Household Food Security and Infant Adiposity

Preventing food insecurity in young children in the United States is critically important; in this study, we investigate longitudinal associations of household food security and infant adiposity.

Abstract

OBJECTIVES:

Food insecurity has been associated with obesity, but previous studies are inconsistent and few included infants. We examined associations between household food security and infant adiposity and assessed the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and Supplemental Nutrition Assistance Program (SNAP) as effect modifiers. We hypothesized that infants from food-insecure households would have greater adiposity, with attenuation by WIC and SNAP.

METHODS:

We repeatedly measured 666 infants from the southeastern United States in 2013–2017. We categorized households as high, marginal, low, or very low using the US Household Food Security Survey Module. Outcomes were BMI z score, subscapular and triceps skinfold-for-age z score, the sum of subscapular and triceps skinfolds, the ratio of subscapular and triceps skinfolds, and BMI z score ≥1 (at risk for overweight). We used covariate-adjusted repeated-measures linear and logistic regressions.

RESULTS:

Of infants, 68.6% were Black and 60.5% had household incomes <$20 000. Interactions between food security and WIC and/or SNAP were not significant. Compared with infants from high food security households, infants from very low food security households had higher BMI z scores (0.18 U; 95% confidence interval [CI] 0.01 to 0.35), higher subscapular skinfold-for-age z scores (0.31 U; 95% CI 0.04 to 0.59), a higher sum of subscapular and triceps skinfolds (0.53 mm; 95% CI 0.002 to 1.07), and greater odds of being at risk for overweight (odds ratio 1.55; 95% CI 1.14 to 2.10). Infants from low food security households had greater odds of being at risk for overweight (odds ratio 1.72; 95% CI 1.17 to 2.10).

CONCLUSIONS:

In larger and longer studies, researchers should examine food security and adiposity in young children.

What’s Known on This Subject:

Food insecurity in young children may be associated with obesity and excess adiposity, but findings are inconsistent. However, in a single previous study of low-income infants, authors found no association between food insecurity and obesity.

What This Study Adds:

Infants from very low food security households, compared with infants from high food security households, had higher odds of being at risk for overweight and had greater adiposity over the course of infancy.

Preventing food insecurity in young children in the United States is an important public health goal because there are both immediate and long-term adverse health consequences.¹ Household food security is defined by the US Department of Agriculture as all members having readily available, nutritionally adequate, and safe foods at all times.² An estimated 12% of households were food insecure in the United States in 2017.³ That same year, nearly 3 million or ∼8% of households included children who experienced food insecurity.³ In a recent report by the US Department of Agriculture, it was estimated that nationally, 7.1% of food-insecure households had children aged 6 years and younger who experienced food insecurity in 2018.³ In another study of caregivers with children aged 5 years and younger in 4 US cities, household food insecurity ranged from ∼10% to 15% in 2018.⁴ Little is known, however, about the prevalence of food insecurity among infants. In one previous study of 144 infants in Ohio, authors found that 31% lived in food-insecure households.⁵ In a study of ∼20 000 infants participating in a federal food assistance program in Massachusetts, authors found that 20% to 25% lived in households that were food insecure.^6,7

Infants may be especially vulnerable to food insecurity because infancy is a time of rapid growth. Food insecurity during infancy has been associated with overall poor health,⁸ but few studies have assessed associations between food insecurity and obesity or excess adiposity in young children. In a recent large cross-sectional study of low-income infants from 5 US cities, authors found no association between food insecurity and obesity, defined as weight-for-age >90th percentile by using World Health Organization (WHO) reference data.⁹ There is, however, some evidence that food insecurity in preschool- and school-aged children may be associated with obesity and excess adiposity.^6,10–15 The results of these studies, however, have been mixed. In one study, authors observed an association with overweight among girls only.¹³ On the other hand, the authors of some studies observed no association,^16–20 mixed associations with obesity was noted in one study,⁹ and the authors of another study found that food insecurity had a protective effect against having overweight.²¹ Additionally, the authors of 2 previous systematic reviews found mixed evidence of an association between food security and weight status in children,^22,23 further adding to the lack of clarity surrounding food insecurity and obesity or excess adiposity in early childhood. Food insecurity may be associated with lower diet quality,^24–26 which, in turn, may increase risk of obesity. In young children, caregiver feeding behaviors may also play a role.^10,24–29

Moreover, the ability of federal food assistance programs to influence the association between food insecurity and obesity or excess adiposity remains unclear. Programs such as the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and the Supplemental Nutrition Assistance Program (SNAP) play a role in alleviating food insecurity and associated health risks in young children.^30–34 In 2 previous studies, authors found a protective effect of WIC on childhood overweight and the BMI z score,^31,35 but in another study, it was suggested that participation may be associated with greater obesity.³⁶ This association may, however, depend on food security status. Kohn et al³⁵ found that participation in WIC and SNAP was associated with higher weight in food-secure but not food-insecure youth. In a large study of children younger than 3 years, authors found that SNAP participation was marginally (approached but did not reach significance) associated with lower odds of being at risk for obesity.³⁴ Similarly, but in school-aged children, Jones et al³⁷ found that girls had a reduced risk of overweight if they participated in the Food Stamp program (now SNAP) and the National School Lunch and Breakfast program. However, in the study by Nguyen et al,³⁸ household food security was not associated with BMI percentile in school-aged children whose families participated in SNAP. The majority of these studies included few or no infants and were limited by their cross-sectional design. Therefore, our purpose for this study was to examine associations between household food security and infant adiposity. Secondarily, we evaluated whether participation in WIC and SNAP modified this potential relation. We hypothesized that infants from food-insecure households would have greater adiposity than those from food-secure households and that WIC and SNAP participation would attenuate this association.

Methods

Study Design and Population

Participants were from the Nurture Study, a birth cohort of 666 racially diverse women and their infants residing in the southeastern United States.³⁹ Between 2013 and 2017, we enrolled pregnant women, reconsented them after delivery, and conducted home visits when infants were 3, 6, 9, and 12 months. Women completed questionnaires via telephone in months 1, 2, 4, 5, 7, 8, 10, and 11. The goal of the Nurture Study was to assess feeding, physical activity, sleep, and stress and associations with weight gain in infancy, exploring the role of food security.

We recruited women from a private prenatal clinic and the local county health department prenatal clinic. To participate, women were required to be between 20 and 36 weeks into their pregnancy, be pregnant with a singleton with no known congenital abnormalities, be at least 18 years old, speak and read English, intend to keep their infants, and plan to stay within the area for 1 year. At the reconsent at birth, we excluded infants born before 28 weeks’ gestation, infants with congenital abnormalities that could affect development, infants in the hospital for ≥3 weeks, and infants not able to take food by mouth at discharge. Women provided written informed consent for their participation and that of their infants. The Duke University Medical Center Institutional Review Board approved this study.

Exposure: Household Food Security

Our primary exposure was household food security status, measured when infants were 3, 6, 9, and 12 months. We used the 18-item US Household Food Security Survey Module.⁴⁰ We asked mothers to consider the past 30 days rather than the previous 12 months when responding to each question because we were interested in assessing household food security status over a shorter and often dynamic life stage. We categorized response scores of 0 as high, 1 to 2 as marginal, 3 to 7 as low, and 8 to 18 as very low food security, consistent with the US Department of Agriculture classifications for households with ≥1 child present.⁴⁰

Outcomes: Infant Adiposity

We measured infant length using a ShorrBoard Portable Length Board to the nearest one-eighth inch and infant weight using a Seca Infant Scale to the nearest 0.1 lb. We calculated infant BMI z scores using WHO age- and sex-specific reference data.⁴¹ We classified infants as at risk for overweight if their BMI z scores were ≥1, per WHO cut points.^42,43 We measured subscapular and triceps skinfold thicknesses to the nearest 0.2 mm at each home visit using Holtain Tanner/Whitehouse Skinfold Calipers.

We used the sum of subscapular and triceps skinfolds (SS + TR) thickness as a proxy for overall fatness, and the ratio of subscapular and triceps skinfolds (SS/TR) as a proxy for central fatness. This approach is consistent with that of previous studies.^44,45 We also calculated subscapular and triceps skinfold-for-age z scores using WHO growth standards.⁴¹ We measured in triplicate, taking an average of the three, using standard techniques.⁴⁶

Other Measures

Maternal variables included race (Black, white, other or >1 race), age, education (high school graduate or less, some college or more), prepregnancy BMI, and number of children in the household. Infant variables included race (Black, white, other or >1 race), sex, and birth weight for gestational age z score. At each home visit, we assessed participation in WIC and SNAP. We also measured breastfeeding monthly and calculated total months of any breastfeeding.

Analysis

Of the 666 women who enrolled their infants, 535 (80.3%) completed the 3-month home visit, 497 (74.6%) completed the 6-month home visit, 457 (68.6%) completed the 9-month home visit, and 468 (70.3%) completed the 12-month home visit. We examined covariate-adjusted associations between household food security and infant adiposity at 3, 6, 9, and 12 months. We fit repeated-measures linear and logistic regression models with an unstructured error covariance (linear) or an unstructured working covariance (logistic) to account for the correlation among repeated measures within infants. We treated food security as a time-varying exposure because families may experience food insecurity differently over time. We evaluated 4 categories of food security, including high, marginal, low, and very low, with high security as the reference. We treated time (ie, each home visit) as a categorical variable to allow for a nonlinear time trend. We omitted an interaction between time and food security after testing for and failing to detect a significant difference of food security effect over time. To examine effect modification, we tested a 2-way interaction between food security and WIC or SNAP separately.

The final model included covariates we identified a priori as potential confounders on the basis of existing literature,^{6,13,20,25,38,47–49} including maternal race, age, education, and prepregnancy BMI; number of children in the household; and infant race, sex, birth weight for gestational age z score, and breastfeeding. However, because we were interested in fat distribution after controlling for overall body size, we further adjusted for BMI z score in our analyses with SS/TR.^44,45 We fit the final multivariable models to both available-case and multiply imputed data sets. For multiple imputation, we generated 1000 imputed data sets and combined parameter estimates and SEs using standard imputation formulas.⁵⁰ We used an F test to assess the interaction between WIC and food security and SNAP and food security and combined results across multiply imputed data sets using the approach described by Grund et al.⁵¹ Results were nearly identical between imputed and nonimputed models. Therefore, we present results from imputed models in terms of parameter estimates, 95% confidence intervals (CIs), and 2-sided P values. We conducted these analyses using SAS version 9.4 (SAS Institute, Inc, Cary, NC), with a significance level of α = .05.

Results

Of infants, 68.6% were Black and 14.9% were white (Table 1). Households had a mean (SD) of 3.7 (1.6) children. Just over half (55.4%) had household incomes ≤$20 000 per year. Among infants from food-secure households (high and marginal food security) the percentage of infants at risk for overweight decreased from 46.8% at 3 months to 33.1% at 12 months, whereas among infants from food-insecure households (low and very low food security), the percentage increased from 53.2% to 66.9%. Household food security status also changed over the course of infancy (Table 2). Of households, 66.4% remained in the high or marginal food security category and 8.8% remained in the low or very low food security category across assessments; the remainder changed categories over time (eg, moved from marginal to low). When we tested for effect modification, we found no evidence of interactions between food security and WIC or SNAP (P = .691 for WIC and P = .22 for SNAP). Therefore, we fit a main effect model only and adjusted for WIC and SNAP in all analyses.

TABLE 1.

Demographic Characteristics of Participants in the Nurture Study, 2013–2017 (N = 666)

	Value
Infant characteristics
Birth length, cm, mean (SD)	49.7 (2.6)
Birth wt, g, mean (SD)	3209.7 (512.5)
Birth wt for gestational age z score, mean (SD)	0.1 (1.0)
Any breastfeeding duration, min, mean (SD)	4.1 (4.3)
BMI z score, mean (SD)
At 3 mo	0.2 (0.9)
At 6 mo	0.3 (1.0)
At 9 mo	0.6 (1.0)
At 12 mo	0.7 (1.0)
At risk for overweight (BMI z score ≥1), No. (%)
At 3 mo	94 (14.1)
At 6 mo	110 (16.5)
At 9 mo	129 (19.4)
At 12 mo	154 (23.1)
Wt gain from birth to 12 mo, kg, mean (SD)	6.6 (1.2)
Female sex, No. (%)	326 (48.9)
Race, No. (%)
Black	457 (68.6)
White	100 (14.9)
Other race or >1 race	99 (14.9)
Ethnicity, Latinx or Hispanic, No. (%)	59 (8.9)
Age, mean (SD), mo
At 3-mo assessment	3.4 (0.5)
At 6-mo assessment	6.3 (0.5)
At 9-mo assessment	9.5 (0.5)
At 12-mo assessment	12.5 (0.8)
Maternal characteristics
Age, y. mean (SD)	27.3 (5.8)
Prepregnancy BMI, mean (SD)	29.9 (9.2)
Parity assessed in pregnancy, No. (%)
0	236 (35.4)
1	190 (28.5)
2	216 (32.4)
Marital status, No. (%)
Married or living with partner	395 (59.3)
Never married, divorced, separated, or other	271 (40.7)
Race, No. (%)
Black	476 (71.4)
White	128 (19.2)
Other race or >1 race	62 (9.3)
Ethnicity, Hispanic, No. (%)	43 (6.5)
Education, No. (%)
Less than high school	135 (20.3)
High school degree	184 (27.6)
Some college	240 (36.0)
College or graduate degree	107 (16.1)
Household characteristics, mean (SD)
No. adults	2.9 (1.2)
No. children	3.7 (1.6)
Annual income, No. (%)
<$20 000	403 (60.5)
$20 000–$39 999	133 (20.0)
$40 000–$70 000	65 (9.8)
>$70 000	49 (7.4)
Missing	16 (2.4)
WIC participation, No. (%)
At 3 mo	391 (58.7)
At 6 mo	357 (53.6)
At 9 mo	323 (48.8)
At 12 mo	318 (47.7)
SNAP participation, No. (%)
At 3 mo	352 (52.3)
At 6 mo	324 (48.6)
At 9 mo	292 (43.8)
At 12 mo	289 (43.4)

TABLE 2.

Household Food Security Status of Nurture Study Infants at 3, 6, 9, and 12 Months

Household Food Security	No. (%)
	3 mo	6 mo	9 mo	12 mo
High	348 (65.3)	358 (72.9)	341 (74.8)	348 (74.7)
Marginal	67 (12.6)	48 (9.8)	43 (9.4)	40 (8.6)
Low	74 (13.9)	50 (10.1)	33 (7.2)	37 (7.9)
Very low	44 (8.3)	35 (7.1)	39 (8.6)	41 (8.8)

After adjustment for potential confounders, infants from very low food security households had higher BMI z scores (0.18 U; 95% CI 0.01 to 0.35; P = .04), higher subscapular skinfold-for-age z scores (0.31 U; 95% CI 0.04 to 0.59; P = .03), higher SS + TR (0.53 mm; 95% CI 0.002 to 1.07; P = .046), and greater odds of being at risk for overweight (odds ratio 1.55; 95% CI 1.14 to 2.10; P = .005) compared with infants from high food security households (Tables 3–5). Additionally, infants from low food security households had greater odds of being at risk for overweight (odds ratio 1.72; 95% CI 1.17 to 2.10; P = .005). We did not observe any other significant associations.

TABLE 3.

Adjusted Estimates and 95% CIs of Longitudinal Associations of Household Food Security and Infant Adiposity: At Risk for Overweight and BMI z Score

Household Food Security	Infant Adiposity
	At Risk for Overweight (BMI z Score ≥1)			BMI z Score
	Odds Ratio	95% CI	P	Estimate	95% CI	P
High	Reference	Reference	Reference	Reference	Reference	Reference
Marginal	1.07	0.78 to 1.49	.66	0.01	−0.11 to 0.19	.57
Low	1.72	1.17 to 2.48	.005	0.09	−0.10 to 0.27	.35
Very low	1.55	1.14 to 2.10	.005	0.18	0.01 to 0.35	.04

Adjusted for infant race, sex, birth weight for gestational age z score, and breastfeeding; maternal age, education, and prepregnancy BMI; and number of children in the household, WIC participation, and SNAP participation.

TABLE 5.

Adjusted Estimates and 95% CIs of Longitudinal Associations of Household Food Security and Infant Adiposity: SS + TR and SS/TR

Household Food Security	Infant Adiposity
	SS + TR			SS/TRa
	Estimate	95% CI	P	Estimate	95% CI	P
High	Reference	Reference	Reference	Reference	Reference	Reference
Marginal	0.24	−0.3 to 0.72	.32	0.01	−0.02 to 0.40	.50
Low	0.23	−0.34 to 0.80	.42	−0.01	−0.05 to 0.03	.55
Very low	0.53	0.002 to 1.07	.046	0.01	−0.02 to 0.04	.53

Adjusted for infant race, sex, birth weight for gestational age z score, and breastfeeding; maternal age, education, and prepregnancy BMI; and number of children in the household, WIC participation, and SNAP participation.

^aAdditionally adjusted for BMI z score.

TABLE 4.

Adjusted Estimates and 95% CIs of Longitudinal Associations of Household Food Security and Infant Adiposity: Triceps Skinfold-for-Age z Score and Subscapular Skinfold-for-Age z Score

Household Food Security	Infant Adiposity
	Triceps Skinfold-for-Age z Score			Subscapular Skinfold-for-Age z Score
	Estimate	95% CI	P	Estimate	95% CI	P
High	Reference	Reference	Reference	Reference	Reference	Reference
Marginal	0.08	−0.22 to 0.38	.60	0.16	−0.08 to 0.40	.20
Low	0.17	−0.18 to 0.52	.34	0.07	−0.24 to 0.38	.66
Very low	0.23	−0.09 to 0.55	.16	0.31	0.04 to 0.59	.03

Adjusted for infant race, sex, birth weight for gestational age z score, and breastfeeding; maternal age, education, and prepregnancy BMI; and number of children in the household, WIC participation, and SNAP participation.

Discussion

In this cohort of infants, we found that infants from low and very low food security households had greater odds of being at risk for overweight compared with infants from high food security households. Moreover, we found that infants from very low food security households had higher BMI z scores, subscapular skinfold-for-age z scores, and SS + TR, which is a marker of overall adiposity. We did not, however, observe these associations for infants from low or marginal food security households. Furthermore, we did not observe any associations between household food security and triceps skinfold-for-age z scores or SS/TR, which may indicate that food security is less influential on extremity versus trunk fat. Alternatively, the effect of food insecurity may need to accumulate throughout both infancy and childhood before a consistent association with skinfolds emerges.

Our findings build on results from a recent study of food insecurity and obesity in the United States that included a large number of infants.⁹ The authors examined food security and obesity and found no associations.⁹ However, the cross-sectional design, the inclusion of low-income children only, and the focus on growth parameters associated with subsequent health and development risks (and not with anthropometric changes within the normal range) may explain the differences in our findings. The authors recommended that subsequent studies investigate the mechanisms linking food insecurity to adverse health conditions. Thus, our study extends and complements the previous work by Drennen et al.⁹

There are several potential explanations for our finding that very low food security was associated with greater adiposity during infancy. One may relate to maternal feeding behaviors in the absence of an adequate supply of healthy foods. Some previous studies suggest that food insecurity may influence obesity and excess adiposity through inappropriate infant feeding practices and less healthy diets.^10,24–29 It is well established that healthier foods cost more than less healthy foods, contributing to socioeconomic disparities in health outcomes.^52–54 Studies in older children suggest associations between food insecurity and lower diet quality.^24–26 In a previous study of 278 mothers and children, Feinberg et al²⁷ found that food-insecure mothers were more likely to give their preschool- and school-aged children high-energy supplements to compensate for food shortages. However, no researchers have examined food security and diet quality in infants.²⁶

In other studies, authors have examined associations between food insecurity and maternal feeding behaviors. In a sample of 201 predominately Hispanic mothers and infants participating in WIC, Gross et al²⁸ found that food insecurity was not associated with breastfeeding or adding cereal to the bottle. Moreover, in a sample of 8693 infants, Bronte-Tinkew et al¹⁰ did not observe associations between household food security and breastfeeding or early introduction of solid foods. However, their results suggested that food security influenced parenting style, which, in turn, was associated with infant feeding behaviors. These studies provide some evidence of a food insecurity–obesity association, but additional research is needed to help identify a relationship and elucidate potential pathways.

Notably in our study, we did not observe an interaction between WIC and food security or SNAP and food security. Therefore, our findings suggest that the effect of food security did not vary by WIC or SNAP participation, which was contrary to our hypothesis. Our hypothesis was based on previous studies in which authors found a protective effect of WIC and SNAP participation on childhood obesity.^31,34,35 Conversely, WIC and SNAP may be associated with greater weight,³⁶ but this may depend on children’s food security status.³⁵ Nonetheless, both WIC and SNAP play an essential role in the lives of food-insecure families in the United States.^30–33,55

This study has limitations. First, we measured food security at the household level rather than in infants. Mothers responded to questions on child food security for all children in the household and not just the index infant enrolled in the Nurture Study. There is limited information on the extent to which household food insecurity reaches infants. Second, we modeled food security in 4 categories rather than dichotomizing households as food secure or food insecure.^{5,6,9,13,23,56,57} There is, however, some previous evidence of associations of marginal household food security and adverse health outcomes in young children, suggesting that marginal food security should be examined separately,⁵⁸ which is why we modeled them separately.

Third, Nurture Study participants were not representative of the general population in the southeastern United States; the demographic composition of our sample included a higher representation of Black infants. However, Black infants are underrepresented in most US birth cohorts,⁵⁹ so this racially diverse cohort makes a unique contribution. Fourth, as with most birth cohorts, we experienced attrition. From birth to the 12-month assessment, we retained 70% of the sample. In a similar birth cohort from the same geographic area, 56% of women completed a 12-month follow-up.⁶⁰ To compensate for modest attrition, we imputed data for this analysis. Fifth, we did not assess the nutritional quality of infants’ diets. Diet quality, for example, may fall within the causal pathway between household food security and infant adiposity and is an area for further exploration. We did, however, control for breastfeeding, which is likely an important factor related to diet.

Finally, we did not use a more objective measure of infant adiposity. A recent study suggests that bioelectrical impedance may be a more appropriate measure of body composition in infants.⁶¹ This may explain the inconsistent associations we observed between household food security and infant adiposity. For instance, very low household food security was associated with higher infant BMI z scores (both dichotomized and continuous), but low household food security was associated with higher BMI z scores as a dichotomized variable only. Moreover, very low household food security was associated with greater subscapular, but not triceps, skinfold thickness; low household food security was associated with neither. Notably, however, the percentage of infants at risk for overweight increased substantially from 3 to 12 months but more so among food-insecure infants.

In previous studies, authors have observed similar inconsistent associations between exposure and BMI or skinfold thickness. For example, Taveras et al⁶² found that infants who slept <12 hours per day had higher BMI z scores and greater SS + TR but not SS/TR. Similarly, Gillman et al⁶³ observed associations between higher sugary beverage consumption in pregnancy and higher BMI z score and SS + TR, but not SS/TR, among children in middle childhood. Skinfold thickness may be more closely associated with body fat than BMI in infants.⁶⁴ However, in another study,⁶⁵ authors found that among infants, skinfold thicknesses and BMI were relatively equivalent in their ability to predict later body fat. Despite a more objective measure of infant adiposity, this study includes both BMI and skinfolds, and both provide useful information about risk of overweight and excess adiposity.

Conclusions

Food insecurity may be especially problematic during infancy because the first year represents a dynamic time for growth and development. We found that food insecurity in infancy was associated with obesity and excess adiposity, but larger and longer studies are needed to assess sustained associations.

Glossary

CI

confidence interval

SNAP

Supplemental Nutrition Assistance Program

SS + TR

sum of subscapular and triceps skinfolds

SS/TR

ratio of subscapular and triceps skinfolds

WHO

World Health Organization

WIC

Special Supplemental Nutrition Program for Women, Infants, and Children