点击显示 收起
1 From the Research Institute of Child Nutrition, Affiliated Institute of the University of Bonn, Dortmund, Germany (ALBG, AEB, and TR), and the Department of Nutrition, Food and Consumer Sciences, Fulda University of Applied Sciences, Fulda, Germany (AK)
2 Supported by the Ministry of Science and Research of North Rhine Westphalia, Germany, and by a research grant from the International Foundation for the Promotion of Nutrition Research and Nutrition Education (to ALBG). 3 Reprints not available. Address correspondence to ALB Günther, Nutrition and Health Unit, Research Institute of Child Nutrition, Heinstueck 11, 44225 Dortmund, Germany. E-mail: guenther{at}fke-do.de.
ABSTRACT
Background: A high early protein intake has been proposed to increase obesity risk.
Objective: We examined whether a critical period of protein intake for later obesity may exist early in childhood and investigated the relation between protein intake from different sources and body mass index SD score and body fat percentage (BF%) at 7 y of age.
Design: The study population included 203 participants of the Dortmund Nutritional and Longitudinally Designed Study with information on diet at 6 mo, 12 mo, 18-24 mo, 3-4 y, and 5-6 y. Life-course plots were constructed to assess when protein intake (% of energy) was associated with body mass index SD score and BF% at 7 y. Mean values were then compared among tertiles (T1-T3) of protein from different sources at the important time points.
Results: The ages of 12 mo and 5-6 y were identified as critical ages at which higher total and animal, but not vegetable, protein intakes were positively related to later body fatness. In fully adjusted models, animal protein intake at 12 mo was associated with BF% at 7 y as follows [
Key Words: BMI • obesity • dairy protein • critical periods • adiposity rebound
INTRODUCTION
Dietary factors during the sensitive period of infancy and early childhood are increasingly recognized as being potentially critical for adult disease and predisposition to obesity (1, 2). In this context, protein has received particular attention, because the "early protein" hypothesis postulates that high protein intakes in the first months of life increase the risk of subsequent obesity, possibly by inducing distinct hormonal responses, such as stimulating the secretion of insulin and insulin-like growth factor-I (IGF-1) (3).
Although the evidence in favor of this hypothesis is still limited, there are 2 main reasons the first years of life may actually represent a critical time window with regard to protein intake and later adiposity. First, infant formula is characterized by an 50–80% higher protein content than human milk (4), which has accordingly been discussed as one mechanism behind the commonly observed increased risk of later obesity in nonbreastfed children (5). Second, during the period of complementary feeding and the transition to the family diet, there is a rapid increase in protein intake. In various populations from 9 mo of age onward, it has been reported to exceed current recommendations (6), up to almost 5-fold in some children (7, 8). In fact, recent analyses of ours suggested that the persistence of a high protein intake at both 12 and 18–24 mo is associated with a higher body mass index SD score (BMI SDS) and body fat percentage (BF%) at 7 y of age (9). Nevertheless, it cannot be excluded that dietary protein in later childhood, eg, during "adiposity rebound" [which itself represents a critical period for obesity development (10)], might play an ongoing role regarding later body fatness.
Particularly in a critical period of intake, if present, it is conceivable that only certain protein qualities or protein intake from distinct food groups may be responsible for any relation with later obesity risk. Most studies in the context of the early protein hypothesis have so far concentrated on total protein intake (3, 9, 11-14); however, the main protein sources once complementary feeding has commenced [meat, dairy, and cereal (15)] are known to exert differential metabolic effects in children. In particular, several studies have suggested that cow milk but not meat intake or vegetable protein stimulates the secretion of insulin and IGF-1 in pediatric age groups (16-18).
Using data from the Dortmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study, the objectives of the present analysis were therefore 1) to evaluate whether certain time points or periods of protein intake in infancy, early childhood, or the preschool years may be most important with regard to BMI and BF% at 7 y of age, and, based on these results, 2) to investigate whether only protein intake from distinct sources (animal, vegetable, dairy, meat, and cereal protein) could be responsible for potential associations between early total protein intake and later body fatness.
SUBJECTS AND METHODS
Study population
The DONALD Study is an ongoing, open cohort study that was started in 1985 in the area of Dortmund, Germany. Details of the study design have been published previously (19). In brief, an average of 40–50 infants are newly recruited each year and first examined at the age of 3–6 mo. From then on, detailed data on nutrition, growth, metabolism, and health status are collected at regular intervals between infancy and young adulthood, ie, up to 3 further visits in the first year of life, 2 in the second, and 1 per year thereafter. The study was approved by the Ethics Committee of the University of Bonn, and all assessments are performed with parental consent.
For the purpose of this analysis, we only used data from term (gestational age 37–42 wk) singletons with a minimum birth weight of 2500 g. Furthermore, several additional criteria had to be met: 1) complete anthropometric measurements (weight, height, 4 skinfold thicknesses) had to be available at ages 6 mo (baseline), 3–4 y, and 7 y (endpoint) (n = 340); 2) this number was reduced to those with plausible dietary records at the ages of 6 mo, 12 mo, 18–24 mo (1 record out of possible 2), 3–4 y (1 out of possible 2), and 5–6 y (1 out of possible 2) (n = 205); and finally, 3) information on potential confounders, eg, maternal overweight (BMI 25; yes or no), maternal educational status (12 y schooling; yes or no), breastfeeding (full breastfeeding for 4 mo; yes or no), firstborn status (yes or no), and siblings in the dataset (yes or no), had to be available (n = 203).
Hence, the subcohort analyzed here consisted of 203 children (102 boys, 101 girls). This number was sufficient to detect a difference of 0.40 BMI SDS and 1.70% body fat (original scale) between 2 equal groups and a mean difference of 0.48 BMI SDS and 2.07% body fat between the highest and the lowest tertile with = 0.05 and a power of 80% (two-tailed) (20). We chose 7 of age as our endpoint because at that age, BMI correlates well with BMI in adulthood (21, 22).
Parental and birth characteristics
On a child's entry to the study, parents are asked to provide information about family characteristics and their educational status and employment, and their weight and height are measured by the same trained nurses who assess the anthropometrics of the participating children. Information on birth weight, birth length, and gestational age are abstracted from a standardized document given to all pregnant women in Germany.
Anthropometric data
At each visit, anthropometric measurements are performed by trained nurses according to standard procedures, with the children dressed in underwear only. The nurses undergo an annual quality-control check in which intra- and interobserver agreement is carefully monitored. Body weight is assessed to the nearest 100 g with a supine infant weighing scale (Mettler PS 15; Mettler Toledo, Columbus, OH) or an electronic scale for subjects in the standing position (Seca 753 E; Seca GmbH & Co KG, Hamburg, Germany). Recumbent length in children up to 24 mo of age is measured to the nearest 0.1 cm by using a Harpenden stadiometer (Holtain Ltd, Crymych, United Kingdom). From 24 mo of age onward, standing height is measured to the nearest 0.1 cm with a digital telescopic wall-mounted stadiometer. 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 by using a Holtain caliper (Holtain Ltd).
For each child, age- and sex-independent SD scores of weight, height, and BMI were calculated by using the German reference curves (23). To correct for general deviations of our sample from the reference data, BMI SDS values were internally standardized (mean = 0, SD = 1; according to age and sex) for the multivariable analyses. Body density and BF% were calculated by using Deurenberg's equations (24). Because BF% values were skewed, we applied log-transformations (lnBF%) in all analyses and present geometric means with 95% CIs throughout. Furthermore, we assessed the proportion of overweight children according to the definition of the International Obesity Task Force (25). In an analogous manner, the 85th percentile of the body fat reference curves published by McCarthy et al (26) formed the basis for the definition of overfat.
Dietary data
In the DONALD Study, dietary intake is assessed by use of 3-d weighed-diet records. Parents are asked to weigh all foods and beverages consumed by their children, including leftovers (eg, in milk bottles), to the nearest 1 g over 3 consecutive days with the use of regularly calibrated electronic food scales (initially Soehnle Digita 8000, Leifheit SG, Nassau, Germany; now WEDO digi 2000, Werner Dorsch GmbH, Muenster/Dieburg, Germany). With regard to breastfeeding, test weighing is performed, which means that the intake of breastmilk is assessed by weighing the infant before and after each meal to the nearest 10 g by using an infant-weighing scale (Soehnle multina 8300, Leifheit SG). In this analysis, 5% was added to the test weighing results to account for insensible water losses (27).
Parents are instructed by trained dietitians. Semiquantitative recording with household measures (eg, number of spoons or scoops) is allowed when exact weighing is not possible (eg, foods eaten away from home); however, 91.5% of the meals reported in the present sample were eaten or prepared at home. Information on recipes or the types and brands of food items is also requested, and at the end of the 3-d record period, a dietitian visits the family and checks the record for completeness and accuracy. The dietary records are analyzed by using the continuously updated in-house nutrient database LEBTAB (28), which incorporates information from standard nutrient tables, product labels, or recipe simulation based on the labeled ingredients and nutrients (eg, commercial weaning foods).
For the purpose of this study, total energy (kcal/d) and protein intakes (g/d) at all time points between 6 mo and 6 y were derived for each participant from the mean of the respective 3 dietary recording days. The reported energy intake was related to the basal metabolic rate (29), and age- and sex-specific cut-offs (30) were used to exclude potentially implausible records (2.6%). To represent diet at 18–24 mo, 3–4 y, and 5–6 y, we took the mean of single standardized energy intakes (mean = 0, SD = 1) or the mean nutrient intakes at the respective time points.
In addition to total protein intake, we also considered animal protein (excluding protein from human milk because its effect on obesity risk was expected to differ from that of other animal sources and might have confounded their associations) and vegetable protein intake. Animal and vegetable protein intakes were further divided into protein from the following food groups that are known to be main contributors once complementary feeding has commenced (15): 1) dairy protein (eg, from cow milk, custard and other milk desserts, yogurt, buttermilk, and cheese), 2) meat protein (eg, from beef, pork, poultry, and meat products like ham and sausages), and 3) cereal protein (eg, from bread, breakfast cereals, pasta, rice, and flour).
To create these food groups, complementary and convenience foods were broken down into their components, eg, commercial baby meals were divided into vegetables and meat. Infant formula was included in the calculation of total and animal protein, but, as is common practice (15), not in the dairy food group. First, infant formula represents milk feeding, whereas dairy products are first introduced during the complementary feeding period. Second, it is unique to the first years of life and of only minor relevance once the family diet has been established. We also did not consider further food subgroups (eg, protein from fluid milk versus solid dairy foods, protein from full-fat versus reduced-fat dairy), because in the first 2 y of life, these foods were not regularly consumed.
Statistical analysis
To evaluate which time points or periods of protein intake might be most important with regard to BMI and BF% at 7 y of age (objective 1), life-course plots were constructed. This approach, which has been described in detail by Cole (31), deals with the question of how to present the relation between an explanatory variable that has been repeatedly measured and changes during childhood, eg, weight or diet, and an outcome. To use this method in the present analysis, total, animal, and vegetable protein intakes at the ages of 6 mo (ie, 0.5 y), 12 mo (1 y), 18–24 mo (1.5–2 y), 3–4, and 5–6 y were corrected for total energy intake by expressing them as nutrient densities (% of energy) and were standardized (mean = 0, SD = 1) to facilitate comparability. They were then entered into multiple linear regression models as independent variables that were adjusted for each other and with BMI SDS or BF% at 7 y of age as the outcomes. To remove its known effect on BF% values, sex was additionally included in the models with BF% as the dependent variable; in addition, the mean standardized energy intake from 0.5–6 y was entered into all models. The resulting regression coefficients were then plotted against age, and both their values (representing the strength of the relations at a distinct time point) and their changes (representing the associations between outcome and change in the explanatory variables over the corresponding time interval) were evaluated to try to identify sensitive time points or periods of protein intake with respect to later body fatness.
To specifically investigate whether only protein intake from distinct sources could be responsible for potential associations between early total protein intake and later body fatness (objective 2), analyses subsequently concentrated on the critical ages identified by the life-course plot approach. Tertiles of animal, vegetable, dairy, meat, and cereal protein intake at those time points were created, and their association with BMI SDS and BF% at 7 y of age were investigated in depth. First, adjusted mean BMI SDS and BF% outcome levels were calculated for each tertile of intake; second, P values for linear trends were derived from multiple linear regression analyses, treating the nutrient intakes as continuous variables.
Tests for interactions between diet and sex did not indicate a significant effect modification. Thus, all analyses were performed with the total sample of 203 children. Relevant confounders were evaluated on an individual basis and in full models or were considered because of a priori interest. These included sex, maternal overweight (BMI 25; yes or no), maternal educational attainment (12 y schooling; yes or no), maternal age at birth of the child, gestational age, firstborn status (yes or no), smoking in the household (yes or no), and breastfeeding characteristics (full breastfeeding for 4 mo, yes or no; still 50 mL human milk/d at ages 6 or 12 mo, yes or no). Because a considerable number of siblings participate in the DONALD Study, we also took the presence of siblings in our subcohort (yes or no) into account. Furthermore, an indicator variable was created to consider whether children had consumed dairy or meat protein throughout the recording days, but was only retained in the models if it tended to be associated with the outcome (P 0.1) or affected the estimates for protein intake (dairy protein at age 12 mo in the model with BMI SDS, meat protein at age 12 mo in the models with BF% as the outcome).
To adjust for total energy and to be consistent with the life-course plot analyses, we chose the nutrient density approach, ie, all protein variables were expressed as percentages of energy and total energy intake was additionally included (32). In addition, we included fat intake (% of energy) in all models, because it has been proposed that it is not (only) a high protein intake, but also the typical simultaneous decrease in fat intake that might predispose children to later obesity (10, 33). All models with vegetable or cereal protein intake as the independent variable further included fiber intake (g/kcal). Finally, the effect of adding the baseline values of BMI SDS and BF% at the preceding time point (eg, 6 mo for diet at 12 mo, 3–4 y for diet at 5–6 y) was evaluated, taking means of sex- and age-standardized lnBF% values if the preceding time point consisted of more than one measurement (eg, 3–4 y). All statistical analyses were carried out by using SAS (version 8.2; SAS Institute Inc, Cary, NC), and a P value of < 0.05 was considered statistically significant.
RESULTS
A general description of the study sample with respect to birth, breastfeeding, and family characteristics is given in Table 1. Overall, the prevalence of overweight at 7 y of age was moderate, with 1 child in 7 having a BMI (in kg/m2) above the International Obesity Task Force cutoffs of 17.92 for boys and 17.75 for girls, respectively (24). The children's energy and protein intakes at 6 mo, 12 mo, 18–24 mo, 3–4 y, and 5–6 y of age are summarized in Table 2. In general, median protein intake at all ages exceeded the current recommendations [1.4 g · kg–1 · d–1 at 6 mo, 1.2 g · kg–1 · d–1 at 12 mo, 1.0 g · kg–1 · d–1 at 18–24 mo, and 0.9 g · kg–1 · d–1 at 3–6 y (6)]. With respect to food groups, dairy, meat, and cereal products were mainly introduced between 6 and 12 mo.
View this table:
TABLE 1. General characteristics of the DONALD Study sample
View this table:
TABLE 2. Total energy and protein intakes of the DONALD Study sample at different ages in infancy, early childhood, and midchildhood1
The results of the life-course plot analyses interestingly yielded similar results for total and animal protein intake. At both 12 mo (ie, 1 y) and 5–6 y of age, intakes were consistently and, except for total protein intake at 5–6 y and later BF% (P = 0.06), statistically significantly associated with a higher BMI SDS and a higher BF% at 7 y of age (Figure 1, A and B). Total and animal protein consumption at 3–4 y tended to be inversely related to the outcomes, but the CI of its regression coefficients included 0 throughout and hence the association was not significant. Because the same applied to the regression coefficients at the other time points (0.5 y, 1 y, and 1.5–2 y), there was no indication of a period during which a change in protein intake was of importance for later body fatness, because this would have made a clear switch of signs of the coefficients mandatory (34).
FIGURE 1.. Life-course plots of multiple linear regression analyses with BMI SD score (SDS) and the natural log of body fat percentage (lnBF%) at 7 y of age as the outcome and standardized A) total protein, B) animal protein, and C) vegetable protein intakes (% of energy) at different time points throughout infancy and childhood as the explanatory variables, adjusted for each other, mean standardized energy intake from 0.5–6 y, and for sex in the case of lnBF%. n = 203 participants of the Dortmund Nutritional and Longitudinally Designed (DONALD) Study. Values are regression coefficients (95% CI).
By contrast with total and animal protein intakes, the life-course plots of vegetable protein and its association with BMI SDS and BF% at 7 y of age did not indicate significance at any time point or period (Figure 1C). Furthermore, additional adjustments for breastfeeding (full breastfeeding for 4 mo; yes or no) did not influence any of the conclusions (data not presented).
To address the possibility that especially those children with a higher animal protein intake at both 1 y and 5–6 y had a higher body fatness later on, we evaluated the interaction of these factors in an additional regression model. However, there was no suggestion of effect modification (P = 0.3–0.8).
On the basis of the results of the life-course plots, we subsequently concentrated on the ages of 12 mo and 5–6 y in separate models because of age-specific confounders (eg, baseline anthropometrics) to investigate the role of different protein qualities and protein from distinct food groups on BMI SDS and BF% at 7 y of age. As shown in Table 3, a higher animal protein intake (% of energy) at 12 mo was related to a higher BMI SDS at 7 y, which was not explained by family and maternal characteristics or the inclusion of BMI SDS at baseline (P for continuous trend = 0.03). A similar, albeit weaker, tendency was obtained by a full model with animal protein intake at 5–6 y (P = 0.07). By contrast, higher protein consumption from vegetable sources at either 12 mo or 5–6 y was not associated with the outcome (P = 0.8 and P = 0.2, respectively).
View this table:
TABLE 3. Adjusted mean BMI standard deviation scores (BMI SDS) at 7 y of age according to tertiles of protein intake (% of energy) at ages 12 mo and 5–6 y for participants of the DONALD Study (n = 203)1
With respect to food groups, a higher protein intake from dairy foods (% of energy) at 12 mo was associated with a higher BMI SDS at 7 y, similar to animal protein intake (P = 0.02). However, dairy protein intake at 5–6 y did not yield comparable results (P = 0.7). For both 12 mo and 5–6 y, there was no relation of meat or cereal protein intake with later BMI SDS.
Regarding BF% at 7 y of age, the results were comparable overall with those for BMI SDS (Table 4). Again, dairy but not meat or cereal protein intake at the age of 12 mo showed similar associations as did animal protein consumption and tended to be related to a higher later BF% (P = 0.07). A higher animal protein intake at 5–6 y was also associated with BF% at 7 y; however, a similar tendency existed for dairy protein also (P = 0.1). In contrast with BMI SDS, a higher vegetable protein intake at 5–6 y (but not 12 mo) showed an inverse relation to later BF% of borderline significance (P = 0.05). Additional adjustments for other potential confounders, such as breastfeeding or maternal educational status, did not change the results notably, nor did the additional inclusion of the protein intakes at the 3 other time points initially examined (6 mo, 18–24 mo, and 3–4 y; data not shown).
View this table:
TABLE 4. Adjusted mean body fat percentages (BF%) at 7 y of age according to tertiles of protein intake (% of energy) at ages 12 mo and 5–6 y for participants of the DONALD Study (n = 203)1
DISCUSSION
This is the first study to compare the relevance for later body fatness of protein intake from different sources at various time points in infancy and childhood. Our results support the hypothesis that the end of the first year of life, when children undergo the transition from breast milk or formula feeding to a diet based on family foods, may represent a critical phase with respect to protein intake and subsequent obesity risk. Our results indicate that animal, in particular dairy, protein intake might actually be responsible for these associations.
To identify critical phases of protein intake, we evaluated closely spaced, repeatedly collected dietary data and used a life-course plot analysis as proposed by Cole (31). To the best of our knowledge, this approach has not been used in nutritional epidemiology so far, although taking a life-course perspective has been explicitly called for when investigating determinants of later disease (35).
Interestingly, the results of the life-course plot analyses suggested that 5–6 y of age may represent a second sensitive period of total as well as animal protein intake for subsequent body fatness. Most pediatric studies describing associations between protein intake and subsequent obesity measures focused on diet in the first 2 y of life and did not include protein consumption later on (3, 12-14, 36). However, Skinner et al (37) reported associations between mean protein intake in mid-childhood (2–8 y) and BMI at 8 y of age in 70 children. By contrast, another prospective study (n = 112) did not see a relation between protein intake at 8 y of age and BMI change over a period of 4 y (38).
In this context, our results underline that it is crucial to discuss nutrients and sustained dietary consequences within distinct developmental or metabolic stages. At the age of 5–6 y, the children in our cohort typically experience adiposity rebound (11), a potentially critical period for obesity development when weight gain starts to regain importance compared with height gain (10). It may well be that, in addition to complementary feeding, a high animal protein intake at this developmental stage also results in adverse effects on body composition and triggers gain in fat mass, as suggested by our findings. It can further be speculated that at time points other than these 2 possibly critical windows, the potentially favorable effects of higher dietary protein intakes, such as increased satiety, which have predominantly been suggested in adults (39, 40), are of greater relevance, and that different, age- and development-dependent protein effects on body composition might exist. The apparently differential associations of vegetable and dairy protein at 12 mo versus 5–6 y with body fatness at 7 y of age may also be interpreted in this way.
In the fully adjusted models, however, the results for protein intake at 5–6 y of age were overall less convincing than those for 12 mo, eg, because of different results depending on whether BMI SDS or BF% was the outcome. Furthermore, the conclusions that can be drawn from our findings may be limited because of the methodologic limitation of a small time gap between the influencing and the dependent variable (diet at 5–6 y, outcome at 7 y), which may result in difficulties in separating our prospective analyses from a cross-sectional approach. It can therefore not be excluded that the short-term effect seen for protein intake at 5–6 y is lost if an endpoint further away is chosen, as opposed to the effect of intake at 12 mo of age. Here, the sustained effect on body fatness suggests a true programming effect of early diet.
The fact that animal, in particular dairy, protein at age 12 mo might be responsible for the associations reported for total protein consumption agrees with both clinical and observational studies that have suggested a specific effect of these protein sources on IGF-1 and insulin secretion. The discussed mechanism behind the early protein hypothesis is that particularly these hormonal responses adversely affect preadipocyte differentiation and multiplication (3, 5). In a 7-d intervention study in 8-y-old boys (n = 24), a higher protein intake from skimmed milk but not meat increased the serum concentrations of IGF-1 and the molar ratio of IGF-1 to its binding protein IGFBP-3 (17). Interestingly, the children in the milk group simultaneously displayed a significant increase in BMI, but it must be mentioned that the intervention also resulted in a difference in energy intake between the milk and the meat group. A stimulatory effect of animal protein in general, and dairy protein in particular, on the secretion of IGF-1 has further been suggested by cross-sectional, observational studies in children aged 2.5 y (n = 90) (18) and 7–8 y (n = 521–538), respectively. In addition to stimulating IGF-1 secretion, it has further been shown in adults that milk seems to have potent insulinotropic properties (41). Correspondingly, in the 7-d intervention in 8-y-old boys cited above, fasting serum insulin and insulin resistance significantly increased in the milk but not in the meat group (16). With respect to this insulinotropic effect, the whey fraction of milk protein might be responsible (42). However, it is still a matter of debate which milk compound actually stimulates IGF-1 secretion (43). Likewise, it is possible that the adverse effects of dairy protein suggested by the present study stem from other, or at least additional, mechanisms.
Only a few studies have prospectively examined the role of cow milk or dairy foods in the first 2 y of life and obesity measures so far. In a large observational study in infants that was nested within the cluster-randomized "Promotion of Breastfeeding Intervention Trial" PROBIT (n = 17 046), a higher consumption of milk other than breast milk or formula at age 9 mo was associated with a larger weight-for-length z score gain between 9 and 12. This effect was greater than that of formula only, which is characterized by a lower protein content than that of whole-fat cow milk. Comparable with our results, cereal intake was not associated with weight-for-length z score in this time period (44).
In general, however, the long-term effects of cow milk in complementary feeding are still under discussion, and recommendations about when it should first be introduced vary considerably among industrialized countries (45). Should the present results be confirmed, the role of cow milk and dairy products during complementary feeding needs further consideration. A potential adverse role with respect to obesity development would contradict recent findings of a beneficial effect of calcium or dairy for adiposity (46), but the evidence in children is controversial (47-50). Thus, as already discussed for protein intake in general, results obtained from studies in adults should probably not be simply adopted for early childhood. For example, patterns of dairy consumption in the first years of life can be expected to differ considerably from later ones, eg, because of an even higher intake of fluid milk instead of solid milk products.
Several limitations of our study should be mentioned. First, in previous analyses of a subcohort of the DONALD Study, a high total protein intake at 12 mo of age was similarly associated with higher later body fatness but was statistically significant only when the intake had stayed high throughout the second year of life (9). Whether such an interaction with subsequent diet also existed for protein intake from different sources could not be answered by the life-course plot, but is not to be excluded given the fact that 93.6% of the children in the present sample were included in the previous study also. Second, the inclusion of blood samples with IGF-1 measurements would have been advantageous to confirm hypotheses on potential mechanisms but are unfortunately not available in the DONALD Study before young adulthood. Third, we also could not extend our analyses to puberty or young adulthood because the ongoing design of the DONALD Study means that many participants have not yet reached that age. This would have been interesting, especially for the purpose of evaluating the longer-term effect of diet at adiposity rebound. However, the closely spaced, detailed data represent a major strength of our study, as do the large number of family and dietary characteristics that we could consider in our analyses.
In summary, our study suggests that a higher intake of animal protein at age 12 mo, in particular from cow milk and dairy products, might be associated with an unfavorable body composition at the age of 7 y. Furthermore, adiposity rebound (age 5–6 y) might represent a second critical period of protein intake for subsequent body fatness.
ACKNOWLEDGMENTS
We thank the staff of the Research Institute of Child Nutrition for carrying out the anthropometric measurements and for collecting and coding the dietary records and all participants of the DONALD Study.
The contributions of the authors were as follows—ALBG: conducted the statistical analyses and wrote the manuscript; ALBG and AEB: conceived the research project; all authors: made substantial contributions to the study design or the interpretation of the results. None of the authors had any financial or personal conflicts of interest.
REFERENCES