Does early nutrition in infants born before term programme later blood pressure?BMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6950.304 (Published 30 July 1994) Cite this as: BMJ 1994;309:304
- A Lucas,
- R Morley
- Correspondence to: Dr Lucas
- Accepted 20 May 1994
Objectives: To test whether nutrition early in20infants' development programmes later blood pressure and whether the reported relation between low birth weight and later high blood pressure is due to poor nutrition or growth before full term.
Design: Prospective randomisation of preterm infants to early diets differing greatly in nutrient content in four parallel multicentre trials, with blinded follow up 7.5-8 years later.
Setting: Neonatal units at Cambridge, Ipswich, King's Lynn, Norwich, and Sheffield.
Subjects: 758 children weighing under 1850 g at birth.
Main outcome measure: Blood pressure at age of 7.5-8 years.
Results: There were major differences in nutrient intake from randomised diets (preterm formula v standard formula and preterm formula v donor breast milk; in each case with or without mother's milk), but follow up showed no differences in later blood pressure. Individual subjects showed large variation in protein and energy intakes and in growth performance, including degrees of growth failure seldom seen in utero, but these factors were also unrelated to later blood pressure.
Conclusion: Extremes of nutritional intake and growth performance in preterm infants do not programme later blood pressure at 7.5-8 years of age. These findings do not support the hypothesis that high blood pressure has early nutritional origins. We suggest that the long term rise in blood pressure reported in individuals who had low birthweight (at full term) is not, as previously speculated, due to poor fetal nutrition or growth as such.
Epidemiological studies have shown that low birth weight, possibly indicating poor early nutrition, is associated with high blood pressure in later life
Infants born prematurely were given controlled diets that varied in nutritional content until the time when they would normally have been born
Infants showed extremes of nutrient intake and poor growth that would seldom occur in fetuses that were born at full term
There was no association between early nutrition or growth and blood pressure measured at 7.5-8 years of age
These results do not support nutritional intervention for pregnant women in order to prevent high blood pressure in their offspring
The possibility that nutrition in early life could influence propensity to adult disease1 is of great concern to public health. This idea originates from the more general concept in developmental biology recently redefined by Lucas as programming,2 the process whereby a stimulus or insult when applied at a critical or sensitive period of development results in a long term or permanent effect on the structure or function of the organism. Animal evidence for programming during critical periods of development dates back over 100 years3; that early nutrition could operate in this way1 is supported by numerous experimental studies in animals.2,4,5 Corresponding evidence in humans has been more difficult to obtain. Our approach is to conduct formal long term prospective outcome studies of individuals randomly assigned to their early diet.6,7 Recent retrospective epidemiological studies have linked possible markers of early nutrition, notably size at birth or in infancy, with cardiovascular disease or its risk factors in adult life.8,9 Several investigators have shown that babies born small have higher blood pressure in later life.9,10
Barker et al suggested that small size at birth reflects suboptimal fetal nutrition and argued that nutritional intervention during pregnancy might favourably influence later blood pressure.11 Since 1982 we have explored the hypothesis that early nutrition programmes later blood pressure in a large multicentre study of preterm infants born in the third or late second trimester. Their diets were strictly randomly assigned until they reached, on average, full term, and their blood pressure was measured at 7.5-8 years of age. This study allowed us to extricate the effects of nutrition before term from other factors that might have influenced later blood pressure had these infants remained in utero. Furthermore, when premature babies reach the equivalent of full term they have often developed severe growth failure,6 commonly compounded by preexisting growth retardation at birth,12 which would enable us to examine effects on later blood pressure of extremes of growth failure seldom seen in the fetus born at term.
Subjects and methods
We enrolled 926 infants with birth weights under 1850 g admitted to the neonatal units in Cambridge, Ipswich, King's Lynn, Norwich, and Sheffield between 1982 and 1985.6 Ethical approval was obtained in each centre. No parent refused consent, and only babies with major congenital anomalies were excluded.
Four parallel randomised trials were conducted. In three centres the infants of mothers who chose not to provide breast milk were randomly allocated to pasteurised banked breast milk donated by unrelated lactating women or to a nutrient enriched preterm formula (Farley's Osterprem, Crookes Health Care) as their sole diet (trial 1). The infants of mothers who provided breast milk (trial 2) were randomly allocated to banked breast milk or preterm formula as a supplement in volumes according to the mother's success in providing breast milk (median intake of mother's milk was 53% of total enteral feed volume (lower and upper quartiles 17% and 91%)). In the two remaining centres, infants were randomly allocated to a standard “term” formula (Farley's Ostermilk) or to the above preterm formula as sole diets (trial 3) or, when mothers provided their breast milk (trial 4), as supplements to mother's milk (median intake of breast milk 39% (5% and 76%)). The trials can be treated independently or combined (see below), still preserving randomisation. The infants were randomised to diets within 48 hours of birth by means of sealed envelopes as previously described.6
Table I shows the protein, energy, and sodium contents of the diets. Compared with standard formula, the preterm formula was also enriched in vitamins, zinc, and copper.13 The assigned diets were given until the infant weighed 2000 g or was discharged home, whichever was sooner. For infants fed maternal milk, protein and energy intakes were estimated from 4935 complete 24 hour collections of milk. Average nutrient values were obtained from 600 samples of pooled banked breast milk. In addition, 200 infants required a period of parenteral nutrition. Total protein and energy consumed from all sources was calculated (per kg body weight) and averaged over hospital stay.
Trained staff weighed the infants daily in hospital. To assess the relation between failure to gain weight and later blood pressure three further variables were calculated: weight gain (g/kg/day) on each diet after birth weight had been regained6; change in standard deviation of body weight from expected mean weight (derived from Gairdner-Pearson growth data14) between birth and hospital discharge (or discontinuation of total diet), incorporating our own birthweight centiles for preterm infants12 (a baby with a birth weight 1 SD below the mean and a discharge weight 3 SD below the mean would have a change in standard deviation of - 2 SD); and standard deviation of body weight from expected weight at hospital discharge, representing a summation of both intrauterine and extrauterine growth failure.
Extensive demographic and clinical data were collected from the 782 subjects who were followed up at 7.5-8 years of age, representing 99% of survivors still in Britain. Blood pressure was recorded in 758 of the subjects followed up and was measured three times. Since the first reading was significantly higher (by a mean of 4.4 mm Hg) than subsequent ones, the last two values were averaged for the analyses. An automated device (Accutor Datascope) was used to measure blood pressure in most subjects (553), but a conventional sphygmomanometer was used when two of the devices became unavailable. Small differences in values obtained were accounted for in the analyses. Study data were analysed with Student's t test and regression analysis.
Table II summarises demographic data. Birth weight (range 663-1847 g), gestation, social class,7 sex ratio, proportion requiring intravenous feeding, and days on assigned diets did not differ between randomised groups.
Analyses according to diet
Table III shows the main comparisons for the four parallel randomised trials. There were no differences in diastolic or systolic pressure in any trial. We also combined all four trials (preserving randomisation) to compare children fed low nutrient diets for preterm infants (donor milk or standard formula used alone or with mother's milk) with those fed the enriched preterm formula (alone or with mother's milk). There were no differences in blood pressure between these two groups.
Table IV shows that systolic blood pressure was unaffected by diet in major subgroups: boys, girls, children born small for their gestational age (birth weights <10th centile). Diastolic pressure was also unrelated to diet (results not shown). Early diet in babies with short gestation (>32 weeks) and prolonged hospital stay (28-120 days; median 42 days) did not affect later blood pressure.
The tendency to lower systolic values seen in the centres using banked breast milk reflected the use of non-automated devices to measure blood pressure in these centres (see methods section). This did not, however, affect comparisons within trials. Regression analysis with adjustment for the type of machine used did not reveal any dietary effect on blood pressure.
Epidemiological (non-randomised) analyses
Table V shows protein and energy intakes during hospital stay (median 30 days; maximum 120 days). There were considerable differences in macronutrient intake between individual infants, reflecting differences in randomised enteral diets and the variation in feed tolerance in these high risk infants. The infants also showed great variation in postnatal growth: 10% showed a fall in relative weight between 2.4 SD below expected birth weight and 4.1 SD below expected weight at discharge from hospital. Steady state weight gain (regressed best fit) after birth weight had been regained varied twofold between the 10th and 90th centiles, and the variation in the range was more than threefold. When the infants were discharged home close to term (mean 37 weeks' gestation) their weights showed extreme variation (full range 2.0 SD above to 5.4 SD below expected weight), reflecting the combined effects of in utero and postnatal (but preterm) growth.
The five intake and growth variables tested in table V showed no association with systolic and diastolic blood pressure at 7.5-8 years of age even after adjustment for possible confounding factors including gestation, size for gestation, device used to measure blood pressure, sickness variables (patent ductus and duration of mechanical ventilation), and anthropometric indices at 7.5-8 years (mid-arm circumference, height, and body mass index). Table VI shows this lack of association.
Our conclusions were unaffected when analyses were confined to a low risk group, which excluded 56 infants with neurological impairment and those who had required ventilatory assistance. In previous analyses we showed that body weight at 7.5-8 years was positively related to both blood pressure at this age and to early growth.15 However, adjustment for body weight at 7.5-8 years did not significantly alter our findings here.
Our data do not support the hypothesis that early nutrition programmes later blood pressure. The relation between small size at birth and later raised blood pressure observed by some investigators*RF 7-10* had suggested an adverse effect of poor fetal nutrition, and hence growth, on long term regulation of blood pressure. The public health and clinical implication of this, in terms of achieving optimal nutrition in the fetus, has stimulated large investment in research in fetal nutrition. Reduced fetal growth, however, might be associated with non-nutritional factors that could explain the later outcomes. In order to test whether poor nutrition or poor growth before full term are critical factors for long term programming of blood pressure it was important to find a model in which these factors could be extricated from other influences on the fetus (for example, those provided by a deranged intrauterine environment). Our study of preterm infants in many respects provides this opportunity, but we did not find that nutritional or growth deficits before full term, even when extreme, programmed blood pressure 7.5-8 years later.
Rationale of study
In our recent analysis of birth weights of babies born before 34 weeks' gestation from this cohort we found that reduced size for gestation was not associated with raised blood pressure at 7.5-8 years of age - indeed, we found the opposite trend.15 Therefore, we hypothesised that the critical window for programming blood pressure might be late in gestation, beyond the time of birth of these preterm infants. We tested this in three ways.
At follow up we examined the blood pressures of children who had been assigned to different diets between preterm birth and, on average, full term (>37 weeks after mother's previous menstruation). We believe that this large randomised study is unique. There were substantial differences between groups in intakes of energy, protein, sodium, calcium, phosphorus, iron, zinc, copper, and several micronutrients,13 which were reflected by significant differences in early growth.6 If any one of these dietary components (such as protein or sodium) programmed later blood pressure then differences between groups might have been expected, yet none was seen. Indeed, we found no effect of early sodium intake on blood pressure in these children at 18 months of age.16 There was also no difference in blood pressure between feed groups within large subpopulations previously shown to be most vulnerable to the effects of early diet,7 notably infants with retarded growth before birth and male infants.7,17 Earlier interim analysis had showed that boys in one subgroup (fed standard formula) had higher systolic blood pressure, but this was not confirmed here. In a balanced combination of our four trials, preserving randomisation and comparing low nutrient with high nutrient diets, we found no difference in blood pressure at 7.5-8 years despite the sample size (758 subjects) allowing us to detect a change of only 1.8 mm Hg in systolic pressure between groups at 5% significance and 80% power.
Protein and energy intake
We examined the two major components of nutrition, protein and energy intake, in relation to later blood pressure. Diets for the preterm infants differed greatly in protein and energy content (table I). Moreover, because of the difficulties feeding these infants, actual achieved protein and energy intakes in individuals averaged over hospital stay were even more variable, the lowest values resulting in changes in body composition seen in malnutrition.18 However, later blood pressures seemed unaffected by such large differences in intake.
We assessed growth from birth (mean 31 weeks' gestation) to attainment, on average, of full term. During this time body weight status in 20% of our cohort fell, in relation to growth charts for preterm infants, by between 2 SD and 4 SD, providing an opportunity to explore the long term consequences of a degree of growth failure rarely seen in utero. Even extreme growth failure was not associated with programmed changes in later blood pressure. Failure to gain body length, common in preterm infants,6 was also unrelated to later blood pressure (unpublished results). Furthermore, the standard deviation of body weight from mean expected weight for age at discharge from hospital (a measure of the combined impact of prenatal and postnatal growth to near term) varied greatly, from 2 SD above to over 5 SD below the mean. This massive range of growth performance did not seem to influence later blood pressure.
Comparison with other studies
We considered why our findings might not have supported previous suggestions of the early nutritional origins of high blood pressure11 based on its epidemiological association with low birth weight. Possibly 7.5- 8 years of age was too early to detect a dietary effect since early influences on later blood pressure may be amplified throughout life.19 Nevertheless, “tracking” for blood pressure has been observed even from infancy,20 and some studies have shown a relation between birth weight (in full term babies) and childhood blood pressure.9 Continued follow up of our cohort will resolve this. As in other studies, adjustment for current weight did not affect our findings.
Preterm infants might be demographically atypical, although social class distribution in our cohort (table II) was similar to that of the British population.21 Arguably, preterm infants might be intrinsically abnormal with an unusual pattern of blood pressure programming. However, if children with congenital defects are excluded (as was the case in our study) there is no evidence that premature babies are generally abnormal; rather, their postnatal illnesses usually reflect complications of immaturity. Moreover, our conclusions were unaffected even in infants who remained well (never needed respiratory assistance) and who were subsequently neurologically unimpaired. Also, we calculate that Whincup's values for systolic blood pressure (based on two readings) in children aged 5-7 born at term22 were the same as those obtained by us with our first two readings (mean 104 mm Hg in each case). The reasons for growth retardation here might differ from those in full term infants. However, we wished to test if early nutritional or growth deficits in themselves programmed blood pressure; in that respect our cohort was uniquely suitable.
In previous studies of birth weight and later blood pressure, growth to full term occurred in utero; in our study part of this growth occurred postnatally. As a result of some complex interaction, possibly involving the placenta,23 high blood pressure may be programmed only if poor nutrition and growth occur in utero. Alternatively, poor fetal growth may be associated with adverse non-nutritional factors such as hypoxia that could programme blood pressure. Regardless of mechanism, our data generate a new hypothesis, that preterm delivery of the growth retarded fetus (often practised when fetal morbidity is likely) could be beneficial by removing it from an adverse environment for blood pressure programming. Even if this is substantiated, however, the many risks of preterm delivery might outweigh any advantage.
This study does not negate the idea of nutritional programming. Indeed, our evidence from the same cohort of subjects shows that nutrition during a brief period before term may have profound long term consequences.2,7,17,24 Furthermore, our work does not challenge the observed association between birth weight (at full term) and later blood pressure, but it does not suggest that early growth failure or inadequate nutrition is in itself the explanation. If size at birth and later high blood pressure are causally related and not explained by unidentified confounding demographic factors then the wider search for aetiological clues must include further exploration of the relation between fetus and placenta and of the intrauterine environment found with retardation of fetal growth. However, in our view, it would be premature to devise nutritional interventions for mother or fetus in an attempt to prevent later hypertension.
We thank Dr Lister and Dr Leeson-Payne, who measured blood pressures in three of the centres; staff of the special care baby units at Cambridge, Ipswich, King's Lynn, Norwich, and Sheffield and Dr Bamford, Dr Dossetor, Dr Crowle, Dr Boon, and Dr Pearse for help and collaboration; Farley Health Products for financial assistance; the many schools which generously provided facilities; the children and their families who participated in this project; and Evelyn Smith for preparing the manuscript.