Household food insecurity and home food availability in relation to youth diet, body mass index, and adiposity

Abstract

Background.

Food security status is related to food types available in the home, which may shape youth dietary patterns, with implications for obesity.

Objective.

Investigate whether household food insecurity and home food availability (HFA) are associated with youth fruit and vegetable (FV) consumption and anthropometric outcomes.

Design.

Cross-sectional study. Youth and parents completed questionnaires during in-home visits (2013–2014). Research staff obtained anthropometric measures.

Participants/setting.

Medical record data for 10–15 year-old Pennsylvania youths were used to identify 434 parent-youth dyads, with 408 evaluated after excluding missing data.

Main outcome measures.

Parent-reported household food security was assessed with the 6-item U.S. Department of Agriculture Food Security Scale (dichotomized as high versus low). Healthy and obesogenic HFA scales assessed parent-report of how frequently particular foods were present in the home. Youth self-reported daily average FV consumption. Anthropometric outcomes included age- and sex-standardized z-scores for body mass index (BMIz), waist circumference (WCz), and percent body fat (PBFz).

Statistical analyses.

Associations were evaluated with multivariable linear regression adjusted for youth age, sex, and race/ethnicity, and parent age and income.

Results.

Compared to food secure counterparts, youth from food insecure households had higher mean (beta [standard error]) BMIz (0.30 [0.15]), WCz (0.27 [0.12]) and PBFz (0.43 [0.16]). Food insecure households had lower mean healthy HFA scores (−1.23 [0.54]); there was no evidence obesogenic HFA differed between food secure and insecure households. Youth from lower healthy HFA or higher obesogenic HFA households reported fewer mean daily FV servings (healthy HFA: 0.08 [0.02]; obesogenic HFA: −0.06 [0.02]). Food security status was not associated with FV consumption, nor was there evidence HFA modified associations between food insecurity and anthropometric outcomes.

Conclusions.

Despite an observed association between healthy HFA and youth FV consumption, this study did not provide evidence that HFA explained associations between food insecurity and youth anthropometric outcomes.

INTRODUCTION

Approximately 20.5% of U.S. adolescents were obese in 2015–2016, a prevalence that increased over the decade prior.¹ Youth overweight/obesity is predictive of obesity later in life and associated with early onset of high blood pressure, type-2 diabetes, and cardio-metabolic disease.^2,3 Given its prevalence and long-term health impacts, understanding the determinants of youth obesity is critical, particularly among high-risk groups.

As dual consequences of socioeconomic disadvantage, obesity and food insecurity frequently coexist.^4,5 In 2016, 17.5% of U.S. children lived in food-insecure households, in which members lack sufficient food for an active, healthy life.⁶ Food insecurity disproportionately impacts groups at elevated risk of obesity, including low-income households and members of racial and ethnic minority groups,^6–8 and the prevalence of overweight is consistently elevated among food-insecure youth.⁵ Yet research examining the association of food insecurity and youth overweight/obesity has shown inconsistent results, with findings differing by age, race/ethnicity, sex, and poverty level.^5,7,9–12

Differential dietary patterns among youth in food insecure households could in part explain the coexistence of food insecurity and obesity; however, past literature demonstrates no clear association between food security status and diet quality. Some studies found that youth in food insecure households consumed a less healthy diet than their food secure counterparts,^13,14 while others have reported few differences in dietary patterns or nutrient intake.^15,16 The relationship between food security status and youth diet is complicated by participation in federal nutrition assistance programs that are designed to improve food security (but that do not necessarily improve diet quality), opportunities for youth to supplement their diets outside the home, and protective measures taken by parents to ensure children have sufficient food.^16–20

Household food insecurity could shape youth diet quality through its influence on the home food environment, an important context for influencing children’s diet.²¹ Even as adolescents gain greater autonomy in their food choices, the home food environment remains a major source of calories and plays a significant role in their current and future dietary patterns.^22–24 Parents can influence their children’s food intake through directive and non-directive control; directive practices include restricting access to food and pressure-to-eat, while non-directive practices involve parental modeling of food choices and the availability of healthy or obesogenic foods at home.²⁵ In two studies that disentangled these components of the home food environment, home food availability (HFA) emerged as a more consistent determinant of youth dietary outcomes than directive parenting practices.^25,26 These studies provide a more nuanced understanding of the home food environment and build upon past research that has repeatedly found a relationship between the availability and accessibility of foods in the home and youth intake of these foods, whether assessing fruits and vegetables or obesogenic foods.^23,27–33 Of the few studies that have evaluated HFA in relation to youth anthropometric measures, none have observed associations.^25,26,30,34

While a complex body of literature has examined linkages between food security status, healthy and obesogenic HFA, youth dietary patterns, and obesity risk, their interrelationships have yet to be fully elucidated. To the authors’ knowledge, no prior study has examined youth anthropometric measures in the context of household food security status, HFA, and youth eating behavior, nor assessed whether HFA modifies potential relations between household food security and youth eating behavior or weight status. Thus, the aim of this study was to evaluate how food insecurity and HFA may be associated with fruit and vegetable consumption and obesity risk among youth living in a geographically and socioeconomically diverse region of Pennsylvania. Based on prior literature, we hypothesized the following:

1. Household food security would be associated with greater mean daily fruit and vegetable servings and lower mean anthropometric measures among youth (as measured by body mass index z-score [BMIz], waist circumference z-score [WCz], and percent body fat [PBFz]).

2. Healthy HFA would be associated with greater mean daily fruit and vegetable servings and lower mean BMIz, WCz, and PBFz among youth, whereas obesogenic HFA would be associated with fewer mean fruit and vegetable servings and higher mean anthropometric outcomes.

3. Healthy or obesogenic HFA would modify associations between household food security and youth anthropometric outcomes.

METHODS

This study was the second phase of a project entitled Understanding Obesity from Epigenetics to Communities ^36,37. Institutional Review Boards from the Johns Hopkins Bloomberg School of Public Health and Geisinger approved the study procedures and all participants provided written informed consent or child assent.

Study Design

Electronic health record data were used to recruit youth ages 10–15 years and their parents from the Geisinger service area to participate in the study. Geisinger is an integrated healthcare system serving central and northeastern Pennsylvania. Youth were sampled from communities geographically distributed across Geisinger’s service area that represented a range of obesogenic and obesoprotective environments, based on low/high average youth body mass index (BMI) and quintiles (high versus low) of three features previously associated with obesity risk: diversity of physical activity establishments (e.g., fitness clubs, parks), population density, and community socioeconomic deprivation.³⁵ Based on these criteria, households of all youth with an electronic health record from 28 communities, which ranged from rural to urban, were contacted to determine eligibity (n = 3365). Of these, 89 were not eligible (e.g. did not live in study community) and eligibility was not determined for 791 households (e.g., disconnected phone). Among remaining eligible households (n = 2051), 434 parent-youth dyads completed data collection, for a response rate of 17.5%. Non-response was largely due to active refusal (43.3%) and inability to contact (i.e., repeatedly reaching an answering machine, 26.4%). Further details on community selection and recruitment were previously described.^36,37

Data were collected during in-home visits conducted in 2013–2014. Behavioral data and demographic covariates were obtained through self-administered questionnaires completed separately by one parent and one youth per household. Questionnaires included items related to the home food environment, eating behavior, and other obesity-related topics.³⁶ Trained staff took anthropometric measures of parents and youth as described below.

Measures

Food security status.

Food security status was assessed based on parent response to the short form (6-items) of the U.S. Department of Agriculture (USDA) Food Security Scale, which has been previously validated with diverse populations.³⁸ Questionnaire responses were coded based on USDA specifications and summed to calculate a food security score. Scores on the scale were dichotomized as high food security (0–1) versus low/very low food security (2–6) (few households had low or very low food security).

Home food availability.

Based on two scales from the Active Where Parent-Adolescent Survey, HFA was assessed using parents’ report of how frequently particular foods were present in their homes. Though not formally validated, in a prior study, both scales demonstrated independent associations with children’s fruit and vegetable intake, suggesting good predictive validity.²⁶ HFA was characterized as “healthy” or “obesogenic” depending on whether available foods aligned with key recommendations of the Dietary Guidelines for Americans³⁹ (e.g., fruit and vegetable intake is aligned, whereas high solid fat and added sugar intake is not aligned). To calculate healthy food availability, six lower calorie/more nutrient dense food items were assessed: four items included in the scale used by Couch et al.²⁶ (raw fruit, raw vegetables, baked chips, unsweetened cereal) and two items added to enhance the scale’s alignment with the Dietary Guidelines for Americans³⁹ (1% or fat free milk and whole wheat bread). Response options ranged from 0 (never present) to 4 (always present). This resulted in a score that ranged from 0–24 (six items multiplied by the response options for the frequency with which foods were present). The adapted scale (with two new items) has not been validated. To calculate obesogenic food availability, the eight high calorie/nutrient poor food items from Couch et al.’s scale²⁶ were assessed (chocolate candy, other candy, cakes/brownies/muffins/cookies, sweetened cereal, regular chips, juice drinks, sugar-sweetened soda, and sports drinks), resulting in a score that ranged from 0–32. Cronbach’s alpha for the two HFA scales were consistent with previously published values (healthy HFA: 0.59; obesogenic HFA: 0.75).²⁶ Previous evaluation of HFA scales demonstrated mostly moderate or good test-retest reliability for individual food items (intra-class correlation coefficients ranged from 0.466 to 0.830).⁴⁰

Youth report of fruit and vegetable consumption.

The PACE+ Fruit and Vegetable Measure, which has demonstrated reliability and significant correlation with 3-day food record data,⁴¹ was used to assess consumption. Responses were combined to create a single continuous measure of youth fruit and vegetable consumption (“In a typical day, how many servings of fruit/vegetables do you eat?”; examples of serving sizes provided). Response options included 0, 1, 2, 3, and 4 or more.

Anthropometric measures.

Trained staff followed national guidelines to measure youth and parent height, weight, and waist circumference.⁴² Youth PBF was measured through bioelectrical impedance analysis (BIA; model TBF-310, TANITA Corporation of America, Inc.). BIA has been shown to be a valid estimate for youth,^43,44 though a lack of research has limited the establishment of standardized BIA guidelines for children.⁴⁵ Measurements were taken three times each and averaged. BMI (kg/m²) was calculated using height and weight values and transformed into age- and sex-specific z-scores (BMIz) using the CDC’s 2000 growth charts.⁴⁶ Waist circumference z-scores (WCz) and PBF z-scores (PBFz) were calculated from age- and sex-specific mean values based on data from the National Health and Nutrition Examination Survey (data from 1998–1994 for WCz and 1999–2004 for PBFz).^47–50 Four implausible PBFz values (defined as five standard deviations above and below the mean) were deleted, thus sample sizes for PBFz models are slightly smaller.

Statistical Analyses

To evaluate study hypotheses, multivariable linear regression models assessed relations among household food security status and three sets of outcomes (youth fruit and vegetable consumption, youth anthropometrics [BMIz, WCz, PBFz], and healthy and obesogenic HFA) and among healthy and obesogenic HFA and two sets of outcomes (youth fruit and vegetable consumption and anthropometrics). In subsequent analyses, healthy and obesogenic HFA were individually added to models with food security and anthropometric outcomes to assess whether controlling for HFA attenuated associations. To further assess effect modification of relations between food security status and youth anthropometrics by HFA, cross-product terms with food security status and healthy and obesogenic HFA variables were included in separate models. All models controlled for common predictors of youth diet quality and weight status,²⁶ identified a priori, including youth age (continuous), youth sex, youth race/ethnicity (white versus non-white), parent age (continuous), and family income (categorical: $0-$24,999; $25,000-$49,999; $50,000-$74,999; $75,000 or more; missing). Models with alternative measures of socioeconomic status were also evaluated, including parent education and youth history of Medical Assistance (Medicaid), to assess the likelihood of residual confounding by socioecomomic status; inferences were unchanged and are not reported. The model evaluating food security status and healthy HFA also controlled for a centered-squared term for parent age. In sensitivity analyses, parent BMI (kg/m², centered) was included as a covariate in models assessing youth anthropometric outcomes; these models differed slightly in sample size due to three dyads with missing information on parent BMI. Given the common behavioral, cultural, and environmental conditions that contribute to both parent and child weight status, parent BMI was not included in primary models out of concern about over-controlling for these common causal pathways. Models were checked for outliers, influential points, multicollinearity, non-linearity, and heteroscedasticity. To assist with the interpretation of linear regression results for models evaluating healthy and obesogenic HFA and fruit and vegetable consumption and anthropometric outcomes, HFA variables were rescaled such that the beta coefficient represented the change in the outcome per an interquartile range (IQR) change in HFA. The resulting coefficients represented differences in outcomes comparing youth in households in the 75^th to the 25^th percentiles of the HFA variables.

To address potential non-response bias, we compared demographic data available in electronic health records of the 434 youth respondents and the 2,051 eligible non-respondents, including sex, age, race/ethnicity, and history of Medical Assistance (a measure of low socioeconomic status⁵¹). There were no significant differences in these variables comparing respondents and eligible non-respondents. From the original 434 youth-parent dyads, 26 dyads with missing data on food security status, HFA, youth anthropometrics, fruit and vegetable consumption, or parent age were excluded. Healthy HFA scores were lower among missing dyads (mean [standard deviation]) (12.0 [5.1] vs. 14.2 [4.4], t-test P = 0.01); otherwise there were no differences between the two groups in terms of demographic variables, food security status, obesogenic HFA, youth fruit and vegetable consumption, or youth anthropometrics. Results are reported as beta coefficients (β) with 95% confidence intervals (95% CI). Results were considered significant at P < 0.05 (two-tailed). Stata version 14.0 was used for data analyses.⁵²

RESULTS

The 408 youth in the analysis had a mean (standard deviation) age of 12.8 (1.7) years, 48% were male, and most were white non-Hispanic (Table 1). The majority of youth were under the 85^th percentile (age- and sex-standardized) for BMI, weight circumference, and percent body fat. A greater proportion of youth from food secure (versus insecure) households were female and white non-Hispanic, and a greater proportion of parents from food secure (versus insecure) households had a family income greater than $50,000 and had earned a bachelor’s or graduate degree.

Table 1.

Demographics and study outcomes of 408 Pennsylvania youth and parents who completed in-home questionnaires related to the home food environment, eating behavior, and other obesity-related topics, by household food security status^a

Characteristic	Food secure	Food insecure	Total
Youth	321 (100)	87 (100)	408 (100)
Age in years, median (interquartile range [IQR])	12.9 (11.4, 14.2)	12.8 (11.1, 14.4)	12.8 (11.3, 14.3)
Male sex, n (%)	148 (46)	49 (56)	197 (48)
Race/ethnicity, n (%)
White non-Hispanic	297 (93)	75 (86)	372 (91)
Black	14 (4)	11 (13)	25 (6)
Hispanic	3 (1)	0 (0)	3 (1)
Otherb	7 (2)	1 (1)	8 (2)
History of Medical Assistance (Medicaid), n (%)	92 (29)	49 (56)	141 (35)
Body mass index, kg/m2, median (IQR)
Male youth	20.0 (17.5, 23.4)	20.5 (18.0, 25.1)	20.0 (17.6, 23.7)
Female youth	19.7 (17.6, 23.5)	21.5 (20.0, 27.5)	20.2 (17.8, 23.9)
Body mass index percentilesc
<85th percentile	222 (69)	46 (53)	268 (66)
85th to 95th percentile	49 (15)	17 (20)	66 (16)
≥95th percentile	50 (16)	24 (28)	74 (18)
Waist circumference, inches, median (IQR)
Male youth	28.8 (26.6, 32.4)	29.7 (25.4, 34.0)	29.1 (26.2, 33.0)
Female youth	27.6 (25.6, 31.2)	29.9 (27.4, 34.7)	28.1 (26.0, 32.5)
Waist circumference percentilesd
<85th percentile	237 (74)	48 (55)	285 (70)
85th to 95th percentile	52 (16)	22 (25)	74 (18)
≥95th percentile	32 (10)	17 (20)	49 (12)
Percent body fat, median (IQR)
Male youth	15.1 (10.9, 24.5)	17.6 (12.5, 28.7)	15.6 (11.5, 25.9)
Female youth	24.1 (17.5, 33.0)	29.2 (23.5, 38.4)	25.2 (18.3, 33.7)
Percent body fat percentilese
<85th percentile	259 (81)	60 (71)	319 (79)
85th to 95th percentile	42 (13)	12 (14)	54 (13)
≥95th percentile	18 (6)	13 (15)	31 (8)
Self-reported fruit and vegetable servings per day, mean (standard deviation)	3.7 (1.9)	3.6 (1.9)	3.7 (1.9)
Parents	322 (100)	87 (100)	409 (100)
Age, years, median (IQR)	42 (38, 47)	40 (35, 46)	42 (37, 47)
Male sex, n (%)	47 (15)	10 (11)	58 (14)
Family income, n (%)
$0 – $9,999	20 (6)	8 (9)	28 (7)
$10,000 – $24,999	36 (11)	24 (28)	60 (15)
$25,000 – $49,999	54 (17)	31 (36)	85 (21)
$50,000 – $74,999	64 (20)	12 (14)	76 (19)
$75,000 – $99,999	55 (17)	3 (3)	58 (14)
$100,000 – $149,000	44 (14)	1 (1)	45 (11)
$150,000 or more	37 (12)	0 (0)	37 (9)
Missing	11 (3)	8 (9)	19 (5)
Highest level of education,f n (%)
Less than high school	14 (4)	8 (9)	22 (5)
High school or GED	109 (34)	46 (53)	155 (38)
Associates or trade degree	36 (11)	10 (12)	46 (11)
Bachelor’s degree	104 (32)	20 (23)	124 (30)
Graduate degree	58 (18)	2 (2)	60 (15)
Body mass index, kg/m2, median (IQR)	27.0 (22.9, 32.4)	30.4 (25.2, 37.5)	27.8 (23.3, 33.4)
Healthy home food availability score,g median (IQR)	15 (12, 18)	12 (10, 14)	14 (11, 17)
Obesogenic home food availability score,h median (IQR)	16 (13, 20)	16 (13, 19)	16 (13, 20)

^aFood security status assessed using the short form of the USDA Food Security Scale and dichotomized as high versus low/very low food security.

^b‘Other’ category included American Indian, Alaskan Native, Asian, Pacific Islander, and an option of ‘Other.’

^cBody mass index percentiles based on CDC growth charts.⁴⁴

^dWaist circumference and percent body fat percentiles based on NHANES data subjected to the LMS method.^46,48

^eFour biologically implausible percent body fat observations were deleted.

^fParent education missing one response.

^gHealthy home food availability assessed based on the presence (0-never to 4-always) of six food items (raw fruit, raw vegetables, baked chips, unsweetened cereal, 1% or fat free milk, whole wheat bread). The resulting score had a possible range of 0–24.

^hObesogenic home food availability assessed based on the presence (0-never to 4-always) of eight food items (chocolate candy, other candy, cakes/brownies/muffins/cookies, sweetened cereal, regular chips, juice drinks, sugar-sweetened soda, sports drinks). The resulting score had a possible range of 0–32.

Household food insecurity was not associated with youth’s mean number of daily servings of fruits and vegetables (Table 2). In unadjusted models, youth from food insecure households had significantly higher mean BMIz, WCz, and PBFz than youth from food secure households; inferences were unchanged in adjusted models except for BMIz, which crossed the inferential boundary, from P = 0.018 to 0.051 (Table 2). In sensitivity analyses, adding parent BMI to models attenuated associations between food security status and anthropometric outcomes (β [95% CI], P) (BMIz: 0.20 [−0.09, 0.49], 0.172; WCz: 0.19 [−0.03, 0.43], 0.095; PBFz: 0.34 [0.04, 0.65], 0.027). Compared to food secure households, food insecure households had significantly lower mean healthy HFA scores after adjustment for covariates (Table 2); dropping the 19 individuals with missing data for family income resulted in an attenuation of 23% (from −1.23 [−2.29, −0.18] to −0.95 [−2.04, 0.13]). There was no evidence of a difference in mean obesogenic HFA scores.

Table 2.

Estimated associations between household food security status^a and anthropometric outcomes, fruit and vegetable consumption, and home food availability among 408 Pennsylvania youth ages 10–15 years

	Food insecure vs. secure households
Study outcomes	Beta coefficient (95% confidence interval)	P-value
Daily fruit and vegetable servings
Unadjusted	−0.06 (−0.52, 0.40)	0.804
Adjustedb	−0.06 (−0.56, 0.43)	0.805
Body mass index z-score
Unadjusted	0.33 (0.06, 0.61)	0.018
Adjustedb	0.30 (−0.00, 0.60)	0.051
Waist circumference z-score
Unadjusted	0.26 (0.04, 0.48)	0.020
Adjustedb	0.27 (0.03, 0.50)	0.027
Percent body fat z-scorec
Unadjusted	0.45 (0.15, 0.75)	0.003
Adjustedb	0.43 (0.12, 0.75)	0.006
Healthy home food availability
Unadjusted	−2.34 (−3.36, −1.33)	< 0.001
Adjustedb	−1.23 (−2.29, −0.18)	0.022
Obesogenic home food availability
Unadjusted	−0.54 (−1.75, 0.67)	0.381
Adjustedb	−1.02 (−2.32, 0.28)	0.122

^aFood security status assessed using the 6-item short form of the USDA Food Security Scale and dichotomized as high versus low/very low food security.

^bAdjusted models controlled for youth age, youth sex, youth race/ethnicity, parent age, and family income.

^cPercent body fat assessed among 404 youth after removing four biologically implausible values.

Youth from households with higher healthy HFA scores reported significantly greater mean daily fruit and vegetable servings in unadjusted and adjusted models (Table 3). Youth living in households that scored in the 75^th percentile of healthy HFA had an additional 0.5 serving of fruits and vegetables, on average, as compared to youth living in households that scored in the 25^th percentile. There were no associations between healthy HFA scores and anthropometric outcomes in unadjusted or adjusted models (Table 3). Youth from households with higher obesogenic HFA scores reported fewer mean daily fruit and vegetable servings (Table 3); youth living in households that scored in the 75^th percentile of obesogenic HFA had an average 0.4 servings of fruits and vegetables less than youth living in households that scored in the 25^th percentile. After adjustment for covariates, higher obesogenic HFA scores were associated with lower mean WCz (Table 3): youth living in households that scored in the 75^th percentile of obesogenic HFA had an average of 0.13 standard deviation units lower WCz than did youth living in households in the 25^th percentile. Further adjustment for parent BMI did not change inferences. There were no associations between obesogenic HFA scores and BMIz or PBFz in unadjusted or adjusted models.

Table 3.

Estimated associations between healthy and obesogenic home food availability and anthropometric outcomes and fruit and vegetable consumption among 408 Pennsylvania youth ages 10–15 years

	Healthy home food availabilitya		Obesogenic home food availabilityb
Study outcomes	Beta coefficient (95% confidence interval)	P-value	Beta coefficient (95% confidence interval)	P-value
Daily fruit and vegetable servings
Unadjusted	0.08 (0.04, 0.12)	< 0.001	−0.06 (−0.10, −0.02)	0.001
Adjustedc	0.08 (0.03, 0.12)	0.001	−0.06 (−0.09, −0.02)	0.003
Body mass index z-score
Unadjusted	−0.02 (−0.05, 0.01)	0.134	−0.01 (−0.03, 0.01)	0.305
Adjustedc	−0.01 (−0.03, 0.02)	0.659	−0.02 (−0.04, 0.01)	0.143
Waist circumference z-score
Unadjusted	−0.01 (−0.04, 0.01)	0.155	−0.02 (−0.03, 0.00)	0.089
Adjustedc	−0.01 (−0.03, 0.01)	0.495	−0.02 (−0.04, −0.00)	0.042
Percent body fat z-scored
Unadjusted	−0.02 (−0.05, 0.01)	0.115	−0.01 (−0.03, 0.01)	0.366
Adjustedc	−0.01 (−0.03, 0.02)	0.717	−0.02 (−0.04, 0.00)	0.112

^aMultivariable linear regression models controlled for youth age, youth sex, youth race/ethnicity, parent age, and family income, as described in Methods.

^bHealthy home food availability assessed as a continuous variable; based on the presence (0-never to 4-always) of six food items (raw fruit, raw vegetables, baked chips, unsweetened cereal, 1% or fat free milk, whole wheat bread). The resulting score had a range from 0–24.

^cObesogenic home food availability assessed as a continuous variable; based on the presence (0-never to 4-always) of eight food items (chocolate candy, other candy, cakes/brownies/muffins/cookies, sweetened cereal, regular chips, juice drinks, sugar-sweetened soda, sports drinks,). The resulting score had a possible range from 0–32.

^dPercent body fat assessed among 404 youth after removing four biologically implausible values.

There was no evidence that healthy or obesogenic HFA modified (attenuated, increased, or changed) the food security/youth anthropometrics associations. Given a lack of associations between food security status and youth fruit and vegetable consumption, this model was not evaluated for effect modification by HFA.

DISCUSSION

This study examined associations of household food security, healthy and obesogenic HFA, fruit and vegetable consumption, and anthropometric outcomes among youth ages 10–15 years living in a geographically and socioeconomically diverse region of Pennsylvania. Consistent with past observations that obesity and food insecurity frequently coexist,^4,5 mean BMIs among youth and parents were higher in food insecure (versus food secure) households. Although findings from past research have not been consistent, growing evidence suggests a link between food insecurity and youth obesity.⁹ The current study found that food security status was related to youth anthropometric outcomes after controlling for common predictors of youth diet quality and weight status. Youth from food insecure households had, on average, 0.30 (P = 0.051), 0.27 (P = 0.027) and 0.43 (P = 0.006) greater standard deviations of BMI, waist circumference and PBF, respectively, compared to their food secure counterparts. Controlling for parent BMI weakened associations between food security status and youth anthropometric outcomes, potentially due to over-control for underlying factors that may link household food insecurity and parent and youth obesity such as diet, food preferences, and cycles of food deprivation and overconsumption.¹⁰ Although parent and child weight status are related in part due to genetics and perinatal factors,^53–56 parent BMI is not likely to be a cause of food insecurity, thus it seems unlikely that parent BMI was a confounder of the relation between food security status and youth adiposity.

This study hypothesized that household food insecurity may be associated with increased risk of obesity among youth due to differences in HFA, which prior research has shown to be associated with youth dietary intake.²⁷ Parents from food insecure households use a variety of shopping strategies and informal food sources to stretch scarce resources and provide food for their families, some of which may negatively impact dietary quality.⁵⁷ Nutrient-dense foods such as fresh fruits and vegetables and lean meats are generally more expensive than foods with refined grains, added sugars, and fats,⁵⁸ and past studies have reported that food insecure households face difficulty accessing affordable healthy foods.^57,59 Studies have also reported lower availability of fruits and vegetables among individuals from households with food insecurity or lower socioeconomic resources as compared to food secure households or those with higher socioeconomic status.^33,59–61 Similarly, this study found that food insecure households had significantly lower mean healthy HFA scores than did food secure households. Excluding 19 individuals with missing data for family income attenuated this association, which may be due to altering the study sample or could indicate the association is subject to residual confounding by socioeconomic status. There was no evidence that obesogenic HFA differed by household food security status.

Although availability does not guarantee consumption, it may be more difficult for youth to meet recommendations for fruit and vegetable intake if availability at home is limited, and consumption of highly palatable, energy-dense food may be more likely if it is easily available and accessible. Findings showed that youth from households with lower healthy HFA scores or higher obesogenic HFA scores reported fewer mean daily fruit and vegetable servings. These findings are consistent with those of Couch et al., who used an identical scale to measure obesogenic HFA and a comparable scale to measure healthy HFA in association with fruit and vegetable intake using food recalls among children aged 6 to 11 years.²⁶ There was not a direct association between food security status and youth fruit and vegetable consumption, nor evidence that HFA modified the association between food security status and youth anthropometric measures, and thus no evidence that HFA and its relation with youth fruit and vegetable consumption explained the association between food security status and higher youth anthropometric outcomes. Study results could be confounded by unmeasured factors such as participation in US nutrition assistance programs, which are intended to alleviate food insecurity;¹⁸ however, lower dietary quality has been observed among participants in such programs.²⁰ Alternatively, the HFA measures may not have sufficiently captured the availability of healthy and obesogenic foods in the home. For example, the measures did not evaluate the volume or varietal quantity of each food type available, nor temporal variations in HFA.⁶² Further, the HFA measures did not assess food accessibility within the home; an important consideration since parents can restrict or promote access to food.²⁵ Without validation of the HFA scales, it is unclear whether they represented household eating patterns. Higher anthropometric measures observed among youth in food insecure households may also relate to differential patterns in foods eaten outside the home or other behavioral factors (e.g., physical activity).

Of the few studies that have evaluated HFA in relation to youth anthropometric measures, none have observed associations,^25,30,34 including the aforementioned study by Couch et al. that utilized HFA measures comparable to the scales used in the current study.²⁶ Most evaluated younger children (elementary school age) and only evaluated BMI, with the exception of Bauer and colleagues,³⁴ who assessed healthy and obesogenic HFA in relation to both BMI and PBF (as measured by dual-energy X-ray absorptiometry) among adolescents. However, their study is not directly comparable to the current study as it focused on a markedly different sample, consisting of girls with diverse racial/ethnic composition and in an intervention context. The current study demonstrated an association between higher obesogenic HFA scores and lower mean WCz, contradicting the hypothesis that obesogenic HFA would be associated with higher adiposity given its association with higher intake of these foods by youth.^29,30 Youth from households with obesogenic HFA scores in the 75^th percentile had, on average, 0.13 greater standard deviation units of waist circumference compared to those with scores in the 25^th percentile. As highlighted in previous cross-sectional studies of the home food environment, the temporal direction of these associations remains unclear.^25,36 For example, findings may reflect a pattern in which parents limit availability of obesogenic foods in response to concerns about their children’s weight as an effort to improve their children’s health.⁶³ Determining whether changes to the home food environment might ultimately result in concomitant changes in youth’s weight requires longitudinal studies.

The study findings are subject to limitations. Although the sample is reflective of the region’s population, the lack of racial and ethnic diversity limits the study’s generalizability. The measure of household food security was not specific to the youth residing in the household, as parents may take measures to ensure children have adequate food at home even when experiencing food insecurity themselves.^6,17 The measure also failed to capture differences in whether and how individual children within a single household experience food insecurity (i.e., personal food insecurity). This potential for misclassification bias could have weakened associations between food security status and youth behavioral and anthropometric outcomes. For example, prior research observed personal food insecurity was associated with obesity among children aged 6 to 11 years, whereas aggregate food-insecurity measures showed no association, in part because the aggregate measure accounted for all children in the household as compared to an individual child.⁶⁴ The HFA scales may not have captured all food types important for characterizing a healthy or obesogenic home food environment. Since they were not formally validated, it is unknown as to whether the HFA scales adequately represented the intended constructs; however, the scales aligned with evidence-based dietary recommendations and the obesogenic HFA scale and a four-item version of the healthy HFA demonstrated good predictive validity in a prior study.²⁶ Additionally, although the measure of fruit and vegetable consumption was shown to correlate with 3-day food record data, it has low sensitivity to food variety (an indicator of dietary quality).⁴¹ Assessment of a typical day rather than actual recall of multiple days could also have contributed to under- or over-reporting. BIA methodology did not standardize on factors such as hydration and body position, which may have affected accuracy of PBF measurements,⁴⁵ though this was unlikely to have occurred differentially by exposure status (i.e., food security status or HFA). Finally, as with all cross-sectional research, the observational study design limited our ability to determine the temporal sequencing of associations.

CONCLUSIONS

In one of the first studies to examine youth anthropometric measures in the context of household food security status, HFA, and youth eating behavior, youth living in food insecure households had higher mean z-scores for body mass index, waist circumference, and percent body fat compared to youth in food secure households. Similar to prior research, food security status and mean healthy HFA scores were associated, as were healthy and obesogenic HFA scores and youth’s mean daily fruit and vegetable servings. However, overall findings from this study did not provide evidence to support the hypothesis that differences in the availability of healthy and obesogenic foods in the home and their influence on youth dietary patterns explained the association of food insecurity and youth anthropometrics. Future longitudinal studies evaluating children’s personal food security status alongside rigorous assessment of diet and home food availability may better elucidate relationships between parenting practices, child behaviors, the home food environment, and anthropometrics.

RESEARCH SNAPSHOT.

Research Question:

Is household food insecurity and home availability of healthy and obesogenic foods related to youth fruit and vegetable consumption and anthropometric outcomes?

Key Findings:

In this cross-sectional study of 408 youth ages 10–15 years, youth in food insecure (versus secure) households had higher mean body mass index, waist circumference, and percent body fat z-scores. High healthy and low obesogenic home food availability were associated with greater youth mean daily fruit and vegetable servings and food insecure households had lower mean healthy home food availability scores. There was no evidence home food availability modified associations between food insecurity and youth anthropometric outcomes.