A Longitudinal Study of Food Insecurity on Overweight in Preschool Children

Year: 2007

Research Center: Department of Nutrition at the University of California, Davis

Investigator: Metallinos-Katsaras, Elizabeth, Jan Kallio, Aviva Must, Parke Wilde, and Kathleen Gorman

Institution: Simmons College

Project Contact:
Liz Metallinos-Katsaras
Department of Nutrition
Simmons College
300 The Fenway
Boston, MA 02115
Phone: 617-521-2708


Childhood overweight and household food insecurity (HFI) represent urgent public health problems in the United States. Food insecurity is the lack of access to enough food for an active healthy life that results from the limited or uncertain access to nutritionally adequate and safe foods in socially acceptable ways. Low-income households are more likely be food insecure, and paradoxically, low-income adults, specifically women, are more likely to be overweight. In children, however, studies of this association have yielded conflicting results, perhaps because study designs (cross-sectional versus longitudinal) and populations (ages and income levels) have varied. Additionally, most of the prior work did not examine this association in the most relevant population—that is, low-income households.

The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) serves low-income women and children at high nutritional risk. Anthropometric, sociodemographic, and other health-related data are routinely collected. The purpose of this research is to (1) examine the effect of WIC participation on household food security status in women and children and (2) assess the relationship between HFI with/without hunger in infancy and later childhood weight status in 2- to 5-year-old WIC participants. Strengths of this study lie in the use of the large and diverse WIC population on which longitudinal anthropometric and food security data are available.

This longitudinal study includes data collected from 2001-06 on children and their mothers who participated in Massachusetts WIC. WIC data are collected every 6 months and prepared for submission to the Centers for Disease Control and Prevention (CDC) for inclusion in the Pregnancy and Pediatric Surveillance Systems. The addition of household food security measures to the WIC management information system was undertaken as part of a previously CDC-funded cooperative agreement granted to the Massachusetts Department of Public Health (1996-2000). Inclusion of the full-length food security module was not feasible due to time constraints. For this study, a subscale to measure food security status was used. It addressed the following areas: (1) not having enough money to buy food for a balanced meal, (2) adults cutting the size of or skipping meals, (3) frequency of cutting meal size or skipping meals, and (4) adults not eating for a whole day. Household food security status was defined by the number of positive (“yes”) responses to the questions: food security = 0 positive responses; HFI without hunger = 1–2 positive responses; and HFI with hunger = 3–4 positive responses.

Household food security status at both time points were combined to create a dynamic food security variable that comprised the following categories: persistently food insecure, food insecure at first visit and secure at the second, food secure at the first visit and insecure at the second, and persistently food secure (referent). The age- and sex-specific body mass index (BMI) percentile and z-score of children was based on their directly measured height and weight relative to the CDC growth reference. At-risk for overweight and overweight were defined, as recommended, as sex-specific BMI-for-age of greater than or equal to (1) 85th percentile and (2) 95th percentile, respectively.

Multinomial logistic regression was used to assess the relationship between duration of WIC participation and household food security status (for example, food secure, HFI without hunger, and HFI with hunger) at the last visit for women (n=21,863) and children (n=57,377), adjusting for race, maternal education, household size, and initial household food security status. Both general linear model and logistic regression techniques were used to examine the relationship between household food security status at the first and last visit and child weight status. Children meeting the following criteria were included (n=25,186): (1) first-visit data available, (2) first WIC visit within first 12 months of age, (3) at least four-WIC-visits data available, (4) complete data on household food security status at first and last visit and on covariates (birth weight, age, sex, race/ethnicity, maternal education, household size, and maternal weight status), (5) child of non-Hispanic White, Hispanic, Black non-Hispanic, or Asian race/ethnicity, (6) complete anthropometric data at both time points and ages 24-60 months, and (7) birth weight of the child available.

The association between duration of WIC participation and household food security status depends on household food security status at the initial visit. For both women and children from initially (that is, at first WIC visit) food-secure households, WIC duration had no effect on later household food security status. On the other hand, among women who were from households that were initially food insecure with hunger, early prenatal certification into WIC produced the greatest improvement in household food security status by the postpartum period. Among children who were from initially food insecure households (with or without hunger), longer WIC participation was associated with the greatest improvements in their households’ food security status.

Preliminary results suggest that the relationship between household food security status and children’s weight status depends on other factors. Significant effect modification (p < 0.05) in the fully adjusted model was noted for the dynamic HFI variable and maternal education, maternal pre-pregnancy weight status, and child’s birth weight. Thus, analyses were adjusted or stratified by each of these variables. Stratification of the analyses by birth weight, using a median split for this sample (3,291.5 grams), yielded a significant association between household food security status and weight status among children whose birth weight was less than the median (but not those greater than or equal to the median). Persistent HFI was associated with a 27 percent higher odds (p < 0.01) of attaining a BMI-for-age greater than or equal to the 85th percentile and a 31 percent greater odds (p < 0.01) of becoming overweight by the time they were 2-5 years old compared with children whose households were persistently food secure. Among children whose mother’s pre-pregnancy weight classified them as overweight or obese (BMI > 25), persistent HFI was associated a 22 percent higher odds (p < 0.01) of their children attaining a BMI-for-age greater than or equal to the 85th percentile and a 19 percent greater odds (p < 0.05) of 2- to 5-year-old children becoming overweight compared with those whose households were persistently food secure. No association was found among those children whose mother’s pre-pregnancy weight was normal.

From a policy perspective, these findings suggest that enrolling mothers in WIC earlier in pregnancy could reduce later risk of overweight among their children by improving household food security status once their children are born. The results also imply that certain subgroups of children are particularly vulnerable to the adverse effects of household food insecurity on overweight risk, thus targeting these groups may be necessary.

Direct inquiries about this study to the Project Contact listed above.