or standard term formula.
Objective: Our objective was to test the hypothesis that for preterm infants, the neonatal period is a critical one for programming growth performance and that early diet influences long-term growth.
Design: A total of 926 preterm infants were recruited into 2 parallel, randomized trials of neonatal diet. In trial 1, infants were fed either banked donor breast milk or preterm formula whereas in trial 2, infants were fed either standard term formula or preterm formula. Within each trial, the allocated milk was the sole diet for some infants (study A), whereas for others it was a supplement to maternal breast milk, given when not enough expressed breast milk was available (study B). We followed up 781 of 833 survivors (94%) to age 7.5–8 y. Trained assessors obtained anthropometric measurements according to a standard protocol.
Results: Despite significantly better neonatal growth performance in infants fed preterm formula (compared with either banked donor breast milk or standard formula), early diet had no influence on weight, height, head circumference, or skinfold thicknesses at 9 or 18 mo postterm or at age 7.5–8 y.
Conclusions: These findings suggest that the preterm period is not a critical window for nutritional programming of growth, which contrasts with evidence from these trials showing that early diet influences later neurodevelopment.
Key Words: Preterm infants • premature infants • infant nutrition • milk • human milk • breast milk • infant growth • growth • infant feeding • neonatal diet • preterm formula • term formula • standard infant formula
INTRODUCTION
Children who are preterm are at higher risk of growth deficits in infancy and throughout childhood than those born at term (1–3). In a recent epidemiologic study, low weight at 1 y of age was associated with increased risk of cardiovascular disease in adult life, suggesting that growth in infancy may be of lifelong significance (4).
Animal studies suggest that nutrition-related growth restriction during a critical window in infancy may have long-term so-called programming effects on body size (5–7). In humans, fetal growth is greatest in absolute terms during the last trimester of pregnancy, a period when preterm infants are no longer in utero. Growth of preterm infants who fail to thrive as neonates may therefore be subject to the same influences as that of children born growth-retarded at term. In a recent US study of children aged 2–47 mo, percentage body fat was higher in those born small for gestational age (<10th percentile; 8). Young adult offspring of mothers exposed to the Dutch famine during the third trimester of pregnancy had a lower than baseline prevalence of obesity (9). There is also evidence that higher adult waist-to-hip ratio (a risk factor for cardiovascular disease) is associated with lower birth weight for gestational age (10, 11). However, among 18-y-old male military conscripts in Sweden, risk of obesity rose with size for gestational age (12).
It is possible that nutritional interventions designed to optimize the growth of preterm infants in the neonatal period could permanently place these children on a more normal growth trajectory, thereby reducing the prevalence of growth deficits or unfavorable fat distribution later in life. We elected to test the hypothesis, in formal randomized trials, that the postnatal period for preterm infants (generally born in the third trimester) is a critical window for nutrition in terms of later growth and body proportions, and therefore later health. We have already shown that this same period in preterm infants can be critical for nutrition in terms of later neurodevelopment (13, 14). In 2 randomized trials, we examined the long-term impact of nutritional interventions during the neonatal period on the subsequent growth of preterm children.
SUBJECTS AND METHODS
Two randomized trials of neonatal diet in preterm infants were conducted in 5 centers in the United Kingdom; we enrolled 926 infants with birth weights <1850 g. The design of the trials is shown in Figure 1. The randomization sequences were prepared by using permuted blocks of random length and were held in numbered, sealed, opaque envelopes in each neonatal unit. Trial 1 was conducted in 3 centers with donor breast milk banks (Cambridge, Ipswich, and King's Lynn) and trial 2 was conducted in Norwich and Sheffield, centers without breast milk banks. These trials were approved by the ethics committees of all participating centers. The parents or guardians of the infants gave informed, written consent.
FIGURE 1. . Trial design. Randomization was stratified by birth weight. Infants that weighed 1200 g were randomly assigned separately.
In each trial, mothers were asked whether they wished to provide their own expressed breast milk for their infants. If a mother elected to provide her milk, her infant was assigned to receive the trial diet as a supplement, to ensure adequate enteral intake when the amount of expressed breast milk was not sufficient (study B). If the mother did not wish to express her milk, the infant was assigned to receive the trial diet as its sole milk source (study A).
In trial 1, the dietary comparison was between banked donor breast milk and a preterm formula designed to meet the nutritional needs of preterm infants (Osterprem; Farley Health Products, a division of HJ Heinz Company Ltd, Uxbridge, United Kingdom). In this trial, 159 subjects were randomly assigned a diet to be given as the sole milk source and 343 were randomly assigned a diet to be given as a supplement to mother's milk. In trial 2, the comparison was between standard term formula (Osterfeed, now Ostermilk; Farley Health Products, a division of HJ Heinz Company Ltd) and the preterm formula described above, with 160 subjects in the sole diet group and 264 in the supplement group. The contents of major nutrients in these diets are shown in Table 1. Values for the infants' mothers' expressed breast milk were derived from analyses of 24-h milk samples; those for donor breast milk were from pooled samples. Mean nutrient contents per 100 mL, for expressed maternal milk and donor milk, respectively, were 1.5 and 1.1 g protein, 3.0 and 1.7 g fat, 7.0 and 7.1 g carbohydrate, 260 and 193 kJ, 23 and 16 mg Na, and 35 mg Ca and 15 mg P in both types of milk.
View this table:
TABLE 1.. Major nutrient constituents of the trial formulas per 100 mL
Infants in both trials were fed the assigned diet until they reached a weight of 2000 g or were discharged from the neonatal unit, whichever occurred first. After that time, the study had no further influence on the infants' diets; they were fed as their parents and medical advisors chose.
Extensive social, demographic, antenatal, and perinatal data, and detailed data on the neonatal clinical course were recorded by trained research nurses (15). Social class was coded into 6 categories by using the United Kingdom Registrar General's classification. The nurses also measured neonatal growth by using electronic balances (Sartorius MP, Edgewood, NY) to weigh the infants daily to ±1 g. Length was measured with neonatometers (Holtain, Crymych, United Kingdom) and head circumference was determined with paper tapes; both were measured twice weekly to the next succeeding millimeter. Calipers (Holtain) were used to measure skinfold thicknesses at all ages.
Children enrolled in trial 1 were followed up at 9 and 18 mo postterm and at 7.5 y (up to 8 y if they were not available for study at 7.5 y). Subjects in trial 2 were seen at 18 mo postterm and 7.5–8 y. Weight was measured to the nearest 10 g with digital baby scales (Seca, North Bend, WA) for the early assessments and to the nearest 100 g with digital scales (Soehnle, Murrhardt, Germany) at 7.5 y. Length was measured at 9 and 18 mo with a stadiometer (Holtain) and height at 7.5–8 y was also measured with a stadiometer (Harpenden stadiometer; Holtain). At all ages, head circumference was measured with a paper tape. All these measurements were made to the next succeeding millimeter. Head circumference was measured 3 times and the largest of the 3 values was used in the analyses.
Waist and hip circumferences were measured at 7.5–8 y with the same tape; the former was measured at the minimum circumference between the iliac crest and rib cage and the latter was measured at the maximum width over the greater trochanters.
Gains in weight, length, and head circumference in the neonatal unit were calculated by linear regression of all available measurements made over 2 wk after the infant regained its birth weight; this date was defined as the first of 3 consecutive days when weight was above the birth weight (15). Body mass index (BMI) was calculated as weight in kg divided by length or height in m squared, and the waist-to-hip ratio was calculated as the waist circumference divided by the hip circumference.
Student's t test was used to compare group mean measurements after we determined that the data for each outcome measure were approximately normally distributed. We also performed regression analyses, adjusting the results for sex and for age at the time when measurements were made. Data from studies A and B (the sole diet and supplement groups, respectively) were analyzed separately. We also analyzed data from studies A and B combined for each trial. Regression models including the interaction term "study x diet" were used to check whether the effect of diet on each outcome measure was significantly different in studies A and B. Regression models including the interaction term "sex x diet" were used to check whether there was an interaction between the sex of the infant and randomized diet with respect to all of the outcome measures.
RESULTS
In trial 1, 400 of 459 surviving subjects (87%) were seen at 9 mo postterm, at a median age of 40.4 wk (25th and 75th percentiles, 39.6 and 41.9 wk), and 438 of 457 (96%) were seen at 18 mo postterm, at a median age of 80.7 wk (25th and 75th percentiles, 79.4 and 83.6 wk). A total of 421 of 456 survivors aged 7.5–8 y (92%) were seen at a median age of 7.56 y (25th and 75th percentiles, 7.51 and 7.64 y). Of those not seen, 4 (from traveling families) were not traced, 6 refused follow-up, and 24 had moved overseas. Thus, of 432 subjects still residing in the United Kingdom, 421 (97%) were seen. Two children had severe cerebral palsy and we could not obtain weight or length measurements.
In trial 2, 334 of 377 surviving subjects (89%) were seen at 18 mo postterm at a median age of 82.6 wk (25th and 75th percentiles, 80.1 and 88.3 wk). At 7.5–8 y, 360 of 377 subjects (95%) were seen at a median age of 7.63 y (25th and 75th percentiles, 7.53 and 7.87 y); 6 refused and 11 were overseas. Of the 360 children seen (98% of the 366 subjects still residing in the United Kingdom), one child with severe cerebral palsy was not measured. Children not seen at 9 and 18 mo are described elsewhere (13, 16, 17), as are the demographic characteristics of children measured at those ages. Characteristics of children measured at 7.5–8 y are shown in Tables 2 and 3.
View this table:
TABLE 2.. Characteristics of children measured at 7.5–8 y of age in trial 1
View this table:
TABLE 3.. Characteristics of children measured at 7.5–8 y of age in trial 2
As described previously (15) and shown in Table 4, infants in trial 1 that were fed preterm formula either as their sole diet or as a supplement to mother's milk (studies A and B combined) had significantly faster gains in weight, length, head circumference, and both subscapular and triceps skinfold thicknesses than those fed donor breast milk. In trial 2, which compared standard term formula with preterm formula as the sole diets or supplements, infants fed the preterm formula had significantly faster gains in weight, head circumference, and subscapular skinfold thickness (Table 5) (13). In trial 1, the weight gain advantage of infants fed preterm formula was significantly greater in study A than in study B (P < 0.002 for the interaction). Similarly, in trial 2, preterm formula was associated with a significantly greater advantage in terms of head circumference gain in study A than in study B.
View this table:
TABLE 4.. Growth in the neonatal period according to randomized diet for children measured at 7.5–8 y of age in trial 11
View this table:
TABLE 5.. Growth in the neonatal period according to randomized diet for children measured at 7.5–8 y of age in trial 21
We looked for evidence that the effect of diet on neonatal growth was different in boys than in girls. In trial 1, the advantage in terms of neonatal weight gain that was associated with preterm formula as the sole diet (study A) was significantly greater in boys than in girls (5.1 compared with 1.9 g•kg-1•d-1, respectively; P < 0.03 for the interaction). For all the other comparisons shown in Tables 3 and 4, there was no significant difference between boys and girls with regard to the effect of diet on growth.
In contrast, there were no significant differences in anthropometric measures or in BMI in either of the trials at any of the follow-up points (Tables 6 and 7), except for the significantly lower waist-to-hip ratio in children fed preterm rather than term formula as the sole diet in trial 2 (0.86 ± 0.05 compared with 0.89 ± 0.07; 95% CI for difference: 0.006, 0.046; P < 0.01). This significant association remained after adjustment for sex and age at the time of measurement. Such adjustment did not reveal associations between any other anthropometric variables and randomized diet at any of the 3 ages when children were seen. We also looked for evidence that the effect of diet group on each measure was different in boys and girls, but found no evidence of interaction in any of the comparisons shown in Tables 6 and 7. We performed further exploratory analyses that included only children without cerebral palsy (because of the known growth problems in cerebral palsy) or included only children who received most of their intake as the assigned diet. The latter was defined as receiving full enteral feedings for 2 wk or, in the supplement group, receiving <50% of the diet as maternal milk. In all of these analyses (data not shown), no significant associations were found between early diet and any other measures of later growth or body proportions.
View this table:
TABLE 6.. Anthropometry at follow-up according to randomized diet in trial 11
View this table:
TABLE 7.. Anthropometry at follow-up according to randomized diet in trial 2
DISCUSSION
We found that growth performance in the neonatal unit was significantly better in infants fed preterm formula rather than either banked donor breast milk (trial 1) or standard term formula (trial 2). Differences were seen in the studies that compared these diets as sole diets and as supplements to mother's milk. The subjects in our study were a relatively unselected population, and maternal success in providing breast milk was variable. The analysis of combined data from the sole diet and supplement studies (A and B) indicates the overall impact in these nurseries of using preterm formula, rather than donor breast milk or standard formula, as a back-up diet (when the mother cannot supply any or enough milk).
Despite the effect of diet on neonatal growth, differences between the diet groups in weight, length or height, head circumference, and skinfold thicknesses had disappeared by 9 mo postterm and no differences emerged by 7.5–8 y. Likewise, no between-group difference in BMI was seen at any follow-up examination. However, we found that in trial 2, children fed preterm rather than term formula as their sole neonatal enteral diet had significantly lower waist-to-hip ratios. Although this finding would be consistent with data showing that adult subjects born small for gestational age have higher waist-to-hip ratios, there was no consistent trend across the trials that would indicate a benefit for the group fed nutrient-enriched preterm formula. This suggests that this was a chance finding, especially because there were multiple comparisons and this was a post hoc analysis, unlike the comparisons of weight, height, head circumference, and skinfold thicknesses, which were planned a priori. However, this study compared 2 diets which differed only in nutrient content, whereas in all of the other diet comparisons some children received breast milk, either from their own mother or from a donor. It is therefore possible that breast milk in the diet in some way modified the effect of nutrition in infancy on later body fat distribution. This needs to be tested prospectively in another cohort.
Thus, although our previous studies clearly showed that the early postnatal period in preterm infants is a sensitive one for nutrition in terms of later neurodevelopment (13, 14, 16, 17), this same period was not shown to be a critical one for nutritional programming of long-term growth or body fatness.
Some evidence suggests that early life events may program later growth in humans. For instance, although children born growth retarded often exhibit catch-up growth, a long-term deficit was reported in many studies (18–21) and preterm infants as a population show reduced long-term growth (1–3). We hypothesized that there would be a nutritional basis for the long-term growth deficit generally seen in small-for-gestational-age infants born at term (often growth retarded during the last trimester in utero) and in preterm infants in general (ex utero in the last trimester). Given our data indicating the marked effect of early diet on growth in the neonatal period, it is surprising that we found no lasting differences in body size or fatness. The randomized diets differed substantially in protein, energy, and a wide range of mineral and micronutrient contents (Table 1) and, in the case of breast milk compared with formula, also differed in a wide variety of nonnutrient factors. If suboptimal nutrition in general, or the intake of any of these specific nutrient or nonnutrient factors, in the preterm period were important for programming long-term growth in humans, we would have expected to find that the difference in body size between randomized diet groups in the neonatal period persisted into childhood.
Most interestingly, early nutrition had no effect on later body fatness, as measured by BMI, during childhood. There is much interest in the possible early nutritional origins of obesity. For instance, low birth weight, which might reflect impaired fetal nutrition, has been shown to be inversely related to body fatness in childhood and adult life (10, 22). In contrast, preterm infants as a population are noted to be thin. When our data from preterm children at 18 mo were compared with unpublished data (A Lucas and R Morley, 1999) from 779 term children, mean BMI in the preterm children was significantly lower (16.0 ± 1.5 compared with 16.7 ± 1.4; P < 0.0001). However, data from baboons (7) show that differences can emerge in adult life. Continued follow-up will be needed to determine whether our early nutritional interventions influenced the propensity to obesity in adulthood.
It is possible that the duration of dietary manipulation in the present study (4 wk on average), although long enough to induce effects on long-term neurodevelopment, was not long enough to have a lasting effect on growth. We are now exploring the possibility that a more prolonged period of nutritional intervention, extending throughout the first 9 mo postterm, could program long-term growth. The study is a formal, randomized intervention in preterm children and in full-term children born growth retarded. Preliminary unpublished data (MS Fewtrell, R Morley, and A Lucas, 1999) suggest that nutrition after hospital discharge may indeed influence growth to 18 mo postterm. This suggests that in humans, as in experimental animals (6), the period after full term is a more critical one than the period before full term for nutritional programming of growth.
As in other studies of preterm children, the children in our study, as a group, failed to catch up with children born at term. At 7.5–8 y, mean SD scores (±SD) when compared with recent British norms (23) were as follows: for boys, -0.67 ± 1.24 for weight and -0.45 ± 1.09 for height, and for girls, -0.67 ± 1.24 for weight and -0.50 ± 1.13 for height. From a biological point of view, the explanation for long-term effects on growth or body fatness in individuals who were growth-retarded in utero or born before term may not simply relate to nutrition. Further fundamental research is needed here.
From a clinical viewpoint, it might seem reassuring that the poor neonatal growth that frequently results from suboptimal early nutrition might resolve during infancy. Nevertheless, our previous observation that improved nutrition in this same period may have beneficial effects on long-term development (14) remains an important incentive to pay close attention to nutrition in hospitalized preterm infants.
ACKNOWLEDGMENTS
We thank the staff of the neonatal units in Cambridge, Ipswich, Kings Lynn, Norwich, and Sheffield; the children who took part in this study and their families; and the schools that provided the facilities.
REFERENCES