点击显示 收起
1 From the Research Institute of Child Nutrition, Affiliated Institute of the University of Bonn, Dortmund, Germany
2 Supported by the Ministry of Science and Research of North Rhine Westphalia, Germany, and 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, D-44225 Dortmund, Germany. E-mail: guenther{at}fke-do.de.
ABSTRACT
Background: A high protein intake during infancy and early childhood has been proposed to increase the risk of subsequent obesity.
Objective: We analyzed the association of different protein intakes during 6–24 mo with body mass index (BMI; in kg/m2) and percentage body fat (%BF) at 7 y of age.
Design: The analyses included 203 participants of the DOrtmund Nutritional and Longitudinally Designed (DONALD) Study with complete information on early diet (6, 12, and 18–24 mo) and anthropometric data at the age of 7 y. The median of energy-adjusted protein intakes (in g/d) was used to distinguish different patterns of low and high protein intakes throughout the first 2 y of life, which were then related to BMI SD scores (SDSs), %BF, and the risk of overweight and overfatness at 7 y of age.
Results: Although protein intake at 6 mo of age was not associated with the outcomes, a consistently high protein intake at the ages of 12 and 18–24 mo was independently related to a higher mean BMI SDS and %BF at the age of 7 y [BMI SDS: 0.37 (95% CI: 0.12, 0.61) compared with 0.08 (95% CI: –0.09, 0.26), P = 0.04; %BF: 18.37 (95% CI: 17.29, 19.51%) compared with 16.91 (95% CI: 16.19, 17.66%), P = 0.01] and a higher risk of having a BMI or %BF above the 75th percentile at that age [odds ratio for BMI: 2.39 (95% CI: 1.14, 4.99), P = 0.02); odds ratio for %BF: 2.28 (95% CI: 1.06, 4.88), P = 0.03].
Conclusions: High protein intakes during the period of complementary feeding and the transition to the family diet are associated with an unfavorable body composition at the age of 7 y.
Key Words: Dietary protein • complementary feeding • children • obesity • body fatness • cohort study
INTRODUCTION
In view of the rising prevalence of obesity in children and adults (1-3), the identification of dietary risk factors in early life is of paramount importance for prevention. Although nutritional epidemiology has mainly focused on the protective effect of breastfeeding, later infancy and early childhood have received less attention in this context. The complementary feeding period, characterized by both fundamental changes in the child's dietary environment and a high growth velocity, may especially represent a sensitive phase during which nutritional influences exert long-term effects on later development and health (4).
One dietary factor that might play a decisive role during this period of continual introduction of weaning and family foods is protein intake. The early protein hypothesis postulates that a high protein intake during infancy and early childhood results in a metabolic programming of later obesity as a result of adverse hormonal responses (5, 6). Accordingly, a lower protein intake from human milk in comparison to formula was discussed as one possible reason for the beneficial effect of breastfeeding seen in various meta-analyses (7, 8), a hypothesis which is currently being tested in a large European multicenter intervention trial (9). Of potential equal importance is that many infants experience a rapid increase in protein intake during complementary feeding and the transition to a diet based on family foods. Protein intakes typically exceed recommendations 2- to 4-fold during this period, reaching intakes of 5 g/kg body weight or 20% of energy in some populations (10). Compared with the protein content of human milk of 7% of energy, such excessive protein intakes may represent an imbalance that contributes to later obesity risk. In fact, a high protein intake at various times in the first 2 y of life was associated with an earlier adiposity rebound (AR) or a higher body mass index (BMI; in kg/m2) in later childhood in some (4, 5, 11-13), but not all (14, 15), epidemiologic studies.
To date, no study has considered the importance of different "patterns" or combinations of protein intakes during the whole period of complementary feeding and early childhood. In view of the continual changes in the infant and toddler diet however, it might be decisive not to restrict analyses to one single time point. Furthermore, studies that have linked early intake of protein to measures of obesity other than BMI are lacking. One of the main limitations of BMI is that it does not distinguish between lean and fat mass (16), but it is excess fat mass that is regarded as most relevant for the development of obesity-related health problems (17, 18).
We aimed to investigate the impact of protein intakes at 6 mo of age (milk feeding, introduction of complementary feeding), 12 mo of age (complementary feeding, introduction of family diet), and 18–24 mo of age (establishment of family diet) on later BMI and percentage body fat (%BF) in healthy children with prospective data on diet and growth. Specifically, we addressed the question of whether it is a consistently high protein intake during the first year of life, an excessive increase during the weaning period, or both that is associated with a higher BMI and %BF at 7 y of age. We then extended our focus to the second year of life and assessed the impact of maintaining a high protein intake in early childhood.
SUBJECTS AND METHODS
Study population
The DOrtmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study is a longitudinal, open cohort study that was started in 1985 in the area of Dortmund, Germany. Details of the study design were published previously (19). In brief, recruitment of 40 healthy infants each year starts at 3–6 mo of age. From then on, detailed data on nutrition, growth, metabolism, and health status are collected at regular intervals between infancy and young adulthood, namely up to 3 further visits in the first year of life, 2 visits in the second year, and 1 visit 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, data from term (gestational age: 37–42 wk) singletons with a minimum birth weight of 2500 g were used. Furthermore, children had to meet 3 additional criteria: 1) complete anthropometric measurements (weight, height, measurements of skinfold thickness at 4 sites) at 6 mo (baseline) and 7 y (endpoint) of age; 2) plausible dietary records at 6, 12, and 18–24 mo; and 3) information on all relevant confounders. The use of these criteria resulted in a subcohort of 203 children (104 boys, 99 girls) out of 322 children who fulfilled the first criterion and 205 children who met both the first and second criteria. We chose 7 y of age as our endpoint because at that age BMI correlates well with BMI in adulthood (20, 21).
Anthropometric data
At each visit, anthropometric measurements are performed by nurses who are trained according to standard procedures (22). The children are 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 the use of a supine infant weighing scale (PS 15; Mettler, Columbus, OH) or an electronic scale (753 E; Seca, Hamburg, Germany) for subjects in the standing position. Recumbent length in children aged <24 mo is measured to the nearest 0.1 cm with the use of a Harpenden stadiometer (Holtain, Crymych, United Kingdom). Starting at 24 mo of age, standing height is measured to the nearest 0.1 cm with the use of a digital telescopic wall-mounted stadiometer. Starting at 6 mo of age, skinfold thicknesses of the biceps, triceps, subscapular, and suprailiac sites are measured on the right side to the nearest 0.1 mm with the use of a Holtain caliper.
For each child, age- and sex-independent SD scores (SDSs) were calculated for weight, height, and BMI with the use of the German reference curves (23) and for measurements of scapular and triceps skinfold thicknesses with the use of the Tanner and Whitehouse reference data (24). To correct for general deviations of our sample from the reference data all SDS values were then internally standardized (
In the DONALD Study, dietary intake is assessed by 3-d weighed dietary 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 during 3 consecutive days with the help of regularly calibrated electronic food scales [initially Soehnle Digita 8000 (Leifheit AG, Nassau, Germany), now WEDO digi 2000 (Werner Dorsch GmbH, Muenster/Dieburg, Germany)]. Weekdays (71.8%) and weekend days (28.2%) were proportionally distributed in our sample. With regard to breastfeeding, test weighing is performed (ie, weighing the infant before and after each meal) to the nearest 10 g with the use of an infant-weighing scale (Soehnle multina 8300).
Parents are instructed by trained dietitians, and semiquantitative recording (eg, number of spoons, scoops) is allowed when exact weighing is not possible. Information on recipes or the types and brands of food items is also requested. The dietary records are analyzed with the use of the continuously updated in-house nutrient database LEBTAB (27), which is based on information from standard nutrient tables (eg, mature human milk), product labels (eg, most infant or follow-on formula), or recipe simulation based on the labeled ingredients and nutrients (eg, commercial weaning foods).
For the purpose of this study, absolute intakes of energy (in kcal/d) and protein (in g/d) at the ages of 6, 12, 18, and 24 mo were calculated for each participant from the mean of the dietary recording days. The reported energy intake was used as a surrogate measure for the total quality of the dietary records by relating it to the basal metabolic rate. Basal metabolic rate was estimated with the use of the equations of Schofield that include weight and height (28). A cutoff of 0.97, which was suggested for children aged 1-5 y and considers age-specific CVs of energy intake and levels of light physical activity (29), was used to identify implausible records (1.9% of all records between 6 and 24 mo).
To adjust total protein for energy intake and sex, we used the residual method separately for each of the time points and with log-transformed nutrients to improve homoscedasticity (30). With the use of the median of the obtained residuals, the children were then divided into groups of low and high protein intake at 6, 12, and 18–24 mo, respectively. By doing this, we were able to examine different patterns of protein intakes throughout the first 2 y of life. First, only the infancy period (6 and 12 mo) was considered, when 4 elementary patterns of intakes [low-low (L-L), low-high (L-H), high-low (H-L), and high-high (H-H)] existed. On this basis, analyses were then extended to the second year of life, and mean protein intake at 18–24 mo was further included.
Parental characteristics and additional information
On a child's entry to the study, parents were asked to provide information about family characteristics, educational status, and employment. The parents' weight and height were measured by the same trained nurses who assessed the anthropometrics of the participating children. Information on birth weight and length as well as gestational age was abstracted from a standardized document given to all pregnant women in Germany.
Statistical analysis
Descriptive statistics of the study sample across the 4 elementary patterns of protein intake mentioned above at 6 and 12 mo of age were calculated. The differences between groups were analyzed with the use of analysis of variance or Kruskal-Wallis tests for continuous data and chi-square tests for categorical data.
To investigate the association between early protein intake and later obesity in depth, we compared adjusted mean BMI SDS and %BF at 7 y of age according to different protein intakes with the use of analysis of covariance and calculated adjusted odds ratios for being at risk of overweight and overfatness at 7 y of age with the use of logistic regression procedures. Tests for interactions between diet and sex or breastfeeding status did not indicate a significant effect modification. Thus, all analyses were performed with the total sample of 203 children. Relevant confounders, which were evaluated both on an individual basis and in full models, included sex, maternal overweight (yes or no), maternal educational attainment (12 y schooling; yes or no), gestational age, firstborn status (yes or no), smoking in the household (yes or no), and breastfeeding characteristics at 6 mo of age (still receiving 50 mL human milk; yes or no). Because a considerable number of siblings participate in the DONALD Study, we also considered the presence of siblings in our subcohort (yes or no). Furthermore, total energy intakes were logged and standardized within each age group separately, and means for the respective time period were included in every model, because in the analyses covering the whole period from 6 to 24 mo of age not all children provided data from exactly the same time points. Finally, to adjust for simple tracking of BMI SDS and %BF, the effect of adding the baseline values at 6 mo of age was also evaluated.
All statistical analyses were performed with the use of SAS (version 8.2; SAS Institute Inc, Cary, NC). A P value of <0.05 was considered statistically significant.
Power considerations
Sample size analysis was performed with the use of G*POWER (31). This analysis showed that the sample size was adequate to detect a mean difference of 0.40 BMI SDS and 1.10% in body fat between 2 equal groups, and a mean difference of 0.43 BMI SDS and 1.12% in body fat between our final groups of 62 and 141 children, with = 0.05 and a power of 80% (2-tailed).
RESULTS
General characteristics of the study sample by protein intakes at 6 and 12 mo of age (L-L, L-H, H-L, H-H) are presented in Table 1. Children with a low protein intake at 6 mo of age had been fully breastfed more often and for a longer duration than were children with a high protein intake at that age. In addition, the 4 groups differed with respect to maternal education (P < 0.01) and tended toward a difference in maternal weight status (P = 0.08). However, no differences were observed in birth, family, or paternal characteristics among the 4 groups (P > 0.2).
View this table:
TABLE 1. General characteristics of the study sample by protein intake during the first year of life: DOrtmund Nutritional and Longitudinally Designed (DONALD) Study, n = 203
About dietary protein intake during the first 24 mo of age, most children in our study sample exceeded current recommendations (32). However, marked differences were observed between the low and the high protein intake groups at 6 mo (7–8% of energy compared with 12% of energy) and 12 mo (11–12% of energy compared with 14–15% of energy) of age, respectively (Table 2). Furthermore, protein intake did not seem to track during the second year of life in all children, because the differences apparent in the first year of life, although still discernable to some extent, were weaker later on.
View this table:
TABLE 2. Dietary characteristics of the study sample by protein intake in the first year of life: DOrtmund Nutritional and Longitudinally Designed (DONALD) Study, n = 2031
By 7 y of age, the 4 patterns of protein intakes at 6 and 12 mo of age did not differ significantly with the proportion of children who were (at risk of) overweight or overfatness. No significant differences were observed in BMI SDS or %BF at 7 of age either (Table 3). However, we then fit a basic model that separately included diet (H compared with L) at both age groups and an interaction term. For BMI SDS, no interaction was observed between protein intake at 6 and 12 mo of age (P = 0.76) and no significant main effect of diet at 6 mo of age (P = 0.51), but a high protein intake at 12 mo of age was associated with a higher BMI SDS at 7 y of age (P = 0.02). Similar results were obtained for %BF in a model that additionally adjusted for sex (P < 0.0001), thus indicating that it was a high protein intake at 12 mo of age that was decisive, regardless of intake at 6 mo of age (ie, the L-H or H-H groups). Stepwise addition of further confounders to the model and the inclusion of baseline anthropometrics, however, reduced the differences notably (P = 0.05 for BMI SDS and P = 0.19 for %BF, respectively) (data not presented).
View this table:
TABLE 3. Anthropometric characteristics at 7 y of age by protein intake in the first year of life: DOrtmund Nutritional and Longitudinally Designed (DONALD) Study, n = 2031
Next, we extended our analyses to the second year of life and also considered mean protein intake at 18–24 mo of age. Here, our rationale was to investigate whether children with a consistently high protein intake at 12 mo and also 18–24 mo of age (H-H) would be those with the highest risk of later obesity, regardless of protein intake at 6 mo of age (ie, the L-H-H and H-H-H groups). At first, 3-factor analyses with BMI SDS and %BF at 7 y of age as the outcomes were used, including protein intakes at all 3 time points separately, as well as their 2- and 3-factor interactions (data not presented). Because the term representing the potential 3-factor interactions was not significant in both cases (P = 0.42–0.46), it was removed subsequently. In the resulting models with all 2-factor interactions, there were suggestions of positive main effects of both protein intake at 12 mo of age (H compared with L) and at 18–24 mo of age (H compared with L) (P = 0.09–0.13) with regard to BMI SDS and %BF at 7 y of age, respectively, but none of protein intake at 6 mo of age (H compared with L) (P = 0.76–0.78) and no significant interactions (P = 0.31–0.87). However, those children with a consistently high protein intake at 12 and 18–24 mo of age were heavier and fatter than the remainder at 7 y of age, and they were the only ones who were different from those whose intake was low throughout these 2 time points (P = 0.07 for BMI SDS and P = 0.04 for %BF, Dunnett's test for post hoc comparisons of means). We therefore compared these 62 children with a consistently high protein intake (H-H) during the second year of life to the remaining 141 children who had either shown a consistently low or a changeable protein intake throughout early childhood (L-L, L-H, H-L intake at 12 and 18–24 mo of age). To control for any effect attributable to protein intake at 6 mo of age, it was still included in these models (H compared with L) but not its nonsignificant interactions with diet in the second year of life.
As shown in Table 4, BMI SDS and %BF at 7 y of age were significantly higher in the H-H group of children than in the reference group, both in simple and multivariate-adjusted analyses (models 1 and 2). In addition, the significant differences remained after adjustment for anthropometrics at 6 mo of age (model 3; P = 0.04 and P = 0.01, respectively). Protein intake at 6 mo of age had no effect in either model, and further consideration of breastfeeding characteristics at 6 mo of age, gestational age, firstborn status, or smoking in the household did not alter these results.
View this table:
TABLE 4. Adjusted mean BMI SD scores (BMI SDS) and percentage body fat (%BF) at 7 y of age by protein intake in the second year of life: DOrtmund Nutritional and Longitudinally Designed (DONALD) Study, n = 2031
We then investigated the relation of protein intakes during the first 24 mo to obesity risk at 7 y of age in logistic regression models, again comparing children with a consistently high intake (H-H) at 12 and 18–24 mo of age with the remainder. The children in the H-H group had a >2-fold higher risk of having a BMI SDS or %BF >75th percentile at 7 y of age than children with a low or changeable protein intake during the second year of life (Table 5).
View this table:
TABLE 5. Adjusted odds ratios (and 95% CIs) of being at risk of overweight and overfatness according to protein intake in the second year of life: DOrtmund Nutritional and Longitudinally Designed (DONALD) Study, n = 203
Finally, we were interested in how the children with the consistently high protein intake (H-H) at 12 and 18–24 mo of age differed from their reference group with regard to weight and height at 7 y of age and which component was responsible for the differences in BMI SDS. The results suggest that these differences were mainly due to the higher body weight of the children in the H-H group (data not presented in tables). After adjustment for sex, total energy intake, presence of siblings in the data set, maternal weight category (weight no more than reference for height; yes or no), breastfeeding status, and weight SDS at 6 mo of age, those children were still significantly heavier (P = 0.02) than the remainder. In basic analyses, they also appeared to be taller (P = 0.05), but the inclusion of maternal height, which can be interpreted as a person's genetic potential in this case, decreased the differences (P = 0.11). In a subcohort with available data (n = 156), additional adjustment for paternal height resulted in an even further reduction (P = 0.22).
DISCUSSION
To our knowledge, this is the first study to indicate that early protein intakes are associated not only with a higher BMI but also with a higher %BF in mid-childhood. As suggested by our results, a high protein intake at 12 mo of age may be unfavorably related to later adiposity if high protein intakes are maintained throughout the second year of life, whereas protein intake at 6 mo of age showed no association with the outcomes.
The early protein hypothesis, postulating that a high early protein intake predisposes for a higher obesity risk, was first introduced by Rolland-Cachera et al (5). In their French study (n = 112), weak-to-moderate associations were found between a higher protein intake (% of energy) at 2 y of age and both a higher BMI at 8 y of age and an earlier AR. Subsequently, results from small prospective studies in Italy (n = 147) (11) and Iceland (n = 90) (12) supported a positive association between protein intakes at 12 mo of age and obesity status at 5 y of age and protein intake from 6 to 12 mo of age and BMI at 6 y of age in boys, respectively. But the overall evidence in favor of the hypothesis is limited. In a previous analyses of the DONALD Study (n = 313), a higher protein intake between 12 and 24 mo of age was not consistently related to age at AR. However, an association was observed with BMI SDS at AR but only in girls (13). Others could not replicate the findings with regard to age at AR or later BMI, but these studies were characterized by shortcomings such as a small sample size (14, 33) or a follow-up which was probably too short to reliably determine age at AR (15, 34).
Most studies conducted so far have in common that they only considered diet at single time points, and none of them encompassed the whole period of interest (ie, both infancy and early childhood). Thus, this study not only confirms an impact of early protein intake on BMI in mid-childhood as already indicated by others, but it also extends previous work because of the longitudinal character of its dietary data and the evaluation of different patterns of protein intakes during the first 24 mo of life. According to our results, high protein intakes during the period of complementary feeding and the transition to the family diet might be decisive, irrespective of protein intake at 6 mo of age, which did not have an impact in any of our analyses but was considered throughout for reasons of consistency.
With regard to the importance of diet during complementary feeding and thereafter, note that changes during this period are further characterized by a decrease in fat intake from 50% of energy in human milk to intakes 30–35% of energy (4, 10, 35). Consequently, some researchers have argued that a high protein-to-fat ratio might exacerbate coping with a high-fat diet later in life and thus also contribute to obesity development (34, 36, 37).
In addition, no study so far has convincingly shown that a higher intake of protein during infancy or childhood might result in a true accretion of fat mass. On the contrary, dietary protein intake may selectively stimulate an increase of lean body mass (38), highlighting the importance of considering endpoints that assess adiposity more directly than does BMI. In the French study, protein intake (% of energy) at 2 y of age was weakly correlated with measurements of subscapular but not triceps skinfold thickness at 8 y of age after adjustment for parental BMI (r = 0.20; P = 0.04) but not when energy intake or BMI at 2 y of age was accounted for (5). In another study (n = 77 children aged 1.5–4.5 y), no correlation existed between protein intake and %BF assessed by isotope analyses (39). However, the explanatory power of that study was limited because of its cross-sectional design, the small sample size, and insufficient consideration of diet before 2 y of age. In a Danish cohort (n = 105), no association was seen between protein intake at 9 mo of age and both BMI and body fat (assessed by dual-energy X-ray absorptiometry) at 10 y of age (14). However, a relation to height at 10 y of age was observed, which disappeared when body size at baseline was adjusted for. The prevalence of overweight at 10 y of age was low in this sample, with only 8 children being overweight and none being obese. In contrast, our study suggests an association between early protein intakes and later BMI and %BF but not height. In the DONALD Study, %BF is estimated from skinfold-thickness measurements. Disadvantages of this technique lie in a probably limited transferability of the equations to populations other than those from which they were derived and its susceptibility to measurement error (40, 41). However, measurements were conducted by trained personnel who were monitored for quality, which has been shown to reduce intra- and interobserver variability considerably (42).
The mechanism behind the early protein hypothesis is thought to be that an increased secretion of insulin and insulin growth factor I triggers the multiplication and differentiation of preadipocytes (5, 6, 9), and infancy and early childhood may represent a "time window" when high protein intakes can exert these adverse effects. Interestingly, researchers from a recent British study also suggested that especially dietary factors in the second year of life might be decisive for obesity risk and pointed out that adverse BMI values seem to be well established by 3 y of age (43). Hence, an initial stimulus of a high protein intake on adiposity, probably followed by simple tracking in mid-childhood, could be responsible for the higher obesity risk. This idea is supported by studies suggesting that protein intake during the complementary feeding period might stimulate short-term weight gain (44, 45), as well as the identification of rapid growth during infancy and early childhood as a risk factor for obesity (46-49).
Our study has several limitations. First, we are aware that our results are based on a relatively small study sample and that our consistent but overall not highly significant associations may stem from this fact. Because of the sample size, it was also not possible to follow all potential combinations of protein intakes at 6, 12, and 18–24 mo of age (L-L-L, H-L-L, L-L-H, H-L-H, L-H-L, H-H-L, L-H-H, and H-H-H) separately throughout or to statistically confirm interactions among intakes which were indicated by absolute numbers. For comparisons of our final comparison groups, however, our study was sufficiently powered. Second, the participants of the DONALD Study are characterized by a higher socioeconomic status than the German population (19). Thus, they probably do not represent extremes of behavior or weight status. The non-representativeness of our sample, however, should be of minor importance when looking at internal validity and cause-effect relations. Besides, several evaluations showed no or only minor deviations between the dietary habits seen in the DONALD Study and those in previous or more recent German surveys (50, 51). Third, our relatively homogeneous sample may have limited the range in protein intakes of the participants. In fact, we did not see protein intakes as excessive as those reported in other European studies (37). However, the homogeneity of our sample might have reduced our vulnerability to residual confounding by behavioral or socioeconomic factors.
A clear strength of our analyses lies in the carefully collected, repeated data on growth and in a large number of possible confounders such as parental characteristics and diet. Weighed dietary records may provide limited information about habitual diet when only collected for 3 d, but 3–4 d were found to be acceptably accurate to classify infants and toddlers into groups of macronutrient intake because of the fairly uniform diet of that age group (52). In summary, our results suggest an association of high protein intakes during complementary feeding and the transition to the family diet with both a higher BMI and a higher body fatness at 7 y of age.
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 Nadina Karaolis-Danckert for proofreading the final draft. We also thank all participants of the DONALD Study.
The authors' responsibilities were as follows—ALBG: study design, statistical analyses, interpretation of results, and writing of the manuscript; AEB and AK: study design and interpretation of results. None of the authors had any financial or personal conflicts of interest.
REFERENCES