AJCN Cancer Health Disparities Conference
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Johnson, L.
Right arrow Articles by Jebb, S. A
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Johnson, L.
Right arrow Articles by Jebb, S. A
Agricola
Right arrow Articles by Johnson, L.
Right arrow Articles by Jebb, S. A
American Journal of Clinical Nutrition, Vol. 87, No. 4, 846-854, April 2008
© 2008 American Society for Nutrition

ORIGINAL RESEARCH COMMUNICATION

Energy-dense, low-fiber, high-fat dietary pattern is associated with increased fatness in childhood1,2,3,4

Laura Johnson, Adrian P Mander, Louise R Jones, Pauline M Emmett and Susan A Jebb

1 From the Medical Research Council (MRC) Human Nutrition Research, Cambridge, United Kingdom (LJ, APM, and SAJ), and the Unit of Paediatric and Perinatal Epidemiology, Department of Social Medicine, University of Bristol, Bristol, United Kingdom (LRJ and PME)

2 This article is the work of the authors, and Laura Johnson and Susan Jebb serve as guarantors for the content.

3 Supported by the UK Medical Research Council. LJ holds a UK Medical Research Council PhD studentship. The UK Medical Research Council, the Wellcome Trust, and the University of Bristol provide core support for the Avon longitudinal study of parents and children.

4 Address reprint requests to L Johnson, MRC Human Nutrition Research, Elsie Widdowson Laboratory, 120 Fulbourn Road, Cambridge, CB1 9NL, United Kingdom. E-mail: l.johnson{at}public-health.ucl.ac.uk.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Background: Evidence for the dietary determinants of obesity in children is limited.

Objective: The objective was to identify a dietary pattern that explained dietary energy density (DED), fiber density (FD), and percentage of energy intake from fat and analyze its relation to fatness in children.

Design: The subjects were 521 (at ages 5 and 9 y) and 682 (at ages 7 and 9 y) children participating in the Avon Longitudinal Study of Parents and Children. Diet was assessed with the use of 3-d diet diaries at ages 5 and 7 y. Reduced rank regression derived a dietary pattern with the use of DED, fiber, and fat intake as intermediate variables. Fat mass was measured at age 9 y with the use of dual-energy X-ray absorptiometry. Fat mass index (FMI) was calculated, and excess adiposity was defined (as the top quintile of logFMI).

Results: Pattern score at ages 5 and 7 y was correlated with DED (r = 0.8), FD (r = –0.7), and percentage of energy intake from fat (r = 0.5). An increase of 1 SD of pattern score at ages 5 and 7 y, respectively, was associated with a 0.15-kg (95% CI: –0.1, 0.45 kg) and a 0.28-kg (95% CI: 0.05, 0.53 kg) higher fat mass at age 9 y, after controlling for confounders. The adjusted odds of excess adiposity at age 9 y for children in quintile 5 compared with quintile 1 of dietary pattern score at ages 5 and 7 y, respectively, were 2.52 (95% CI: 1.13, 6.08) and 4.18 (95% CI: 2.07, 9.38).

Conclusion: An energy-dense, low-fiber, high-fat diet is associated with higher fat mass and greater odds of excess adiposity in childhood.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The World Health Organization has identified energy density and fiber as important dietary factors for determining obesity risk (1). However, robust evidence from prospective studies for those specific determinants of obesity in children is limited (2, 3). Analyzing the effect of single nutrients or food groups alone ignores the inherent complexity of diet. Nutrients are not eaten in isolation; they are integrated within foods with other nutritional features. For example, energy-dense foods tend to be high in fat and low in fiber. Thus, a dietary pattern based on the energy density and fiber and fat contents of the diet may be more informative of obesity risk than looking at each factor in isolation.

Traditionally, 2 approaches were used for analyzing patterns in dietary data. Hypothesis-driven approaches (eg, diet indexes) are based on existing evidence, whereas data-driven methods (eg, principal components analysis) are exploratory. In the first approach, prior knowledge prevents new hypotheses from being generated, and, in the second, prior knowledge is ignored completely (4). Reduced rank regression (RRR) uses both existing knowledge and exploratory statistics and is a more powerful technique for extracting dietary patterns that are related to a specific disease (5). A limitation of RRR is that it requires existing evidence that diet affects disease by a set of continuous intermediate variables. Previous applications of RRR have used nutrients or biomarkers as intermediate variables; however, a method for identifying the best set of intermediates for a particular disease is unknown (4). If evidence for a set of appropriate intermediates is lacking or if appropriate intermediates are known but data are not available, then RRR cannot be used (6).

Studies of dietary patterns and obesity to date have been largely limited to adults. Cross-sectional studies report conflicting relations (7). Prospective designs yield more consistent associations, indicating that dietary patterns high in fruit, vegetables, and fiber and low in high-fat dairy products, sweets, and processed meat are associated with smaller gains in weight, body mass index (BMI; in kg/m2), and waist circumference (8-10). Few studies have characterized dietary patterns in European children (11, 12). In the Avon Longitudinal Study of Parents and Children (ALSPAC) a "junk" pattern, including soft drinks, crisps, chocolate, and takeaway food, and a "healthy" pattern, including fruit juice, salad, pulses, rice, and pasta, were not associated with obesity at 7 y of age after adjustment for confounders (13). A recent study of dietary patterns in 5-y-old Korean children found that a pattern including fast foods and animal products increased the risk of overweight (defined by BMI > 85th Korean reference percentiles) by 77% (range: 6–294%) (14). One further study of dietary patterns in US adolescents showed that a cluster of girls eating a healthy dietary pattern had a lower waist circumference after 10 y of follow-up, although this pattern was consumed by only 12% of the sample (15).

The aim of this study was to identify a dietary pattern characterized by 3 risk factors for obesity, namely dietary energy density (DED), fiber density (FD), and percent of energy from fat with the use of RRR and to analyze the association with fatness. This extends previous work based on the ALSPAC cohort (12) by using 3-d diet diaries at ages 5 and 7 y and subsequent fatness measured at age 9 y.


   SUBJECTS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Sample
Data from ALSPAC was used. ALSPAC is a prospective cohort study, started in 1991, to assess all aspects of pregnancy, infancy, and childhood growth and development. A detailed account of study methods can be found elsewhere (16). Briefly, all pregnant women in Avon with an expected date of delivery between 1 April 1991 and 31 December 1992 were eligible for recruitment. A total of 14 541 pregnant women were enrolled in the study. Extensive data were collected on the parents and their children primarily with the use of questionnaires, as well as medical records, biological samples, and clinical data. Children in Focus (CIF) is a random subsample of 1432 children selected from the cohort born in the last 6 mo of the recruitment that were invited for regular clinical assessments from birth. The present analysis includes data on diet at ages 5 and 7 y and body composition at age 9 y. Complete data on diet and body composition were available for 521 (36% of CIF) children at ages 5 and 9 y and 682 (48% of CIF) children at ages 7 and 9 y. Parents provided informed written consent for their child to participate in the study, which was approved by the ALSPAC Law and Ethics Committee and local research ethics committees.

Dietary data
Dietary data were collected at a mean (±SD) age of 5.2 ±0.1 y and 7.4 ± 0.1 y with the use of 3-d unweighed diet diaries. This method of dietary data collection was shown, in another cohort, to be highly correlated with data collected with the use of weighed dietary records (17). Parents completed the diary on behalf of the child before attending the clinic. Diet recordings were requested to cover 2 weekdays and 1 weekend day and were bought to the clinic by parents. Diet coding of all food and supplements was done with the use of DIDO software (MRC Human Nutrition Research, Cambridge, United Kingdom) (18). Portion sizes were coded with the use of household measures, and, when portion sizes were missing, a standard child's portion size for age 5 y or age 7 y was used, based on data from the 1997 UK National Diet and Nutrition Survey (19). Nutrient data were generated with the use of coded data on all food and supplements with an in-house program that used composition information from the 5th edition of McCance and Widdowson's food tables and supplements (20). Food intake data were collapsed into 46 food groups according to usage or differences in energy density and fat and fiber contents (foods listed in Appendix AGo). Average absolute intakes (in g/d) were computed for each food group.


View this table:
[in this window]
[in a new window]

  		APPENDIX A Food groups and their contents

 
RRR was used to derive a dietary pattern score (5). RRR derives patterns that maximize the variation explained in a set of response variables, which are believed to be related to the outcome of interest. DED, FD, and percentage of energy intake from fat were selected as intermediate response variables, based on evidence of a relation to obesity from the literature, primarily in adult studies (21-24). DED was calculated, excluding drinks, by dividing total food energy (in kJ) by total food weight (in g) (25). Percentage of energy intake from fat was calculated by dividing kilojoules from fat by total energy intake (EI; in kJ) and then multiplying by 100. FD (in g/MJ) was calculated by dividing nonstarch polysaccharide fiber intake (in g) by total EI (in MJ). With the use of SAS (version 9; SAS Institute, Cary, NC) PROC PLS with RRR method option, 3 patterns were extracted in accordance with the number of response variables included. The first pattern extracted at both 5 and 7 y of age explained 47% of the response variation at each age. The second and third patterns explained <20% response variation each at each age and were not used in subsequent analyses (age 5 y, second pattern 15% and third pattern 11%; age 7 y, second pattern 16% and third pattern 11%). RRR was run in the whole sample and separately in boys and girls. Because the sex-specific patterns were qualitatively the same as those derived from the whole sample, indicated by similar pattern loadings, the pattern loadings derived from the whole sample were used in subsequent analyses (Table 1Go). Pattern score was calculated for each child as a linear combination of all food group intakes.


View this table:
[in this window]
[in a new window]

  		TABLE 1 Top 5 food groups, according to loadings, positively and negatively associated with pattern score at ages 5 and 7 y

 
Misreporting of EI was assessed, with the use of an individualized method (26), by calculating the ratio of reported EI to estimated energy requirement (EER) (EI:EER). Individual EERs were calculated with the use of equations from the FAO/WHO/UNU Expert Consultation Report on Human Energy Requirements (27). A 95% CI for the accuracy of EI:EER was calculated by taking into account the amount of variation inherent in the methods used to estimate EI and EER (28). The 95% CI for EI:EER, calculated for this data set, was 0.79 and 1.21 at both ages. Therefore, reports of EI between 79% and 121% of EER can be considered to be within the range of normal measurement error associated with estimating EI and EER and were defined as plausible reports. With this definition, 72% and 73% of 5- and 7-y-olds, respectively, had plausible reports of EI. However, the prevalence of overweight was significantly higher among underreporters (25). Excluding implausible reporters would have removed the very children of greatest interest, so the categorical misreporting status variable (underreporter; plausible reporter; overreporter) was used as a covariate in the analysis of the relation between dietary pattern and fatness.

Fat mass data
Fat mass (in kg) was estimated at a mean (±SD) age of 9.8 ± 0.15 y from dual-energy X-ray absorptiometry (DXA) with the use of the Lunar Prodigy DXA fan beam scanner (GE Medical Systems Lunar, Madison, WI). DXA estimates of fat mass in children are highly correlated with estimates derived from a 4-component model (29). Fat mass index (FMI) was calculated by dividing fat mass (in kg) by height (in m5.8) to adjust for body size. The optimal power (5.8) to raise height to was derived from the data so that the relation between fat mass and height was completely removed (30). FMI was log transformed to remove positive skew and was used for descriptive purposes. There is no accepted cutoff to define excess adiposity with the use of fat mass, percentage of body fat, or FMI. With the use of age- and sex-specific BMI cutoffs from the International Obesity Task Force (31), a 20% prevalence of overweight was found in this sample at 9 y of age. For comparability, it was assumed an equivalent percentage of children could be defined as having excess adiposity; therefore, those children in the top 20% or quintile 5 of logFMI were categorized in this way.

Potential confounders
At 5, 7, and 9 y of age, height was measured to an accuracy of 0.1 cm with the use of a Harpenden stadiometer (Holtain Ltd, Crymmych, Pembs, United Kingdom), and weight was measured to 0.1 kg with the use of Tanita (West Drayton, Middlesex, United Kingdom) Body Fat Analyzer weighing scales. BMI was calculated, and overweight status was defined by age- and sex-specific cutoffs for BMI (31). The average time spent by children watching television (TV) was reported by parents in questionnaires sent out at 4.5 y of age and was subsequently defined as <1 h/d, 1–2 h/d, and >2 h/d. Parental socioeconomic information (occupation and education) and prepregnancy heights and weights were self-reported by questionnaires sent to the mother at 32 wk of gestation.

Statistical analyses
Variables are described with the use of mean ± SD when symmetric and median and interquartile range otherwise. Tracking of dietary patterns from age 5 y to age 7 y was assessed by calculating an intraclass correlation coefficient. Pearson's r correlation coefficients were calculated to assess the relation between 2 continuous variables. Differences in dietary pattern scores at ages 5 and 7 y by categorical variables were assessed by one-factor analysis of variance. Linear regression analysis was used to model the effect of dietary pattern score at ages 5 and 7 y on fat mass at age 9 y. In linear regression, log fat mass, rather than logFMI, was used as the dependent variable with height as a covariate in the model to adjust for body size to improve interpretability. Estimates of the difference in log fat mass at age 9 y for a 1-unit difference in pattern score at ages 5 or 7 y were back transformed to show change in fat mass (in kg). Logistic regression analysis was used to model the effect of dietary pattern score at ages 5 and 7 y on the odds of excess adiposity at age 9 y. Misreporting status at baseline was controlled for in a separate model. Other potential confounders were identified, based on a search of the literature for known risk factors for obesity. When there was evidence that these risk factors for obesity may also be related to diet, and data were available, they were included in multiple regression models to assess their effect on the relation between dietary pattern and fat mass. The covariates included in the full model were child overweight status at baseline (age 5 or 7 y), TV watching, socioeconomic status, and parental BMI. Controlling for EI may not be appropriate in studies of diet and obesity because it may lie on the causal pathway rather than confound it; the full model was run with and without EI to establish this. All analyses were completed with the use of SPSS version 11.0 (SPSS Inc, Chicago, IL).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Dietary pattern score in relation to diet
Food groups with the top 5 positive and negative loadings for the dietary pattern at ages 5 and 7 y are presented in Table 1Go. The pattern score at 5 y of age was highly correlated with DED (r = 0.79) and FD (r = –0.74) and moderately correlated with percentage of energy intake from fat (r = 0.49). Similar correlations were observed for the pattern score derived at 7 y of age (DED: r = 0.81; FD: r = –0.73; percentage of energy intake from fat: r = 0.49). Pattern loadings at both ages indicated that a high pattern score was associated with a low consumption of fresh fruit and vegetables and a high consumption of crisps and snacks, chocolate, and confectionery. At 5 y of age, low-fiber bread had the highest positive loading, whereas at 7 y of age it was only third highest behind the crisps and snacks group and the chocolate and confectionery group. Tracking of pattern score between ages 5 and 7 y was strong (intraclass correlation coefficient: 0.66; 95% CI: 0.60, 0.71), indicating that, in general, children with high pattern scores at 5 y of age also had a high pattern score at 7 y of age. Although the pattern scores are not identical for each age, pattern loadings for food groups indicate they reflect broadly similar patterns of consumption.

There was no evidence of an association between EI and quintile of dietary pattern score at 5 y of age (Table 2Go), although there was evidence of a weak direct correlation (r = 0.10, P = 0.01). The increase in EI from quintile 1 to quintile 5 of pattern score at 7 y of age was greater than at 5 y of age and was associated with a stronger direct correlation with pattern score (r = 0.13, P < 0.0001). At both ages 5 and 7 y, those children in quintile 5 of pattern score had a significantly higher DED and lower FD than did children in quintile 1. Consumption of fruit and vegetables decreased dramatically from quintile 1 to quintile 5 of pattern score at both ages 5 and 7 y. In contrast, there was a comparatively small increase in the consumption of crisps and snacks or chocolate and confectionery from quintile 1 to quintile 5.


View this table:
[in this window]
[in a new window]

  		TABLE 2 Descriptive dietary characteristics by quintile (Q) of dietary pattern score at ages 5 and 7 y1

 
Dietary pattern score in relation to child fatness
At 9 y of age, median (interquartile range) fat mass was 7.35 kg (4.79–10.88 kg) and FMI was 1.06 kg/m5.8 (0.74–1.57 kg/m5.8). No direct trend was observed in fat mass at 9 y of age across quintiles of pattern score at 5 y of age, with the highest fat mass observed in quintile 3 (Table 3Go). Similarly, no direct trend was observed in fat mass at 9 y of age across quintiles of pattern score at 7 y of age. After adjusting for body size with FMI, a trend was observed toward increasing adiposity from quintile 1 to quintile 5 of pattern score at 7 y of age. A trend was also observed toward a higher prevalence of excess adiposity with quintile of pattern score at 7 y of age, in which the prevalence of excess adiposity increased from 13% in quintile 1 to 23% in quintile 5.


View this table:
[in this window]
[in a new window]

  		TABLE 3 Child body size by quintile (Q) of dietary pattern score at ages 5 and 7 y

 
Regression analysis indicated that a higher pattern score at 5 y of age was associated with higher fat mass at 9 y of age; however, these results were not statistically significant at P < 0.05, and the effect was attenuated after controlling for potentially confounding variables (Table 4Go). An increase (by 1 SD) in pattern score at 7 y of age was associated with an extra 0.28 kg (95% CI: 0.05, 0.53 kg) of fat mass at 9 y of age, after controlling for misreporting of EI, maternal education, maternal BMI, overweight status at baseline, and TV watching. This association was attenuated to 0.23 kg (95% CI: 0.002, 0.47 kg) by the inclusion of total EI as a covariate.


View this table:
[in this window]
[in a new window]

  		TABLE 4 Relation between fat mass (in kg) at age 9 y and dietary pattern score at ages 5 and 7 y

 
A positive linear trend was observed between the odds of excess adiposity at 9 y of age and quintile of dietary pattern score at 5 y of age (P = 0.03) and at 7 y of age (P < 0.0001) (Figure 1Go). The adjusted odds of excess adiposity at 9 y of age for children per quintile of dietary pattern score at ages 5 and 7 y, respectively, were 1.26 (95% CI: 1.03, 1.57) and 1.43 (95% CI: 1.18, 1.73). At 5 y of age a child in quintile 5 compared with quintile 1 was 2.52 (95% CI: 1.13, 6.08) times more likely to have excess adiposity at 9 y of age. At 7 y of age the effect was larger so that a child in quintile 5 compared with quintile 1 was 4.18 (95% CI: 2.07, 9.38) times more likely to have excess adiposity at 9 y of age.


Figure 1
View larger version (7K):
[in this window]
[in a new window]

  		FIGURE 1. Odds of excess adiposity at age 9 y by quintile of dietary pattern score at ages 5 ({blacktriangleup}) and 7 ({blacksquare}) y. Models were adjusted for sex, misreporting status, maternal BMI, maternal education (5 categories), overweight status (by BMI) at baseline, and television watching at 54 mo (<1 h/d, 1–2 h/d, or >2 h/d). Test for linear trend for pattern score at age 5 y (P = 0.03) and at age 7 y (P < 0.0001) was performed. Test for interaction between age and quintile of dietary pattern score on odds of excess adiposity (P = 0.81) was performed. OR, odds ratio; Q, quintile; FMI, fat mass index.

 
Dietary pattern score in relation to potential confounders
Maternal education was negatively associated with pattern score at both ages, indicating that children of degree-educated mothers are more likely to consume a diet of low-energy density, high-fiber, and low-fat content (Figure 2Go). No direct relation was observed between pattern scores at ages 5 or 7 y and maternal BMI (data not shown). TV watching at age 4.5 y was positively associated with dietary pattern score at both ages, indicating that children watching TV >2 h/d are more likely to consume an energy-dense, low-fiber, high-fat diet (Figure 3Go).


Figure 2
View larger version (24K):
[in this window]
[in a new window]

  		FIGURE 2. Dietary pattern score at ages 5 and 7 y by maternal education. ANOVA was used to test for differences across categories at age 5 y (P = 0.001) and at age 7 y (P = 0.01). Test for interaction between age and maternal education on dietary pattern score (P = 0.14) was performed. CSE, Certificate of Secondary Education.

 

Figure 3
View larger version (25K):
[in this window]
[in a new window]

  		FIGURE 3. Dietary pattern score at ages 5 and 7 y by television (TV) watching at 54 mo. ANOVA was used to test for differences across categories at age 5 y (P < 0.0001) and at age 7 y (P = 0.001). Test for interaction between age and TV watching on dietary pattern score (P = 0.64) was performed.

 
The effect of missing data was assessed (data not shown). Children who attended clinics were more likely to come from more affluent or better-educated families than were children that did not attend clinics. However, there appeared to be no differences in dietary and anthropometric variables between children who attended clinics at ages 5, 7, and 9 y who also had complete data for all confounders than did children without this data, indicating that data on confounders were unlikely to introduce substantial bias.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
In these analyses we identified a dietary pattern at ages 5 and 7 y that was associated with a high DED, low-fiber intake, and a high-fat intake. Higher pattern scores were prospectively associated with greater fat mass and higher odds of excess adiposity at 9 y of age. For each quintile of dietary pattern score at 7 y of age, the odds of excess adiposity rose by 43%, meaning that children in quintile 5 of dietary pattern score were >4 times more likely to have excess adiposity than were children in quintile 1. The effect of this dietary pattern on fatness is much greater than the effect of DED, FD, or percentage of energy intake from fat alone (25, 32). The results presented are the first to show prospectively a direct relation between an energy-dense, low-fiber, high-fat dietary pattern and increased fatness in childhood. The weaker effect observed for diet at 5 y of age compared with diet at 7 y of age may reflect the longer duration between measurement and follow-up; however, analysis of the pattern score at 5 y of age in relation to BMI and overweight (defined by BMI) at 7 y of age did not show a relation. Alternatively, the weaker relation may be explained by a deterioration in the innate ability of children to match EI to energy needs (33, 34), meaning that younger children are more able to compensate at subsequent meals for higher EIs than are older children, thus preventing excess weight or fat gain in the long term.

The pattern we have identified with RRR is similar to those identified in adults as related to weight gain. For example, Schulz et al (10) found a low-fiber, high-fat pattern, using RRR, was associated with greater weight gain. A US study that used cluster analysis found that an "empty calorie" pattern, characterized by high consumption of refined grains, sweets, and desserts as well as low-fiber intake, was associated with overweight status in women (9). With the use of both cluster and principal components analysis, Newby et al (8, 35) found a pattern high in fruit, vegetables, whole grains, and low-fat dairy products was associated with smaller gains in waist circumference in both men and women from the United States.

Previous dietary patterns derived in ALSPAC with food-frequency questionnaire data at 3 y of age and principal components analysis (PCA) (12) were not associated with obesity at 7 y of age (13), which is in contrast to the results in the current study with data from the same sample. The contrast in findings may be explained in several ways. 1) The method used to collect dietary data at 3 y of age was a food-frequency questionnaire rather than a 3-d diet diary, which provides a more detailed measure of individual diet and is less dependent on standard portion sizes and assumptions about foods eaten. 2) PCA was used to derive dietary patterns at 3 y of age, whereas RRR was used in the current analysis. The pattern extracted by RRR may be more pertinent to the outcome than a pattern extracted with the use of PCA, which simply reflects a common pattern of food consumption. 3) The outcome in the previous analysis, obesity, was defined by BMI, which may be less sensitive than a direct estimate of fat mass, such as DXA. This reduced sensitivity of BMI is supported by data from our analysis, which showed a weaker relation between pattern score at 7 y of age and overweight (defined by BMI) at 9 y of age compared with excess adiposity (defined by FMI) at 9 of age. 4) Finally, and most importantly, there may actually be a weaker effect of dietary pattern at 3 y of age on obesity because innate appetite control has been shown to be better in younger than in older children, suggesting that a junk food pattern may not lead to increased EI at this age. Very young children may compensate for high EIs from junk food at subsequent eating occasions, meaning weight gain in the long term may be avoided (33, 34).

We found a relation between dietary pattern score and maternal education, which is similar to the association observed in 2 previous analyses of dietary patterns in the ALSPAC and another analysis of dietary patterns in Spanish children (11, 12, 36). In the present analysis the relation between maternal education and child diet appears weaker at 7 y of age than at 5 y of age (although there was no statistical evidence of an interaction between education and age; P = 0.14), suggesting that older children may take more control over their own diet. TV watching was also associated with dietary pattern score, which was observed in a previous study of Spanish children (11).

Note that the pattern loadings for fruit and vegetables were double that of the loadings for crisps and confectionery, which indicates that, when all other food groups stay the same, a difference of 1 unit in fruit or vegetable intake has double the effect on dietary pattern score than with a similar change in crisps or confectionery. This difference in effect of specific food groups on dietary pattern score emphasizes the importance of consuming low-energy dense and high-fiber foods, such as fruit and vegetables, to achieve a healthier pattern of growth. The difference reinforces efforts to encourage consumption of fruit and vegetables as part of a healthy dietary pattern, rather than focusing solely on the exclusion of high-fat, low-fiber foods such as crisps and confectionery.

The association observed in the current analysis between an energy-dense, low-fiber, high-fat pattern at 7 y of age and increased fat mass at 9 y of age is independent of total EI. Independence from EI may be explained by the error associated with the measurement of EI, although we have attempted to address this by controlling for misreporting of EI. Alternatively, it may reflect the physiologic effects of fiber on the energy value of the diet. Many studies of adults have observed an increased amount of fat excreted in stools when the fiber content of the diet is increased (37). In one study a difference of 10 g/d fiber intake was associated with a 0.7-g and 44-g increase in stool fat content and total stool weight, respectively, equivalent to 44 kcal/d ({approx}184 kJ/d) ingested but not absorbed by the body (38). A difference in fiber intake of {approx}5 g/d was observed between quintile 1 and quintile 5 at 7 y of age in the ALSPAC (data not shown). If the effect of fiber on fat excretion in adults is also true for children, then this difference in fiber intake could be associated with a 22-kcal/d ({approx}92 kJ/d) reduction in energy absorbed. During 1 y, this reduction in energy absorption would be equivalent to 8008 kcal/y ({approx}33 500 kJ/y) and may explain a 1.0-kg difference in fat mass (assuming 1 kg of fat provides 7800 kcal). Future research should establish the effect of fiber on the metabolisable energy of diets in children.

There are several strengths to this study. Dietary patterns were derived with the use of RRR, which incorporates a prespecified pathway to disease, resulting in the extraction of a pattern pertinent to the outcome (39). The data analyzed come from a large population-based cohort, which was followed prospectively for 9 y with detailed records of diet and good measures of a range of potential confounders. Importantly, adiposity was based on a direct measure of fat mass rather than BMI, which can be misleading at the individual level (40). Misreporting of EI was more common among overweight children and may bias results; however, in this study we have attempted to quantify this bias and analyzed its effect on the relation between pattern score and fatness. A limitation of this study is the lack of data on physical activity, a potential source of uncontrolled confounding. Future studies should incorporate data on both diet and activity behaviors collected simultaneously to further investigate the relation between dietary patterns, activity, and fatness. The current study was not representative of nonwhite and less-affluent families because only a small proportion were included in this sample; furthermore, only 36% and 48% of the original random sample (CIF) had a complete set of data for this study at ages 5 and 7 y, respectively. However, the prevalence of overweight and dietary intakes observed in these children were comparable to that observed in nationally representative samples of young UK children (41, 42).

In conclusion, the findings presented indicate that an energy-dense, low-fiber, high-fat diet is associated with greater fatness 2 or 4 y later. The smaller effect size observed for diet at 5 y of age compared with 7 y of age may reflect a deterioration in the innate ability of children to match EIs to energy needs, leading to greater weight gain in response to the same pattern of diet in older children. However, there was evidence of tracking of dietary patterns within childhood, which suggests that interventions to change diet should start at an early age.


   ACKNOWLEDGMENTS
 
We thank all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. The UK Medical Research Council, the Wellcome Trust, and the University of Bristol provide core support for the Avon Longitudinal Study of Parents and Children. We also thank Petra Lahmann for advice on composing food groups.

The author's responsibilities were as follows—LJ and SAJ: had the idea for the analysis and wrote the report; LJ: analyzed data; APM: provided statistical expertise; LRJ: entered, collated, and cleaned the dietary data; PME: designed and managed the collection and entering of the dietary data. All authors were responsible for critical revisions and final approval of the manuscript. Medical Research Council (MRC) Human Nutrition Research is a member of a number of advisory boards for the food industry and conducts collaborative projects with these industries. No external organizations were involved in any part of this analysis, which was funded wholly by the Medical Research Council. None of the authors had a personal or financial conflict of interest.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Diet, nutrition and the prevention of chronic diseases. Report of a Joint WHO/FAO Expert Consultation. World Health Organ Tech Rep Ser 2003:916:i–viii, 1–149.
  2. Rennie KL, Johnson L, Jebb SA. Behavioural determinants of obesity. Best Pract Res Clin Endocrinol Metab 2005;19:343–58.[Medline]
  3. Newby PK. Are dietary intakes and eating behaviours related to childhood obesity? A comprehensive review of the evidence. J Law Med Ethics 2007;35:35–60.[Free Full Text]
  4. Michels KB, Schulz M. Can dietary patterns help us detect diet–disease associations? Nutr Res Rev 2005;18:241–8.
  5. Hoffmann K, Schulze MB, Schienkiewitz A, Nothlings U, Boeing H. Application of a new statistical method to derive dietary patterns in nutritional epidemiology. Am J Epidemiol 2004;159:935–44.[Abstract/Free Full Text]
  6. Kroke A. RE: Application of a new statistical method to derive dietary patterns in nutritional epidemiology. Am J Epidemiol 2004;160:1132–5.[Free Full Text]
  7. Togo P, Osler M, Sorensen TI, Heitmann BL. Food intake patterns and body mass index in observational studies. Int J Obes Relat Metab Disord 2001;25:1741–51.[Medline]
  8. Newby PK, Muller D, Hallfrisch J, Andres R, Tucker KL. Food patterns measured by factor analysis and anthropometric changes in adults. Am J Clin Nutr 2004;80:504–13.[Abstract/Free Full Text]
  9. Quatromoni PA, Copenhafer DL, D'Agostino RB, Millen BE. Dietary patterns predict the development of overweight in women: The Framingham Nutrition Studies. J Am Diet Assoc 2002;102:1239–46.[Medline]
  10. Schulz M, Nothlings U, Hoffmann K, Bergmann MM, Boeing H. Identification of a food pattern characterized by high-fiber and low-fat food choices associated with low prospective weight change in the EPIC-Potsdam cohort. J Nutr 2005;135:1183–9.[Abstract/Free Full Text]
  11. Aranceta J, Perez-Rodrigo C, Ribas L, Serra-Majem L. Sociodemographic and lifestyle determinants of food patterns in Spanish children and adolescents: the enKid study. Eur J Clin Nutr 2003;57(suppl)1:S40–4.[Medline]
  12. North K, Emmett P. Multivariate analysis of diet among three-year-old children and associations with socio-demographic characteristics. The Avon Longitudinal Study of Pregnancy and Childhood (ALSPAC) Study Team. Eur J Clin Nutr 2000;54:73–80.[Medline]
  13. Reilly J, Armstrong J, Dorosty A, et al. Early life risk factors for obesity in childhood: cohort study. BMJ 2005;330:1357.[Abstract/Free Full Text]
  14. Shin KO, Oh S-Y, Park HS. Empirically derived major dietary patterns and their associations with overweight in Korean preschool children. Br J Nutr 2007;16:1–6.
  15. Ritchie LD, Spector P, Stevens MJ, et al. Dietary patterns in adolescence are related to adiposity in young adulthood in black and white females. J Nutr 2007;137:399–406.[Abstract/Free Full Text]
  16. Golding J, Pembrey M, Jones R. ALSPAC–the Avon Longitudinal Study of Parents and Children. I. Study methodology. Paediatr Perinat Epidemiol 2001;15:74–87.[Medline]
  17. Bingham SA, Gill C, Welch A, et al. Comparison of dietary assessment methods in nutritional epidemiology: weighed records v. 24 h recalls, food-frequency questionnaires and estimated-diet records. Br J Nutr 1994;72:619–43.[Medline]
  18. Price GM, Paul AA, Key FB. Measurement of diet in a large national survey: comparison of computerised and manual coding of records in household measures. J Human Nutr 1995;8:417–28.
  19. Gregory J, Lowe S. National Diet and Nutritional Survey: young people aged 4 to 18 years. Volume 1: Report of the diet and nutrition survey. London, United Kingdom: The Stationery Office, 2000.
  20. Holland B, Welch AA, Unwin ID. McCance and Widdowson's the composition of foods. 5th ed. London, United Kingdom: The Royal Society of Chemistry and MAFF, 1991.
  21. Howarth NC, Saltzman E, Roberts SB. Dietary fiber and weight regulation. Nutr Rev 2001;59:129–39.[Medline]
  22. Ledikwe JH, Blanck HM, Kettel Khan L, et al. Dietary energy density is associated with energy intake and weight status in US adults. Am J Clin Nutr 2006;83:1362–8.[Abstract/Free Full Text]
  23. Liu S, Willett WC, Manson JE, Hu FB, Rosner B, Colditz G. Relation between changes in intakes of dietary fiber and grain products and changes in weight and development of obesity among middle-aged women. Am J Clin Nutr 2003;78:920–7.[Abstract/Free Full Text]
  24. Field AE, Willett WC, Lissner L, Colditz GA. Dietary fat and weight gain among women in the Nurses' Health Study. Obesity (Silver Spring) 2007;15:967–76.[Medline]
  25. Johnson L, Mander AP, Jones LR, Emmett PM, Jebb SA. A prospective analysis of dietary energy density at age 5 and 7 years and fatness at 9 years among UK children. Int J Obes (London) 2007 (Epub ahead of print 2 October 2007).
  26. Rennie K, Coward WA, Jebb SA. Estimating under-reporting of energy intake in dietary surveys using an individualised method. Br J Nutr 2007;97:1169–76.[Medline]
  27. Torun B. Energy requirements of children and adolescents. Public Health Nutr 2005;8:968–93.[Medline]
  28. Black AE, Cole TJ. Within- and between-subject variation in energy expenditure measured by the doubly-labelled water technique: implications for validating reported dietary energy intake. Eur J Clin Nutr 2000;54:386–94.[Medline]
  29. Sopher A, Thornton J, Wang J, Pierson R, Heymsfield S, Horlick M. Measurement of percentage of body fat in 411 children and adolescents: a comparison of dual-energy x-ray absorptiometry with a four-compartment model. Pediatrics 2004;113:1285–90.[Abstract/Free Full Text]
  30. Wells JC, Cole TJ. Adjustment of fat-free mass and fat mass for height in children aged 8 y. Int J Obes Relat Metab Disord 2002;26:947–52.[Medline]
  31. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 2000;320:1240–3.[Abstract/Free Full Text]
  32. Johnson L, Mander AP, Jones LR, Emmett PM, Jebb SA. A prospective analysis of dietary fat, fibre and energy intake in relation to fatness in childhood. Proc Nutr Soc (in press).
  33. Johnson SL, Taylor-Holloway LA. Non-Hispanic white and Hispanic elementary school children's self-regulation of energy intake. Am J Clin Nutr 2006;83:1276–82.[Abstract/Free Full Text]
  34. Cecil JE, Palmer CN, Wrieden W, et al. Energy intakes of children after preloads: adjustment, not compensation. Am J Clin Nutr 2005;82:302–8.[Abstract/Free Full Text]
  35. Newby PK, Muller D, Hallfrisch J, Qiao N, Andres R, Tucker KL. Dietary patterns and changes in body mass index and waist circumference in adults. Am J Clin Nutr 2003;77:1417–25.[Abstract/Free Full Text]
  36. Northstone K, Emmett P. Multivariate analysis of diet in children at four and seven years of age and associations with socio-demographic characteristics. Eur J Clin Nutr 2005;59:751–60.[Medline]
  37. Cummings J. The effect of dietary fiber on fecal weight and composition. In: Spiller G, ed. Handbook of dietary fiber in human nutrition. 2nd ed. London, United Kingdom: CRC Press, 1993:263–350.
  38. Southgate D, Durnin J. Calorie conversion factors. An experimental reassessment of the factors used in the calculation of the energy value of human diets. Br J Nutr 1970;24:517–35.[Medline]
  39. Hoffmann K, Zyriax BC, Boeing H, Windler E. A dietary pattern derived to explain biomarker variation is strongly associated with the risk of coronary artery disease. Am J Clin Nutr 2004;80:633–40.[Abstract/Free Full Text]
  40. Piers L, Soares M, Frandsen S, O'Dea K. Indirect estimates of body composition are useful for groups but unreliable in individuals. Int J Obes Relat Metab Disord 2000;24:1145–52.[Medline]
  41. Glynn L, Emmett P, Rogers I. Food and nutrient intakes of a population sample of 7-year-old children in the south-west of England in 1999/2000: what difference does gender make? J Hum Nutr Diet 2005;18:7–19, quiz 21–3.[Medline]
  42. Jotangia D, Moody A, Stamatakis E, Wardle H. Obesity among children under 11. London, United Kingdom: Department of Health and the Information Centre for Health and Social Care, 2005.
Received for publication July 23, 2007. Accepted for publication November 15, 2007.




This article has been cited by other articles:


Home page
 JAOA: Journal of the American Osteopathic AssociationHome page
J. M. Wieting
Cause and Effect in Childhood Obesity: Solutions for a National Epidemic
J Am Osteopath Assoc, October 1, 2008; 108(10): 545 - 552.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Johnson, L.
Right arrow Articles by Jebb, S. A
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Johnson, L.
Right arrow Articles by Jebb, S. A
Agricola
Right arrow Articles by Johnson, L.
Right arrow Articles by Jebb, S. A

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS