Is food insecurity related to overweight and obesity in children and adolescents? A summary of studies, 1995–2009

Summary

In the USA, several nutrition-related issues confront the normal growth, maturation and development of children and adolescents including obesity and food insecurity. The purpose of this paper is to provide a review of the concept of food insecurity and a summary of studies that have examined the association between food insecurity and overweight/obesity in children and adolescents. Besides the initial case report, we review 21 studies (16 cross-sectional and five prospective studies) that have been published on this topic as of December 2009. As there is limited literature in this area, we review studies that sample children and adolescents in the USA. The results are mixed with positive, negative and null associations. The reasons for the mixed results are difficult to disentangle. Among earlier studies, small samples hampered definitive conclusions. More recent studies with larger samples have overcome these limitations and tend to find no associations between these constructs. Nonetheless, all of the studies to date have shown that food insecurity and overweight co-exist – that is, even though there may not be statistically significant differences in overweight between food-insecure and food-secure children, the prevalence of overweight remains relatively high in food-insecure children.

Introduction

Adequate nutrition, along with genetics and hormones, is among the chief determinants of normal biological growth and maturation of children (1,2). For children in the USA, several nutrition-related disorders confront normal growth and maturation. Two issues of concern are obesity as well as food insecurity (e.g. children who do not get enough food). In the USA approximately 16% of children and adolescents are considered obese with an additional 18% classified as overweight (3). Similarly, 21% of US children live in a food-insecure household (4) which means children in these households live without the financial means to access enough food for active, healthy living.

Although fully aware of the vast research on the topic of obesity, paediatric obesity researchers may be unfamiliar with the concept of food insecurity, perhaps because it is at the other end of the food consumption spectrum. A PUBMED search as of December 2009 revealed more than 300 papers on the topic of food insecurity of which many document the negative health consequences in children. These negative outcomes include poor general health and health limitations; increased hospitalization and psychological visitation, clinical levels of psychosocial dysfunction, including anxiety and depression; and academic problems (5–11). Recently, a report on the relationship between food insecurity and obesity was published by the Robert Wood Johnson Foundation under its Healthy Eating Research programme (12). In their publication, studies were included for non-elderly adults and children (many of which are also reviewed herein). Although similar to this review paper, we provide a summary of each study that examines the association between food insecurity and child overweight from a historical and study design perspective. In addition, the review begins with a brief overview of the concept, measurement and epidemiology of food insecurity.

What is food insecurity and how is it measured?

The United States Department of Agriculture (USDA) defines food insecurity as limited or uncertain availability of nutritionally adequate and safe foods, or limited or uncertain ability to acquire acceptable food in a socially acceptable way (13). The measurement of food insecurity by the US government officially began in 1995 with the addition of a food security supplement, the Core Food Security Module (CFSM), to the Current Population Survey. Based on responses to 18 items (Table 1), the food security status of households can be determined.

Table 1.  Food insecurity questions in the Core Food Security Module

Responses in bold indicate an affirmative response.

1. ‘We worried whether our food would run out before we got money to buy more’. Was that often, sometimes or never true for you in the last 12 months?

2. ‘The food that we bought just didn't last and we didn't have money to get more’. Was that often, sometimes or never true for you in the last 12 months?

3. ‘We couldn't afford to eat balanced meals’. Was that often, sometimes or never true for you in the last 12 months?

4. ‘We relied on only a few kinds of low-cost food to feed our children because we were running out of money to buy food’. Was that often, sometimes or never true for you in the last 12 months?

5. In the last 12 months, did you or other adults in the household ever cut the size of your meals or skip meals because there wasn't enough money for food? (Yes/No)

6. ‘We couldn't feed our children a balanced meal, because we couldn't afford that’. Was that often, sometimes or never true for you in the last 12 months?

7. In the last 12 months, did you ever eat less than you felt you should because there wasn't enough money for food? (Yes/No)

8. (If yes to Question 5) How often did this happen –almost every month, some months but not every month or in only 1 or 2 months?

9. ‘The children were not eating enough because we just couldn't afford enough food’. Was that often, sometimes or never true for you in the last 12 months?

10. In the last 12 months, were you ever hungry, but didn't eat, because you couldn't afford enough food? (Yes/No)

11. In the last 12 months, did you lose weight because you didn't have enough money for food? (Yes/No)

12. In the last 12 months, did you ever cut the size of any of the children's meals because there wasn't enough money for food? (Yes/No)

13. In the last 12 months did you or other adults in your household ever not eat for a whole day because there wasn't enough money for food? (Yes/No)

14. In the last 12 months, were the children ever hungry but you just couldn't afford more food? (Yes/No)

15. (If yes to Question 13) How often did this happen –almost every month, some months but not every month or in only 1 or 2 months?

16. In the last 12 months, did any of the children ever skip a meal because there wasn't enough money for food? (Yes/No)

17. (If yes to Question 16) How often did this happen –almost every month, some months but not every month or in only 1 or 2 months?

18. In the last 12 months did any of the children ever not eat for a whole day because there wasn't enough money for food? (Yes/No)

Across time, there have been changes in the nomenclature of food insecurity. Earlier studies report food insecurity with hunger, a more severe category within food insecurity, which was defined as the uneasy or painful sensation caused by lack of food or recurrent and involuntary lack of access to food. In 2006, the USDA changed the label ‘food insecurity with hunger’ to ‘very low food security status’. Currently, households are classified as food-secure if they respond affirmatively to two or fewer questions in the CFSM or food-insecure if they respond affirmatively to three or more CFSM questions. Within the food-insecure category, households with children can either be classified as low food-secure (i.e. three to seven affirmative CFSM responses) or very low food-secure (i.e. eight or more affirmative responses). This method of categorizing households does not distinguish between the difference in the severity of questions. For example, someone responding affirmatively to only questions 14, 16 and 18 is treated the same as someone responding affirmatively to only questions 1 through 3 – both are food-insecure. The modal respondent, however, who responds affirmatively to a particular question will have also responded affirmatively to the previous questions. The literature has tended to use these categories, although some work has used continuous measures (14,15). Consistent with the majority of the food insecurity literature, research on the relationship between food insecurity and childhood obesity has examined relationships between categorical measures rather than between continuous measures of food insecurity and body mass index (BMI).

Besides food insecurity, other studies have reported on a related construct – food insufficiency. The food sufficiency question asks respondents to describe their food intake in terms of the following: which of these statements best describes the food eaten in your household in the last month? Respondents have four choices: enough of the kinds of food we want to eat; enough but not always the kinds of food we want to eat; sometimes not enough to eat; or often not enough to eat. Although not consistent across all studies, those households reporting that they sometimes or often do not get enough to eat are considered food-insufficient. In studies reporting both food insufficiency and food insecurity, the extent of food insufficiency is substantially lower than food insecurity. As a consequence, it is generally seen as a more serious negative health outcome than food insecurity. The food insufficiency construct, while still available in some datasets, has largely been replaced with the food insecurity construct.

Prevalence of food insecurity

The extent of food insecurity in the USA for households with children in 2008 (4) is shown in Table 2. Among all households, 14.6% are food-insecure and 5.7% are very low food-secure. (As discussed above, very low food-secure households are a subset of food-insecure households.) Among households with children, the rates are higher – 21.0% and 6.6%, respectively. There is a great deal of variation in the prevalence of food insecurity in the USA across different demographic categories. As expected, food insecurity is more likely to occur in low-income households, households headed by a black or Hispanic person, and being in a household headed by a single mother or single father.

Table 2.  Prevalence (%) of food insecurity rates for US households by various demographic categories, 2008

	Food-secure	Food-insecure	Very low food-secure
Values are %. All analyses use weights provided in the Current Population Survey.
All	85.4	14.6	5.7
Presence of children
With children present	79.0	21.0	6.6
Without children present	88.7	11.3	5.3
Income/poverty ratio
<1.00	57.8	42.2	19.3
<1.30	61.0	39.0	17.4
<1.85	66.1	33.9	14.6
≥1.85	92.3	7.7	2.6
Income unknown	90.3	9.7	3.4
Residence
Non-metro	85.8	14.2	5.6
Metro	85.3	14.7	5.7
Race/ethnicity
White non-Hispanic	89.3	10.7	4.5
Black	74.3	15.6	10.1
Hispanic	73.1	26.9	8.8
Other	86.3	13.7	5.0
Household head status among households with children
Married couple	85.7	14.3	4.1
Single mother	62.8	37.2	13.3
Single father	72.4	27.6	7.2

The beginning of the study of food insecurity and child overweight: a case report

Most papers examining food insecurity and child overweight reference the case report published in 1995 by Dietz (16) as the initial paper on this topic. This report was a case study of a 7-year-old black girl who presented to a weight control clinic in Boston. At the time of the clinical examination, her BMI was 38.6 kg m⁻²; however, the only clinical finding associated with obesity was acanthosis nigricans as other usual comorbidities (e.g. lipids, etc.) were normal. The patient lived with her mother and her family received food stamps. The child's mother indicated that consumption of healthy, low-caloric-density foods was difficult, particularly during the time period just prior to receiving their Aid to Families with Dependent Children (a federal cash assistance programme for poor single mothers with children). The distribution of food stamps once a month also affected the mother's ability to positively influence the child's consumption patterns.

Dietz posited two possible mechanisms that may underlie the food-insecure-obese child: (i) increased dietary fat consumption and (ii) physiological adaptations to episodic food insufficiency. He concluded that ‘. . . confirmation of this hypothesis requires the demonstration of obesity associated with food insufficiency in larger cross-sectional and prospective studies’. The following sections review the literature on cross-sectional and longitudinal studies of food insecurity and overweight of children and adolescents since this initial case report in 1995.

A review of studies: association between food insecurity and overweight in children and adolescents

As of December 2009 we identified 21 pertinent studies (excluding the case report by Dietz) on this topic using a search of PUBMED. We limited our search to include studies that specifically examined children and adolescents 2–19 years of age from mainly the USA (one study from Canada). In this section, we begin with the first empirical reports and then discuss cross-sectional results from large, national surveys followed by smaller studies (n < 500). Finally, longitudinal or prospective cohort studies are considered. The results of the work are summarized in Tables 3 and 4.

Table 3.  Measures of overweight/obesity and food insecurity in published studies

First author	How height and weight established	Criteria for weight status	Measure of food insecurity	Level of food insecurity measurement
Studies are arranged in alphabetical order by last name of first author. BMI, body mass index; CDC, Centers for Disease Control and Prevention; CFSM, Core Food Security Module; IOTF, International Obesity Task Force; P85, age- and sex-specific 85th centile; P95, age- and sex-specific 95th centile; TSF, triceps skinfold thickness; WC, waist circumference; WHZ, weight-for-height z-score.
Alaimo (2001)	Measured	CDC P85	Food insufficiency	Household
Bhargava (2008)	Measured	IOTF	CFSM	Household
Bhattacharya (2004)	Measured	CDC P95	CFSM	Household
Bronte-Tinkew (2007)	Measured	CDCP95 (weight-for-length)	CFSM	Household
Casey (2006)	Measured	CDC P85 P95	CFSM	Household and children
Casey (2001)	Self-report	Raw BMI; CDC P85	Food insufficiency	Household
Cutts (1998)	Measured	CDC growth centiles	Food insufficiency	Household
DuBois (2006)	Measured	CDC and IOTF	Food insufficiency	Household
Feinberg (2008)	Measured	CDC P85	Six-item food insecurity scale	Household
Gundersen (2008a)	Measured	CDC P85, P95; IOTF	Three questions from CFSM	Child
Gundersen (2008b)	Measured	CDC P85	CFSM	Household and children
Gundersen (2009)	Measured	CDC P85; WC >90th percentile for age and sex; TSF >95th percentile for age and sex; high trunk fat mass for age and gender; whole body fat >95th percentile for age and gender	CFSM	Household
Jones (2003)	Self-report weight; measured height	CDC P85	CFSM	Household
Jyoti (2005)	Measured	n.a.	CFSM	Household
Kaiser (2002)	Measured	WHZ P85 P95	Radimer/Cornell scale	Household
Lohman (2009)	Measured	CDC P85; IOTF	3 questions from CFSM	Child
Martin (2007)	Measured	CDC P85 P95	CFSM	Household
Matheson (2002)	Measured	n.a.	CFSM	Children
Rose (2006)	Measured	CDC P 95	CFSM	Household
Whitaker (2006)	Measured	CDC P95	CFSM	Household and children
Winicki (2003)	Measured	CDC P85 P95	CFSM	Household

Table 4.  A summary of published studies on food insecurity and overweight/obesity in children and adolescents

First author	Study design	Dataset	n	Age of children	Age of children	Race/ethnicity of children	Main findings
Studies are arranged in alphabetical order by last name of first author. BMI, body mass index; CSFII, Continuing Survey of Food Intakes by Individuals; ECLS-B, Early Childhood Longitudinal Study – Birth Cohort; ECLS-K, Early Childhood Longitudinal Study – Kindergarten; NHANES, National Health and Nutrition Examination Survey.
Alaimo (2001)	Cross-sectional	NHANES-III	9 196 (4 185 boys and 5011 girls)	2–16 years	2–16	White, Hispanic, black, Asian, other	For younger girls, food insufficiency was negatively associated with overweight status; for non-Hispanic white older girls, food insufficiency was negatively associated with obesity status
Bhargava (2008)	Longitudinal	ECLS-K (1999–2003)	11 479 (5 797 boys and 5 682 girls)	Children from kindergarten to fifth grade	Children from kindergarten to fifth grade	White, Hispanic, black, Asian, other	No association
Bhattacharya (2004)	Cross-sectional	NHANES-III	9 502	2–17 years	2–17	White, Hispanic, black, Asian, other	No association
Bronte-Tinkew (2007)	Cross-sectional	ECLS-B	8 693	2 years	2	White, black, Asian, native American	No direct association; linked through parenting practices and infant feeding practices to influence overweight
Casey (2006)	Cross-sectional	NHANES 1999–2002	6 995 (3 553 boys and 3 442 girls)	3–17 years	3–17	White, Hispanic, black, Asian, other	Positive association between food insecurity and overweight status
Casey (2001)	Cross-sectional	CSFII 1994–1996	5 669	0–17 years	0–17	White, Hispanic, black, Asian, other	No direct association between children's food sufficiency and nutrient intake
Cutts (1998)	Cross-sectional	Survey of patients from an ambulatory paediatrics clinic	2 578	0–5 years	0–5	White, black, native American, other (English-speaking)	No differences in weight-for-height or weight-for-age between hunger groups
DuBois (2006)	Longitudinal	Longitudinal Study of Child Development in Quebec (1998–2002)	2 103 (49% girls)	1.5–4.5 years	1.5 – 4.5	White	There was a 2–3× odds of overweight in the presence of family food insufficiency at some point during pre-school years. There was an interaction between birthweight and family food insufficiency in relation to being overweight at 4.5 years
Feinberg (2008)	Cross-sectional	Community survey of black mothers	278 (mother-child dyads)	2–13 years	2–13	Black	No association
Gundersen (2008a)	Cross-sectional	Three-City Study	1 031 (533 girls and 498 boys)	10–15 years	10–15	White, black, Hispanic, other	No association
Gundersen (2008b)	Cross-sectional	NHANES 1999–2003	841 (457 boys and 384 girls)	3–17 years	3–17	White, Hispanic, black, Asian, other	No direct association; negative association for interaction of food insecurity and cumulative stress index for children aged 10–17
Gundersen (2009)	Cross-sectional	NHANES 2001–2004	2 516 (1 239 girls and 1 277 boys)	8–17 years.	8–17	White, Hispanic, black, Asian, other	No association
Jones (2003)	Cross-sectional	Panel Study of Income Dynamics, Child Development Supplement	772	5–12 years	5–12	White, Hispanic, black, Asian	No direct association; indirect association through food assistance programme participation; negative interaction of programme participation and risk of overweight
Jyoti (2005)	Longitudinal	ECLS-K	11 180 (5 682 boys and 5 498 girls)	Children entering kindergarten in 1998 and followed through 3rd grade	Children entering kindergarten in 1998 and followed through 3rd grade	White, Hispanic, black, Asian	Positive association between food insecurity and overweight, and weight gains among girls
Kaiser (2002)	Cross-sectional	Survey of low-income Mexican–American families with children of pre-school age	211 families	3–6 years	3–6	Mexican–American	No association
Lohman (2009)	Cross-sectional	Three-City Study	1011 (486 boys and 525 girls)	10–15 years	10–15	White, black, Hispanic	No direct association; positive interaction of food insecurity and maternal stress index
Martin (2007)	Cross-sectional	community survey in Hartford, Connecticut	212 (108 girls, 104 boys)	2–12 years	2–12	Hispanic, white, black, West Indian, other	No association
Matheson (2002)	Cross-sectional	Community survey	124 (65 boys, 59 girls)	Fifth grade students	Fifth grade students	Predominantly Hispanic	Food insecurity was negatively associated with the BMI.
Rose (2006)	Longitudinal	ECLS-K	16 889 (estimated: 8 681 boys and 8 208 girls)	Children in their kindergarten year	Children in their kindergarten year	White, black, Latino, Asian–American, other	No association
Whitaker (2006)	Cross-sectional	Fragile Families and Child Well-being Study	2 452 (information not available by gender)	3 years	3	White, Hispanic, black, Asian, other	No association
Winicki (2003)	Longitudinal	ECLS-K	18 847	Children in their kindergarten year	Children in their kindergarten year	White, black, Latino, Asian–American, other	No association

The first empirical studies

Although previous papers addressed this issue in adult women (17), the first empirical study in children appears to have been published in 1998 (18). Among 2578 0- to 5-year-old children attending an ambulatory paediatric clinic in Minneapolis, MN, hunger status was not associated with mean percentiles for weight-for-height or weight-for-age. Shortly thereafter, an abstract was published in 1999 and then as a full manuscript in 2002 (19) on 211 Mexican–American 3- to 6-year-old children and their parents living in northern California. There were no significant differences in weight-for-height z-score or per cent overweight by level of food security. At about the same time as the Kaiser et al. (19) study, another northern California study examined the association between food insecurity and BMI in Hispanic fifth grade children and their mothers (n = 124, 65 boys and 59 girls) (20). Children from food-secure households had a significantly higher mean BMI (21.2 kg m⁻²) compared with food-insecure children (19.8 kg m⁻²).

Nationally representative cross-sectional surveys of food insufficiency and overweight

Continuing Survey of Food Intakes by Individuals (CSFII)

In 2001, Casey et al. (21) were among the first (along with Alaimo et al. – see below) to examine issues of food insufficiency and nutritional status, including weight status, using a large, representative sample – the CSFII (1994–1996). The CSFII was a nutritional survey conducted by the USDA on a nationally representative sample of individuals of all ages that provided detailed food and nutrient intakes on the general population and low-income populations. Overall, 3% of households with children experienced food insufficiency with the rate being higher (7.5%) in low-income households with children. The per cent of youth classified as overweight was similar (47%) between low-income food-insufficient and low-income food-sufficient youth. However, when low-income food-insufficient households were compared with higher-income food-sufficient households, low-income food-insufficient households had a significantly higher percentage of overweight subjects (46.7% vs. 31.5%).

National Health and Nutrition Examination Survey (NHANES) III

Alaimo et al. (22) and Bhattacharya et al. (23) reported on a large, representative sample of white, black and Mexican–American children and adolescents from NHANES III (1988–1994). Both studies report that for the full sample there was no significant associations between food insufficiency and overweight even when broken down into various categories. One reason for the lack of significance may be that the number of youth classified as food-insufficient within age-, race- and gender-specific groups was small (e.g. n = 17 for 2- to 7-year-old white boys and 8- to 16-year-old white girls, etc.).

Early Childhood Longitudinal Study – Birth Cohort (ECLS-B)

Akin to the Early Childhood Longitudinal Study – Kindergarten (ECLS-K) (see below), the ECLS-B is a longitudinal study of a nationally representative sample of children to be examined at 9 months, 2 years, 4 years, entry into kindergarten and first grade. Bronte-Tinkew et al. (24) examined 8693 subjects at 24-months (cross-sectional analysis). The authors employed structural equation modelling (SEM) to examine direct and indirect associations between food insecurity and overweight (weight-for-length > 95th centile). Although food insecurity was not directly associated with overweight, results from SEM indicated that food insecurity worked indirectly through parenting practices and infant feeding to influence overweight. More specifically, positive parenting practices influenced good infant feeding (beta coefficient = 0.067) and, in turn, good infant feeding practices were negatively associated with overweight (beta coefficient = −0.361).

Results from other large-scale surveys: measuring food insecurity at the level of children in the household

One methodological issue of the studies reported to this point is that food insecurity was measured at the household level. The following studies report results for food insecurity defined at the level of the children in the household. This measure of ‘child food insecurity’ is defined for all children in the household and is based on eight questions in the CFSM that ask specifically about the children in the household. So, in general, analyses of the relationship between ‘child food insecurity’ and childhood obesity are not using a measure of food insecurity defined for a particular child. The exception to this is when there is only one child in the household. The other exception, as noted below, is in two papers where the authors examined child-specific food insecurity.

National Health and Nutrition Examination Survey 1999–2002

Using the child food insecurity measure and NHANES 1999–2002, Casey (25) examined the relationship between food insecurity and childhood obesity for several age groups (3–5, 6–8, 9–11 and 12–17 year) and by gender and race/ethnicity (white, black and Hispanic) (total n = 6995). Overall, the prevalence of combined overweight and obesity was 28.8% among children in food-secure households vs. 38.8% in children in food-insecure households (P < 0.01). These results were also fairly consistent by age and race categories, although not all comparisons were statistically significant. Results of logistic regression showed a 32% increased odds of overweight/obesity for child food insecurity controlling for age, gender, race and poverty.

Gundersen et al. (26) also used data from NHANES 1999–2002. In contrast to the paper by Casey et al. this study focused on low-income children. The sample included 841 children (3 to 17 years) in households with incomes below 200% of the poverty line. In addition, the study provided a more in-depth examination of the relationship as it sought to examine the links between food insecurity, stress and child obesity. Results of the probit regression analysis showed that child food insecurity was not associated with overweight for 3- to 10-year-olds or 11- to 17-year-olds. However, the interaction of food insecurity and maternal stressors was significantly linked to the probability of being overweight among 3- to 10-year-olds; more specifically, an increase in maternal stressors amplified a food-secure child's probability of being overweight or obese. This result was not found among 11- to 17-year-old youth. Sex- and race-specific analyses were not conducted but these variables were used as covariates.

Another paper by Gundersen et al. (27) examined a sample of 8- to 17-year-old children in households with incomes below 200% of the poverty line who participated in the 2001–2004 NHANES. This work adds to the literature in that five measures were used to classify a child as obese – BMI, waist circumference, triceps skinfold, trunk fat mass and percentage of body fat – compared with previous work which used BMI only. There was no significant relationship between food insecurity and obesity for any of the five measures in either bivariate analyses or results from multivariate regression models. This held when the sample was broken down by gender, by race/ethnicity, and by gender and race/ethnicity.

Fragile Families and Child Well-being Study

The Fragile Families and Child Well-being Study enrolled approximately 4900 children at birth from 20 large US cities to address various issues related to child development in low-income families. Whitaker and Orzol (28) examined the relationship between household and child food insecurity in 2452 children who were 3 years old (mean age 38.6 ± 3.3 months). In the total sample, there was no significant association between food security status and overweight; however, children in households that were very low food-secure showed a slightly lower rate of overweight/obesity (13.4%) compared with food-secure children (18.6%) (P = 0.23). When the association between food security and prevalence of overweight was examined by race, blacks showed a statistically significant lower percentage of overweight and obesity than whites among both food-insecure categories.

Welfare, children and families: a Three-City Study

Recently, two papers have been published using data from the Three-City Study (29,30). The Three-City Study is a multi-method investigation designed to understand the effects of the Personal Responsibility and Work Opportunities Reconciliation Act of 1996 (‘Welfare Reform’) on the well-being of low-income children and families. The Three-City Study includes a household-based stratified random sample survey with over 2400 low-income children and their mothers in low-income neighborhoods in Boston, Chicago and San Antonio. In these papers, data from approximately 1000 10- to 15-year-olds were examined. Central to both papers is that food insecurity was defined for a focal child for whom BMI was collected rather than for all children in the household.

In the first paper (29), bivariate analyses and logistic regression analyses indicated that there were no statistically significant differences in the prevalence of overweight and obesity between food-secure and food-insecure children. Similar patterns were shown by gender, race and income. In the second paper (30), higher levels of individual stressors (e.g. poor academic performance, drug and alcohol use, lack of future orientation, as well as externalizing and internalizing symptoms) increased the probability of being overweight or obese for adolescents, whereas there was no direct association between food insecurity, maternal, or family stressors and overweight or obesity. The interaction of food insecurity and maternal stressors (e.g. maternal employment status, self-esteem, psychological distress) was significantly linked to the probability of being overweight or obese; more specifically, an increase in maternal stressors amplified a food-insecure adolescent's probability of being overweight or obese.

Other cross-sectional studies

Martin and Ferris (31) report on 212 children 2 to 12 years of age from Hartford, CT. The prevalence of overweight and obesity in food-secure children was 43% and in food-insecure children 57%, but results from logistic regression did not show a significant association between food insecurity and overweight/obesity. Feinberg et al. (32) found similar prevalence of overweight between food-secure (58%) and food-insecure (60%) black children. The notably high prevalence rates in this sample probably derive from the uniqueness of the sampled sub-populations (urban population; half of which were Haitian children and half African–American) and are not representative of black children generally.

Do food assistance programmes moderate the association between household food security status and child overweight?

Jones et al. (33) noted that neither of the two large-scale studies published in 2001 (21,22) controlled for participation in food assistance programmes [e.g. Supplemental Nutrition Assistance Program (SNAP), formerly known as the Food Stamp Program, the Special Supplemental Nutrition Program for Women, Infants, and Children, the National School Lunch Program]. This is an important consideration as about 20 million households with children participate in at least one of these three programmes (4). This omission is perhaps surprising in light of the extensive work looking at the impact of SNAP on food insecurity (34–38) and obesity (33,39–42).

Jones et al. hypothesized that (i) the prevalence of child overweight varies by household food security status and participation in food assistance programmes in low-income households and (ii) food assistance programmes moderate the association between household food security status and child overweight. In a sample of 5- to 12-year-old low-income children (185% income–poverty ratio) (n = 772) from the 1997 Panel Study of Income Dynamics Child Development Supplement, 34% were overweight and 24% were food-insecure. A majority (84%) of low-income families participated in at least one food assistance programme. The prevalence of overweight was lower in food-insecure boys and girls (36% and 23%, respectively) compared with food-secure boys and girls (46% and 29%, respectively). Furthermore, food-insecure children participating in food assistance programmes had a lower prevalence of overweight compared with food-secure children participating in food assistance programmes. Results of logistic regression also indicated that food-insecure girls, but not boys, who participated in food assistance programmes had a significantly reduced odds (OR 0.32; 0.12–0.77 95% confidence interval) of being overweight when compared with food-insecure girls in nonparticipating households. However, it should be noted that the odds ratios for food-insecure boys were less than 1.0. In addition, even though the authors found that ‘Girls (and boys) in food-secure households generally had no greater or less risk of overweight if they participated in any or all programs’, the odds ratios were also less than 1.0. Therefore, food assistance programmes may play a protective role for risk of overweight in low-income children.

While the results from this paper are suggestive, they are by no means conclusive. Previous research on food assistance programmes has demonstrated that food assistance recipients are different from non-recipients in both observed and unobserved ways (36). In terms of the former, non-recipients tend to have higher incomes and be in households headed by a married couple and a homeowner. The methodology employed by the authors controls for some of these observed factors; however, the methodology does not control for unobserved factors that may lead individuals to participate in these programmes. As a consequence, the true effect of food assistance programmes may not be portrayed in their analyses. Future work may wish to more clearly delineate the effect of food assistance on obesity in the presence of food insecurity.

Nationally representative longitudinal or prospective cohort studies

The papers summarized to this point have been cross-sectional studies. Although they show the magnitude of the association, they cannot determine time-order. The papers summarized in this section are longitudinal or prospective cohort studies.

Early Childhood Longitudinal Study – Kindergarten

Four papers (43–46) have been published using data from the ECLS-K. In the first studies using the ECLS-K, both Winicki and Jemison (46) and Rose and Bodor (45) included a large sample (n > 15 000) of young children (mean age approximately 6 years). Winicki and Jemison (46) examined changes in BMI during the kindergarten year and found no differences between food-secure and food-insecure children. Rose and Bodor (45) defined excess weight gain as a gain in BMI from wave 2 (Spring 1999) to wave 4 (Spring 2000) (1 year) in the top 15% of age- and sex-specific groups. Although descriptive characteristics indicated that there were no significant differences in mean BMI or prevalence of overweight/obesity between food-insecure and food-secure children in any race or gender groups, results from logistic regression analyses, controlling for several covariates, showed decreased odds (OR = 0.80, 0.66–0.98 95% CI) of overweight for food-insecure children compared with food-secure children. In sub-analyses, the authors used various expressions of both food insecurity (e.g. child food insecurity) and overweight (e.g. BMI > 85th centile) and also continuous measures in linear regressions and found similar results (i.e. OR < 1.0 or negative (inverse) beta coefficient). Lastly, the longitudinal analyses also indicated an OR = 0.73 (95% CI not provided) for risk of excessive weight gain during 1 year of follow-up.

Jyoti et al. (44) utilized the longitudinal data and examined children from baseline (kindergarten) through grade 3. The authors use fixed-effects models which allow the researcher to control for all unobserved, time invariant factors. Results indicated that food-insecure children at baseline had a 0.27 kg m⁻² greater gain in BMI than food-secure children (P < 0.001); however, when stratified by gender only girls showed a significant increase (beta coefficient = 0.503 kg m⁻²) (P < 0.001). These results held after controlling for a host of both time-varying and invariant covariates (e.g. child age, gender, socio-demographic characteristics, child exercise, parent depression, etc.) (0.428 kg m⁻²). Longitudinal analyses using persistent food insecurity and transitions into and out of food insecurity also showed similar results – that is, significant associations in food-insecure girls (beta coefficient of 0.55 kg m⁻²) (P = 0.02) compared with persistently food-secure girls, but not boys.

Bhargava et al. (43) conducted analyses on 7635 children from the ECLS-K from grade 1 through grade 5 using random-effects models and other econometric models. Instead of treating food insecurity as a dichotomous variable, the authors created a continuous measure of household food insecurity by summing the affirmative responses to the 18 CFSM food insecurity items. In contrast to the results of Jyoti et al. (44), household food insecurity was not associated with body mass, BMI or BMI z-score. The authors attributed this finding to the statistical models and measurement of household food insecurity.

Longitudinal Study of Child Development in Quebec

This study examined 1514 4- to 5-year-old children who participated in the nutrition portion of a longitudinal study from the Canadian province of Quebec (47). The mean BMI was significantly higher in children from food-insufficient families (17.1 kg m⁻²) compared with food-sufficient children (16.2 kg m⁻²) controlling for birthweight, parental BMI, parental education and income. Logistic regression results indicated that food-insufficient children were twice and 3.4 times, respectively, as likely to be classified as obese based on CDC and International Obesity Task Force cut-points. A subsequent analysis showed a food sufficiency and birthweight interaction. Food-insufficient children with either low or high birthweight were almost 30 times and six times, respectively, more likely to be overweight compared with children of normal birthweight.

Summary, conclusions and recommendations for future research

When considering the question of food insecurity and weight status of children, it is difficult to imagine that food-insecure children would be overweight given that early growth studies of poor children clearly showed retarded growth and underweight (2). The impetus for examining the relationship between food insecurity and child overweight began with a case report of a low-income, food-insecure, obese girl published in 1995 (16). Here, we have reviewed 21 studies (16 cross-sectional and five prospective studies) that have examined this topic since the initial case report. In general, the results are mixed with positive, negative and null associations. Therefore, it is difficult to conclude that food insecurity is associated with child and adolescent overweight because of the lack of consistent results and measurement in the published literature. Likewise, the American Dietetics Association, in a note published in 2003, concluded that ‘household food insecurity does not appear to be associated with overweight among children, a finding that may be due, in part, to the fact that a comprehensive measure of child food insecurity was not used in most studies’ (http://www.adaevidencelibrary.com/conclusion.cfm?conclusion_statement_id=39). However, more recent studies have begun to employ comprehensive food insecurity measures, including at the child level. Clearly, more research is needed as the majority of studies still found that obesity and food-insecurity co-exist in children and adolescents, especially those living in poverty.

The reasons for the mixed results are difficult to disentangle. In some of the early studies, there appeared to be age-, sex- and race-variation in these associations. Small analytic samples from earlier studies have also hampered definitive conclusions. In addition, previous studies that included household food insecurity were not specific to the child of interest, and therefore, the child him or herself may not have been food-insecure. Most of the more recent studies that have employed larger samples and more rigorous statistical analyses have shown that there is no relationship between food insecurity and obesity (e.g. (26,27). Thus, the recent findings support the conclusion by the American Dietetics Association (see above) without the caveat that it might be due to not measuring food insecurity at the child level. Regardless of whether food insecurity and obesity are related after controlling for other factors, all of the studies to date have definitely shown that food insecurity and overweight/obesity co-exist. Thus, even though the paradoxical relationship may not exist at the population level, the paradoxical food-insecure-obesity phenotype exists at the individual level.

Future research should address methodological limitations inherent in addressing the issue of the acute and chronic nature of exposure to periods of food insecurity. In addition, prospective studies utilizing child-specific measures of food insecurity are also warranted to establish time-order. Hopefully, additional studies will clarify the direction of the relationship and the mechanisms that underlie the co-existence of the food-insecure-obese phenotype in children. One line of inquiry that requires further attention is the possible mediating effects of parenting and feeding behaviour and family stressors, including maternal depression, in the relationships between food security and obesity in children. An additional area that requires attention is the possible effect of physical inactivity as a confounding factor. As food insecurity is more prevalent among lower socioeconomic status, excessive television watching, use of motor vehicles for short transport, desire to reduce any physical effort and so on could have an additive effect on food insecurity leading to increased prospects of obesity. Furthermore, given the complex psychosocial, economic and biological aspects of food insecurity and obesity among low-income children, a ‘systems-oriented’ type approach is needed (48).

Conflict of Interest Statement

None.

Acknowledgements

This work is supported by the USDA, Cooperative State Research, Education, and Extension Service grant 2007-35215-17871.