Meeting the Dietary Reference Intakes for Fiber: Sociodemographic Characteristics of Preschoolers With High Fiber Intakes

Abstract

To better understand the public health impact of the National Academy of Sciences’ Dietary Reference Intakes (DRIs) for fiber in preschoolers, I analyzed data from the United States Department of Agriculture Continuing Survey of Food Intake in Individuals for 5437 preschoolers and examined sociodemographic predictors of meeting the DRIs. Overall, only 12% of the children met the DRIs.

Older children (age 4 and 5 years) were less likely than younger children, girls were less likely than boys, and children from medium-income families (those earning 186% to 350% of the poverty guidelines, with poverty set at 100%) were least likely to meet the DRIs. Low-income children participating in the Special Supplemental Nutrition Program for Women, Infants, and Children were twice as likely as nonparticipants to meet the DRIs. The public should be educated about the importance of increasing fiber density in the diet.

The Dietary Reference Intakes (DRIs) for energy, macronutrients, and fiber from the National Academy of Sciences suggest that Americans aged 2 years and older should consume 14 g of total fiber per 4200 kJ of total energy intake.1 At this level, fiber seems to protect against constipation and reduce the risk of cardiovascular disease, some cancers, obesity, and diabetes.2^–6

The DRIs are based on total fiber intake, which includes dietary and functional fiber. A more detailed explanation of the distinction between the 2 kinds of fiber can be found elsewhere.7^,8 In short, dietary fibers are naturally occurring nondigestible carbohydrates and lignin, and functional fibers are components of foods that can be isolated or extracted with chemical, enzymatic, or aqueous processes.1 Functional fiber intake is difficult to estimate because it is not included in food or nutrient databases and is generally not listed on Nutrition Facts labels. Adults consume an estimated 5 g per day of functional fiber on average. Based on average energy intakes, the National Academy of Sciences recommendation for dietary fiber is 14 g per day (DRI of 19 g less 5 g of functional fiber) for children aged 2 to 3 years and 20 g per day (DRI of 25 g less 5 g of functional fiber) for children aged 4 to 5 years. Previous dietary fiber intake recommendations were much lower, ranging from the 0.5 g per kg of body weight recommended by the American Academy of Pediatrics Committee on Nutrition9 to the “age plus 5” rule developed by the American Health Foundation in 1995.10^,11

On average, preschoolers do not eat high-fiber foods but consume large amounts of low-fiber items.7 Most high-fiber sources (such as whole grain products, fruits, and vegetables) are relatively expensive (e.g., whole grain bread is more expensive per serving than refined grain bread).12 To examine the sociodemographic background of children meeting the DRIs for fiber, I analyzed a nationally representative sample of American preschoolers from the 1994–1996 and 1998 United States Department of Agriculture (USDA) Continuing Survey of Food Intake in Individuals (CSFII) and compared those children who met the DRIs with those who did not meet the DRIs.

USDA DATA EXAMINED

The 1994–1996 USDA CSFII survey data were collected using stratified, multistage, area probability sampling to obtain a nationally representative sample of persons living in households. Day 1 dietary intake information was collected via a household interview, and day 2 dietary intake information was collected during a phone interview or a home visit. Adult interview respondents reported on the diets of children aged younger than 6 years. In 1998, the same methods were used for a supplemental survey of children. I combined the data to include all survey waves (CSFII 1994–1996 and 1998; n = 21 662).

The sample included 5686 preschoolers. Children who were breastfeeding (n = 10), those for whom no head of household information was available (n = 24), and those who had missing sociodemographic variables (n = 215) were excluded, resulting in a final sample size of 5437. The survey requested sociodemographic information, such as age; race/ethnicity; gender; urbanicity of and region of residence; years of education; employment status; Food Stamp Program and Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) participation; day care or school attendance; and total household income. Children with missing parental information were not included in the regression analysis (n = 138).

The head of household is the person identified by the interview respondent as the man or woman in charge of the house. This study used the education and employment data of the female heads of household. I substituted the information for the male head of household for the 57 children who did not have a female head of household. I categorized education (measured in years of school completed) into 3 groups: less than high school, high school (12 years), and more than high school.

Employment status was coded as employed or not employed. To determine the effect of federal food programs (WIC and Food Stamp) on the odds of meeting the DRIs for fiber in American preschoolers, I categorized household income as low (≤185% of the poverty guidelines), medium (186%–350% of the poverty guidelines), or high (> 350% of the poverty guidelines) in the model that included the full sample. I developed 3 income-stratified models that reflected eligibility for federal food programs: more than 185% of the poverty guidelines (not program eligible; n = 2765), less than or equal to 185% of the poverty guidelines (WIC eligible; n = 2534), and at or less than 130% of the poverty guidelines (Food Stamp eligible; n = 1818).

In an effort to capture cultural differences, the variables race (Black, White, and other) and Hispanic ethnicity (Mexican, Puerto Rican, Cuban, other Spanish subgroup, or none of the above) were recoded to define 4 distinct ethnic groups: non-Hispanic White, non-Hispanic Black, non-Hispanic other, and Hispanic. Similar coding has been employed by others.13 Region of residence corresponded with the US Census regions: Northeast, Midwest, South, and West. The urbanicity categories were non–metropolitan statistical area (MSA), MSA central city, and MSA non–central city.

I analyzed the data using Stata 8.2 (StataCorp LP, College Station, Tex), which allows adjustment for sample design effect and weighting to maintain the nationally representative character of the data. I employed the Huber correction to correct estimates for multiple children from one household.14 I conducted descriptive analysis and calculated means and standard errors to describe the sample. The data set provided average fiber intake in g per day, and I established a dichotomous dependent variable of “meets the DRI” (coded 1; n=646) or “does not meet the DRI” (coded 0; n=4791) to indicate whether children’s diets included 14 g per day and 20 g per day of dietary fiber for children aged 2 to 3 years and those aged 4 to 5 years, respectively. I calculated dietary fiber density (g of dietary fiber per 4200 kJ total energy) to indicate the amount of fiber in relation to the overall energy consumed. I used the Student t test to determine significant differences between children meeting the DRIs for fiber and children not meeting the DRIs for fiber. I developed binary logistic regression (probit) models (the P value for inclusion of covariates was .25) to examine the association between sociodemographic variables and the ability to meet the DRIs. I used the log likelihood test for model fit to determine the best regression model.

To investigate the combined effect of the variables that were collinear, I generated interaction terms by creating dummy variables for both covariates (e.g., low-income non-Hispanic White vs medium- and high-income non-Hispanic White and low-income non-Hispanic Black vs medium-and high-income non-Hispanic Black). I tested the following interaction terms in the models: income and education level, income and work status, income and race/ethnicity, work status and day care attendance, and Food Stamp and WIC participation. I report the results only for statistically significant covariates (P≤ .05) as odds ratios and 95% confidence intervals (CIs) with P values.

FINDINGS FROM USDA DATA

Half of the children in the sample were girls, and the 4 ages examined (2, 3, 4, and 5 years) were equally represented. Approximately 48% of households surveyed were in the WIC-eligible income group, and 35% were eligible for food stamps. The majority of the children in the sample was non-Hispanic White and lived in households with an income equal to or below 185% of the poverty guidelines. A more detailed description of the sample can be found in a previous publication.15

Overall, boys consumed significantly more energy than did girls, but had fiber density similar to the girls’ (6775 kJ and 6300 kJ per day, or 6.94 g and 6.88 g of fiber per 4200 kJ for boys and girls, respectively; data not shown). Only 12% (n = 646) of the entire preschool population met the DRIs for fiber (Table 1 ▶). Children meeting the DRIs for fiber had significantly higher total energy intakes and higher fiber density than children with low fiber consumption levels (P≤ .001). Within the group of children meeting the DRIs, there was no significant difference in energy intake or fiber density between boys and girls (data not shown).

TABLE 1—

Mean Energy Intake (Joule) and Fiber Density (g fiber/4200 kJ) of Preschool-Aged Children: USDA Continuing Survey of Food Intake in Individuals, 1994–1996 and 1998

Population	No, %	Energy Intake (kJ/d), Mean (SE)	Fiber Density (g/4200 kJ), Mean (SE)
All children aged 2–5	5437 (100)
Met DRIs	646 (12)	8342.0 (8.1)	10.09 (0.134)
Did not meet DRIs	4791 (88)	6303.8* (28.1)	6.49* (0.039)
Boys	2755 (100)
Met DRIs	380 (14)	8358.9 (34.5)	10.09 (0.164)
Did not meet DRIs	2375 (86)	6511.3* (11.7)	6.42* (0.050)
Girls	2682 (100)
Met DRIs	266 (10)	8314.7 (42.9)	10.09 (0.228)
Did not meet DRIs	2416 (90)	6098.0* (10.4)	6.57* (0.055)

Note. DRIs = Dietary Reference Intakes; USDA = United States Department of Agriculture.

* P ≤ .001 (Student t test).

Boys meeting the DRIs (14%) had significantly higher energy intake and fiber density than did boys not meeting the DRIs; the same relationship was found for girls. On average, children meeting the DRIs had a fiber density of 10 g per 4200 kJ of total energy compared with only 6.5 g per 4200 kJ in the children not meeting the DRIs. Children meeting the DRIs for fiber achieved this level of fiber density because they consumed more fiber-rich foods than did children who were not meeting the DRIs.

The regression models included the following variables as covariates or confounders: age, gender, race/ethnicity, day care attendance, education level, income level, census region, and the interaction terms of income and race/ethnicity (total sample); age, gender, day care attendance, employment status, education status, and the interaction term of employment status and day care attendance (not eligible for WIC or Food Stamp); age, gender, day care attendance, education level, census region, and WIC participation (eligible for WIC); age, gender, education level, census region, urbanicity, Food Stamp and WIC participation, and the interaction between Food Stamp and WIC participation (eligible for Food Stamp).

Older age in children corresponded with a lower likelihood of meeting the DRIs for fiber (Table 2 ▶), and boys were more likely to meet the DRIs than were girls. In the model containing data from the full sample, day care attendance improved children’s odds of meeting the DRIs by approximately 30%. Children in households where the head of household had a high-school education and children living in families with medium incomes (186%–350% of the poverty guidelines) were the least likely to meet the DRIs for fiber.

TABLE 2—

Sociodemographic Characteristics of Preschool-Aged Children Meeting the DRIs for Fiber: USDA Continuing Survey of Food Intake in Individuals, 1994–1996 and 1998

Characteristic	Odds Ratio (95% CI)	P
Total population (n = 5299)
Age	0.65 (0.58, 0.72)	≤ .001
Boys (Reference: girls)	1.68 (1.36, 2.08)	≤ .001
Attends day care (Reference: does not attend day care)	1.31 (1.04, 1.64)	.02
Education level of head of household, y (Reference: < 12)
12	0.68 (0.49, 0.94)	.02
> 12	0.97 (0.70, 1.35)	.87
Household income levela (Reference: low)
Medium	0.66 (0.49, 0.86)	.01
High	0.76 (0.56, 1.04)	.08
Race/ethnicity of child (Reference: non-Hispanic other and Hispanic)
Non-Hispanic White	0.90 (0.50, 1.64)	.74
Non-Hispanic Black	1.12 (0.44, 2.83)	.81
Ethnicity × income interactions
Non-Hispanic White × medium income (Reference: non-Hispanic White × low income)	0.65 (0.44, 0.97)	.03
Non-Hispanic White × high income (non-Hispanic White × low income)	0.78 (0.39, 1.59)	.50
Non-Hispanic Black × medium income (Reference: non-Hispanic Black × low income)	0.86 (0.37, 1.97)	.72
Non-Hispanic Black × high income (Reference: non-Hispanic Black × low income)	1.38 (0.54, 3.50)	.50
Non-Hispanic other and Hispanic × medium income (Reference: non-Hispanic other and Hispanic × low income)	0.67 (0.35, 1.25)	.21
Non-Hispanic other and Hispanic × high income (Reference: non-Hispanic other and Hispanic × low income)	0.62 (0.21, 1.81)	.39
Medium or high household income (not eligible for WIC or Food Stamps; n = 2 765)
Age	0.60 (0.51, 0.72)	≤ .001
Boys (Reference: girls)	1.58 (1.16, 2.17)	≤ .001
Low household income (eligible for WIC; n = 2 534)
Age	0.70 (0.60, 0.83)	≤ .001
Boys (Reference: girls)	1.78 (1.32, 2.38)	≤ .001
US Census region (Reference: Northeast)
Midwest	1.24 (0.74, 2.08)	.41
West	1.65 (1.01, 2.70	.05
South	0.98 (0.59, 1.61)	.94
Household eligible for Food Stamps (income – 130% of the poverty guidelines; n = 1 818)
Age	0.77 (0.65, 0.92)	≤ .001
Boys (Reference: girls)	1.76 (1.28, 2.43)	≤ .001
Urbanicity: MSA central city (Reference: non-MSA)	1.17 (0.73, 1.87)	.52
Urbanicity: MSA non–central city (Reference: non-MSA)	1.63 (1.00, 2.66)	.05
Enrolled in WIC (Reference: not enrolled in WIC)	2.04 (1.24, 3.35)	.01

Note. DRIs = Dietary Reference Intakes; USDA = United States Department of Agriculture; CI = confidence interval; WIC = Special Supplemental Nutrition Program for Women, Infants, and Children; MSA = metropolitan statistical area.

^aLow-income households earn 185% or less of the poverty guidelines; medium, 186% to 350%; and high, more than 350%.

In the WIC-eligible population, living in the West compared with the Northeast census region increased the odds of meeting the DRIs by 65%. In the Food Stamp–eligible group, children living in urban areas (MSA non–central city) were 63% more likely to meet the DRIs than were children living in rural areas (non-MSA), and enrollment in the WIC program increased the likelihood of meeting the fiber DRIs by 104%. It is noteworthy to point out that enrollment in the WIC and Food Stamp programs simultaneously (n = 203) did not significantly increase the likelihood of meeting the DRIs.

CONCLUSIONS

I examined a nationally representative sample of American preschoolers to determine the sociodemographic characteristics of children who meet the DRIs for fiber. As I expected, the results showed that American children are not equally likely to consume dietary fiber at the recommended level.

Children were less likely to meet the DRIs as their age increased, probably because of their increasing independence in food selection as they aged. Although boys were more likely than girls to consume fiber at the recommended level, they also had higher energy intakes and their average fiber density was not significantly different from the girls. Thus, because boys ate more, they consumed greater amounts of fiber.

As the models indicated, the effect of family income was mediated by race/ethnicity, which was not statistically significant once income was added as a factor. However, an examination of the interaction terms between income and race/ethnicity showed that income was not mediated by race/ethnicity in non-Hispanic White children. The combined effect of income level and race/ethnicity was significant in medium-income non-Hispanic White children, who were less likely to meet the DRIs for fiber than were low-income non-Hispanic White children. In contrast to our findings, others found significant differences of fiber intake by race/ethnicity, such as lower fiber consumption in the African American child population.16 Hispanic mothers following cultural eating patterns have been found to have higher intakes of fiber than a comparable population that did not adhere to the traditional diet.17 However, with increasing acculturation and adaptation to the US diet, those eating habits might disappear.17

My examination of household income revealed that children in the medium income group were the least likely to meet the DRIs for fiber. This result confirms findings of others,12 who explained this phenomenon as the difference in purchasing power between the income groups. Although low-income families consumed fewer processed and refined foods because of financial restrictions, high-income families (with more purchasing power) bought and consumed more fresh fruits, vegetables, and whole grain products; either pathway could lead to higher fiber intake than observed in medium-income families.

Although the employment status of female heads of household did not affect children’s likelihood to meet DRIs, children attending day care were more likely to meet the DRIs for fiber than were children who did not attend day care. Enrollment in the WIC program had a strong beneficial effect on preschoolers’ likelihood of meeting the DRIs for fiber; it more than doubled the likelihood to meet the DRIs in the Food Stamp–eligible group. Thus, WIC participation is leading to increased fiber consumption among preschoolers. This effect could be because of the foods provided in the WIC package, especially the ready-to-eat cereals, or the nutrition education component delivered to the participants at their monthly WIC clinic visits.

One limitation to this study was that interviewers collected children’s dietary intake data from adults via 24-hour recall. In addition to the problems associated with diet assessment by an interviewer, such as under-reporting and overreporting of some foods, the respondent might not have been aware of the child’s diet on the previous day. Information might have been missing because some of the foods consumed may not have been provided by the caretaker, some of the foods were exchanged with friends at day care, or some of the foods were lost because they were spilled or because day care workers discarded the leftovers, leading to overestimation of consumption.

The DRIs for fiber are higher than previous fiber intake recommendations. Research has shown that high fiber intake in preschoolers is indicative of an overall healthy diet.7 Most of the fiber consumed by those preschoolers was from relatively low-fiber foods. A direct link between dietary fiber consumption and measured health outcomes in preschool-age children has yet to be researched.

As results of this study show, the likelihood of meeting the DRIs is associated with certain sociodemographic characteristics. Participation in WIC and attending day care were beneficial to children’s fiber levels. Improving the diets of preschoolers of all income groups and encouraging them to adopt a more fiber-dense diet may decrease childhood obesity. Providing easy-to-prepare high-fiber foods, such as unsweetened high-fiber cereals, whole grain pasta or rice, or beans, might help promote the fiber density of preschoolers eligible for WIC or food stamps. There is an urgent need to develop public health messages to provide information on high-fiber sources and possible ways to introduce high-fiber foods into preschoolers’ diets to increase fiber consumption in this population.