Home Snack Environments in the United States: Latent Class Analysis Findings From a Home Food Environment Survey

Abstract

Snacking occasions have increased in frequency and energy density in recent decades, with considerable implications for diet. Studies have linked presence of foods in the home with intake of those foods. This study examines home snack food inventories among a large sample of U.S. adults using latent class analysis findings to present latent classes of home snack food inventories and multinomial regression to model classes as correlates of percent of calories from fat. Participants (n = 4,896) completed an online household food environment survey including presence of 23 snack foods in the home and demographics. Less healthy snack foods were more commonly reported than healthier snack foods (M = 4.3 vs. M = 3.5). Among White and Latinx participants, high-income households reported greater numbers of both healthier and less healthy snack foods than lower income households, with larger income-based differences in inventory sizes for healthier snack foods. Latent class analysis revealed three classes by inventory size (Small, Medium, and Large) and three classes by inventory content (Healthy Snacks, Standard American, and Limited Standard American). Compared with the Small Inventory class, the Healthy Snacks class had lower caloric intake from fat (p = .002), the Large and Medium Inventory classes had much higher caloric intake from fat (p < .0001), and Standard American and Limited Standard American class members had somewhat higher caloric intake from fat (p < .0001, and p = .0001, respectively). Future research should explore the role of snacks in Americans’ diets, their impact on diet quality and health, and how interventions can support healthy home food and snack food environments to foster healthy eating.

Keywords

dietary fat intake home food availability home food inventory household income race/ethnicity snacks

Snacking occasions have increased in both frequency and energy density leading to greater contributions to overall dietary intakes among U.S. children (Dunford & Popkin, 2018) and adults (Dunford & Popkin, 2017; Zizza et al., 2001) in recent decades. Since the 1970s, studies have tied increases in energy intake from snack foods to increases largely in salty snacks, particularly among non-Hispanic Blacks and households of lower socioeconomic status (SES; Dunford & Popkin, 2017, 2018; Zizza et al., 2001). Evidence suggests overconsumption of discretionary calories (i.e., salty snacks, cookies, candy) may be directly contributing to the obesity epidemic (Cohen et al., 2010).

Snack foods are typically included in assessments of home food inventories but they are rarely the focus. Studies have documented positive associations between having certain foods in the home and consumption of such foods, including high-fat foods, sweets, and fruits and vegetables (Cullen et al., 2003; Ding et al., 2012; Ranjit et al., 2015; Spurrier et al., 2008; Trofholz et al., 2016; Wyse et al., 2011). Other studies have noted associations between presence of foods high in fat, sugar, and/or salt and various measures of fat intake (Gattshall et al., 2008; Grant et al., 2017; Hermstad et al., 2010; Kegler et al., 2016; Patterson et al., 1997) and overweight (Gorin et al., 2011).

Demographic characteristics, such as income or the presence of children, are potentially important factors influencing home food environments. Though they are rare, studies exploring race/ethnicity have noted associations between race/ethnicity and snack foods in the home. Masters et al. (2014) found Whites more likely to have a higher prevalence of salty snack foods than other groups, while Nackers and Appelhans (2013) found Blacks more likely than Latinx to report greater home food availability of obesity-promoting foods, including snacks and sweets. Using varied measures of SES (household/neighborhood income, food insecurity), studies have had mixed findings regarding unhealthy home food environments, including snack foods. While Chai et al. (2018) and Masters et al. (2014) each found positive relationships between salty snack foods in the home and income, other studies have shown negative relationships of SES with salty snacks and sweets (Green & Glanz, 2015; Nackers & Appelhans, 2013).

The current study examines home inventories of snack foods across demographic groups among a large sample of U.S. adults. Specific study aims are to identify whether there exist sociodemographic differences in healthy and less healthy snack food inventories, to explore whether distinct types or classes of snack food inventories exist, and to understand if dietary fat intake is correlated with snack food inventory classes among U.S. adults. This study posits that there may be consistent patterns in snack food inventories across demographic groups and that those patterns may be tied to eating behaviors. Identifying these groups will aid in developing social marketing campaigns or structural interventions to promote healthy eating in priority populations.

Method

Study Design, Participants, and Recruitment

This cross-sectional study used data from an online survey exploring home food environments of U.S. adults. Of interest were community, home, interpersonal, and individual-level influences on eating behaviors. Eligible participants were 18 to 75 years of age, living in the United States, and able to read English. During enrollment, quotas were used to ensure that the final sample matched Census statistics for age, gender, income, race/ethnicity, and geographic region.

Participants were recruited from Lightspeed, an online survey panel service, in fall, 2015. Using our study eligibility criteria and enrollment quotas, Lightspeed identified eligible panelists and invited them by e-mail to participate. The informed consent process and survey took approximately 30 minutes. Participants were compensated by Lightspeed with points to be redeemed for online gift certificates, merchandise, or cash. The Emory University Institutional Review Board approved this research.

We obtained 4,942 completed surveys (39.9% of those consented). Of 12,396 individuals who consented to participate, reasons for not completing the survey included not satisfying enrollment quotas (n = 3,811, 30.7%), discontinuation of the survey (n = 2,994, 24.2%), and termination by Lightspeed for failing in-survey quality control checks (n = 649, 5.2%).

Instrument and Measures

The home food environment survey used in this study was developed based on an expanded version of our conceptual model of home food environments (Kegler, Alcantara, et al., 2014). Questions were added to the survey based on review of existing measures in the literature and measures used in our previous work (Kegler, Alcantara, et al., 2014; Kegler et al., 2016; Kegler, Swan, et al., 2014; Woodruff et al., 2017). Measures used in this study are described below.

Home Snack Food Environment

A household food inventory assessed whether 23 healthier and less healthy sweet and salty snack foods were in the home in the past 7 days, with response options of yes or no. The inventories, originally adapted from Gattshall et al. (2008), have been used in our work in southwest Georgia (Kegler, Alcantara, et al., 2014; Kegler et al., 2016; Kegler, Swan, et al., 2014; Woodruff et al., 2017). For this study, we added several new snack foods to reflect a more comprehensive list of snack foods applicable across the U.S. Inventories were divided into healthier snack foods (n = 13, range 0–13) including baked, reduced fat, and naturally low-fat/low-salt snack foods and less healthy snack foods (n = 10, range 0–10) including full-fat, fried, and salted versions of common snack foods.

Household Income and Race/Ethnicity

To assess household income, participants were asked “What is your total yearly household or family income before taxes?” Participants indicated their income by selecting one of seven categories, which were collapsed into three categories for analysis: low (<$35,000), middle ($35,000–$74,999), and high (≥$75,000) income. Race/ethnicity was assessed by asking “Which of the following best represents your race or ethnicity?” Response options were White, African American or Black, Hispanic (hereafter referred to as Latinx), Asian, and Other. Question text was adapted from the Behavioral Risk Factor Surveillance System (Centers for Disease Control and Prevention, 2015). Participants reporting “Other” race/ethnicity (n = 46) were excluded from all analyses, resulting in a sample size of n = 4,896 for this study.

Caloric Intake From Fat

We used the National Cancer Institute’s Quick Food Scan (Fat Screener) to assess percent of calories from fat based on eating habits in the past year. Following scoring guidelines, we calculated percentage of daily calories from fat based on frequency of consuming 15 food items. This measure has modest validity, with correlations ranging from r = .36, to r = .77, compared with 24–hour dietary recall data (Thompson et al., 2004). In our own more recent work consisting of cross-sectional and longitudinal studies, this measure has consistently shown high construct validity with caloric intake reduction and weight change (Alcantara et al., 2015).

Demographics

Questions assessing individual and household demographic characteristics included age, gender, state of residence (to classify by region), educational attainment, household composition (presence of children, living alone), and rurality (urban, suburban, small town, and rural). Except for rurality, which was created for this study and informed by previous qualitative work distinguishing small towns from rural areas (Kegler et al., 2015; Kegler et al., 2008), demographic measures were drawn from the Behavioral Risk Factor Surveillance System (Centers for Disease Control and Prevention, 2015) or the U.S. Census Bureau’s (2015) American Communities Survey.

Data Analysis

Frequencies and descriptive statistics were run for all demographic variables, home inventory items, and subtotals for healthier and less healthy snack foods using SPSS 24. Analyses of variance and t tests were used to explore bivariate associations between individual snack foods and demographic variables. Income and race/ethnicity showed the greatest differences in preliminary analyses and thus became the focus for more in-depth analyses. We used chi-square tests to compare individual item frequencies by household income category and post hoc pairwise comparisons to assess differences by household income (reference = high-income) in individual item frequencies. Analysis of variance was used to compare mean snack food scores by household income (reference = high-income), and household income by race/ethnicity (reference = White).

Snack food inventories were exported to Mplus 8.0 for latent class analysis (LCA), a statistical method to identify unobservable subgroups within a population based on (most often) binary indicator variables, in our case the snack food inventory items. We used LCA to determine if distinct classes of home snack food inventories exist and if classes differ by sociodemographic characteristics. We analyzed a sequence of models asking to extract solutions for 1 through 8 classes. The negative log likelihood, Akaike’s information criterion, and Bayesian information criterion assessed model fit. The Vuong–Lo–Mendell–Rubin assessed whether a C-class solution was better than a C + 1 class solution. For the best solution, classes were named using the graphical output in Mplus 8.0. The final best class assignment was exported and merged with demographic data in SAS 9.4.

A multinomial logistic regression examining demographic characteristics as possible correlates of inventory type was conducted to describe class membership. To assess the relationship between class membership and fat intake, regression analyses modeled class membership as predictors of percent calories from fat first using a crude model with class membership as the only covariate, followed by an adjusted model adding age group, race/ethnicity, income, education, region, and rurality as covariates. To compare the predictive ability of inventory class membership with simple inventory size (i.e., count of snack foods in the home), we modeled inventory size and compared the variances explained by inventory class and size. An alpha of .05 was used to establish statistical significance for all analyses.

Results

Participant Demographics

The overall sample matched census estimates of the U.S. population for age, gender, race/ethnicity, household income, and geographic region, as intended by the sampling method (Table 1). Additional demographic data include educational attainment, household composition, and rurality. Half of participants (50.0%) reported having graduated college, with more than one-third of those also reporting a postgraduate or professional degree. Nearly one-third of participants reported children living in the home (32.9%), while almost one-fourth (23.0%) lived alone. A plurality of participants reported living in suburban areas (40.8%), followed by 29.0% living in urban areas, and smaller proportions of participants living in small towns (15.8%) or rural areas (14.4%).

Table 1.

Participant Demographics and Household Characteristics (N = 4,896).

Demographics	n	%
Age category (years)
18–34	1,578	32.2
35–54	1,888	38.6
55–75	1,430	29.2
Gender
Female	2,532	51.7
Male	2,364	48.3
Race/ethnicity
White	3,335	68.1
Black	604	12.3
Latinx	703	14.4
Asian	254	5.2
Income category ($)
<35,000	1,791	36.6
35,000–74,999	1,517	31.0
≥75,000	1,588	32.4
Region
West	1,137	23.2
Midwest	1,077	22.0
Northeast	895	18.3
South	1,787	36.5
Education
Some high school or less	126	2.5
High school diploma or GED	817	16.7
Some college/technical school	1,502	30.7
College graduate	1,793	36.6
Postgraduate/professional degree	658	13.4
Children in the home	1,611	32.9
Lives alone	1,126	23.0
Rurality
Urban	1,420	29.0
Suburban	1,996	40.8
Small town	775	15.8
Rural	705	14.4

Home Snack Food Availability

Table 2 shows individual snack food frequencies for healthier and less healthy snacks in the home in the past 7 days, overall and stratified by income level, and mean snack totals for healthier and less healthy snacks stratified by income and race/ethnicity. Participants reported 7.9 snack foods in the home on average (SD = 4.6; range 0–23). Less healthy snack foods were more commonly reported in the home than healthier snacks (M = 4.3 vs. M = 3.5), despite the inventory list being composed of 13 healthier and 10 less healthy snacks.

Table 2.

Home Snack Availability (Past 7 Days) by Income and Income by Race/Ethnicity (N = 4,896).

Snack item frequency by income level	% ^a
Snack item frequency by income level	All participants (N = 4,896)	<$35,000 (n = 1,791)	$35,000-$74,999 (n = 1,517)	≥$75,000 (n = 1,588) (ref)
Healthier snacks
Popcorn	61.6	57.6^b	63.7	64.2
Granola or granola bars	50.9	41.1^b	52.7^c	60.2
Pretzels	38.3	30.4^b	36.7^c	48.7
Whole grain snack crackers	33.0	25.4^b	31.6^c	42.9
Unsalted nuts	28.2	20.8^b	28.3^c	36.6
Sports/protein/meal replacement bars	27.3	19.7^b	26.8^c	36.5
Baked potato chips	23.4	17.4^b	23.9^c	29.5
Sherbet, sorbet, or frozen fruit juice bars	17.2	14.5^b	16.0^c	21.5
Reduced fat cookies, cakes, snack cakes, donuts	14.6	11.3^b	13.2^c	19.5
Baked tortilla chips	14.5	10.0^b	14.4^c	19.6
Reduced fat ice cream	13.8	8.7^b	13.1^c	20.2
Reduced fat frozen yogurt	13.2	8.5^b	12.9^c	18.8
Baked cheese curls or puffs	12.9	9.9^b	11.9^c	17.3
Less healthy snacks
Chocolates, candy, or candy bars	70.0	66.2^b	70.7^d	73.6
Regular cookies, cakes, snack cakes, donuts	62.2	59.9^e	63.9	63.0
Regular potato chips	61.6	58.4^e	64.9^d	61.9
Regular ice cream	52.9	50.3^f	53.1	55.7
Regular snack crackers	48.2	45.8^f	48.2	50.8
Regular tortilla chips	44.3	37.7^b	44.5^c	51.5
Salted nuts	43.2	34.8^b	42.5^c	53.4
Regular cheese curls or puffs	25.3	24.6^f	22.9^c	28.4
Regular fat frozen yogurt	15.5	11.7^b	16.1^d	19.3
Pork rinds	11.6	10.3^f	10.8^g	13.9
Mean snack totals by income and race/ethnicity	M ± SD^h
Mean snack totals by income and race/ethnicity	All participants (N = 4,896)	<$35,000 (n = 1,791)	$35,000-$74,999 (n = 1,517)	≥$75,000 (n = 1,588) (ref)
Healthier snacks
White (n = 3,335) (ref)	3.6 ± 2.7ⁱ	2.6 ± 2.3^b,j	3.5 ± 2.7^c	4.5 ± 2.9^i,k
Black (n = 604)	2.9 ± 2.8	2.6 ± 2.8	3.1 ± 3.0	3.2 ± 2.6
Latinx (n = 703)	3.8 ± 3.1	3.4 ± 3.0^b	3.6 ± 2.8^g	4.7 ± 3.5
Asian (n = 254)	3.1 ± 2.7	2.6 ± 2.3	3.2 ± 2.4	3.4 ± 3.1
Total (N = 4,896)	3.5 ± 2.8	2.8 ± 2.5^b	3.4 ± 2.7^c	4.4 ± 2.9
Less healthy snacks
White (n = 3,335) (ref)	4.3 ± 2.4^l	3.9 ± 2.4^b,l	4.3 ± 2.2ⁱ	4.7 ± 2.4
Black (n = 604)	4.2 ± 2.6	4.1 ± 2.6	4.3 ± 2.6	4.6 ± 2.4
Latinx (n = 703)	4.6 ± 2.6	4.4 ± 2.6^b	4.6 ± 2.4	5.1 ± 2.8
Asian (n = 254)	4.2 ± 2.3	3.9 ± 2.2	4.3 ± 2.3	4.2 ± 2.4
Total (N = 4,896)	4.3 ± 2.4	4.0 ± 2.4^b	4.4 ± 2.3^c	4.7 ± 2.5

Chi-square test used to assess differences in frequencies of individual items.

Significant difference between <$35,000 and ≥$75,000 at P < .001.

Significant difference between $35,000-74,999 and ≥$75,000 at P < .001.

Significant difference between $35,000-74,999 and ≥$75,000 at P < .05.

Significant difference between <$35,000 and ≥$75,000 at P < .05.

Significant difference between <$35,000 and ≥$75,000 at P < .01.

Significant difference between $35,000-74,999 and ≥$75,000 at P < .01.

Analysis of variance used to assess differences in means.

Significant difference between White and Black at P < .001.

Significant difference between White and Latinx at P < .001.

Significant difference between White and Asian at P < .01.

Significant difference between White and Latinx at P < .05.

Home Snack Food Availability by Household Income

High-income participants reported the greatest mean number of healthier and less healthy snack foods compared with the middle- and low-income groups, with larger differences by income group for healthier snacks than for less healthy snacks. High-income participants reported significantly greater frequencies for all 13 healthier snack foods than low-income participants, and significantly greater frequencies for all healthier items except popcorn than middle-income participants (all p < .001). A similar, but less consistent pattern was observed for less healthy snack foods. High-income participants reported the greatest number of healthier snack foods (M = 4.4 ± 2.9), significantly more than middle- (M = 3.4 ± 2.7, p < .001) and low-income (M = 2.8 ± 2.5, p < .001) participants. Similarly, high-income participants reported the greatest mean number of less healthy snack foods (M = 4.7 ± 2.5), significantly more than middle-income (M = 4.4 ± 2.3, p < .001) and low-income (M = 4.0 ± 2.4, p < .001) participants (Table 2).

Home Snack Food Availability by Household Income and Race/Ethnicity

Mean inventory size increased significantly with income level among both White and Latinx participants (Table 2). Among White participants, there were significant differences between low- and high- as well as middle- and high-income participants for healthier snacks (both p < .001), and for less healthy snacks, between low- and high-income participants (p < .001), and middle- and high-income participants (p < .01). Among Latinx participants, there were significant differences for mean number of healthier snacks between low- and high- as well as middle- and high-income participants (both p < .001), and for less healthy snacks, between low- and high-income participants (p < .001). A similar positive relationship was seen among Black and Asian participants, but differences across income categories were not significant (Table 2).

Across income categories, Latinx participants reported the greatest mean number of healthier and less healthy snack foods (Table 2). Among low-income participants, Latinx participants reported a significantly greater mean number of healthier (p < .001) and less healthy snacks (p < .05) than White participants. There were no significant differences in the middle-income group by race/ethnicity. Among high-income participants, White participants reported a significantly greater mean number of healthier snacks than Black (p < .001) and Asian (p < .01) participants.

Latent Class Analysis Findings

Class Types

The LCA revealed the best model having six classes based on Bayesian information criterion, Vuong–Lo–Mendell–Rubin, and entropy. The patterns indicated two class types, with three classes within each type. The first class type, characterized by inventory size, had three classes: Small (n = 989), Medium (n = 502), and Large (n = 129) reported mean snack scores of 2.3 (SD = 1.30), 14.0 (SD = 2.67), and 21.3 (SD = 1.72), respectively. For these three classes, probabilities for having items in the home were on average 93% (SD = 7%, range 78% to 100%) for the Large Inventory class, 60% (SD = 9%, range 45% to 83%) for the Medium Inventory class, and 11% (SD = 10%, range 2% to 32%) for the Small Inventory class.

The second class type had distinct patterns showing preferences for certain types of snack foods. Standard American class (n = 767) members reported an average of 11.6 items (SD = 1.79). Members of this class showed high probabilities (>70%) for typical U.S. snack foods, that is, those that are less healthy. The remaining two classes showed opposing patterns for several items. The Healthy Snacks class (n = 1,119) reported an average of 7.4 snack foods (SD = 2.0), with medium to lower probabilities for less healthy items in the home, and a higher probability of having several healthier snack foods in the home. The largest class (n = 1,390), Limited Standard American, reported an average of 6.6 items (SD = 1.60) and was similar to the Standard American class except that it tended to have fewer types of snacks overall. This class showed a pattern opposite the Healthy Snacks class with higher probabilities for less healthy snack foods and distinctly lower probabilities for healthier snacks (see Figures 1 and 2).

Figure 1.

Snack scores and latent class analysis probabilities Classes 1–3: inventory size classes.

Figure 2.

Snack scores and latent class analysis probabilities Classes 4–6: distinct inventory pattern classes.

Demographic Characteristics of Class Membership

Multinomial regression findings showed class differences for several demographic characteristics (Table 3). Most pronounced were differences in household income and presence of children in the home. Compared with Small Inventory class members, those with large snack food inventories were significantly younger, more likely to live in urban areas, had children at home or lived alone, and had higher household incomes. Medium Inventory class members were also more likely to have higher incomes and children in the home but less likely to live alone than those with Small inventories.

Table 3.

Characteristics of Class Members from Multinomial Regression Compared With the Small Inventory Class (N = 4,896).

Variable	Large Inventory				Medium Inventory				Standard American				Limited Standard American				Healthy Snacks
Variable	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p
Age category (years)
55–75 (ref)	1				1				1				1				1
18–34	8.64	3.59	−20.80	<.0001	1.09	0.83	−1.43	.54	4.86	3.38	−6.99	<.0001	0.82	0.66	−1.03	.08	1.09	0.86	−1.38	.48
35–54	4.45	1.85	−10.70	.00	0.78	0.60	−1.00	.05	1.42	0.98	−2.06	.06	0.80	0.65	−0.98	.03	0.85	0.68	−1.06	.15
Race/ethnicity
White (ref)	1				1				1				1				1
African American/Black	1.28	0.71	−2.30	.41	0.58	0.42	−0.81	.00	0.90	0.62	−1.29	.56	0.89	0.70	−1.13	.35	0.55	0.41	−0.73	<.0001
Latinx	1.77	1.09	−2.86	.02	1.00	0.74	−1.34	.99	1.14	0.81	−1.59	.45	1.06	0.82	−1.37	.67	0.93	0.71	−1.23	.62
Asian	0.49	0.19	−1.26	.14	0.52	0.33	−0.81	.00	0.64	0.40	−1.04	.07	0.72	0.50	−1.03	.07	0.48	0.32	−0.71	.00
Income category
<$35,000 (ref)	1				1				1				1				1
$35,000–$74,999	1.79	1.06	−3.01	.03	1.47	1.15	−1.89	.00	2.18	1.60	−2.98	<.0001	1.25	1.02	−1.53	.03	1.44	1.15	−1.80	.00
⩾$75,000	4.22	2.41	−7.38	<.0001	2.49	1.88	−3.31	<.0001	4.84	3.44	−6.81	<.0001	1.10	0.86	−1.41	.46	2.62	2.03	−3.37	<.0001
Education
Some high school/less (ref)	1				1				1				1				1
High school or GED	1.56	0.43	−5.59	.50	1.37	0.70	−2.67	.35	0.61	0.31	−1.23	.17	0.95	0.58	−1.55	.82	3.72	1.40	−9.91	.01
Some college/tech. school	0.68	0.19	−2.49	.56	1.51	0.78	−2.89	.22	0.45	0.23	−0.89	.02	0.99	0.61	−1.60	.97	6.37	2.42	−16.72	.00
College graduate	0.87	0.24	−3.13	.83	1.11	0.57	−2.14	.76	0.68	0.35	−1.34	.27	0.75	0.46	−1.21	.24	5.52	2.10	−14.55	.00
Postgrad./prof. degree	0.99	0.26	−3.79	.99	0.84	0.41	−1.71	.63	0.65	0.32	−1.35	.25	0.60	0.35	−1.02	.06	5.33	1.98	−14.36	.00
Children in the home	5.80	3.41	−9.87	<.0001	2.72	2.13	−3.49	<.0001	3.98	2.96	−5.36	<.0001	1.66	1.33	−2.09	<.0001	1.33	1.04	−1.68	.02
Lives alone	1.88	1.03	−3.43	.04	0.35	0.26	−0.47	<.0001	1.21	0.88	−1.67	.24	0.57	0.47	−0.70	<.0001	0.60	0.48	−0.75	<.0001
Rurality
Urban (ref)	1				1				1				1				1
Suburban	0.73	0.39	−1.40	.34	1.57	1.12	−2.19	.01	1.00	0.70	−1.44	.99	1.31	1.00	−1.72	.05	1.05	0.77	−1.42	.77
Small town	0.68	0.37	−1.26	.22	1.53	1.11	−2.11	.01	0.60	0.41	−0.87	.01	1.34	1.03	−1.74	.03	1.22	0.91	−1.62	.18
Rural	0.51	0.33	−0.81	.00	1.46	1.13	−1.87	.00	0.50	0.38	−0.66	<.0001	1.02	0.83	−1.25	.87	1.20	0.97	−1.49	.10
Region
West (ref)	1				1				1				1				1
Midwest	0.64	0.36	−1.16	.14	1.28	0.96	−1.71	.09	0.89	0.63	−1.26	.51	1.24	0.96	−1.60	.10	0.94	0.72	−1.22	.65
Northeast	0.85	0.50	−1.46	.55	0.85	0.62	−1.16	.30	1.10	0.78	−1.54	.59	1.17	0.90	−1.53	.24	0.89	0.68	−1.17	.41
South	0.79	0.48	−1.28	.33	1.00	0.76	−1.30	.97	0.89	0.66	−1.22	.48	1.36	1.08	−1.70	.01	0.95	0.75	−1.20	.65

Note. Significant p values bolded.

Compared with the Small Inventory class, Standard American class members were less likely to be Black or Asian, but more likely to have higher household incomes, live in urban settings, and to live with children or other adults but not alone. Members in the Limited Standard American class were more likely to be in the middle to older age groups, middle-income, and live with children. They tended to live in small towns and in the South. Healthy Snacks class members were less likely to be Black or Asian compared with the Small Inventory class. They were more likely to have higher incomes and were the only class that reported higher levels of education compared with the Small Inventory class.

Snack Food Inventory Class Membership and Caloric Intake From Fat

In the subsequent regression analysis, class membership was significantly associated with percent of calories from fat. Compared with the Small Inventory class, only the Healthy Snacks class had lower caloric intake from fat (M = 33.5%; b = −0.66%, p = .002). The Large and Medium Inventory classes had much higher caloric intake from fat (M = 41.0%; b = 6.22%, and M = 39.1%; b = 4.38%, respectively, p < .001). This is in addition to an already high mean proportional fat intake by the Small Inventory class of 34.2%, which is at the high end of the recommended intake (U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015). Standard American and Limited Traditional class members had only somewhat higher mean caloric intake from fat (M = 35.2%; b = 1.00%, p < .0001, and M = 35.0%; b = 0.77%, p = .0001, respectively) compared with the Small Inventory class (Table 4).

Table 4.

Regression Results Assessing the Relationship between Snack Inventory Class Membership and Caloric Intake from Fat (N = 4,892).

Variable	β	SE	p
Intercept	34.23	0.52	<.0001
Snack inventory class
Small Inventory (ref)
Large Inventory	6.22	0.46	<.0001
Medium Inventory	4.38	0.28	<.0001
Standard American	1.00	0.24	<.0001
Limited Standard American	0.77	0.20	.0001
Healthy Snacks	−0.66	0.22	.002
Age category (years)
55–75 (ref)	0
35–54	1.28	0.18	<.0001
18–34	1.67	0.19	<.0001
Race/ethnicity
White (ref)	0
African American/Black	−0.17	0.22	.43
Latinx	−0.67	0.21	.001
Asian	–1.94	0.32	<.0001
Income category
<$35,000	0
$35,000–$74,999	−0.10	0.18	.58
≥ $75,000	0.09	0.20	.66
Education
Some high school or less (ref)	0
High school or GED	−0.12	0.46	.80
Some college/technical school	−0.05	0.46	.91
College graduate	−0.36	0.46	.43
Postgraduate/professional degree	−0.65	0.49	.19
Children in the home	0.30	0.17	.09
Lives alone	0.57	0.19	.003
Rurality
Urban (ref)	0
Suburban	−0.94	0.17	<.0001
Small town	−0.36	0.22	.11
Rural	0.05	0.23	.81
Region
West (ref)	0
Midwest	−0.41	0.21	.52
South	−0.14	0.19	.45
Northeast	−0.41	0.21	.03
R ²	15.80%

Note. Significant p values bolded.

In the crude model, snack food inventory class membership explained 12.17% of the variance in percent of calories from fat. When controlling for demographics, total variance explained was 15.80%. In an alternate model, the snack inventory score (i.e., sum of snack foods reported in the home) was modeled instead of snack inventory class membership and explained only 7.01% of the variance in calories from fat, indicating that class membership has higher predictive validity than inventory size.

Discussion

This study is unique in that it characterizes home snack food environments of households across the United States. Though home inventories of snack foods have been measured in specific geographic populations, no known studies have reported estimates of the type or mean number of items in U.S. households. Furthermore, this study is the first to use LCA to explore distinct patterns in home snack food inventories and their associations with demographics and dietary intakes. Overall, households had fewer healthier snacks than less healthy snacks. There was a strong relationship between income and snack food inventory size and content, such that higher income households had a greater number of both healthier and less healthy snacks, with larger income-based differences for healthier snacks than for less healthy snacks. This is somewhat consistent with prior research that showed household income was positively associated with greater availability of less healthful foods in the home (Chai et al., 2018; Masters et al., 2014). However, other studies have found a negative relationship between salty snacks in the home and SES (Green & Glanz, 2015; Nackers & Appelhans, 2013).

Research concerning racial/ethnic differences in home food inventories is lacking. Masters et al. (2014) found Whites had a significantly higher prevalence of salty snacks always available compared with Latinx and Blacks, similar to what we found. In our study, snack food inventories varied by income and race/ethnicity, where high-income White and Latinx participants reported greater mean snack scores for healthier and less healthy snack foods. Within each income category, Latinx participants reported the greatest mean snack scores, and Black and Asian participants reported the lowest mean snack scores. This latter finding demonstrates a need to explore whether Black and Asian participants tended to report smaller inventories than White and Latinx participants because they have fewer snack foods in the home and why that may be the case, or if typical inventories like we used do not fully capture the home snack food environment for these groups. Beyond proximal cultural and economic factors, there may be upstream factors contributing to demographic differences in home snack food inventories. Intervention on upstream determinants, such as access to healthy foods or better paying jobs, may minimize differences.

The LCA revealed six inventory classes, with three inventory classes based on number of items reported (Small, Medium, and Large), and three classes related to inventory composition (Healthy Snacks, Standard American, and Limited Standard American). Multinomial regression analyses of these classes yielded notable differences. Income and presence of children in the home were consistently related to class membership, with the Small Inventory class less likely to have children in the home and more likely to have lower incomes than any other class. The Large Inventory class members were much more likely to be younger and to report having children in the home. The strongest relationship for educational attainment was seen in the Healthy Snacks class, which was much more likely to have higher levels of educational attainment than those in the Small Inventory class. These findings suggest a possible social marketing approach to targeting messages and other components of behavior change interventions for different groups based on inventory class types and the demographic characteristics among the different classes.

Exemplifying the relationship between home snack food inventories and diet, both the Large and Medium Inventory classes had significantly higher levels of caloric intake from fat than the Small Inventory class. Notably, even the Small Inventory class reported average fat intakes at the high end of the range (20% to 35% for adults aged 19 years and older) recommended by the Dietary Guidelines for Americans (U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015). Smaller differences in percent of calories from fat were also seen among the pattern classes, with Healthy Snacks members showing lower proportions of fat intake and both the Standard American and Limited Standard classes showing slightly increased percent of calories from fat compared with the Small Inventory class. While past studies of home food inventories have not focused exclusively on snack foods, prior research has demonstrated that availability of healthy foods in the home is positively associated with healthier dietary behaviors, availability of less healthy foods with less healthy diets (Ding et al., 2012), and absence of unhealthy foods in the home associated with healthier eating (Florez et al., 2017).

Our study adds to the literature on snack food intake and diet quality. Studies investigating energy contributions of snacks to overall diet have found that total calories from snacks have increased in recent decades in Americans of all ages, and that salty snack consumption increased significantly (Dunford & Popkin, 2017, 2018; Zizza et al., 2001). Dunford and Popkin (2018) analyzed data from eight nationally representative studies from 1977 and 2014 and found groups with the largest increases included non-Hispanic Blacks and children in the lowest income and household education categories. A similar study of adults found increased contributions of snack foods to total daily energy intake due to increases in energy intake per snack and increased number of snacks (Zizza et al., 2001). A national study of Brazilian youth and adults found that increased snacking frequency was associated with increased overall caloric intake (Duffey et al., 2013). While we were unable to explore overall energy intake, the positive relationship between snack foods in the home and fat intake we observed is generally consistent with findings from these prior studies.

Strengths and Limitations

While this study has notable strengths, including the large sample and use of tested and validated measures, it is not without limitations. By conducting a cross-sectional survey, we cannot make causal claims. Furthermore, participants were members of a panel service that completes online surveys and may differ from the overall U.S. population, especially those who lack internet access and/or do not read English. Although the sample was constructed to reflect the U.S. on key demographic variables, it may not be fully representative. The home snack food inventory is not comprehensive and excludes fruits and vegetables, possibly an important kind of healthier snack food for certain groups. Related to the measurement of snack foods, we quantified availability by type of snack, not amount of snacks, in the home. In addition, less healthy snack foods were often distinguished from healthier snacks based solely on fat content, despite increasing recognition that processed foods are unhealthy due to myriad factors beyond total fat content, including type of fat, sugar, and various additives (Fiolet et al., 2018). Finally, this survey was conducted in fall, 2015 and inventory results may be subject to seasonality.

Implications for Research and Practice

Findings from this study may inform future studies of underlying causes of demographic differences in home snack food inventories and dietary intake. Race/ethnicity, income, and other demographic traits may serve as proxies for modifiable structural factors influencing home food environments, meriting examination. Furthermore, researchers may explore how snack inventory classes can be used in a social marketing approach to targeting messages and intervention components for different groups. For example, if some groups’ diets are tied to cultural or family traditions, while others are more health-oriented, then approaching them differently may make sense. Demographic differences could inform how to target different groups with tailored health messages. Future research should explore the role of snack foods in Americans’ diets, their impact on overall diet quality and health status, and how interventions can support healthy home food and snack food environments. Understanding the contribution of snack foods to overall caloric intake and weight status may be important in addressing rising obesity rates among U.S. adults.

Footnotes

Acknowledgements

We thank the survey participants for their contributions to this research and Lightspeed GMI for their support in administering the survey.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this study was provided by the Dean’s Office at the Emory University Rollins School of Public Health. IGR was supported by NHLBI T32HL130025 (PI: Vaccarino) and NHLBI T32HL007034 (PI: Gardner).

ORCID iDs

April Hermstad

Michelle C. Kegler

Ilana G. Raskind

Regine Haardörfer

References

Alcantara

Haardörfer

Gazmararian

J. A.

Hartman

T. J.

Greene

Kegler

M.C.

(2015). Relative validation of fruit and vegetable intake and fat intake among overweight and obese African-American women. Public Health Nutrition, 18(11), 1932–1940. https://doi.org/10.1017/S1368980014002547

Centers for Disease Control and Prevention. (2015). Behavioral Risk Factor Surveillance System Survey Questionnaire. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention.

Chai

Fan

J. X.

Wen

(2018). Association of individual and neighborhood factors with home food availability: Evidence from the National Health and Nutrition Examination Survey. Journal of the Academy of Nutrition and Dietetics, 118(5), 815–823. https://doi.org/10.1016/j.jand.2017.11.009

Cohen

D. A.

Sturm

Scott

Farley

T. A.

Blumenthal

(2010). Not enough fruit and vegetables or too many cookies, candies, salty snacks, and soft drinks? Public Health Reports, 125(1), 88–95. https://doi.org/10.1177/003335491012500112

Cullen

K. W.

Baranowski

Owens

Marsh

Rittenberry

de Moor

(2003). Availability, accessibility, and preferences for fruit, 100% fruit juice, and vegetables influence children’s dietary behavior. Health Education & Behavior, 30(5), 615–626. https://doi.org/10.1177/1090198103257254

Ding

Sallis

J. F.

Norman

G. J.

Saelens

B. E.

Harris

S. K.

Kerr

Rosenberg

Durant

Glanz

(2012). Community food environment, home food environment, and fruit and vegetable intake of children and adolescents. Journal of Nutrition Education and Behavior, 44(6), 634–638. https://doi.org/10.1016/j.jneb.2010.07.003

Duffey

K. J.

Pereira

R. A.

Popkin

B. M.

(2013). Prevalence and energy intake from snacking in Brazil: Analysis of the first nationwide individual survey. European Journal of Clinical Nutrition, 67(8), 868–874. https://doi.org/10.1038/ejcn.2013.60

Dunford

E. K.

Popkin

B. M.

(2017). Disparities in snacking trends in US adults over a 35 year period from 1977 to 2012. Nutrients, 9(8), 809–818. https://doi.org/10.3390/nu9080809

Dunford

E. K.

Popkin

B. M.

(2018). 37 year snacking trends for US children 1977-2014. Pediatric Obesity, 13(4), 247–255. https://doi.org/10.1111/ijpo.12220

10.

Fiolet

Srour

Sellem

Kesse-Guyot

Alles

Mejean

Deschasaux

Fassier

Latino-Martel

Beslay

Hercberg

Lavalette

Monteiro

C. A.

Julia

Touvier

(2018). Consumption of ultra-processed foods and cancer risk: Results from NutriNet-Sante prospective cohort. BMJ, 360, k322. https://doi.org/10.1136/bmj.k322

11.

Florez

K. R.

Richardson

A. S.

Ghosh-Dastidar

M. B.

Beckman

Huang

Wagner

Dubowitz

(2017). Improved parental dietary quality is associated with children’s dietary intake through the home environment. Obesity Science and Practice, 3(1), 75–82. https://doi.org/10.1002/osp4.81

12.

Gattshall

M. L.

Shoup

J. A.

Marshall

J. A.

Crane

L. A.

Estabrooks

P. A.

(2008). Validation of a survey instrument to assess home environments for physical activity and healthy eating in overweight children. International Journal of Behavioral Nutrition and Physical Activity, 5, 3. https://doi.org/10.1186/1479-5868-5-3

13.

Gorin

A. A.

Phelan

Raynor

Wing

R. R.

(2011). Home food and exercise environments of normal-weight and overweight adults. American Journal of Health Behavior, 35(5), 618–626. https://doi.org/10.5993/AJHB.35.5.10

14.

Grant

Gearry

Wilson

Pearson

Skidmore

(2017). Home availability of fruit and vegetables and obesogenic foods as an indicator of nutrient intake in 50 year olds from Canterbury, New Zealand. Asia Pacific Journal of Clinical Nutrition, 26(3), 524–530. https://doi.org/10.6133/apjcn.042016.11

15.

Green

S. H.

Glanz

(2015). Development of the Perceived Nutrition Environment Measures Survey. American Journal of Preventive Medicine, 49(1), 50–61. https://doi.org/10.1016/j.amepre.2015.02.004

16.

Hermstad

A. K.

Swan

D. W.

Kegler

M. C.

Barnette

J. K.

Glanz

(2010). Individual and environmental correlates of dietary fat intake in rural communities: A structural equation model analysis. Social Science & Medicine, 71(1), 93–101. https://doi.org/10.1016/j.socscimed.2010.03.028

17.

Kegler

M. C.

Alcantara

Haardörfer

Gazmararian

J. A.

Ballard

Sabbs

(2014). The influence of home food environments on eating behaviors of overweight and obese women. Journal of Nutrition Education and Behavior, 46(3), 188–196. https://doi.org/10.1016/j.jneb.2014.01.001

18.

Kegler

M. C.

Alcantara

Haardörfer

Gemma

Ballard

Gazmararian

(2015). Rural neighborhood walkability: Implications for assessment. Journal of Physical Activity and Health, 12(S1), S40–S45. https://doi.org/10.1123/jpah.2013-0431

19.

Kegler

M. C.

Escoffery

Alcantara

Ballard

Glanz

(2008). A qualitative examination of home and neighborhood environments for obesity prevention in rural adults. International Journal of Behavioral Nutrition and Physical Activity, 5, 65. https://doi.org/10.1186/1479-5868-5-65

20.

Kegler

M. C.

Haardörfer

Alcantara

I. C.

Gazmararian

J. A.

Veluswamy

J. K.

Hodge

T. L.

Addison

A. R.

Hotz

J. A.

(2016). Impact of improving home environments on energy intake and physical activity: A randomized controlled trial. American Journal of Public Health, 106(1), 143–152. https://doi.org/10.2105/AJPH.2015.302942

21.

Kegler

M. C.

Swan

D. W.

Alcantara

Feldman

Glanz

(2014). The influence of rural home and neighborhood environments on healthy eating, physical activity, and weight. Prevention Science, 15(1), 1–11. https://doi.org/10.1007/s11121-012-0349-3

22.

Masters

M. A.

Stanek Krogstrand

K. L.

Eskridge

K. M.

Albrecht

J. A.

(2014). Race/ethnicity and income in relation to the home food environment in US youth aged 6 to 19 years. Journal of the Academy of Nutrition and Dietetics, 114(10), 1533–1543. https://doi.org/10.1016/j.jand.2014.04.023

23.

Nackers

L. M.

Appelhans

B. M.

(2013). Food insecurity is linked to a food environment promoting obesity in households with children. Journal of Nutrition Education and Behavior, 45(6), 780–784. https://doi.org/10.1016/j.jneb.2013.08.001

24.

Patterson

R. E.

Kristal

A. R.

Shannon

Hunt

J. R.

White

(1997). Using a brief household food inventory as an environmental indicator of individual dietary practices. American Journal of Public Health, 87(2), 272–275. https://doi.org/10.2105/AJPH.87.2.272

25.

Ranjit

Wilkinson

A. V.

Lytle

L. M.

Evans

A. E.

Saxton

Hoelscher

D. M.

(2015). Socioeconomic inequalities in children’s diet: The role of the home food environment. The International Journal of Behavioral Nutrition and Physical Activity, 12(1), S4. https://doi.org/10.1186/1479-5868-12-S1-S4

26.

Spurrier

N. J.

Magarey

A. A.

Golley

Curnow

Sawyer

M. G.

(2008). Relationships between the home environment and physical activity and dietary patterns of preschool children: A cross-sectional study. International Journal of Behavioral Nutrition and Physical Activity, 5, 31. https://doi.org/10.1186/1479-5868-5-31

27.

Thompson

F. E.

Midthune

Subar

A. F.

Kahle

L. L.

Schatzkin

Kipnis

(2004). Performance of a short tool to assess dietary intakes of fruits and vegetables, percentage energy from fat and fibre. Public Health Nutrition, 7(8), 1097–1105. https://doi.org/10.1079/PHN2004642

28.

Trofholz

A. C.

Tate

A. D.

Draxten

M. L.

Neumark-Sztainer

Berge

J. M.

(2016). Home food environment factors associated with the presence of fruit and vegetables at dinner: A direct observational study. Appetite, 96(January), 526–532. https://doi.org/10.1016/j.appet.2015.10.019

29.

U.S. Census Bureau. (2015). American Community Survey. https://www2.census.gov/programs-surveys/acs/methodology/questionnaires/2015/quest15.pdf

30.

U.S. Department of Health and Human Services and U.S. Department of Agriculture. (2015). 2015–2020 Dietary guidelines for Americans. https://health.gov/dietaryguidelines/2015/guidelines/

31.

Woodruff

R. C.

Haardörfer

Gazmararian

J. A.

Ballard

Addison

A. R.

Hotz

J. A.

Tucker

R. B.

Kegler

M.C.

(2017). Home environment-focused intervention improves dietary quality: A secondary analysis from the Healthy Homes/Healthy Families randomized trial. Journal of Nutrition Education and Behavior, 18(30635-3), S1499–S4046. https://doi.org/10.1016/j.jneb.2018.06.007

32.

Wyse

Campbell

Nathan

Wolfenden

(2011). Associations between characteristics of the home food environment and fruit and vegetable intake in preschool children: A cross-sectional study. BMC Public Health, 11, 938. https://doi.org/10.1186/1471-2458-11-938

33.

Zizza

Siega-Riz

A. M.

Popkin

B. M.

(2001). Significant increase in young adults’ snacking between 1977-1978 and 1994-1996 represents a cause for concern! Preventive Medicine, 32(4), 303–310. https://doi.org/10.1006/pmed.2000.0817