点击显示 收起
1 From the Department of Social Medicine, University of Bristol, Bristol, United Kingdom (RMM, AM, GDS, and YB-S), and the Department of Child Health, University of Wales College of Medicine, Cardiff, United Kingdom (DPD).
2 RMM was supported by the Wellcome Trust, and the follow-up portion of the Barry Caerphilly cohort was supported by grants from the British Heart Foundation and Diabetes UK. 3 Address reprint requests to RM Martin, Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, United Kingdom. E-mail: richard.martin{at}bristol.ac.uk.
ABSTRACT
Background: Evidence suggests that environmental factors acting early in life may affect blood pressure in adulthood.
Objective: The objective was to test the hypothesis that dried formula milk (derived from cow milk) intake in infancy is positively associated with blood pressure in early adulthood.
Design: We conducted a long-term follow-up (1997–1999) of the Barry Caerphilly Growth study cohort (1972–1974) into which mothers and their offspring had originally been randomly assigned to receive a milk supplement or usual care. Participants were the offspring, who were aged 23–27 y at follow-up. The main outcome measures were systolic and diastolic blood pressure.
Results: The social and demographic characteristics of the subjects who were (n = 679) and were not (n = 272) followed up were similar. For each increase in quartile of dried milk consumption (in oz) at 3 mo of age, there was a 1.28-mm Hg (95% CI: 0.46, 2.10 mm Hg) increase in systolic and a 0.63-mm Hg (95% CI: 0.04, 1.22 mm Hg) increase in diastolic blood pressure after adjustment for sex, intervention group, birth weight z scores, social class in childhood, age at follow-up, alcohol consumption, and pack-years of smoking. These coefficients were attenuated when adult body mass index and height were included in the models, but the association of dried milk consumption at 3 mo of age with systolic pressure remained significant (1.07 mm Hg; 95% CI: 0.27, 1.87 mm Hg).
Conclusions: Our findings are consistent with the hypothesis that high blood pressure in later life is influenced by early postnatal nutrition. Thus, interventions to optimize infant nutrition may have important long-term health benefits.
Key Words: Infant nutrition • formula feeding • blood pressure • hypertension • cardiovascular disease • cohort studies
INTRODUCTION
Elevated blood pressure is an established determinant of premature cardiovascular disease (1). Several factors are associated with blood pressure, including body mass (2) and diet (3) in adulthood, but a growing body of evidence suggests that environmental influences acting much earlier in life may affect future blood pressure (4). The role of prenatal factors is underlined by inverse associations of birth weight with blood pressure later in life (5). It is hypothesized that intrauterine factors operating during fetal development program permanent changes in the structure or function of developing organs and the vascular tree (6). Others emphasize the importance of postnatal exposures such as childhood diet (7), nutritional status (4, 8–11), and social circumstances (12, 13). Some studies have shown associations of birth weight with blood pressure later in life, which often get stronger or change direction after adjustment for current weight or body mass index (BMI) (14). Because early size adjusted for later size could be interpreted as a measure of change in size over time, the association between birth weight and blood pressure may arise from postnatal nutritional influences (15). An influence of postnatal exposures is further suggested by cohort studies examining associations between ischemic heart disease mortality and height, where height is used as a marker of nutritional status and adverse exposures during childhood (16). These studies show inverse associations between cardiovascular disease mortality and stature, which are independent of birth weight (17, 18). A substantial downward trend in blood pressure levels was observed among young adults attending Glasgow University between 1948 and 1968 (19), supporting the importance of early-life factors on future blood pressure levels.
Two sources of experimental evidence in humans suggest that infant nutrition influences blood pressure later in life. First, preterm infants randomly assigned to be fed donor human milk had lower systolic (by 2.7 mm Hg) and diastolic (by 3.2 mm Hg) blood pressure at ages 13–16 y than did infants fed preterm formula (20), but the relevance of these results to term babies is unclear (21). Second, systolic blood pressure in infants randomly assigned to be fed a low-sodium diet was 2.1 mm Hg lower than in control subjects fed a normal-sodium diet at 6 mo of age (22). This effect persisted into adolescence (23). Observational data, however, are equivocal. One study found an inverse relation between breastfeeding and systolic blood pressure in childhood and adolescence; the mean blood pressure was 4.5 mm Hg lower in females and 6.5 mm Hg lower in males breastfed for > 3 mo than in infants not breastfed (24). Another study found that systolic blood pressure at 7 y of age was 3.9 mm Hg higher in those who were exclusively formula-fed than in those who were breastfed as infants (25). Others showed no influence of infant feeding on blood pressure later in life (4, 8, 26–28).
Our aim was to determine whether dried milk intake in infancy positively influences blood pressure in early adulthood in a prospective cohort study. The public health importance of this question arises from the possibility that knowledge of any specific postnatal nutritional influence on adult health may provide an opportunity for intervention during infancy (15).
SUBJECTS AND METHODS
The Barry Caerphilly Growth study
The MRC Epidemiology Unit (South Wales) undertook a population-based randomized controlled trial to examine the effects of milk tokens [equivalent to the price of half a pint (284 mL) of milk per day and acceptable to milk delivery men] given to pregnant women and their offspring on birth weight and childhood growth up to 5 y of age (29). The study involved 1163 consecutively born infants from the towns of Barry and Caerphilly in South Wales between March 1972 and October 1974 (the Barry Caerphilly Growth study). The provision of milk tokens led to a nonsignificant increase in birth weight compared with unsupplemented control subjects (in keeping with the results of other studies of this issue in similar populations) and had no effect on height, weight, or skinfold thickness at 5 y (29). The study population can, therefore, be treated as a single cohort for the present purposes (30).
Of the 1163 children born in the trial, 951 (82%) completed the first 5 y of follow-up, which ended in 1979 (29). Detailed data were collected prospectively by nurses who visited the infants in their home at frequent intervals up to 5 y of age. The birth weight and gestational age of the children were extracted from hospital records. There were no missing data on birth weight, but data on gestational age were missing for 45 subjects. Where there were missing data or extreme values for gestational age, we calculated gestational age from the date of the last menstrual period, which had also been extracted from hospital records. At the home visits at 10 d, 6 wk, and 3, 6, 9, and 12 mo of age, nurses recorded the children’s total dried formula milk consumption per day (in oz), whether (yes or no) the children were mixed feeding (ie, solid foods had been introduced), the age at which mixed feeding started, and the children’s cow milk consumption per day (31). Dried milk was defined as artificial foods in powdered form that were reconstituted by dilution with water to prepare a formula feed for the infants. In the early 1970s, dried milk was derived from full-fat cow milk (32). Information about breastfeeding was not collected. Data on the social class of the children’s fathers were available when the children were 18 mo of age.
Follow-up study
Subjects
The present analysis is based on the follow-up in early adulthood of the 951 subjects who completed the original 5-y study period (29). Between 1997 and 1999, we attempted to contact all subjects who had completed the original study, at which time the subjects were aged 25 y on average (range: 23–27 y). The subjects were traced either through their parents or by invitation from the relevant health authority after the subjects had been flagged through the National Health Service Central Register. A total of 679 of 951 (71%) subjects were successfully followed up. Ethical approval was given by the Bro Taf Health Authority Local Research Ethics Committee.
Measurements
At follow-up in adulthood, several ischemic heart disease risk factors and anthropometric variables were measured (30). Two blood pressure and heart rate measurements were made on the left arm after the subjects had rested supine for ≥ 5 min (before and between readings). The measurements were made with an automated oscillometric monitor (Omron HEM-705CP; Omron Matsusaka Co, Japan) that had been calibrated at the start of the study and then after approximately every 300 patients. Height was measured twice to the last complete millimeter with the use of a portable stadiometer. Body weight was measured twice to the last complete 1.0 kg while the participants were wearing a hospital gown. Participants filled in a questionnaire, including detailed questions on their lifetime smoking habits and alcohol consumption.
Statistical analysis
The baseline characteristics (sex, birth weight, gestational age, dried milk consumption, and father’s social class) of the subjects who were followed up and included in the final data set were compared with the characteristics of all remaining subjects (ie, those not followed up or not in the final data set) by using a two-sample t test or Wilcoxon’s rank-sum test for continuous variables and a chi-square test for categorical variables. STATA (release 7.0; Stata Corp, College Station, TX) was used for the analyses.
For the analysis of the association of infant feeding with blood pressure later in life, the total dried milk consumption per day (in oz) at 10 d, 6 wk, and 3 mo was recoded into quartiles of this variable. With the use of data from the original study records, internally sex-standardized z scores for birth weight were calculated within weeks of gestation (36–44 wk). The z score indicates the number of SDs each subject’s measurement of birth weight is from the mean, adjusted for sex and gestational age. This enables a comparison across groups independent of sex and gestational age.
At follow-up in adulthood, weekly units of alcohol intake were derived from detailed questions about weekday and weekend consumption of beer, spirits, sherry, and wine. Pack-years of smoking were derived from detailed questions about current and previous smoking habits. Systolic and diastolic blood pressure, heart rate, body weight, and height were based on the average of 2 readings. BMI [weight (kg)/height2 (m)] was computed for inclusion in the models as a measure of weight independent of height at 25 y of age.
Univariable associations of infant feeding at age 10 d, 6 wk, and 3 mo with height, body mass index, and systolic and diastolic blood pressure were investigated by using one-factor analysis of variance. This provides a variable estimate (mean and SD) for each quartile of dried milk quantity and an overall P value for the comparison of means. Associations of infant feeding with these outcomes were then investigated by using multiple linear regression. Quartiles of dried milk quantity (in oz) were entered into the models as a continuous variable, and the assumption of linearity tested by comparing this model with a model in which quartiles of dried milk quantity was entered as an ordinal variable. Assumptions of normality underlying these models were checked by investigating the distribution of residuals by fitted values and examination of the normal probability plot. One previous study has suggested that associations of infant feeding with blood pressure are stronger in those with a high heart rate (23). Sex-related differences have also been noted in associations of early feeding and blood pressure later in life (24, 33, 34). Overweight and obesity have been associated with infant feeding (35), and we hypothesized that associations of infant feeding with blood pressure may vary according to current BMI. Likelihood ratio tests were used to assess effect modification between heart rate (divided into a binary variable above and below the mean) and infant feeding, sex and infant feeding, and current BMI and infant feeding on blood pressure. All multiply adjusted associations included terms for age at follow-up, sex, and intervention group (ie, subjects supplemented with milk tokens or control subjects in the original trial). Additional models were developed that also adjusted for 1) birth weight z score; 2) birth weight z score plus alcohol consumption, smoking status, and father’s social class during childhood; and 3) the above potential confounding factors plus current BMI and height. We investigated the relative importance of early and later dried milk intake by using the analytic strategy proposed by Lucas et al (15). Regression models were developed that 1) simultaneously included intake at 10 d (early intake) and 3 mo (later intake) and 2) included the interaction of early and later feeding (calculated as the product of quartiles of milk intake at 10 d and 3 mo) to the combined model. To assess the effect of weight gain in early infancy, we adjusted associations of dried milk intake with blood pressure separately and in combination for weight at 6 wk and 3 mo and for rate of weight gain between 6 wk and 3 mo (calculated as the weight at 3 mo minus the weight at 6 wk divided by the time between examinations and multiplied by 4.36 to express weight gain in grams per month).
The analysis was restricted to the 656 of 679 adults who were followed up and who were born at term (defined as those born after the 35th and before the 45th gestational week). Furthermore, all models were computed in subjects for whom a complete data set was available at each age at which dried milk consumption was measured. Thus, associations of dried milk consumption at 10 d, 6 wk, and 3 mo with outcomes in early adulthood were investigated in 504 (74%), 544 (80%), and 560 (83%) subjects, respectively. Models that simultaneously adjusted for dried milk intake at both 10 d and 3 mo were based on 478 subjects. Associations of dried milk intake at 3 mo with blood pressure, adjusted for weight gain in early infancy, were based on 548 subjects.
RESULTS
The social and demographic characteristics of the subjects who were (n = 679) and were not (n = 272) followed up were similar (Table 1). At 10 d of age, 2% of the study participants received no dried milk, 6% received 284 mL dried milk, and 45% received 568 mL dried milk. Only 36 (5%) infants were given unmodified cow milk at 3 mo of age. As noted previously (31), a large proportion of infants (454/679; 67%) had started mixed feeding at or before 6 wk, which is in line with patterns of infant feeding in the United Kingdom (32).
View this table:
TABLE 1 . Comparison of subjects who were and were not followed up, by baseline characteristics: Barry and Caerphilly Growth study (1972–1974)1
The distribution of baseline characteristics measured during childhood were compared according to the quantity of dried milk consumed (Table 2). There was a positive association between quantity of dried milk at 10 d and birth weight and a weaker positive trend at 6 wk (P = 0.06). This association disappeared by 3 mo. There was a weak trend (P = 0.07) that, at 10 d, infants from lower social classes consumed less dried milk than did those from higher social classes. At 6 wk, but not at 10 d or 3 mo, males consumed more dried milk than did females. There was little evidence that dried milk consumption in infancy was associated with intervention group (Table 2), mean age at follow up, alcohol consumption, or cigarette consumption (Table 3).
View this table:
TABLE 2 . Baseline characteristics of the subjects included in the study, by quantity of reconstituted dried milk consumed at 10 d, 6 wk, and 3 mo of age
View this table:
TABLE 3 . Sociodemographic characteristics of the subjects at follow-up (1997–1999), by quantity of reconstituted dried milk consumed at 10 d, 6 wk, and 3 mo of age
The quantity of dried milk consumed at 10 d and 6 wk was positively associated with height in adulthood and at 3 mo it was positively associated with BMI (Table 4). There was a clear dose-response relation between dried milk consumption at 3 mo and systolic blood pressure. Dried milk consumption at 3 mo also tended to be positively associated with diastolic blood pressure (P = 0.06).
View this table:
TABLE 4 . Association of quartiles of total reconstituted dried milk consumption per day at 10 d, 6 wk, and 3 mo of age in relation to height, weight, and systolic and diastolic blood pressures in later life (ages 23–27 y)
The association of dried milk consumption with height was attenuated after adjustment for intervention group, age at follow-up, and sex, and disappeared after adjustment for birth weight (Table 5). There was strong evidence for a positive association of dried milk consumption at 3 mo with BMI, which was little affected by adjustment for other factors. In the fully adjusted model, BMI increased by 0.48 (95% CI: 0.12, 0.84) per unit increase in quartile of dried milk.
View this table:
TABLE 5 . Multiple linear regression coefficients (and 95% CIs) showing unit increases in height, weight, and systolic and diastolic blood pressures at follow-up (ages 23–27 y) per increase in quartile of total reconstituted dried milk consumption per day at 10 d, 6 wk, and 3 mo of age1
In the fully adjusted models, there was weak evidence of an inverse association of dried milk consumption at 10 d with systolic blood pressure later in life. There was a strong positive association of dried milk consumption at 3 mo and systolic blood pressure in early adulthood (regression coefficient: 1.31 mm Hg; 95% CI: 0.49, 2.14 mm Hg) after adjustment for intervention group, age at follow-up, and sex, which was little attenuated by further adjustment for birth weight, father’s social class, alcohol consumption, and smoking. There was some attenuation of the association after adjustment for BMI and height in adulthood (95% CI: 1.07 mm Hg; 0.27, 1.87 mm Hg). There was a positive association of dried milk consumption at 3 mo and diastolic blood pressure, which was only attenuated after adjustment for BMI and height in adulthood.
The mean difference in systolic blood pressure between the highest and lowest quartiles of dried milk consumption at 3 mo was 6.1 mm Hg (95% CI: 2.7, 9.5 mm Hg) in the univariate model and was 3.7 mm Hg (95% CI: 1.0, 6.4 mm Hg) in the fully adjusted model. The respective mean differences in diastolic blood pressure were 2.3 mm Hg (95% CI: 0.3, 4.3 mm Hg) and 1.4 mm Hg (95% CI: -0.5, 3.3 mm Hg).
There was no strong evidence of an interaction between sex and consumption of dried milk at 10 d on systolic blood pressure (P for interaction = 0.27). There was weak evidence that the association of dried milk consumption at 3 mo with systolic blood pressure differed between males and females (P for interaction = 0.05). The regression coefficients in fully adjusted models were 1.55 (95% CI: 0.42, 2.68) in males (P = 0.008) and 0.39 (95% CI: -0.74, 1.52) in females (P = 0.49). There was little evidence of an interaction between sex and dried milk consumption at 3 mo on diastolic blood pressure (P = 0.6). There was little evidence that associations of dried milk consumption with blood pressure differed by BMI (P = 0.8) in adulthood.
In post hoc analyses, we explored further the findings of inverse and then positive associations of dried milk intake at 10 d and 3 mo on adult blood pressure (Table 5). Simultaneous adjustment for both 10-d and 3-mo intakes did not affect the estimates, suggesting that the blood pressure effects of early and later intakes were independent of each other. In an interaction model (see Subjects and Methods) we found a negative interaction of intake with systolic (P = 0.07) and diastolic (P = 0.005) blood pressure, suggesting that adults who had been fed the least amount of dried milk at 10 d but the most at 3 mo had the highest blood pressure. In stratified analyses, the regression coefficients for systolic blood pressure on 3-mo intake in the subjects in the lowest and highest quartiles of dried milk intake at 10 d were 3.19 (95% CI: 1.45, 4.92) and 0.56 (95% CI: -1.66, 2.78), respectively. Similar results were found for diastolic blood pressure (data not shown). The rate of weight gain between 6 wk and 3 mo attenuated but did not negate the association of dried milk intake at 3 mo with systolic blood pressure later in life.
DISCUSSION
We found a positive association between dried formula milk consumption in infancy and blood pressure at ages 23–27 y. The long-term importance of infant feeding was shown by Barker et al (36), who found that formula feeding was associated with increased cardiovascular disease mortality in men and women (37) born between 1911 and 1930. Our finding is consistent with the hypothesis that high blood pressure in later life is influenced by early postnatal nutrition, independent of both birth weight and BMI later in life. We found a positive association between formula feeding and BMI, in agreement with other studies (35, 38).
Limitations
We considered many potential limitations in interpreting these findings. First, selection bias may have occurred because of losses to follow-up. However, the baseline characteristics of those subjects who were not followed up were similar to those who were followed up, and there is no obvious reason why the association between infant feeding and blood pressure would be different among those not followed up. Exposure information bias is unlikely because infant feeding data were collected prospectively, and outcomes were measured many years later. Unmeasured or poorly measured confounding may explain the observed associations between infant feeding and blood pressure. Our questionnaires, however, provided detailed measures of smoking and alcohol consumption. Father’s social class measured at 18 mo may have poorly characterized early social environment. The lack of attenuation of effect estimates when included in the models argues against residual confounding by father’s social class.
Comparison with other studies
The findings of our study are in keeping with experimental (20, 22, 23) and some observational (24, 25) evidence that formula feeding is associated with elevated blood pressure. Males who were never breastfed had a mean systolic blood pressure that was 6.5-mm Hg higher than that of males breastfed > 3 mo (24). We found that 3-mo-old infants fed > 33 oz (976 mL) dried milk daily had a mean systolic blood pressure that was 6.1 mm Hg greater than that of infants fed < 24 oz (710 mL). Others observed no effect of infant feeding on blood pressure (4, 8, 26–28). The possibility that associations between infant nutrition and blood pressure may not be manifest until later in childhood or in adolescence (20, 39, 40) could explain the absence of an observed effect of type of infant feeding on blood pressure in younger children (4, 26). Blood pressure in late middle age may be more influenced by determinants of atherosclerosis, which operate independently of any possible formula feeding mechanism (see below), attenuating associations with infant nutrition (28).
An earlier study that reported stronger associations between infant nutrition and blood pressure in males than in females did not present a formal test for interaction (24). We also found a stronger association of dried milk consumption with systolic blood pressure later in life in males than in females (P for interaction = 0.05). This finding may have occurred by chance or could suggest that males may be more sensitive to early nutritional influences than are females. We found little evidence that the association between dried milk and blood pressure differed according to baseline heart rate, as suggested by others (23).
Mechanisms
An association between formula feeding and blood pressure later in life is biologically plausible. First, differences in sodium content between breast milk and formula foods may have had a role. Sodium intake in childhood is positively associated with blood pressure in childhood (7) and adolescence (23). Evidence exists for early tracking of blood pressure, perhaps even from the first months of life (41), suggesting a role for early nutrition as a determinant of tracking. Other studies have suggested that in the first few weeks of life there is little difference in the estimated amount of ingested sodium from breast milk (4–5 mmol/d) compared with formula milks of the type in use in the 1970s (6 mmol/d) (42). However, the sodium concentration of breast milk decreases in the first weeks of life, paralleling an increasing average daily intake of human milk, so that the total amount of ingested sodium during the first 4 mo of breastfeeding remains constant (4–5 mmol/d) (42). In contrast, the sodium content in formula milks is steady throughout the full period of formula feeding, so the average intake of sodium rises from an estimated 6 mmol/d in the first week of life to 18–23 mmol/d in the fifth week (42). These differences are similar to the average differences in sodium intake in infants randomly assigned to a low-sodium (5 mmol/d) or normal-sodium (15 mmol/d) diet during the first 6 mo of life (22). In that trial, the adjusted systolic blood pressure at 15 y of age was 3.6 mm Hg lower, and the adjusted diastolic blood pressure was 2.2 mm Hg lower in children in the low-sodium group than in the control subjects (23). These trial results are remarkably similar to the fully adjusted mean differences in systolic (3.7 mm Hg) and diastolic (1.4 mm Hg) blood pressure observed in our study between the highest and lowest quartiles of dried milk consumption at 3 mo. Differences in the amount of sodium ingested at 3 mo by those fed high volumes of dried milk compared with those fed lower volumes (who may have been partly breastfed) may explain the blood pressure effect seen in our study.
In the 1970s, when our cohort was established, there was concern about high blood osmolality (solute concentration) in formula-fed infants and consequent risks of hypernatremia dehydration complicating gastroenteritis (43). The view then held was that the combination of a very early introduction of solid foods along with the use of milk formulas with a high sodium content (often made worse through the incorrect reconstitution of milk powder) was responsible (44). This concern provided the impetus to modify the composition of milk formulas to follow more closely that of human milk and to delay the introduction of solid foods in the diet.
Second, breastfeeding may protect against overfeeding, calorie excess, and—in the long-term—future obesity (35), which is a risk factor for elevated blood pressure (2). Adjustment for current body weight attenuated but did not remove the observed association between dried milk intake and blood pressure, suggesting that an influence of infant nutrition on blood pressure later in life may partly operate via an effect on body weight later in life. Third, formula-fed infants have higher total energy (45), protein (46), and micronutrient (47) intakes than do breastfed infants, and this may stimulate an increased secretion of insulin (48) and a higher output of hepatic glucose (49) and insulin-like growth factor I (50, 51). It is possible that these different physiologic responses to infant feeding method may promote permanent changes in the structure and function of tissues at a critical period of development (52). Fourth, the possibility exists that mothers who choose a particular method of feeding their babies might adopt different dietary habits and food preferences for their children in later years. Mothers in the United Kingdom who breastfeed are likely to be better educated and to encourage healthier eating habits for their children than are mothers who do not breastfeed (53). Although we are not aware of any study that has shown that formula-fed children eat less healthy foods than do breastfed children, studies have shown that formula-fed infants are more likely to be overweight in later childhood and adolescence than are breastfed infants (25, 35, 38), which suggests dietary influences. It is therefore possible that links between formula feeding and higher blood pressure reflect a continuum effect of a less healthy diet throughout childhood rather than early programming.
Finally, Stettler et al (54) showed a positive association of infant weight gain with weight at age 7 y. An infant’s need for catch-up growth may increase their milk intake so that the association between milk intake and blood pressure arises from catch-up growth in low-birth-weight babies (55) rather than milk intake per se. Alternatively, weight gain may be on the biological pathway between dried milk intake and blood pressure. However, adjustment for early infant weight gain did not negate the association between milk intake and blood pressure, supporting the importance of the influence of dried milk intake on blood pressure later in life. The finding that this association was not altered by adjustment for birth weight suggests that the association between birth weight and blood pressure operates independently of postnatal dried milk intake.
Public health implications
The implications of these findings relate first to the possible health effects for those fed formula milk in the 1970s and second to the degree to which this finding is relevant to infant feeding formula and patterns today. In our sample we found an unadjusted difference of 6.1 mm Hg in systolic and of 2.3 mm Hg in diastolic blood pressure between the highest and lowest quartiles of dried milk consumption at 3 mo of age. Others showed that reducing the population mean systolic (56) and diastolic (57) blood pressure levels to levels lower than these would be as or more effective in reducing blood pressure–related morbidity or mortality than would targeting interventions at high-risk persons only (58).
It could be argued that these findings have no relevance to current infant feeding practices because there have been changes in the nutrient content of infant formulas since the 1970s (59). Human milk is highly variable in composition within and between individuals and over time (60), however, and it contains many hormones, trophic factors, and other active substances that influence neonatal physiology (60). Such characteristics have not yet been matched by formula foods. Furthermore, the difficulties of judging the most appropriate nutrient composition of formula milk have been highlighted (59). Although some modern infant formulas have a sodium content similar to that of breast milk, others have a higher content (61), and errors in reconstitution may lead to higher than anticipated intakes of sodium (44). Until data are available on the long-term effects of modern formula foods, the data from this and other studies are all that exist to guide policy. Formula feeding was positively associated with BMI in adulthood, as shown elsewhere (25, 35, 38). Because this association is plausibly related to bioactive factors in breast milk (35) rather than to the salt content, a possible adverse effect of bottle milk on BMI remains of contemporary public health relevance regardless of the association between formula feeding and blood pressure. Our data support recommendations to breastfeed for ≥ 4 mo (61).
Breastfeeding rates declined steadily between the 1940s and 1970s and were followed by an upward trend (62). In 1975, 49% of infants in England and Wales were never breastfed, and only 4% of mothers exclusively breastfed at 6 wk (63). Although breastfeeding rates have since increased, 32% of infants are currently never breastfed (64), and breastfeeding rates are lower in those disadvantaged populations (64) who are at increased risk of ischemic heart disease. The possibility of postnatal nutritional influences on adult health (15) offers the prospect that interventions to optimize infant nutrition may have important long-term benefits.
ACKNOWLEDGMENTS
We are very grateful to the subjects who participated in the original survey and who were willing to continue to be followed up in early adulthood. We thank the Bro Taf Health Authority for help with contacting the subjects, Carol Hopkinson for help with the fieldwork and data entry, and Dan Dedman for advice on the derived variables in the Barry Caerphilly Growth cohort data set.
GDS, YB-S, AM, and DPD designed the cohort follow-up study of the original trial. AM ran the screening clinics supervised by YB-S and GDS. RMM, YB-S, and GDS developed the hypothesis for this analysis. RMM undertook the initial analysis, wrote the first draft of the paper, and coordinated the completion of the paper. RMM, YB-S, GDS, AM, and DPD contributed to the final version. YBS and RMM will act as guarantors. None of the authors had a conflict of interest.
REFERENCES