Factors Influencing the Sugar-Sweetened Beverage Intake of Caregivers of Adolescents in Appalachia

Abstract

Objective:

To identify factors that influence the sugar-sweetened beverage (SSB) intake of caregivers of middle school-aged adolescents.

Design:

Cross-sectional

Setting:

Southwestern Virginia, U.S, part of Central Appalachia.

Participants:

Caregivers (n=362) of adolescents enrolled in the Kids SIPsmartER trial. Participants were mostly female (91%) and non-Hispanic White (96%), and 21% received SNAP benefits.

Main Outcome Measures:

Caregiver daily SSB intake and demographic, personal level, interpersonal level, and environmental level determinants.

Analysis:

Descriptive statistics, one-way ANOVAs, step-wise regression.

Results:

On average, caregivers consumed 25.7 (SD=33.2) fluid ounces of SSB per day. In the final model which included all variables, age (B=−0.41, P<0.05), receiving SNAP benefits (B=14.19, P≤0.01), behavioral intentions (B=−5.48, P≤0.001), affective attitudes (B=−2.15, P<0.05), perceptions of whether their adolescent frequently consumes high amounts of SSB (B=1.92, P≤0.001) and home availability (B=7.43, P≤0.01) were significantly associated with SSB intake.

Conclusion and Implications:

Caregivers of Appalachian middle school students are high SSB consumers. Findings highlight the importance of implementing behavioral interventions for caregivers of adolescents that target multiple levels of influence, including demographic, personal-, interpersonal-, and environmental-level factors. Interventions may be particularly important for communities and groups that may have higher SSB intakes, such as those in Appalachia and who receive SNAP.

Introduction

Sugar-sweetened beverages, i.e., soda/pop, fruit drinks, energy drinks, sports drinks, and calorically sweetened coffee and tea, have high amounts of added sugar and few nutrients.¹ Intake of SSBs is high across all age groups, with SSBs accounting for 7% and 8% of daily energy intake among children and adults, respectively.² Intakes are high among specific populations, such as those living in rural areas.^{3, 4} As excessive SSB intake increases risk for numerous preventable chronic diseases, including obesity, cancer, diabetes, and dental caries,⁵ reducing SSB intake has considerable public health implications.

Behavioral interventions targeting dietary behaviors among adolescents are more effective when they target multiple levels of the socio-ecologic model (SEM).⁶ The SEM postulates that behavior change is influenced at proximal and distal levels, including the personal, interpersonal, and environmental levels.⁷ For example, adolescents’ caregivers (i.e., parents, guardians) should be included in interventions, as they are role models and gatekeepers of food and beverages within the home and directly influence adolescents’ behaviors.^8–10 Furthermore, including caregivers may change their personal behaviors, which could have further impact on their adolescent’s behaviors. Therefore, there have been many calls to action related to using a multi-level approach and, specifically, engaging caregivers in interventions addressing sugar-sweetened beverage (SSB) intake among children and adolescents.^8–11 However, multi-level approaches involving caregivers are recommended to reduce adolescent SSB intake, yet are infrequently utilized.¹²

Existing evidence of adult SSB intake identifies influences related to demographics (e.g., age, sex, race/ethnicity, income, educational attainment, participation in the Supplemental Nutrition Assistance Program (SNAP)) and personal-level factors (e.g., psychosocial variables, health literacy status).^{3, 13–19} Additionally, evidence suggests that at the interpersonal level intakes (both caregiver and child) are often similar and child behavior could influence parental intake.^{20, 21} Lastly, environmental influences, such as availability in the home,^{9, 10} have been shown to impact caregiver intake. However, no known study has comprehensively assessed influences of SSB intake across the SEM among adults, let alone caregivers of children and adolescents. This lack of evidence is in stark contrast to established factors associated with child and adolescent SSB intake, which have been comprehensively assessed across the SEM.^{22, 23}

Exploring multi-level factors that influence SSB intake among adolescent caregivers in regions with recognized high intakes of SSBs and health disparities, such as Central Appalachia,^{24, 25} may be of particular importance for reducing SSB intake for both the caregiver and adolescent. This understanding would help support the design of multi-level interventions targeting caregiver SSB intake. The purpose of this present research was to examine demographic, personal-, interpersonal-, and environmental-level factors that were hypothesized to influence the SSB intake of caregivers of adolescents in rural Appalachia.

Methods

This study was a secondary, cross-sectional analysis of baseline data collected from the on-going Kids SIPsmartER trial (Clincialtrials.gov: NCT03740113; 2018–2022). Kids SIPsmartER was a theory-based, multi-level school-based behavior and health literacy intervention designed to improve SSB behaviors among 7^th grade middle school students and their caregivers.²⁶ The 6-month intervention used strategies that are guided by the Theory of Planned Behavior (TPB)²⁷ and health and media literacy concepts.^{28, 29} The primary intervention aim is centered on improving student’s individual-level SSB behaviors. Students received a 12-lesson curriculum as part of their health/physical education class. The secondary intervention aim focused on improving the caregiver’s SSB behaviors, SSB role modeling, SSB parenting strategies, and SSB home environment. To accomplish this, enrolled caregivers participated in a short message service (SMS) intervention, which included about 2 SMS messages per week. Complete intervention and study protocol details are presented elsewhere.²⁶ This study was approved by the University of Virginia Institutional Review Board.

Study Sample and Recruitment

This study utilizes baseline data from the 11 cohorts enrolled during the first 2 years of the Kids SIPsmartER trial.²⁶ These cohorts come from 8 middle schools located in 4 rural Appalachian counties in southwest Virginia. These counties are predominantly rural³⁰ and score “low or very poor” on the Health Opportunity Index.³¹

To be eligible to enroll in the SMS component, an adult had to be the legal caregiver of a 7^th grade student from 1 of the schools participating in the Kids SIPsmartER trial. One caregiver per student was allowed to enroll. Only caregivers with complete data were included in this secondary analysis.

Caregivers and their 7^th grade student were recruited using a study flyer and an informational letter signed by their school’s principal. Schools also utilized tailored recruitment strategies, such as presentations at back-to-school nights and direct calls to caregivers who had not returned consent forms. Caregivers completed written or verbal informed consent. They received a $10 gift card for completing the baseline survey.

Data Collection

Following consent, caregiver paper surveys were sent home with the 7^th grade student or mailed to their home. A handout was included to help caregivers accurately complete the SSB-related measures. Using a standardized script and contact protocol, research staff called or texted caregivers who had not returned their survey.

Measures

All survey measures were validated and/or previously used in this population.

SSB behaviors.

Caregiver SSB behaviors were assessed using the validated Beverage Intake Questionnaire (BEVQ-15).¹ The BEVQ-15 measures frequency and amount for 5 SSB categories: regular soft drinks, sweetened juice drinks, energy/sports drinks, sweetened tea, and sweetened coffee. Total SSB ounces were calculated using standardized scoring procedures.¹

Demographic characteristics.

Eight demographic variables, which have previously been associated with adult SSB intake, were self-reported. These variables were: (1) sex: male or female; (2) age: continuous variable, (3) race: American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, White or Caucasian, or Other (specify); (4) ethnicity: Hispanic, non-Hispanic, or not sure; (5) household income: 14 categories ranging from $5,000 to $100,000 or more; (6) enrollment in the Supplemental Nutrition Assistance Program (SNAP): yes or no; (7) highest level of educational attainment: 7 categories from grades 0–8 to graduate degree; and (8) marital status: married, divorced, widowed, separated, never married, unmarried, or living with romantic partner.

Theory of Planned Behavior constructs.

Five constructs from the TPB were measured using a validated instrument.^{13, 32} Caregivers were asked about their behavioral intentions (2 items), affective attitude (i.e., emotional judgment of the behavior, 1 item), instrumental attitude (i.e., benefits/costs associated with a behavior, 1 item), subjective norms (1 item), and perceived behavioral control (1 item). Each construct was asked in context of consuming 1 cup or less of SSBs per day. Items were assessed on a 7-point Likert agreement scale.

Health literacy.

Caregiver subjective health literacy was measured using 3 single item measures.^33–35 Each item was assessed on a 5-point Likert scale.

Media literacy.

A shortened version of the validated Sugar Sweetened Beverage Media Literacy (SSB-ML) survey was used to assess media literacy.¹⁵ Items were measured on a 7-point Likert scale (1=strongly disagree to 7=strongly agree).

Perceived adolescent SSB intake.

A single item assessed caregiver perception of the whether their adolescent frequently consumes high amounts of SSB.³⁶ Responses were on a 7-point Likert frequency scale (1= strongly agree to 7 = strongly disagree).

Home environment.

Home availability was assessed for each of the 5 SSBs measured on the BEV-Q: regular soft drinks, sweetened juice drinks, energy/sports drinks, sweetened tea, and sweetened coffee.^{1, 37} Responses were on a 5-point Likert scale (1=never to 5=always).

Analysis

Descriptive statistics and ANOVAs were conducted using SPSS Version 26.0 (SPSS Inc, Chicago, IL 2020). Stata 16.0 (Stata Corp, College Station, TX, 2019) was used to conduct the regression models.

Three demographic variables were transformed for analyses. Race and ethnicity were combined and collapsed into 2 categories: non-Hispanic White and not non-Hispanic White. Household income was transformed into a continuous variable. Marital status was collapsed into 2 categories: married/living with a partner or single/not living with a partner.

Four constructs had multiple items and were assessed as scales: behavioral intentions, health literacy, media literacy, and home environment. For each construct, the scale was calculated as a mean score of the associated items.^{15, 32, 36}

Bivariate relationships between the dependent variable (SSB ounces) and independent variables were assessed. The intention of the bivariate analysis was to describe differences in SSB intake by categories within the variables, not to test the variables. Therefore, all variables were broken into categories so that simple ANOVAs could be used. For the 4 scaled variables, mean values were rounded to the nearest appropriate Likert category (e.g., on a 7-point Likert scale 4–4.99 = neither agree nor disagree) for inclusion into the descriptive ANOVAs. Statistically significant post hoc relationships were determined using Tukey’s test at P ≤ 0.05.

The independent variables were entered into a modified 2-part model in sequential steps to assess their influence on caregiver SSB intake.³⁸ There were 4 steps, reflecting the levels of the SEM – Step 1: personal level: demographics, Step 2: personal level: psychosocial and health literacy factors (i.e., TPB, health literacy, media literacy variables), Step 3: intrapersonal level (perceived adolescent intake), and Step 4: environmental level (home environment). Variables were entered into the model based on the proximity of their hypothesized influence on caregiver SSB intake and following the conceptual model of the Kids SIPsmartER intervention.²⁶ The means of the 4 scaled items were treated as continuous variables in the sequential regression.³⁹

A 2-part regression model was used due to the sizable number of participants reporting zero ounces of SSB intake. These true zeros, as opposed to missing or censored data, cause significant skewness to the SSB outcome distribution. The modified 2-part model generalizes the Tobit model to analyze data with true zeros and is more robust to distribution assumptions.⁴⁰ The 2 parts of the estimated model were (1) a probit model that handled the nonlinear process of generating consume versus not-consume decisions and (2) a log-link generalized model that estimated the nonzero continuous SSB consumption. Beta weights (B) and standard errors (SE) are reported, with SE adjusted to be school-year cohort robust. Akaike Information Criterion (AIC) and Bayesian Information Criteria (BIC) were calculated to assess model fit.

Results

Participants

Of the 1360 caregivers of 7^th grade students in the participating schools, 485 (36%) enrolled in the caregiver SMS component of Kids SIPsmartER. For this secondary analysis, data from 362 caregivers with complete data were included. Table 1 presents the caregiver demographics, along with the tested bivariate comparisons. The sample was predominantly female (91%), 33 to 44 years old (66%), non-Hispanic White (96%), and lived in households that include a spouse or partner (73%). Importantly, 25% lived in households with a yearly income less than $25,000, 29% of caregivers had a high school degree or less, and, 20% received SNAP. On average, caregivers consumed 25.7 (SD=33.2) ounces of SSB per day, whereas the median intake was 15.2 ounces.

Table 1:

Simple ANOVAs of the Bivariate Associations between Caregiver Daily Sugar Sweetened Beverage (SSB) Intake and Demographic, Personal-, Interpersonal-, and Environmental-level Variables (n=362).

Variable		N	(%)	Daily SSB intake (ounces)		F Statistic (p-value)
				Mean†	(SD)
Demographic
Sex	Male	31	(8%)	38.6	(52.2)	5.2 (0.02)
	Female	331	(92%)	24.4	(30.7)
Age	≤ 32 years old	24	(7%)	26.9	(32.6)	0.02 (0.98)
	33 – 44 years old	240	(66%)	25.5	(30.8)
	≥ 45 years old	98	(27%)	25.8	(38.9)
Race and Ethnicity	Non-Hispanic White	348	(96%)	25.9	(33.5)	0.2 (0.65)
	Other racial/ethnic background	14	(4%)	20.4	(27.3)
Household Income	< $25,000	93	(26%)	40.9a	(44.9)	12.2 (<0.001)
	$25,000 – $ 49,999	69	(19%)	26.8b	(29.9)
	$50,000 – $74,999	84	(23%)	23.5b,c	(29.2)
	≥ $75,000	116	(32%)	14.3b,c	(19.7)
Educational Attainment	High School, GED§ or less	106	(29%)	35.8 a	(38.4)	12.4 (<0.001)
	Some college/associate degree	145	(40%)	27.0a	(34.1)
	4-year college degree or higher	111	(31%)	14.2b	(21.4)
Marital Status	Married/Live-in partner	261	(72%)	21.8	(25.3)	13.2 (<0.001)
	Not Married/No live-in partner	101	(28%)	35.7	(46.7)
Receiving Supplemental Nutrition Assistance (SNAP) benefits	No	287	(79%)	20.5	(26.6)	37.0 (<0.001)
	Yes	75	(21%)	45.5	(46.4)
Personal-level
Caregiver Behavioral Intentions	Extremely negative	53	(15%)	52.0a	(42.3)	17.8 (<0.001)
	Quite negative	54	(15%)	36.2a,b	(25.1)
	Slightly negative	31	(9%)	43.9a,c	(52.0)
	Neither	43	(12%)	20.6b,d	(19.5)
	Slightly positive	31	(9%)	22.5b,c,d	(35.3)
	Quite positive	62	(17%)	17.6d	(24.8)
	Extremely positive	88	(24%)	6.1d	(9.9)
Caregiver Affective Attitudes	Extremely unenjoyable	26	(7%)	33.4a,b,c	(39.0)	6.0 (<0.001)
	Quite unenjoyable	45	(12%)	45.0a,c	(36.5)
	Slightly unenjoyable	55	(15%)	31.8a,c	(25.9)
	Neither	103	(28%)	23.2b,c	(31.1)
	Slightly enjoyable	37	(10%)	26.7a,b,c	(48.1)
	Quite enjoyable	38	(10%)	15.9b,c	(27.2)
	Extremely enjoyable	58	(16%)	11.4b	(18.3)
Caregiver Instrumental Attitudes	Extremely unhealthy	16	(4%)	34.1	(55.9)	1.0 (0.42)
	Quite unhealthy	14	(4%)	22.0	(28.3)
	Slightly unhealthy	19	(5%)	29.2	(25.0)
	Neither	32	(9%)	20.0	(20.8)
	Slightly healthy	43	(12%)	31.8	(45.7)
	Quite healthy	110	(30%)	21.4	(24.9)
	Extremely healthy	128	(35%)	27.5	(34.9)
Caregiver Subjective Norms	Strongly disagree	40	(11%)	38.2	(56.6)	1.4 (0.19)
	Moderately disagree	20	(6%)	21.6	(16.6)
	Slightly disagree	20	(6%)	15.3	(18.0)
	Neither	135	(37%)	24.5	(29.5)
	Slightly agree	28	(8%)	29.2	(31.9)
	Moderately agree	38	(10%)	23.6	(23.5)
	Strongly agree	81	(22%)	24.8	(33.1)
Caregiver Perceived Behavioral Control	Strongly disagree	29	(8%)	30.6a,b	(52.4)	3.5 (0.002)
	Moderately disagree	14	(4%)	30.8a,b	(20.3)
	Slightly disagree	21	(6%)	42.3a	(39.2)
	Neither	22	(6%)	30.8a,b	(34.2)
	Slightly agree	41	(11%)	30.2a,b	(24.4)
	Moderately agree	41	(11%)	32.4a,b	(34.2)
	Strongly agree	194	(54%)	18.9b	(29.5)
Health Literacy	Okay	13	(4%)	37.7	(46.8)	1.0 (0.36)
	Good	65	(18%)	27.1	(29.6)
	Excellent or very good	284	(78%)	24.8	(33.3)
Media Literacy	Strongly/moderately disagree‡	3	(1%)	40.5	(49.6)	1.2 (0.29)
	Slightly disagree	13	(4%)	32.0	(26.6)
	Neither	53	(15%)	29.0	(27.2)
	Slightly agree	95	(26%)	30.4	(37.6)
	Moderately agree	117	(22%)	21.4	(28.8)
	Strongly agree	81	(23%)	22.5	(33.2)
Interpersonal-level
Perceived Frequency of High Adolescent Intake	Strongly agree	56	(15%)	35.8a	(39.4)	2.5 (0.02)
	Moderately agree	40	(11%)	22.3a,b	(19.6)
	Slightly agree	71	(20%)	27.9a,b	(28.9)
	Neither disagree	23	(6%)	35.9a,b	(46.7)
	Slightly disagree	47	(13%)	26.4a,b	(42.8)
	Moderately disagree	58	(16%)	15.3b	(20.1)
	Strongly disagree	67	(19%)	21.7a,b	(32.0)
Environmental-level
Home Availability of SSB	Never	40	(11%)	6.6a	(8.2)	14.5 (<0.001)
	Rarely	156	(43%)	18.8a	(26.1)
	Sometimes	141	(39%)	35.7b	(39.6)
	Often/Always‡	25	(7%)	42.5b	(33.9)

^†Post hoc analyses conducted using Tukey method. Values that do not share the same superscript letter (a,b,c) are significantly different (P<0.05).

^‡Categories were combined due to the extreme low/ high category (strongly disagree, always) having 1 response

^§GED = Test of General Education Development

Bivariate analyses

There were significant associations between 5 demographic variables and SSB intake, with SSB intake higher among caregivers who were male (P=0.02), had lower household income (P=<0.001), had lower educational attainment (P<0.001), were single/not living with a partner (P<0.001), and received SNAP (P=<0.001). There were no significant relationships between daily SSB intake and age or race/ethnicity.

At the personal level, there were significant associations between 3 TPB variables and daily SSB intake. Caregivers had higher intake if they held weaker behavioral intentions (P<0.001), lower affective attitudes (P<0.001), and weaker perceived behavioral control (P=0.003) related to consuming 1 cup or less of SSBs per day. Instrumental attitudes, subjective norms, health literacy, and media literacy were not significantly related to SSB intake.

At the interpersonal level, caregiver intake was higher among those who perceived their adolescent consumed SSBs more frequently (P=0.02). Lastly, at the environmental level, caregiver SSB intake was higher among those who reported SSBs were more frequently available in their home (P<0.001).

Two-part step-wise regression models

The final regression model included all demographic, personal-level, interpersonal-level, and environmental-level variables. Each model showed increasing fit based on decreasing Akaike Information Criterion (AIC) and Bayesian Information Criteria (BIC) values. This signals that the added groups of variables are statistically meaningful in terms of explaining the data.

Age, receiving SNAP benefits, behavioral intentions, affective attitudes, perceptions of their adolescent’s SSB intake, and home availability of SSBs each contributed to the model. For each year increase in caregiver age, a caregiver consumed 0.4 ounces fewer SSBs (P<0.05). Caregivers who received SNAP benefits consumed 14.2 ounces more than those who did not (P≤0.01). For each point higher a caregiver rated their behavioral intentions, they consumed 5.5 ounces fewer sugary drinks (P≤0.001). Similarly, for each point higher a caregiver rated their enjoyment (affective attitude) towards drinking fewer SSBs, they consumed 2.2 fewer ounces per day (P<0.05). For each point less a caregiver perceived the frequency of their adolescent’s SSB intake, they consumed 1.9 fewer ounces of SSBs (P≤0.001). For each point higher they rated the frequency in which SSBs were available in their home, caregivers consumed about 7.4 more ounces of SSBs (P≤0.01).

Sex, household income, and receiving SNAP benefits each significantly contributed to the Step 1 of the model, which only included demographic variables. Educational attainment, receiving SNAP benefits, behavioral intentions, and affective attitudes all significantly contributed to Step 2, which included demographic and personal-level variables. In Step 3, which included demographic, personal-level, and interpersonal-level variables, SSB intake was significantly influenced by age, receiving SNAP benefits, behavioral intentions, affective attitudes, subjective norms, and perceptions of adolescent intake. Variables associated with race, marital status, instrumental attitudes, perceived behavioral control, health literacy, and media literacy did not significantly contribute to any of the models.

Discussion

Caregivers of 7^th grade students from rural Appalachian schools participating in the Kids SIPsmartER trial consumed SSBs at levels 3 times greater than the recommended intake of less than 8 ounces per day for adults. This intake level mirrors findings from other studies in the Appalachian region,^41–43 This high intake level underscores the importance of understanding factors that influence caregivers’ SSB intake. Findings from this study identify demographic, personal-level, interpersonal-level, and environmental level influences of SSB intake among caregivers of adolescents living in rural Appalachia. Importantly, this is the only known study to explore 4 levels of influence on adult SSB intake across the SEM in a single model. Study outcomes have implications for interventions for caregiver and caregiver-adolescent dyads.

In the final model, 6 factors that significantly influenced caregiver SSB intake were identified across the levels of SEM: age, receiving SNAP benefits, behavioral intentions, affective attitudes, perceived frequent high adolescent SSB intake, and home availability. These findings mirror findings of facilitators of SSB intake among adults in previous studies; however, the majority of these studies only included behavioral influences at one level. Lower age has been identified as being associated with higher SSB intake in national level cross-sectional studies assessing SSB intake by demographic characteristics (e.g., Rosinger et al)¹⁶ and by personal level factors, while controlling for demographic factors (e.g., Park, et al).¹⁷ Studies conducted by Nguyen and colleagues¹⁸ and Bleich and colleagues⁴⁴ have identified that SSB intake is higher among SNAP recipients compared to both SNAP-eligible nonparticipants and SNAP-ineligible participants. While both these studies controlled for other demographic variables, neither included other personal-, interpersonal-, or environmental-level factors in the models. Regarding behavioral intentions, Zoellner and colleagues found significant relationships between the behavioral intentions of Appalachian adults and their SSB intake in 2 separate studies.^13,31 In both, lower intentions related to reducing SSB intake were associated with higher SSB intake. Also, one of these studies identified a significance relationship between SSB intake and attitudes towards drinking SSBs, with more positive attitudes being associated with greater intake.³¹ However, neither of these trials include other influences of SSB intake as control variables or other independent variables in the model. While it is established that parental SSB intake influences their child’s SSB intake,⁴⁵ there is also emerging evidence caregiver perceptions of their child’s intake can impact their own intake.²¹ Finally, in relation to home availability, the majority of the literature on home availability and SSB intake is within the context of adolescent intake, not parental;^{10, 22, 23} however, evidence does suggest that parental parenting practices around food, including availability of SSBs and other unhealthy foods in the home, is associated with their own intake.⁹

Of the 6 factors that were statistically significant in the final model, 3 emerged as the leading explanatory variables: receiving SNAP benefits (B=14.2), behavioral intentions (B=−5.5), and home availability (B=7.4), suggesting that these factors may be important to address in future interventions targeting SSB intake of caregivers of adolescents. For example, personal-level intervention components for caregivers could engage enrolled caregivers in goal setting activities to promote improvement in their behavioral intentions to decrease SSB intake. Furthermore, findings specific to the higher intake of SSBs among caregivers receiving SNAP add to the debate about whether SSBs should be among the food items eligible to be purchased with SNAP benefits.⁵ Specifically, study findings highlight that SNAP participants may purchase more SSBs which could lead to, rather than prevent or alleviate, chronic diseases, such as obesity and diabetes, which is contrary to the program’s purpose of supplementing food budgets so more healthy foods can be purchased.⁴⁶

Several factors with hypothesized relationships to SSB (i.e., race, marital status, instrumental attitudes, perceived behavioral control, health literacy, and media literacy) were not confirmed in the final model. This may suggest a less influential relationship among these factors and SSB when simultaneously considering other factors, such as age, receiving SNAP benefits, behavioral intentions, affective attitudes, perceived frequent high adolescent SSB intake, and home availability. However, the lack of significance for health and media literacy may be explained by the distribution of the responses, as responses for these variables skewed towards positive ratings of skill level.

Limitations

This study has several limitations that should be taken into account when interpreting the findings. First, all caregiver data was self-reported. However, all the self-reported measures were validated and/or had been previously used in the region. Second, this study was cross-sectional, thus cause and effect cannot be determined. Third, readers should consider the impact of the representativeness of the study sample on generalizability of the findings. The sample included about 25% of eligible caregivers, and male caregivers are under-represented in the study. However, the diverse range of SES is representative of the region. Also, a higher representation of females is consistent with literature indicating that females are frequently the primary food gatekeepers within a household and much of rural Appalachia.⁴⁷ While the study sample lacks racial/ethnic diversity, which limits generalizability to other communities, the racial/ethnic breakdown is representative of Central Appalachia.⁴⁷ Fourth, the small sample of males and non-White participants in our sample may have also limited the power of the model to capture significant associations in the models. Importantly, these limitations are balanced by the study strengths: strong theoretical approach, use of previously validated questionnaires, and an adequate sample size.

Table 2:

The Modified Two-Part Model (1^st Part: Probit Model, 2^nd Part: Log-Link Generalized Model) of the Influence of Demographic, Personal-level, Interpersonal-Level, and Environmental-Level Factors on Caregiver Sugar Sweetened Beverage (SSB) Intake (n=362)

Variable	Step 1	Step 2	Step 3	Step 4
Akaike Information Criterion (AIC)	8.7	8.5	8.5	8.5
Bayesian Information Criteria (BIC)	−1414.8	−1469.7	−1478.0	−1484.7
	B (SE)	B (SE)	B (SE)	B (SE)
Demographics
Femalea	−12.6 (6.4)*	−11.0 (6.7)	−14.1 (7.7)	−10.9 (6.7)
Ageb	−0.4 (0.2)	−0.3 (0.2)	−0.5 (0.2)*	−0.4 (0.2)*
Non-Hispanic Whitea	9.3 (5.7)	7.4 (5.4)	7.3 (5.7)	4.5 (7.4)
Household Incomec	−1.2 (0.6)*	0.1 (0.6)	−0.0 (0.6)	0.1 (0.6)
Educational attainmentd	−3.1 (1.7)	−2.6 (1.3)*	−2.4 (1.4)	−2.3 (1.4)
Married/lives with partnera	−5.7 (4.4)	−4.3 (3.8)	−3.7 (3.6)	−5.8 (3.8)
Receiving SNAP benefitsa	8.0 (3.5)*	16.9 (6.1)**	15.4 (5.9)**	14.2 (5.1)**
Personal-level
Behavioral intentionse		−5.8 (0.7)***	−6.0 (0.6)***	−5.5 (0.8)***
Affective attitudesf		−3.0 (0.9)**	−2.3 (1.0)*	−2.2 (1.0)*
Instrumental attitudesg		0.8 (0.8)	0.4 (0.9)	0.6 (0.8)
Subjective normsh		1.2 (0.7)	1.3* (0.7)	1.2 (0.6)
Perceived behavioral controlh		−0.2 (1.1)	−0.6 (1.1)	−0.3 (1.0)
Health literacyi		5.0 (4.5)	5.3 (4.7)	5.6 (4.7)
Media literacyh		−1.5 (1.8)	−1.8 (1.7)	−1.7 (1.5)
Interpersonal-level
Perceived high adolescent SSB intakeh			−2.5 (0.7)***	−1.9 (0.6)***
Environmental-level
Home availabilityj				7.4 (2.6)**

^*P = ≤0.05

^**P = ≤0.01

^***P = ≤0.001

^a1= yes, 0 = no;

^bcontinuous variable

^c14-point scale, 1 = <$5,000 to 14 = ≥ $100,000

^d7-point scale, 1 = grade 0–8 to 7 = graduate degree

^{e =}7-point scale, 1=extreme negative; 7=extreme positive

^{f =}7-point scale, 1=extreme unenjoyable; 7=extreme enjoyable

^{g =}7-point scale, 1=extreme unhealthy; 7=extreme healthy

^{h =}7-point scale, 1=strongly disagree; 7=strongly agree

^{i =}7-point scale, 1=terrible/very poor; 5=excellent

^{j =}5-point scale, 1=never; 5=always

Implications for Research and Practice.

Study findings provide insight into demographic, personal-level, interpersonal-level, and environmental level influences of SSB intake among caregivers of adolescents living in rural Appalachia that could be used to inform the design of future SSB focused interventions for caregivers. First, the results highlight the need for interventions for caregivers of adolescents that target behavioral intentions and build healthier home environments by reducing SSB availability within the home. Second, findings suggest that SSB-focused nutrition education may be of particular benefit for SNAP recipients, as findings show caregivers who receive SNAP benefits consume more SSBs than caregivers who did not. This emphasizes the need for SNAP-ED and the Expanded Food and Nutrition Education Program (EFNEP) to prioritize behaviorally-focused nutrition education related to SSB intake among their participants receiving SNAP. Lastly, as this study was conducted with rural Appalachian caregivers, future studies are needed to assess influences of SSB intake among caregivers and other adults across multiple levels of the social ecological model in other populations.