Correlates of Energy Intake and Body Mass Index among Homeless Children in Minnesota

Abstract

Background:

This study evaluated environmental, personal, and behavioral correlates of BMI-for-age percentiles, dietary intake (kilocalories, carbohydrates, protein, fat, and Food Guide Pyramid food groups), and physical activity variables among homeless children.

Methods:

A 74-item survey, using social cognitive theory as the theoretical framework, height, weight, and one 24-hour recall were collected from homeless children aged 9–13 (n=159) at two shelters in Minneapolis, MN. Principal component analysis was performed on the subsections of the survey. Independent t-tests, Fisher exact tests, and chi-squared statistics evaluated sociodemographic and BMI percentile variables. Nonparametric tests evaluated dietary data. Stepwise regression models evaluated correlates of BMI percentiles, physical activity, and dietary intake variables.

Results:

Approximately 45% were overweight or obese (≥85^th percentile). Dietary data represented intake on a given day, with children consuming a median 1.2 servings from the fruits and vegetables food group, 17.3 servings from the fats and sweets food group (one serving=grams in 1 Tbsp. fat/1 tsp. sugar), and the percent of calories from fat varying significantly between shelter 1 (S1) versus shelter 2 (S2) boys (37.1% vs. 31.7%, p<0.001). Factors identified from survey items and sociodemographic variables accounted for between 6% and 14% of the variance in energy intake and other dietary and physical activity variables (p range, 0.008 to <0.001).

Conclusions:

Parental role modeling of eating behaviors and getting enough food were associated with less favorable food choices among homeless children. Policy interventions and program initiatives in the homeless environment could promote healthier food choices among children.

Introduction

Over 636,000 people experience homelessness on any given night in the United States, with 37% consisting of families with children.¹ Homelessness can dramatically impact children's dietary intake, exercise habits, and health.^2,3 Tarasuk et al.⁴ found that homeless adolescents had inadequate intakes of folate, vitamins A, C, and B₁₂, iron, magnesium, and zinc. Among homeless preschool children, Taylor and Koblinsky⁵ found inadequate intakes of calcium and zinc and excessive sugar intake. Furthermore, other studies have shown overweight and obesity to be relatively common among children experiencing homelessness.^3,4,6–10 These adverse health effects are most likely related to the transient state of not having permanent housing, poverty, inadequate cooking and food storage facilities, limited access to transportation, and poor nutritional quality of shelter dining facility meals or restrictive shelter rules such as prohibiting perishable food items, encouraging snack food items, and limiting access to the shelter dining room.^11–13 Two previous qualitative studies among homeless children have suggested the negative impact of the shelter environment and surrounding community infrastructure on homeless children's BMI, food access, and food choice.^3,9

To our knowledge, little quantitative research has evaluated dietary intake among homeless school-aged children and their perceptions of how homelessness influences dietary intake, physical activity, and BMI status. Because of recent concerns about rising rates of obesity among children in the United States and the disparities in food accessibility in low-income communities,¹⁴ an understanding of the effects of homelessness on children's nutritional status can help health professionals in developing strategies to promote optimal health in this vulnerable population. Thus, using social cognitive theory (SCT) as the theoretical framework, this study addressed the following research question: How are environmental, personal, and behavioral factors correlated with homeless children's nutritional status (defined as energy, macronutrient and Food Guide Pyramid (FGP) food group intake and BMI-for-age percentiles) and level of physical activity? Environmental factors in this study referred to children's perceived influence of the shelter, media, family, and community on their food choices and health. Behavioral factors included children's perceived knowledge and skills about healthy eating and exercise behaviors and the anticipated health effects of engaging in these behaviors. Personal factors referred to the importance children placed on engaging in healthy behaviors, their confidence in making healthy food choices and exercising, their perceived control over performing these behaviors, and their emotional coping responses to shelter living and food choices. Because of previous research suggesting a negative impact of the shelter environment and surrounding neighborhood on food choice and food access, we hypothesized in our study that environmental factors would be associated with more of the nutritional status and physical activity variables compared to personal and behavioral factors and that the association would be in a manner less favorable for optimal health.

Methods

Subjects and Study Design

Cross-sectional survey data were collected from children aged 9–13 years (n=159) at two family-based homeless shelters in Minneapolis, MN [designated S1 (n=75) and S2 (n=84)] during 2005 and 2006. S1 and S2 serve approximately 400 and 500 individuals per night, respectively. Children from S1 had access to an in-house dining facility and an outdoor playground (∼10′×15′), but there was limited storage space and no cooking facilities in living quarters. Shelter rules permitted only nonperishable goods, defined as chips, candy, and soda pop, in resident rooms, and residents had no access to shelter-provided food after 5:30 p.m.^3,11,12 Food establishments in the surrounding neighborhood (five-block radius) included two convenience stores, 20 restaurants or restaurant/bars, three fast-food restaurants, three coffeehouses, and one soup kitchen, with the convenience store selling foods at 60–99% above available market basket prices of the Thrifty Food Plan [used to calculate Supplemental Nutrition Assistance Program (SNAP) monthly allotments for low-income households].¹² Children from S2 resided in either a small or large room, depending on family size. Small rooms had a small refrigerator/freezer, two stovetop burners, and a microwave, whereas larger rooms included a regular-sized refrigerator/freezer, regular-sized stove/oven, and a microwave. All children had access to a soup kitchen across the street, an on-site exercise facility, and an outdoor playground (∼48′×32′). Food establishments in a five-block radius of S2 included one convenience store, seven restaurants or restaurants/bars, and three coffeehouses, with convenience store food items at 10–294% above available market basket prices.¹²

Children were recruited via posting flyers, placing flyers in resident mailboxes, and by shelter staff. Only one child per family was permitted to participate in the study. Children signed an assent form before survey administration, and mothers provided demographic information and signed a consent form. Survey response rates of children were not calculated because predetermined days and times in which we would be conducting survey data collection at the shelter were often advertised through flyers and shelter staff, and thus, a parent and their child would just show up to participate. The researchers kept a running list of former participants to ensure only one child per family participated in this study. Mothers also completed the USDA's six-item (short form) 12-month Food Security Scale.¹⁵ Responses of “often,” “sometimes,” “yes,” “almost every month,” and “some months but not every month” were coded as affirmative responses.¹⁵ Affirmative responses were summed to give an overall score, with 0–1 being marginal or high food security, 2–4 being low food security, and 5–6 being very low food security.¹⁵ The University of Minnesota's Institutional Review Board approved this study.

Procedure

A 74-item survey was developed using SCT as the theoretical framework, which suggests the existence of a dynamic triadic interrelationship between environmental, behavioral, and personal factors, and by using formative information from preliminary interviews with homeless children 9–13 years of age.^16,17 Specifically, the preliminary interviews (n=56) with children³ provided formative information about how environmental, behavioral, and personal factors were related to dietary intake, physical activity, and BMI. Children indicated that their dietary choices and physical activity patterns were related to factors such as restrictive shelter food policies, inadequate cooking and storage facilities, limited access to the in-house shelter dining facility, and neighboring food stores.³

The survey was evaluated for face validity and pilot tested (n=24). The survey used a five-point Likert scale, with response options as “Always true,” “Kind-of true,” “Don't know,” “Kind-of not true,”, and “Never true,” coded from +2 to −2. Cronbach alpha scores were 0.66 for environment, 0.66 for behavior, and 0.75 for personal, indicating good internal consistency.¹⁸

Children's height and weight were measured with heavy outer clothing and shoes removed. Sex-specific BMI-for-age percentiles were calculated using Epi Info^™, version 3.3.2 (CDC 2005) and children classified as normal weight (5^th to <85^th percentile), overweight (≥85^th percentile to <95^th percentile), and obese (≥95^th percentile). No children were classified as underweight (≤5^th percentile).

One 24-hour recall was collected from each child, with the use of three-dimensional food models and standard measuring utensils.¹⁹ Mothers were present and assisted children as needed. Research has shown that a consensus technique between parent and child in recalling dietary intake more accurately reflects children's actual intake than solely obtaining parental recalls.²⁰ Furthermore, although a single 24-hour recall is less accurate at the individual level, this method has been shown to be effective in assessing group dietary intake.^19,21 Dietary intake data [defined as kcals, carbohydrate (g), protein (g), fat (g), and FGP food groups (number of servings)] were analyzed using ESHA's The Food Processor SQL, v. 9.8. Exercise was defined on the survey as “things that you do for 30 minutes or longer in which your heart beats faster, you breathe hard, and you sweat. Like fast walking, running, dancing, swimming, playing sports or playing outside.” Based on this definition, children responded to the question, “How many times per week do you exercise?” and responses were “None” to “7 or more” times per week.

Statistical Methods

Independent t-tests, Fisher exact tests, and chi-squared statistics were used to compare sociodemographic variables such as age, gender, and race or ethnicity across shelter locations (S1 vs. S2). BMI percentiles and the physical activity variable were compared within (S1 girls vs. S1 boys and S2 girls vs. S2 boys) and across (S1 girls vs. S2 girls and S1 boys vs. S2 boys) shelters using chi-squared statistics. Nonparametric analyses (Mann–Whitney) were used to examine dietary intake data within shelters and across shelters because of the nonnormal distribution of the dietary intake variables. One child reported consuming >5000 kilocalories (an outlier) and was excluded from analyses.

The survey questions included all three constructs (environment, personal, behavior) of the SCT. A principal component analysis (PCA) was performed for each set of construct questions. Factor loadings >|0.30| were identified under each factor, with each item loading only on a single factor based on its highest loading. Factors were summed and averaged to create a mean factor score. Using mean factor scores and selected sociodemographic variables (gender, shelter residency, and number of children in the household), stepwise regression analyses were performed using BMI percentiles, physical activity, and log-transformed dietary intake data as dependent variables. To account for multiple comparisons bias, a p value <0.001 was considered statistically significant based on a pseudo Bonferroni correction. Descriptive statistics were examined using Statistical Package for the Social Sciences (IBM SPSS, v. 20, 2011) and multivariate regression and factor analyses were performed in SAS 9.2 (SAS Institute Inc. Cary, NC, USA, 2007).

Results

Children had a mean age of 11±1.4 years (Table 1) and approximately 45% of children were overweight or obese (Table 2). No differences in sociodemographic and dietary variables were noted among BMI percentile categories (data not shown). On a given day of intake, median energy intake was 1718 kilocalories (Table 2). Children consumed a median of 1.2 servings from the fruit and vegetable food groups and 17.3 servings from the fats, oils, and sweets food group. Only the percent of calories coming from fat was significantly higher among boys at S1 compared to those residing at S2. Most children (63.9%) reported exercising 4 or more days a week.

Table 1.

Sociodemographic Characteristics of Homeless Children (9–13 Years Old) by Shelter Location

	Total (n=159)	S1 (n=75)	S2 (n=84)	p
Age (years), mean±SD ^a	11.0±1.4	10.9±1.5	11.1±1.3	0.50
Adults in household, mean±SD ^a	1.4±0.6	1.4±0.7	1.3±0.6	0.17
Children in household, mean±SD ^a	3.6±1.6	3.2±1.5	3.9±1.7	0.004
Gender, no. (%) ^b
Boys	74 (46.5)	38 (50.7)	36 (42.9)	0.32
Girls	85 (53.5)	37 (49.3)	48 (57.1)
Race/ethnicity, no. (%)^c,d
Caucasian	9 (6.5)	5 (7.4)	4 (5.6)	—
African American	98 (70.5)	50 (73.5)	48 (67.6)	1.00
American Indian	11 (7.9)	3 (4.4)	8 (11.3)	0.36
Hispanic	4 (2.9)	1 (1.5)	3 (4.2)	0.56
Asian	2 (1.4)	1 (1.5)	1 (1.4)	1.00
Other	15 (10.8)	8 (11.8)	7 (9.9)	1.00
Currently using food stamps (household), no. (%) ^b
Yes	122 (76.7)	57 (76.0)	65 (77.4)	0.84
No	37 (23.3)	18 (24.0)	19 (22.6)
Household food security status (past year), no. (%) ^b
Marginal or high food security	43 (27.0)	18 (24.0)	25 (29.8)	0.55
Low food security	49 (30.8)	26 (34.7)	23 (27.4)
Very low food security	67 (42.1)	31 (41.3)	36 (42.9)
Length of homelessness, no. (%) ^b,e
<7 days	25 (19.1)	13 (21.0)	12 (17.4)	0.08
8–30 days	53 (40.5)	26 (41.9)	27 (39.1)
31–180 days	41 (31.3)	14 (22.6)	27 (39.1)
>180 days	12 (9.2)	9 (14.5)	3 (4.3)

Independent t-tests used to compare S1 vs. S2.

Chi-squared analyses to compare S1 vs. S2.

Missing data [n=20 (S1=7 and S2=13)].

Fisher exact test used to compare S1 vs. S2.

Missing data [n=28 (S1=13 and S2=15)].

S1, shelter 1; S2, shelter 2; SD, standard deviation.

Table 2.

BMI Percentiles, Dietary Intake, and Physical Activity Level of Homeless Children (9–13 Years Old) with Comparisons between Shelter Location and Gender

		S1			S2
	Total (n=159)	Girls (n=37)	Boys (n=38)	p ^*	Girls (n=48)	Boys (n=36)	p ^**
BMI percentiles,^a,b no. (%)
<85th percentile (normal weight)	88 (55.3)	16 (43.3)	22 (57.9)	0.44	30 (62.5)	20 (55.6)	0.81
≥85th to <95th percentile (overweight)	22 (13.8)	6 (16.2)	5 (13.2)		6 (12.5)	5 (13.9)
≥95th percentile (obese)	49 (30.8)	15 (40.5)	11 (28.9)		12 (25.0)	11 (30.6)
Selected nutrients,^c median (25^th–75th percentiles)
Total energy	1718.2 (1319.7–2321.8)	1719.3 (1362.9–2147.1)	1700.8 (1294.1–2428.3)	0.95	1628.1 (1318.9–2264.0)	1873.0 (1288.9–2341.8)	0.47
Carbohydrates, g	213.2 (164.3–285.0)	203.9 (161.0–255.4)	225.2 (170.2–289.6)	0.51	195.4 (157.6–285.3)	241.4 (163.2–309.1)	0.33
Total fat, g	66.5 (48.0–89.6)	76.6 (55.2–96.2)	68.9 (49.6–92.7)	0.61	59.8 (41.1–72.2)	66.7 (37.9–91.3)	0.63
% calories from fat	34.6 (29.6–39.6)	38.1 (32.5–44.9)	37.1 (33.1–40.6)^†	0.37	32.0 (26.4–38.5)	31.7 (24.3–35.2)	0.36
Protein, g	62.9 (44.8–86.7)	61.6 (39.9–87.0)	60.8 (38.6–83.9)	0.81	61.7 (48.4–85.6)	71.7 (43.7–96.5)	0.66
% calories from protein	14.5 (12.1–17.6)	13.8 (10.8–16.7)	13.7 (10.6–17.7)	0.85	15.0 (12.8–19.8)	15.3 (13.1–18.1)	0.95
Food group servings,^c median (25–75th percentiles)
Bread, cereal, rice & pasta^d	6.2 (4.0–8.5)	6.2 (3.9–8.4)	5.7 (4.3–7.4)	0.50	6.3 (3.9–10.0)	6.5 (4.4–9.2)	0.95
Fruit^e	0 (0–1.2)	0.6 (0–1.1)	0.2 (0–1.3)	0.40	0 (0–1.2)	0 (0–1.3)	0.88
Vegetable^f	1.2 (0.5–2.6)	1.2 (0.1–2.7)	1.4 (0.3–3.2)	0.83	1.1 (0.6–2.2)	1.4 (0.4–2.7)	0.81
Meat, poultry, fish, nuts, and eggs^g	1.7 (0.8–2.9)	1.6 (0.7–3.0)	1.6 (0.8–2.3)	0.75	1.8 (0.8–3.1)	1.7 (0.9–3.5)	0.98
Milk, yogurt, and cheese^h	1.5 (0.5–2.5)	1.8 (0.4–2.6)	1.9 (1.0–2.7)	0.67	1.1 (0.4–2.3)	1.5 (0.5–2.5)	0.50
Fats, oils, and sweetsⁱ	17.3 (10.7–27.0)	19.1 (11.7–26.0)	17.6 (12.1–30.7)	0.76	14.7 (9.6–25.2)	19.4 (10.6–28.4)	0.29
Physical activity (days),^b,j no. (%)
0	12 (7.7)	2 (5.7)	4 (10.8)	0.05	5 (10.6)	1 (2.8)	0.10
1–3	44 (28.4)	16 (45.7)	6 (16.2)		14 (29.8)	8 (22.2)
4–5	37 (23.9)	7 (20.0)	9 (24.3)		14 (29.8)	7 (19.4)
6–7	62 (40.0)	10 (28.6)	18 (48.6)		14 (29.8)	20 (55.6)

No children classified as <5th BMI percentile.

Chi-squared analysis to compare within and between shelter differences.

Mann–Whitney tests performed with energy intake >5000 kilocalories excluded [n=1 (S1 girl)].

One serving=1 slice of bread, 1 oz. ready-to-eat cereal, ½ cup cooked cereal, rice or pasta.

One serving=1 piece of medium-sized fruit or ½ cup fruit.

One serving=½ cup raw or cooked vegetables, 1 cup leafy vegetables (e.g., lettuce).

One serving=2.5 oz. meat, fish and poultry, 1 egg, ½ cup cooked beans.

One serving=1 cup (8 fl. oz.) of milk or yogurt and 1½ oz. natural cheese.

For fat, ESHA Food Processor SQL, vs. 9.8 defined one serving as “The number of grams in 1 Tbsp. of fat for butter, margarine, oils, and shortening. For meats, an additional fat serving is reported as a multiple of the fat standard for the specific meat. For milk products and mixed foods, an additional fat serving is reported as a multiple of 12.8 grams, weight of 1 Tbsp. of shortening.” For sugar, one serving was defined as “the number of grams in 1 tsp. of sugar (4 grams).”

Missing data [n=4 (S1=3 and S2=1)].

S1 girls vs. S1 boys.

S2 girls vs. S2 boys.

†

p<0.001 (S1 vs. S2 boys).

S1, shelter 1; S2, shelter 2.

The PCA for the three constructs—environment, behavior, and personal—resulted in the identification of 13 factors (Table 3). There were five factors for environment that explained 42% of the total variance. These factors were related to children's perceptions about the influence of the shelter environment on health; social influences on food choices and exercise behaviors; role-modeling food choice behaviors of peers, siblings, and teachers; the influence of media and the shelter environment on food choices; and role-modeling food choice behaviors of parents and perceptions of food adequacy. The three factors for behavior explained 33% of the variance. These factors included children's perceptions about consuming snack foods and its impact on health; the influence of nutrition education via TV, parents, and teachers on food choices; and the ability to make food choices and engage in exercise behaviors. The five factors for personal explained 52% of the variance. These factors were related to children's perceptions about consuming snack foods, making healthy food choices, engaging in healthy behaviors such as exercise and eating fruits and vegetables, consuming any type of food to alleviate hunger and the influence of snack food consumption on weight gain, and being willing to try new foods.

Table 3.

Principal Components Factor Loadings of Survey Items by the Social Cognitive Theory Theoretical Constructs, Based on Homeless Children's (9–13 Years Old) Responses

	Variance explained	Factor loadings (range)
Environment	0.42
1. Shelter and health^a		0.61–0.85
2. Social influence on food choice and exercise^b		0.30–0.86
3. Role modeling peers and teachers^c		0.67–0.72
4. Media, shelter, and food^d		0.35–0.74
5. Role modeling parents and food adequacy^e		0.32–0.70
Behavior	0.33
1. Snack foods and health^f		0.39–0.83
2. Nutrition education and food choice^g		0.45–0.71
3. Self-efficacy and healthy behaviors^h		0.44–0.81
Personal	0.52
1. Snack food consumersⁱ		0.43–0.76
2. Healthy food consumers^j		0.57–0.79
3. Health-promoting activities^k		0.36–0.77
4. Hunger, snack foods, and perceived weight gain^l		0.37–0.68
5. Willingness to try foods^m		0.71–0.75

Four survey items: Since my family moved to the shelter, I go to bed hungry; The food I eat at the shelter makes my stomach hurt; The food I eat at the shelter makes my teeth hurt; and The food I eat at the shelter makes my head hurt.

Five survey items: Since my family moved to the shelter, I exercise less; I eat chips, candy, and pop because people who I care about do; I exercise because people who I care about do; I eat fruits and vegetables because people who I care about eat them; and I eat foods that are good for my body because people who I care about do.

Three survey items: I eat the same foods that I see my brothers or sisters eat; I eat the same foods that I see my teacher eat; and I eat the same foods that I see my friends eat.

Five survey items: Since my family moved to the shelter, I eat more chips, candy, and pop; Since my family moved to the shelter, I eat foods that are good for my body; I get hungry when I watch TV; I eat snack foods like chips, candy, and pop when I watch TV; and When I see chips, candy, or pop on TV, I ask my mom or dad if we can buy it.

Four survey items: Living in the shelter makes it hard for me not to eat chips, candy, and pop; I eat the same foods that I see my mom or dad eat; When I see my mom and dad eat a lot, I do too; and I can eat the same foods here as I did before we moved to the shelter.

Six survey items: Living in the shelter makes it hard for not to eat chips, candy, and pop; My friends tell me to eat foods that are bad for my body; I like to eat chips, candy, and pop better than fruits and vegetables; If I am hungry at bed time, I eat chips, candy, and pop; Eating chips, candy, and pop is not good for my teeth; and Eating chips, candy, and pop is good for my body.

Four survey items: TV teaches me about foods that are good for my body; When I go to the store, I can pick out foods that are good for my body; My family talks to me about foods that are good for my body; and My teacher talks to me about foods that are good for my body.

Three survey items: I am able to try new food I have not eaten before; I get enough exercise; and When I go to the store, I get to pick out foods I like to eat.

Eight survey items: I like watching TV better than exercising; When there is nothing to do, I like to eat; I like to eat while I watch TV; I like to eat chips, candy, and pop because I see them on TV; I eat chips, candy, and pop to make me feel better; I like to buy chips, candy, and pop when I go to the store; When I am hungry, I eat snacks like chips, candy, and pop; and I eat chips, candy, and pop because we have them at home.

Three survey items: I am able to make my own food choices; I can eat fruits and vegetables instead of candy; I can drink juice or water instead of pop.

Four survey items: I exercise a lot because it makes me feel better; I eat fruits and vegetables because they are good for my body; I am able to choose foods to eat that are good for my body; and Eating fruits and vegetables is important for my body.

Four survey items: It is hard not to eat chips, candy, and pop each day; If I am hungry, I will eat foods that I do not like; When I am really hungry, I will eat anything; and If I eat chips, candy, and pop I gain weight.

Two survey items: If someone in my family likes a food, I will try it; and If my friend likes a food, I will try it.

Stepwise regression analysis resulted in an adjusted R² of 6% (p=0.008) for BMI percentiles with four variables in the model (Table 4). These variables included three environmental factors (shelter and health, role-modeling peers and teachers, and role-modeling parents and food adequacy) and the number of children in the household. For total energy intake, the adjusted R² was 8% (p=0.0002) with a single-variable model, which was the environmental factor related to role-modeling parental food choice behaviors and perceptions of food adequacy. Adjusted R² values for total carbohydrates, total protein, the meats, poultry, fish, nuts, and eggs food group, and fats, oils, and sweets group were 9%, 9%, 8%, and 8%, respectively, with a common environmental factor variable—role-modeling parents' food choice behaviors and food adequacy—included in each of the models.

Table 4.

Step-Wise Regression Model of Principal Component Factors (Independent Variables) and BMI Percentiles, Dietary Intake,^a and Physical Activity (Dependent Variables)

Dependent variable=BMI percentiles	Adj. R²=0.06, p=0.008
Factors	ß±SE	t-value	p ^†
Intercept	89.2±4.9	18.1	<0.0001
E1: Shelter and health	3.6±2.0	1.8	0.07
No. children in household	−1.9±1.2	−1.6	0.11
E3: Role modeling peers and teachers	3.6±1.9	1.9	0.06
E5: Role modeling parents and food adequacy	−3.8±2.3	−1.6	0.10

Dependent variable=log(Total energy intake)	Adj. R²=0.08, p=0.0002
Factors	ß±SE	t-value	p ^†
Intercept	3.2±0.02	165.1	<0.0001
E5: Role modeling parents and food adequacy	0.08±0.02	3.9	0.0002

Dependent variable=Physical activity	Adj. R²=0.14, p=< 0.0001
Factors	ß±SE	t-value	p ^†
Intercept	1.0±0.7	1.4	0.16
P3: Health-promoting activities	0.9±0.3	3.1	0.002
Child's gender	1.1±0.4	3.1	0.003
B3: Self-efficacy and healthy behaviors	0.5±0.2	2.5	0.01

Dependent variable=log(carbohydrates, g)	Adj. R²=0.09, p=0.0003
Factors	ß±SE	t-value	p ^†
Intercept	2.3±0.02	94.4	<0.0001
E5: Role modeling parents and food adequacy	0.07±0.02	3.6	0.0005
B3: Self-efficacy and healthy behaviors	0.03±0.02	1.5	0.13

Dependent variable=log(protein, g)	Adj. R²= 0.09, p=0.0003
Factors	ß±SE	t-value	P ^†
Intercept	1.6±0.05	32.3	<0.0001
E5: Role modeling parents and food adequacy	0.1±0.03	3.5	0.0005
Currently using food stamps	0.1±0.05	2.3	0.03

Dependent variable=log(Bread, cereal, rice, and pasta food group)	Adj. R²=0.06, p=0.0083
Factors	ß±SE	t-value	p ^†
Intercept	0.7±0.05	15.8	<0.0001
E5: Role modeling parents and food adequacy	0.05±0.02	2.2	0.03
B3: Self-efficacy and healthy behaviors	0.04±0.02	1.9	0.06
Currently using food stamps	0.07±0.05	1.5	0.14

Dependent variable=log(Fruit food group)	Adj. R²=0.07, p=0.0026
Factors	ß±SE	t-value	p ^†
Intercept	0.2±0.05	3.3	0.001
P4: Hunger, snack foods, and perceived weight gain	0.05±0.02	2.7	0.009
P2: Healthy food consumers	0.06±0.03	2.5	0.01
No. Children in household	−0.02±0.01	−2.0	0.05

Dependent variable=log(Vegetable food group)	Adj. R²=0.04, p=0.0148
Factors	ß±SE	t-value	p ^†
Intercept	0.4±0.02	17.0	<0.0001
E4: Media, shelter, and food	0.05±0.02	2.0	0.05
P5: Willingness to try foods	0.03±0.02	1.7	0.10

Dependent variable=log(Meat, poultry, fish, nuts, and eggs food group)	Adj. R²=0.08, p=0.0005
Factors	ß±SE	t-value	p ^†
Intercept	0.3±0.04	8.1	<0.0001
E5: Role modeling parents and food adequacy	0.07±0.02	3.4	0.0007
Currently using food stamps	0.09±0.04	2.2	0.03

Dependent variable=log(Milk, yogurt, and cheese food group)	Adj. R²=0.04, p=0.0262
Factors	ß±SE	t-value	p ^†
Intercept	0.3±0.03	9.9	<0.0001
E5: Role modeling parents and food adequacy	0.06±0.02	2.5	0.02
Current residency (shelter)	0.08±0.04	2.1	0.04
P1: Snack food consumers	−0.04±0.02	−1.7	0.09

Dependent variable=log(Fats, oils, and sweets food group)	Adj. R²=0.08, p=0.0002
Factors	ß±SE	t-value	p ^†
Intercept	1.2±0.03	41.2	<0.0001
E5: Role modeling parents and food adequacy	0.1±0.03	3.9	0.0002

Total fat (g) was not included in this table because no factors entered the regression model.

†

p<0.001 considered statistically significant.

SE, standard error; BMI, body mass Index; E, environment; B, behavior; P, personal.

Discussion

To our knowledge, this is the first quantitative study to assess homeless school-aged children's perceptions about the effect of homelessness on food choices and its association with nutritional status, including dietary intake, physical activity behaviors, and BMI status. Results from our study showed that obesity is relatively common among homeless children residing in shelters, that children's dietary intake comes less from fruits and vegetables and more from fats, oils, and sweets, and that one environmental factor—children's perceptions of role-modeling parental behaviors and having an adequate amount of food in the shelter environment—was related to children's food choices. Overall, these findings offer some support for our initial hypothesis in that no personal or behavioral factors were significantly associated with any of the dietary intake, exercise, or BMI variables. In contrast, the identified environmental factor was associated with many of the dietary variables, which suggests the environment may be an important aspect associated with homeless children's food choices. Although specific items related to shelter living were not included in the identified environmental factor, because of the homeless condition, it is probable that the reported parental behaviors were influenced by the shelter environment and surrounding neighborhood food environment.

Approximately 45% of children in our study were overweight or obese, which is higher than national statistics of 32.6% and 33.6% among 6–11 and 12–19 year olds, respectively.²² However, when comparing the predominant racial/ethnic group in our study (African American) to national statistics of overweight and obese children, our results are consistent with these estimates.²² Substantial studies have found obesity to be highly prevalent among low-income women^23–29; however, the relationship between low-income status and overweight or obese children has shown mixed results.^26,30–34 A high prevalence of obesity has been evidenced among homeless children,^3,4,6–9 with our study offering further support. The impact of these findings is significant, considering the adverse health consequences associated with childhood obesity, including increased risk of metabolic syndrome, type 2 diabetes, and cardiovascular disease, and the decreased ability of homeless individuals to obtain adequate health care to ameliorate the effects of chronic disease.^35–37 Furthermore, these obesity-related diseases are likely to pose substantial costs to the health care system, including Medicaid and hospitals that provide care to the uninsured.

All homeless children in our study, regardless of shelter residency, consumed inadequate amounts of fruits and vegetables and excess amounts from the fats, oils, and sweets food group, which are similar to national trends.³⁸ Children in our study may have chosen not to consume adequate amounts of fruits and vegetables, lacked access to them, or consumed excess amounts from the fats, oils, and sweets group because of food preferences or familiarity with these foods, selecting foods they had seen at home.³⁹ Previous research has consistently shown that children prefer to eat carbohydrate-rich foods, including sugar-containing sweets, fatty foods, and high-energy-dense foods, and that they dislike consuming vegetables.^40–44 Children in our study also identified eating similar foods in the shelter as they did in the home environment, which was associated with several dietary intake variables, including the fats, oils, and sweets food group. However, because of the homeless condition and shelter residency, it is also possible that the shelter dining facility (S1), soup kitchen (S2), or local neighborhood grocery stores (S1 and S2) contributed to these less healthy dietary behaviors because of the foods that are readily available in these settings.^11,12 For example, previous studies have found that shelter dining facilities fail to provide young children with adequate amounts of grains, fruits, vegetables, and milk, and offer them excess amounts of sweets and fats.^5,13 Furthermore, children's dietary behaviors in our study may have been affected by restrictive shelter rules permitting only nonperishable snack food items in rooms (S1), limited storage and cooking space (S1 and S2), and parents' encouragement to overeat to alleviate subsequent hunger. Results from our study showed that children's perceptions of role-modeling parental eating behaviors, including consuming large amounts of food, was related to several dietary intake variables, including foods from the fats, oils, and sweets food group.

The influence of parental role modeling on children's eating behaviors has been shown substantially in the literature.^45–51 More specifically, previous studies have shown that parents who model the consumption of fruit, vegetables, fruit juice, dairy foods, energy drinks, soft drinks, away-from-home convenience foods, and snack foods such as chips, candy, and bakery food items influence their children's intake of the same foods.^45–51 Our study similarly found an association between perceived role modeling of parental food choices and children's measured intake of carbohydrates (g) and foods found in the fats, oils, and sweets food group; however, unlike other studies, our study also found an association between children's perception of parental role modeling and total energy intake, protein (g), and foods from the meat, fish, poultry, eggs, and beans food groups. Our study also found that children's perceptions of specific role-modeling behaviors of parents, such as eating large amounts of food, were related to these dietary variables.

As a means to prevent hunger, parents may have wanted their children to consume foods that they felt would keep them fuller longer after mealtimes, thus more specifically role modeled consuming carbohydrate- and protein-rich foods. Wiig and Smith⁵² found that low-income women valued meat as the central part of a meal, with starchy foods as a side dish, because they perceived these as filling foods. To prevent subsequent hunger, parents of children in our study may have also role modeled overeating at mealtimes as a means to compensate for unavailability of food from the in-house dining facility (S1) after dinner hours (4:30–5:30 p.m.) or because of limited storage facilities for food (S1 and S2). Previous studies have indicated that homeless parents are willing to do whatever necessary to prevent their children from experiencing hunger and that when food is available, encourage their children to overeat.^3,11,12 Research has also suggested that adults experiencing food deprivation tend to overeat when tasty food is available.^53–55 Because of the positive correlation in our study between eating large amounts of food and energy intake, it is possible that this behavior could lead to weight gain if energy intake consistently exceeded energy expenditure over time. However, it is important to note that the adjusted R² in our regression models only ranged from 8–9%, suggesting other factors may have accounted for the variance in the dietary outcomes. It is possible that children in our study were unable to sufficiently discern the impact of the homeless environment, including shelter rules regarding allowable foods in the rooms or food storage issues, on food choices, or were not fully aware of how the food environment surrounding the shelter may have affected their food choices, when answering our survey. It is also possible that children's food choices were influenced by behaviors learned through the school environment, which was not a focus of this study.

Most children in our study (64%) reported exercising for 30 minutes at least four times per week, which suggests many were getting reasonable amounts of exercise, although still below Dietary Guidelines for Americans (DGA) guidelines.⁵⁶ Because of the homeless nature of these children, it is possible that many were walking with family members as a form of transportation during the week^12,57 or they exercised because of the availability of playground equipment (S1), an on-site exercise facility (S2), or physical education classes in school. However, because our measure of physical activity was based on self-report, it is also possible that children under- or overestimated the amount of activity done during the week.⁵⁸ Future research should be done to determine the actual amount of exercise, including duration and intensity, that children experience in the homeless environment, and how that impacts weight status and overall health.

Study limitations include that dietary intake was assessed using one 24-hour recall, which likely did not represent children's usual intake over time. It is possible that daily energy intake becomes highly variable day-to-day in the transitory state typically seen in the homeless population. However, collecting more than one 24-hour recall was not feasible because of this population's highly transitory state. Because we did not have a large enough sample size of children or shelters, we were unable to stratify our regression models by shelter or other child characteristics. We did adjust our models for gender, shelter residency, and number of children in the household. Our results may not apply to homeless children in other areas of the United States. Because children were recruited from the two largest family-based homeless shelters in the Twin Cities area, we feel our results are generalizable to the population of homeless youth in Minnesota and anticipate similar results among other homeless children.

Conclusions

Obesity and dietary intake among homeless children is a complex issue. Our study provides further evidence that obesity is relatively common among homeless children and that homeless children consume inadequate amounts of fruits and vegetables and excess amounts of foods from the fats, oils, and sweets group. Parental role modeling of eating behaviors, including the overconsumption of food, and feelings of food adequacy in the shelter were related to several dietary intake variables of children in our study, with some less favorable toward health. Because of the homeless condition in which children in our study lived, it is probable that the shelter environment with its restrictive rules for foods permitted in resident rooms, limited food storage and cooking facilities, time limitations for accessing shelter dining facilities, and poor access to food stores in the surrounding neighborhood influenced parental role-modeling behaviors and related food choices by children. Strategies to mitigate the effects of homeless living, such as expanded food shelter policies that encourage consumption of healthier foods, may help to promote healthier dietary patterns. Our study offers further evidence for the importance of promoting future programs and initiatives that increase access to affordable, healthy food. Furthermore, research has found that the provision of urban gardens can expand children's accessibility to fresh fruits and vegetables⁵⁹ and help alleviate food insecurity issues.⁶⁰ Thus, developing strategies to integrate gardens into homeless shelter grounds or connecting homeless shelters with local community gardens may help to promote fruit and vegetable consumption among children, alleviate food insecurity issues, and promote short- and long-term health. Other strategies and policy interventions to increase access to healthy foods could include targeting food banks, soup kitchens, central kitchens, and mobile food carts to provide healthier options to homeless or needy children; changing SNAP benefits to include specific allotments for purchasing healthier foods; and encouraging more farmer's markets across the United States to accept SNAP and Special Supplemental Nutrition Assistance Program for Women, Infants and Children (WIC) benefits. Further research is needed to determine how a child's diet changes from preshelter living to living in the shelter to transitioning out of the shelter, how food cost impacts shopping behavior and dietary intake among this population, and how actual physical activity patterns change during periods of homelessness.

Footnotes

Acknowledgments

This project could not have been conducted without the cooperation of the homeless children and their mothers, and we thank them for participating in this project with interest and enthusiasm. We also thank the shelter staff for their assistance in scheduling and recruitment and for their input in the project. We also thank Jamie Butterfass for her assistance in checking over completed surveys and in data entry.

Author Disclosure Statement

No competing financial interests exist.

References

National Alliance to End Homelessness. State of Homelessness 2012: Chapter 1—Homelessness Counts. www.endhomelessness.org/library/entry/soh-2012-chapter-one-homelessness-counts/. 2013 April 19.

Moore

. Health and home for homeless people in transition. Rev Environ Health, 2000; 15:135–148.

Richards

, Smith

. Environmental, parental, and personal influences on food choice, access, and overweight status among homeless children. Soc Sci Med, 2007; 65:1572–1583.

Tarasuk

, Dachner

, Li

. Homeless youth in Toronto are nutritionally vulnerable. J Nutr, 2005; 135:1926–1933.

Taylor

, Koblinsky

. Dietary intake and growth status of young homeless children. J Am Diet Assoc, 1993; 93:464–466.

Schwarz

, Garrett

, Hampsey

et al. High prevalence of overweight and obesity in homeless Baltimore children and their caregivers: A pilot study. Med Gen Med, 2007; 9:48.

Smith

, Richards

. Dietary intake, overweight status, and perceptions of food insecurity among homeless Minnesotan youth. Am J Hum Biol, 2008; 20:550–563.

Grant

, Shapiro

, Joseph

et al. The health of homeless children revisited. Adv Ped, 2007; 54:173–187.

Dammann

, Smith

. Food-related attitudes and behaviors at home, school, and restaurants: Perspectives from racially diverse, urban, low-income 9- to 13-year-old children in Minnesota. J Nutr Educ Behav, 2010; 42:389–397.

10.

Chui

, Dimarco

, Prokop

. Childhood obesity and dental caries in homeless children. J Pediatr Health Care, 2012 Jan 11[Epub ahead of print].

11.

Richards

, Smith

. The impact of homeless shelters on food access and choice among homeless families in Minnesota. J Nutr Educ Behav, 2006; 38:96–105.

12.

Richards

, Smith

. Shelter environment and placement in community affects lifestyle factors among homeless families in Minnesota. Am J Health Promot, 2006; 21:36–44.

13.

Kelly

. Assessment of dietary intake of preschool children living in a homeless shelter. Appl Nurs Res, 2001; 14:146–154.

14.

Let's Move. America's Move to Raise a Healthier Generation of Kids. www.letsmove.gov/. 2013 April 19.

15.

United States Department of Agriculture. US Household Food Security Survey Module: Six-Item Short Form. www.ers.usda.gov/datafiles/Food_Security_in_the_United_States/Food_Security_Survey_Modules/short2012.pdf/. 2013 April 19.

16.

Bandura

. Social Learning Theory. General Learning Press: New Jersey, 1971.

17.

Bandura

. Social Learning Theory. General Learning Press: New York, 1977.

18.

Bissonette

, Contento

. Adolescents' perspectives and food choice behaviors in terms of the environmental impacts of food production practices: Application of the psychosocial model. JNE, 2001; 33:72–82.

19.

McPherson

, Hoelscher

, Alexander

et al. Dietary assessment methods among school-aged children: Validity and reliability. Prev Med, 2000; 31:S11–S33.

20.

Eck

, Klesges

, Hanson

. Recall of a child's intake from one meal: Are parents accurate? J Am Diet Assoc, 1989; 89:784–789.

21.

Burrows

, Martin

, Collins

. A systematic review of the validity of dietary assessment methods in children when compared with the method of doubly labeled water. J Am Diet Assoc, 2010; 110:1501–1510.

22.

Ogden

, Carroll

, Kit

et al. Prevalence of obesity and trends in body mass index among US children and adolescents, 1999–2010. JAMA, 2012; 307:483–490.

23.

Bell

, Wilbur

, Smith

. Nutritional status of persons using a local emergency food system program in middle America. J Am Diet Assoc, 1998; 98:1031–1033.

24.

Eikenberry

, Smith

. Attitudes, beliefs, and prevalence of dumpster diving as a means to obtain food by Midwestern, low-income urban dwellers. Agric Hum Values, 2005; 22:187–202.

25.

Martin

, Lippert

. Feeding her children, but risking her health: The intersection of gender, household food insecurity and obesity. Soc Sci Med, 2012; 74:1754–1764.

26.

Franklin

, Jones

, Love

et al. Exploring mediators of food insecurity and obesity: A review of recent literature. J Community Health, 2012; 37:253–264.

27.

DeBono

, Ross

, Berrang-Ford

. Does the Food Stamp Program cause obesity? A realistic review and a call for place-based research. Health Place, 2012; 18:747–756.

28.

Leung

, Willett

, Ding

. Low-income Supplemental Nutrition Assistance Program participation is related to adiposity and metabolic risk factors. Am J Clin Nutr, 2012; 95:17–24.

29.

Gibson

. Food stamp program participation is positively related to obesity in low-income women. J Nutr, 2003; 133:2225–2231.

30.

Alaimo

, Olson

, Frongillo

. Low family income and food insufficiency in relation to overweight in US children: Is there a paradox? Arch Ped Adolesc Med, 2001; 155:1161–1167.

31.

Rose

, Bodor

. Household food insecurity and overweight status in young school children: Results from the early childhood longitudinal study. Pediatrics, 2006; 117:464–473.

32.

Shrewbury

, Wardle

. Socioeconomic status and adiposity in childhood: A systematic review of cross-sectional studies 1990–2005. Obesity, 2008; 16:275–284.

33.

Ogden

, Lamb

, Caroll

et al. Obesity and socioeconomic status in children and adolescents: United States, 2005–2008. NCHS data brief no 51. Hyattsville, MD: National Center for Health Statistics, 2010.

34.

Singh

, Siahpush

, Kogan

. Rising social inequalities in US childhood obesity, 2003–2007. Ann Epidemiol, 2010; 20:40–52.

35.

Sun

, Liang

, Huang

TTK

et al. Childhood obesity predicts adult metabolic syndrome: The Fels Longitudinal Study. J Pediatr, 2008; 152:191–200.

36.

Daniels

. The consequences of childhood overweight and obesity. Future Child, 2006; 16:47–67.

37.

Reid

, Vittinghoff

, Kushel

. Association between the level of housing instability, economic standing and health care access: A meta-regression. J Health Care Poor Underserved, 2008; 19:1270–1281.

38.

Foltz

, Cook

, Szilagyi

et al. US adolescent nutrition, exercise, and screen time baseline levels prior to national recommendations. Clin Pediatr (Phila), 2011; 50:424–433.

39.

Dovey

, Staples

, Gibson

et al. Food neophobia and ‘picky/fussy’ eating in children: A review. Appetite, 2008; 50:181–193.

40.

Birch

, Fisher

. Development of eating behaviors among children and adolescents. Pediatrics, 1998; 101:539–549.

41.

Skinner

, Carruth

, Bounds

et al. Children's food preferences: A longitudinal analysis. J Am Diet Assoc, 2002; 102:1638–1647.

42.

Cooke

, Wardle

. Age and gender differences in children's food preferences. Br J Nutr, 2005; 93:741–746.

43.

Bellisle

, Rolland-Cachera

. Kellogg Scientific Advisory Committee. Three consecutive (1993, 1995, 1997) surveys of food intake, nutritional attitudes and knowledge, and lifestyle in 1000 French children, aged 9–11 years. J Hum Nutr Diet, 2007; 20:241–251.

44.

Ventura

, Mennella

. Innate and learned preferences for sweet taste during childhood. Curr Opin Clin Nutr Metab Care, 2011; 14:379–384.

45.

Grimm

, Harnack

, Story

. Factors associated with soft drink consumption in school-aged children. J Am Diet Assoc, 2004; 104:1244–1249.

46.

Hanson

, Neumark-Sztainer

, Eisenberg

et al. Associations between parental report of the home food environment and adolescent intakes of fruit, vegetables and dairy foods. Pub Health Nutr, 2005; 8:77–85.

47.

Cullen

, Baranowski

, Rittenberg

et al. Child-reported family and peer influences on fruit, juice and vegetable consumption: Reliability and validity of measures. Health Educ Res, 2001; 16:187–200.

48.

Pearson

, Biddle

SJH

, Gorely

. Family correlates of fruit and vegetable consumption in children and adolescents: A systematic review. Pub Health Nutr, 2008; 12:267–283.

49.

Campbell

, Crawford

, Salmon

et al. Associations between the home food environment and obesity-promoting eating behaviors in adolescence. Obesity, 2007; 15:719–730.

50.

Gellar

, Schrader

, Nansel

. Healthy eating practices: Perceptions, facilitators, and barriers among youth with diabetes. Diabetes Educ, 2007; 33:671–679.

51.

Hang

, Lin

, Yang

et al. The relationship between snack intake and its availability of 4^th–6^th graders in Taiwan. Asia Pac J Clin Nutr, 2007; 16:547–553.

52.

Wiig

, Smith

. The art of grocery shopping on a food stamp budget: Factors influencing the food choices of low-income women as they try to make ends meet. Public Health Nutr, 2009; 12:1726–1734.

53.

Polivy

, Zeitlin

, Herman

et al. Food restriction and binge eating: A study of former prisoners of war. J Abnorm Psychol, 1994; 103:409–411.

54.

Polivy

. Psychological consequences of food restriction. J Am Diet Assoc, 1996; 96:589–592.

55.

Smith

, Klosterbuer

, Levine

. Military experience strongly influences post-service eating behavior and BMI status in American veterans. Appetite, 2008; 52:280–289.

56.

US Department of Agriculture and US Department of Health and Human Services. Dietary Guidelines for Americans. 2010 7^th. US Government Printing Office: Washington, DC, December 2010.

57.

Bostock

. Pathways of disadvantage? Walking as a mode of transport among low-income mothers. Health Soc Care Community, 2001; 9:11–18.

58.

Ekelund

, Tomkinson

, Armstrong

. What proportion of youth are physically active? Measurement issues, levels and recent time trends. Br J Sports Med, 2011; 45:859–865.

59.

Lautenschlager

, Smith

. Understanding gardening and dietary habits among youth garden program participants using the Theory of Planned Behavior. Appetite, 2007; 49:122–130.

60.

Carney

, Hamada

, Rdesinski

et al. Impact of a community gardening project on vegetable intake, food security and family relationships: A community-based participatory research study. J Community Health, 2012; 37:874–881.