点击显示 收起
1 From the Research Institute of Child Nutrition, Rheinische Friedrich-Wilhelms-Universität Bonn, Dortmund, Germany (NK-D, ALBG, and AEB); the Department of Nutrition, Food and Consumer Sciences, Fulda University of Applied Sciences, Fulda, Germany (AK); and the Faculty of Health Sciences, University of Bielefeld, Bielefeld, Germany (CH)
2 The DONALD Study is funded by the Ministry of Science and Research of North Rhine Westphalia, Germany. ALBG also has a research grant from the International Foundation for the Promotion of Nutrition Research and Nutrition Education. 3 Reprints not available. Address correspondence to N Karaolis-Danckert, Nutrition and Health Unit, Research Institute of Child Nutrition, Heinstueck 11, 44225 Dortmund, Germany. E-mail: karaolis{at}fke-do.de.
ABSTRACT
Background: It is not clear whether the adverse effects of rapid weight gain in infancy are modified by nutrition during the first 2 y of life in term children whose birth weight was appropriate for gestational age (AGA).
Objective: We examined the interaction between rapid weight gain and nutrition in infancy and early childhood and their effect on body fat percentage (BF%) trajectories between 2 and 5 y of age.
Design: The study population comprised 249 (51.4% female) term AGA participants of the Dortmund Nutritional and Anthropometric Longitudinally Designed Study, for whom repeated anthropometric measurements until 5 y of age and information on breastfeeding status and on diet at 12 and 18–24 mo of age were available.
Results: Multilevel model analyses showed that, among rapid growers, those who had been fully breastfed for 4 mo had a lower BF% at 2 y of age than did those who had not been fully breastfed for 4 mo (β ± SE: –1.53 ± 0.59%; P = 0.009). This difference persisted until 5 y. Furthermore, those rapid growers who had a consistently high fat intake at both 12 and 18–24 mo did not show the expected physiologic decrease in BF% between 2 and 5 y seen in those rapid growers with an inconsistent or consistently low fat intake at these time points (0.73 ± 0.26%/y; P = 0.006).
Conclusions: Among rapid growers, full breastfeeding for 4 mo is protective against a high BF% at 2 y of age, whereas a consistently high fat intake in the second year of life "inhibits" the physiologic decrease in BF% between 2 and 5 y.
Key Words: Rapid weight gain • appropriate for gestational age • nutrition • breastfeeding • body fat percentage • children • trajectories
INTRODUCTION
The contribution of rapid weight gain in infancy to the development of overweight and adiposity later in life (1, 2), as well as to several other morbidities, including cardiovascular disease (3) and cancer (4), has now been repeatedly reported for a variety of populations. Recent evidence suggests not only that children born small for gestational age (SGA) subsequently compensate for this in utero growth restriction by growing rapidly (5), but also that this phenomenon is seen in some children whose birth weight and length were appropriate for gestational age (AGA) (6). In both cases, rapid weight gain results in a greater fat mass, and the differences in comparison with normal growers appear to become progressively larger over time (6, 7).
In developed countries, where the prevalence of overweight and obesity is on the rise, most children have a birth weight that is AGA. The failure of currently available methods for treating established obesity has resulted in an emphasis on primary prevention and on identifying modifiable risk factors (and susceptible individuals) as early as possible. With respect to rapid weight gain, it remains unclear which factors stimulate, modify, or amplify its effect on growth and body-composition trajectories. Nutrition is an important regulator of growth, and an effect of prenatal nutrition on subsequent growth, metabolism, and obesity risk is well established (8). Several meta-analyses have also emphasized the role of the postnatal diet in infancy, and breastfeeding status in particular, on obesity risk (9, 10). Because breastfed children grow more slowly than do formula-fed children (11), it is conceivable that breastfeeding status could influence or modify the effect of rapid weight gain. On the other hand, the potential role of diet during the complementary feeding (CF) phase has received comparatively little attention, although this is a period when important quantitative and qualitative changes in the diet occur (12). The transition to the family diet is marked by a considerable increase in protein intake and a reduction in fat intake, the speed, degree, and benefits of which are still being debated (12, 13). Nevertheless, this phase may influence growth, and fat mass development in particular, as was suggested by a recent analysis of ours in which high protein intakes in the second year of life were associated with a larger body mass index SD score (BMI SDS) and body fat percentage (BF%) at age 7 y (14).
Using data from the Dortmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study, we sought to determine whether nutrition in infancy and early childhood modifies the association between rapid weight gain and BF% and BMI SDS trajectories between 2 and 5 y of age in healthy, AGA children.
SUBJECTS AND METHODS
Study population
The DONALD Study is an ongoing, open-cohort study conducted by the Research Institute of Child Nutrition in Dortmund, Germany. This study has been previously described in detail (15). Briefly, since recruitment began in 1985, detailed information on diet, growth, development, and metabolism between infancy and adulthood has been collected from >1100 children. Every year, an average of 40–50 infants are newly recruited and first examined at the age of 3 mo. Each child returns for 3 more visits in the first year, 2 in the second, and then once annually until early adulthood. The study was approved by the Ethics Committee of the University of Bonn, and all examinations are performed with parental consent.
The ages of the children who were initially recruited for the DONALD Study were variable so that information on the first few years of life was not always available. In addition, many children have not yet reached 5 y of age. Therefore, 1) a minimum of anthropometric measurements at 0.5, 2, and 5 y were available for 408 term (37–42 wk gestation) singletons with a birth weight >2500 g; 2) this number was reduced to those whose birth weights and lengths were AGA (n = 326), ie, all birth weights and lengths lay between the 10th and 90th percentiles of the German sex-specific birth weight–for–gestational age curves (16, 17); and 3) finally, all children had to have complete, plausible dietary records at ages 12 mo and 18 or 24 mo (n = 253 remaining) (18), complete information on breastfeeding, and complete information on maternal characteristics (BMI and educational status) (n = 249 remaining). Hence, the subcohort analyzed here included 249 AGA term singletons (51.4% female). The mean number of measurements per child was 7.95 (range: 7–8).
Anthropometry
The DONALD Study participants are measured at each visit according to standard procedures (19). They are dressed in underwear only and are barefoot. The trained nurses who conduct the measurements undergo an annual quality control during which inter- and intraobserver agreements are carefully monitored. Recumbent length in children under 2 y of age is measured to the nearest 0.1 cm with a Harpenden (Crymych, United Kingdom) stadiometer. From the age of 2 y onward, standing height is measured to the nearest 0.1 cm with a digital stadiometer. Weight is measured to the nearest 0.1 kg with an electronic scale (753 E; Seca, Hamburg, Germany). Skinfold thicknesses are measured from the age of 6 mo onward on the right side of the body at the biceps, triceps, subscapular, and suprailiac sites to the nearest 0.1 mm with a Holtain caliper.
On their child's admission to the study, the parents are interviewed by the study pediatrician and are weighed and measured by the study nurses using the same equipment as for children from 2 y onward. Information on birth weight, length and head circumference at birth, gestational age, and maternal weight gain during pregnancy are abstracted from the Mutterpass, a standardized document given to all pregnant women in Germany.
Anthropometric calculations
Sex- and age-independent SDS were calculated by using the German reference curves for weight and body mass index [BMI; in kg/m2 (20)] and the Tanner and Whitehouse reference data for subscapular and triceps skinfold-thickness measurements (21). To remove general deviations of our sample from the reference data, these variables were internally standardized. BF% was calculated by using the Deurenberg equations (22). Overweight at age 5 y was defined according to the International Obesity Task Force BMI cutoffs for children, which correspond to an adult BMI of 25 (23). Overfatness at 5 y of age was defined as a BF% higher than the 85th percentile of the sex-specific body fat reference curves by McCarthy et al (24). Rapid weight gain was defined as an increase in weight SDS of >0.67 between birth and 24 mo, as recommended by Monteiro and Victora (1). The value of 0.67 SDS represents the width of each percentile band on standard growth charts, ie, between the 25th and the 50th percentiles, between the 50th and the 75th, and so on, and indicates clinically significant rapid weight gain (25).
Diet in infancy: breastfeeding
At the infant's first visit (ie, when the infants are 3 or 6 mo old), the study pediatrician asks the mothers about how long (in weeks) they had fully breastfed their child for. The definition of full breastfeeding (breast milk only without any supplemental solid foods or liquids other than tea or water) is carefully explained. If the mother is still fully breastfeeding, this question is repeated at each subsequent visit (eg, 6, 9, and 12 mo) until complementary feeding is initiated. In addition, for 70% of infants, on average, the mothers also keep 3-d weighed dietary records during the first year of life, so that infant feeding can be quantified at 3, 6, 9, or 12 mo. When the study dietitians collect the protocols, they also question the mothers about the introduction of formula or solid foods. Finally, consistency checks to identify possible sources of error are made by comparing the information from the breast milk records with that collected by the pediatricians and the dietitians. Children in this analysis were grouped into those who had been fully breastfed for 4 mo (defined as full breastfeeding for 17 wk) and those who had not. This cutoff was used because it is the lower limit of the German recommendation for the introduction of complementary foods (26).
Diet in early childhood: 12 and 18–24 mo
The dietary intake of DONALD participants is assessed by use of 3-d weighed dietary records. Parents are asked to weigh all foods and beverages consumed by their children, including leftovers, to the nearest 1 g over 3 consecutive days with the use of regularly calibrated electronic food scales [initially Soehnle Digita 8000 (Leifheit AG, Nassau, Germany), now WEDO digi 2000 (Werner Dorsch GmbH, Muenster/Dieburg, Germany)]. The children's parents are instructed by trained dietitians, and semi-quantitative measures (eg, number of spoons, scoops, etc) are allowed when exact weighing is not possible. Information on recipes and on the types and brands of food items consumed is also requested. The dietary records are analyzed by using the continuously updated in-house nutrient database LEBTAB (27), which includes information from standard nutrient tables, product labels, or recipe simulations based on the ingredients and nutrients listed. With regard to those children still breastfeeding, test weighing is performed, ie, the child is weighed before and after each feed to the nearest 10 g with an infant weighing scale (Soehnle Multina 8300). To account for insensible water losses, 5% was added to the test weighing results, as proposed by Reilly et al (28).
For this study, absolute intakes of energy (kcal/d), protein (g/d), and fat (g/d) at the ages of 12, 18, and 24 mo were calculated for each participant from the mean of the 3 dietary recording days. A cutoff of 0.97 for the reported energy intake related to the predicted basal metabolic rate was used to identify implausible dietary records (18), which were then excluded (<2%). The absolute intakes of energy were log-transformed and internally standardized (mean = 0, SD = 1), and the mean of these values was used to represent the mean intake for the whole second year of life. The residual method was used to adjust macronutrient intakes for total energy intake and sex, separately for each time point and with log-transformed nutrients to improve homoscedasticity (29). The medians of the residuals were used to divide the children into those with low and those with high macronutrient intakes at ages 12 and 18–24 mo, the latter of which was based on mean individual intakes at these 2 time points. Finally, the children with consistently high energy-adjusted protein or fat intakes (g/d) throughout the second year of life (ie, high intakes at both 12 and 18–24 mo, or "high-high," in each case corresponding approximately to an intake >13–14% of energy from protein or >35% of energy from fat, respectively) were grouped together, because we previously showed that DONALD children with such a protein intake pattern had a higher BMI SDS and a higher BF% at age 7 y than did those who did not meet this criterion (ie, those with inconsistent intakes or those with consistently low intakes, who were grouped together as "other") (14).
Statistical analysis
Unadjusted associations between the independent variables and rapid weight gain were tested by using chi-square, Student's t test, or Wilcoxon's rank-sum test as appropriate. Because there were equal numbers of boys and girls in the 2 growth groups (chi-square = 0.020, P = 0.89), they were pooled together for all statistical analyses. Linear mixed-effects regression models (using PROC MIXED), including both fixed and random effects, were used to construct longitudinal models of BF% and BMI SDS trajectories subsequent to the period of rapid weight gain (between 2 and 5 y of age) and to investigate the effect of rapid weight gain on baseline BMI SDS or BF% status at age 2 and changes over time. The random component of these models accounts for the nested nature of our data (children within families) and the lack of independence between repeated observations on the same person. Initial models included either BF% or BMI SDS measurements between 2 and 5 y inclusive as the dependent continuous variable and rapid weight gain, time (chronological age and age2), and the interaction between rapid weight gain and time as the independent fixed effects. The effect of adding each of the following variables and their interaction with time (used to decide whether a given variable had a significant effect on the change in BF% or BMI SDS over time) to the initial models was then investigated as follows:
Those variables that modified the coefficient of rapid weight gain by 10% in the initial models (30); had a significant, independent effect on the outcome variables; or both were included in the series of subsequent multivariable models. These began by considering the nutrition variables, to which the variables related to birth, and then maternal, characteristics were added. Akaike's Information Criterion was also used to assess model fit (31). A three-way interaction between time, rapid weight gain, and each of the other fixed variables was also included to consider differential effects of rapid weight gain on the BF% or BMI SDS trajectories of children in various subgroups. When the three-way interaction was significant, it remained in the model, along with all lower-order two-way interactions and main effects. An advantage of the MIXED procedure is that it does not delete children from the analysis if they are missing data for a particular time point, but analyzes all of the data available on the assumption that any missing data are missing at random. A P value < 0.05 was considered statistically significant. All statistical analyses were carried out with SAS version 8.2 (SAS Institute, Inc, Cary, NC).
RESULTS
Overall, 28.5% (71/249) of the children in this sample gained weight rapidly between birth and 24 mo. These children were both significantly lighter and shorter at birth, and a larger proportion were born relatively early compared with the normal growers (Table 1). Rapid growers were also more likely to be first-born. With respect to maternal anthropometry and educational status, no significant differences between the 2 growth groups were found. By the age of 5 y, those children who had gained weight rapidly were significantly heavier and taller and had a higher BMI SDS and BF% than did those children who had gained weight normally. Furthermore, a significantly larger proportion of these children were classified as overweight (17% compared with 7%; P = 0.02) or overfat (27% compared with 15%; P = 0.03).
View this table:
TABLE 1. Birth, maternal, and anthropometric characteristics of those DONALD Study children (n = 249) who gained weight rapidly between the ages of 0 and 24 mo and those who did not1
The dietary characteristics of the 2 growth groups in infancy and early childhood are shown in Table 2. Neither in terms of breastfeeding status (P = 0.5) nor in terms of macronutrient intake at either 12 or 18–24 mo did the differences in median intake between the 2 growth groups reach a statistical significance of < 0.05. Furthermore, there was no significant difference in the proportion of rapid or normal growers in each of the high-high macronutrient intake groups for any of the macronutrient variables: protein (35% compared with 33%; P = 0.8), fat (30% compared with 31%; P = 0.8), and carbohydrate (37% compared with 28%; P = 0.2).
View this table:
TABLE 2. Dietary characteristics during infancy and early childhood of those DONALD Study children (n = 249) who gained weight rapidly between 0 and 24 mo and those who did not1
The results of the linear mixed models analyses for the association between rapid weight gain and BF% trajectories (Table 3) showed that, even after adjustment for birth, maternal overweight, and educational characteristics (model 2), 2 nutrition variables significantly influenced the relation between rapid weight gain and BF%. First, those rapid growers who were fully breastfed for 4 mo had a significantly lower fat mass at age 2 y than did those rapid growers who were not fully breastfed for 4 mo (adjusted difference between rapid growth groups: –1.53 ± 0.59%; P = 0.009; see also Figure 1: solid circles versus solid squares). However, breastfeeding status did not affect the rate of change in BF% between 2 and 5 y of age, ie, there was no three-way interaction between time, rapid weight gain, and breastfeeding status (Figure 1: all symbols). On the contrary, all groups appeared to track at the level of body fat they had reached by 2 y of age. Among normal growers, whether a child had been breastfed for 4 mo appeared to have little effect on BF% at age 2 y (Figure 1: open squares and circles).
View this table:
TABLE 3. Linear mixed models of the association between rapid weight gain, nutrition in infancy and early childhood, and baseline percentage body fat (BF%) at 2 y of age and BF% slope between 2 and 5 y of age (n = 249)
FIGURE 1.. Predicted mean (±SEM) percentage body fat trajectory in subgroups of rapid weight gain and breastfeeding status in 249 children: rapid growers who were not breastfed for 4 mo (, n = 28), rapid growers who were breastfed for 4 mo (•, n = 43), normal growers who were not breastfed for 4 mo (, n = 62), and normal growers who were breastfed for 4 mo (, n = 116). See Table 3 for information on the model used for prediction. The plot shows the 2-way interaction between rapid weight gain and breastfeeding status at age 2 y (P = 0.009 for interaction).
Second, although there was no interaction between a consistently high fat intake at 12 and 18–24 mo and rapid weight gain at 2 y of age (0.28 ± 0.67%; P = 0.7), those rapid growers who had a consistently high fat intake did not show the physiologic decrease in BF% between 2 and 5 y of age seen in those rapid growers with a consistently low fat intake or an inconsistent fat intake at 12 and 18–24 mo (adjusted difference between rapid growth groups: 0.73 ± 0.26%/y; P = 0.006). Instead, as shown in Figure 2, rapid growers in the high-high group (solid squares) maintained the level they had reached by 2 y of age, so that the difference in BF% between them and those rapid growers in the other group (solid circles), whose BF% decreased between 2 and 5 y of age, became progressively larger. The opposite was true of the normal growers. In their case, BF% appeared to decrease faster in those who had had a consistently high fat intake (open squares) than in those with an inconsistent or consistently low fat intake (open circles).
FIGURE 2.. Predicted mean (±SEM) percentage body fat trajectory by subgroups of rapid weight gain and dietary fat consumption at 12 and 18–24 mo of age in 249 children: rapid growers with a consistently high fat intake (, n = 21), rapid growers with an inconsistent or consistently low fat intake (•, n = 50), normal growers with a consistently high fat intake (, n = 55), and normal growers with an inconsistent or consistently low fat intake (, n = 123). See Table 4 for information on the model used for prediction. The plot shows the 3-way interaction between rapid weight gain and fat intake between 2 and 5 y of age (P = 0.006 for interaction).
Of the other nutrition variables shown in Table 3, none interacted significantly with rapid weight gain at either age 2 or between 2 and 5 y of age. However, some had an independent effect on BF%: a higher mean energy intake during the second year of life tended to be associated with a lower BF% at age 2 y (–0.32 ± 0.19%; P = 0.08) but with a gain in BF% between 2 and 5 y (0.20 ± 0.07%/y; P = 0.005). A consistently high protein intake, on the other hand, was associated with a higher BF% at age 2 y (0.67 ± 0.31%; P = 0.03), but did not have any effect on change in BF% over time.
Separate analyses were also conducted by using the sum of the 4 skinfold-thickness measurements, triceps SDS, or subscapular SDS as the outcome variables, respectively. However, the conclusions about the effect of the interactions between rapid weight gain and the nutrition variables and breastfeeding on BF% trajectories remained unchanged, regardless of the outcome variable used.
With respect to BMI SDS (Table 4), none of the nutrition variables considered interacted with rapid weight gain to influence its effect on BMI SDS at 2 y or on the rate of BMI SDS change between 2 and 5 y of age. Nevertheless, a consistently high protein intake resulted in a higher BMI SDS at age 2 y (0.36 ± 0.13 SDS; P = 0.005). There was also an indication that this difference might decrease between 2 and 5 y (–0.06 ± 0.03 SDS/y; P = 0.07). A consistently high fat intake during the second year of life, on the other hand, resulted in a lower BMI SDS at age 2 y (–0.26 ± 0.13 SDS; P = 0.05), but did not affect the change in BMI SDS between the ages of 2 and 5 y.
View this table:
TABLE 4. Linear mixed models of the association between rapid weight gain, nutrition in infancy and early childhood, and baseline BMI SD score (SDS) at 2 y of age and BMI SDS change between 2 and 5 y of age (n = 249)
DISCUSSION
In this study, we showed that the detrimental effect of rapid weight gain on fat mass development in healthy, term AGA children can be modified by dietary factors acting in infancy and early childhood. In particular, being breastfed for 4 mo attenuated the effect of rapid weight gain, resulting in a lower BF% at 2 y of age among rapid growers who had been fully breastfed than in those who had not. Furthermore, a consistently high fat intake during the second year of life modified the subsequent effect of rapid weight gain on the longitudinal development of fat mass, thus inhibiting the normal physiologic decrease, thereby resulting in a larger BF% among exposed rapid growers.
Several studies have described the differences in growth between breastfed and bottle-fed infants (11, 32); to our knowledge, however, this is the first study to investigate in detail the association between breastfeeding and rapid weight gain. We found that full breastfeeding for 4 mo (only) has a protective effect during the period of rapid weight gain. That is to say, full breastfeeding for 4 mo appears to directly influence the extent to which rapid weight gain adversely affects fat mass development in the first 2 y of life, after which the level of BF% reached tracks unchanged. An over-compensatory drive to hyperphagia is one of the mechanisms proposed to play a role in "catch-up fat," ie, the propensity to gain body fat rather than lean tissue during rapid weight gain (33). It is known that differences in appetite control, feeding frequency, and meal size exist between bottle-fed and breastfed infants (34). This could explain why rapid growers who were not fully breastfed showed a more pronounced gain in fat mass. Although BMI is a convenient proxy measure of adiposity, it cannot differentiate between lean and fat mass (35), so it is perhaps not surprising that the beneficial effect of breastfeeding was only discernible in the models of BF%. In fact, BMI was only moderately correlated with BF% in this sample (r = 0.53–0.66 in boys and r = 0.60–0.71 in girls).
Alternatively, the higher protein intakes of bottle-fed children could explain the differences in BF% at 2 y of age. Higher protein intakes may stimulate the secretion of insulin and insulin-like growth factor 1, both of which accelerate growth and increases in fat mass (36, 37). If protein were the major regulatory pathway in the interaction between lack of breastfeeding and rapid weight gain, it would appear that it only has a role to play during a time window in the first year of life. We conclude this because consistently high protein intakes during the second year of life did not modify the effect of rapid weight gain on BF%, despite a consistently high protein intake being independently associated with a higher BF% and a higher BMI SDS at 2 y of age, and a tendency toward a decreased rate of BMI SDS change between the ages of 2 and 5 y.
In this study, among rapid growers, a consistently high fat intake at both 12 and 18–24 mo inhibited the physiologic decline in fat mass that normally occurs between the end of the first year of life and 5–6 y of age (38). Conversely, among normal growers, a consistently high fat intake at these points in time resulted in a greater decrease in BF% between 2 and 5 y than when fat intakes were inconsistent or consistently low. These apparently contradictory associations were observed for fat intakes consistently exceeding 35% during the second year of life. This level of fat intake lies within the recommendations of both national (39) and international (40) health organizations for these age groups. There seems to be a general consensus that fat intake should only be reduced to 30% after the age of 2 y, because there does not appear to be any evidence for a detrimental effect of a high fat intake during the first years of life. On the contrary, it is believed that the rapid transition from breastfeeding (low protein, high fat) to family food (high protein, relatively low fat) results in an inadequately balanced infant diet (13). In a recent review of studies on fat intake in infancy and early childhood and subsequent weight or fatness in infants and children, 11 of 13 studies failed to find an association (41). Those that found a positive association emphasized that the relation was stronger with fat intakes after 2 y.
It remains to be ascertained why a consistently high fat intake during the second year of life was detrimental in those children who gained weight rapidly. A Brazilian study showed a significant, positive association between a higher fat intake (% of energy) at baseline and weight-for-height gain over the follow-up period in those 7–11-y-old girls who were mildly stunted (42). Because high-fat diets may promote excessive weight gain and accumulation of adipose tissue (43, 44), the authors questioned whether the high rate of obesity seen in this group of recovered-malnourished children had to do with "an increased susceptibility to high fat diets." In his article on the control systems that regulate fat storage, Dulloo (33) proposed that "suppression of thermogenesis," ie, when glucose spared from oxidation in muscle is directed toward lipogenesis and storage in white adipose tissue, could be the main mechanism by which rapid weight gain leads to increased fat mass in individuals whose system has been programmed by scarcity. It is possible that exposure to a high-fat diet in the second year of life in susceptible individuals, such as AGA children who gain weight rapidly, further exacerbates this mechanism.
Although we did find a tendency toward a higher energy intake among rapid growers at 12 mo, the linear mixed models did not reveal any interaction with rapid weight gain. We therefore believe that the suggestion of differences in energy intake at 12 mo should be interpreted in the context that rapid growers were already taller and heavier at these ages. In their study of formula- or mixed-fed infants, Ong et al (45) found that dietary energy intake at 4 mo predicted greater subsequent weight gain. We were unable to investigate diet during the first year of life because, although DONALD participants provide accurate information on breastfeeding status, few carry out dietary records before 6 mo of age.
The DONALD Study participants are characterized by a relatively high socioeconomic and educational status. It is possible that the relative homogeneity of the DONALD sample means that extremes of diet or behavior are not represented. We nevertheless observed effect modification by some nutrition variables despite this homogeneity. Although the DONALD Study sample is also unrepresentative, the dietary data in general, and patterns in infancy, in particular in terms of rates of breastfeeding and choice of breast milk substitutes and commercial weaning foods, are similar to several nationwide studies (46, 47). The considerable strengths of this analysis lie in the DONALD Study's prospectively collected, repeated assessments of breastfeeding status and diet in early childhood; the repeated anthropometric measurements from as early as 3 mo of age; the Institute's nutritional database, LEBTAB, which is constantly updated (27); and detailed information on several possible covariates and confounders.
We do not know why some AGA infants gain weight rapidly. As far as the nutrition variables considered in this article are concerned, they simply acted as modifiers of the effect of rapid weight gain. However, on a practical level, the findings of this study support the continued promotion of breastfeeding. Although our study suggests that principally rapid growers would benefit from being breastfed with regards to fat mass development, at the moment it is not possible to say which newborns will grow rapidly and which will not. Moreover, the multitude of other benefits of breastfeeding speak for themselves (48). Our data also suggest that the current recommendations in favor of a higher fat intake during the complementary feeding period may not necessarily be beneficial for all children. This being an observational study, we cannot clearly separate cause from effect. Nevertheless, individualizing recommendations on the basis of a child's growth pattern may be worth considering. We acknowledge that we were unable to investigate the qualitative nature of the fats involved, and more information would therefore be required on the type of fat potentially responsible for both the positive and the negative effects observed in this study.
In conclusion, the influence of certain nutritional factors early in life on later body composition varies by growth pattern: among rapid growers, full breastfeeding for 4 mo exerts a protective effect against high BF%, whereas a consistently high fat intake during the second year of life inhibits the physiologic decrease in BF% between 2 and 5 y of age.
ACKNOWLEDGMENTS
The participation of all children and their families in the study is gratefully acknowledged. We also thank Birgit Holtermann, Ute Kahrweg, and Sabine Twenhöven for carrying out the anthropometric measurements.
The contributions of the authors were as follows—NK-D and ALBG: conceived the project and performed the initial data analyses; NK-D: conducted further analyses and drafted the manuscript; AEB, AK, and CH: supervised the study. All authors contributed to interpretation of the data and revision of the manuscript. None of the authors had any personal or financial conflicts of interest.
REFERENCES