Reinforcing value of food, enriched home environment, and changes in percent overweight in children

Abstract

Objective:

The decision to eat is often a choice made in the context of food and non-food alternatives. However, no research, to the authors’ knowledge, has assessed the combination of the motivation to eat, as indexed by the relative reinforcing value of food (RRV_FOOD), and the enriched home environment, i.e., access to activities that can serve as alternatives to eating on weight gain.

Methods:

This study used a cross-sectional design to study how RRV_FOOD and the enriched home environment predict percent overweight change over 2 years in 291 children aged 6 to 9 years and of varying socioeconomic status.

Results:

Results showed that RRV_FOOD and access to food were positively associated with baseline percent overweight, and an enriched home environment was negatively related to baseline percent overweight. RRV_FOOD and an enriched home environment interacted to predict change in percent overweight. Children with a high relative RRV_FOOD and a relatively non-enriched environment showed the greatest relative weight gain.

Conclusions:

These results suggest that providing an enriched home environment may reduce the effects of food reinforcement and being motivated to eat on weight gain in childhood, and this represents a novel approach to intervention that can be used to strengthen current behavioral approaches to prevent obesity in children.

INTRODUCTION

Energy intake often represents a choice between engaging in eating or non-food alternative behaviors. Although considering choices among foods is a prevalent concern for childhood obesity, choices between eating and alternative behaviors are also part of a child’s leisure time. Substitution describes how easily one behavior can replace a preferred, but less-accessible, behavior with a less-preferred alternative. Childhood obesity risk may be related to individual differences in decreased substitutability of food, as well as fewer available alternative activities and healthy foods in a child’s home environment.

Eating is driven in part by the relative reinforcing value of food (RRV_FOOD), i.e., motivation to eat. Reinforcing value is derived from basic behavioral theory and is measured by assessing how much work a person engages in for access to a reinforcer [1]. The more someone will work for access to a behavior, the more reinforcing that behavior. The behavioral measure of RRV_FOOD has been cross-sectionally and prospectively related to obesity status and weight change in infants [2, 3], children [4, 5], adolescents [6], and adults [7, 8]. Research has shown that relative measures of RRV_FOOD versus a non-food alternative, rather than absolute value of food, is a risk factor for obesity [1]. In other words, high RRV_FOOD is an obesity risk factor if there is also a corresponding low reinforcing value of substitutable behaviors. This concept suggests that individuals with high RRV_FOOD may be protected from developing obesity if they also have access to non-food alternatives that can substitute for eating [2, 7].

An enriched environment for children includes activities that are cognitively or emotionally enriching, such as availability of books, visits to museums, and parental support for hobbies. Two large epidemiological studies have shown that an enriched home environment (ENR_HOME) is a protective factor for childhood obesity [9, 10]. Strauss and Knight showed that low ENR_HOME predicted an increased risk of obesity 6 years later for 0- to 8-year-old children [9], independent from mother’s education and socioeconomic status (SES). East and colleagues showed that low ENR_HOME at age 8 years predicted increased risk of obesity at age 21 years [10].

One factor that may tie childhood obesity, ENR_HOME, and RRV_FOOD together is household SES [11, 12]. Parental education, as an index of SES [11], has been correlated with lower ENR_HOME [13] and greater RRV_FOOD [14], and it was shown to partially mediate the relationship between RRV_FOOD and obesity in adults [14]. It is possible that the greater risk of obesity for individuals in low SES households [11] may be partly due to differences in low levels of environmental enrichment and high levels of RRV_FOOD.

This study examined 2-year weight changes in a group of 291 children aged 6 to 9 years old. Measures of RRV_FOOD, home food availability, and ENR_HOME, including both cognitive/emotional enrichment and engagement in non-food alternative activities, were collected. The primary aims were to examine how RRV_FOOD and aspects of ENR_HOME were associated with baseline percent overweight and predicted 2-year changes in percent overweight in children. A secondary aim was to examine the relationship between household SES markers and child obesity and the possibility that RRV_FOOD or ENR_HOME mediates the relationship among indicators of SES and obesity.

METHODS

Participants

Children (6–9 years old, n = 291) were enrolled in a 2-year longitudinal study (flow diagram in Supporting Information Figure S1; study timeline in Table 1). The average baseline characteristics are shown in Table 2. For follow-up, 284 children (97.6%) completed 1-year measures, and 281 (96.6%) completed 2-year measures. Families were recruited from Buffalo, New York (February 2017 to April 2019) and the surrounding area using a database of previous families with interest in future studies, online advertisements (e.g., Facebook), tabling at local community/school events, and posted flyers/brochures in the community and school districts.

TABLE 1.

Timeline and measurements for baseline assessments

Baseline
Laboratory Sessions			Home Visit
Session 1	Session 2	Session 3
Week 1	Week 2	Week 3	Week 4
Consent/Assent
Height/weight	Preload snack	Preload snack	HOME inventory
Activity sampling	RRV task	RRV task	Home food inventory
Food and activity ratings	Questionnaires	Questionnaires

Abbreviations: RRV, relative reinforcing value; HOME, Home Observation for Measurement of the Environment.

TABLE 2.

Participant characteristics and correlations with child baseline percent overweight

	Mean (SD)	Child percent overweight
n
Federal assistancea	28.2%	−0.08
Sex (male/female)	135/156	−0.09
Minority status (non-White or Hispanic/White non-Hispanic)	87/203	0.10
Race/ethnicity breakdown
Hispanic/Non-Hispanic	26/265
White	218
Asian	6
Black/African American	33
Multiracial (including American India/Alaskan Native, Hawaiian/Pacific Islander, Asian, Black, White, other)	33
Refuse to answer	1
Age (y)	7.8 (1.2)	−0.05
BMI z score	0.435 (1.073)	0.92***
Percent overweightb	−3.75 (17.80)
Percent body fatc	19.78 (8.84)	0.89***
Obesity, >95th BMI percentile 16% (46)	16% (46)
Overweight, >85th BMI percentile, <95th BMI percentile	13% (39)
Incomec	$93,484 ($86,214)	−0.09
Parent education (y)	16.7 (2.4)	−0.16**
Parent BMI	30.2 (8.1)	0.31***
Physical Activity Questionnaire score (range 1 [low] to 5 [high])	2.8 (0.5)	0.01
Reinforcing value (range: 0–3840)
RRVFOOD (responses)	493.1 (456.4)	0.07
Reinforcing value of alternatives (responses)	428.8 (476.4)	−0.01
RRVFOOD (range: 0– 1)	0.55 (0.22)	0.12*
HOME
ENRHOME, enrichment, provision for active stimulation (range: 0–8)	5.88 (1.34)	−0.16**
Learning materials and opportunities (range: 0–8)	5.16 (1.29)	−0.04
Encouragement of maturity (range: 0–7)	5.38 (1.20)	0.03
Responsiveness (range: 0–10)	9.02 (1.42)	−0.01
Emotional climate (range: 0–8)	6.02 (1.08)	−0.04
Family participation in stimulating experiences (range: 0–6)	5.05 (0.83)	−0.07
Physical environment (range: 0–8)	7.39 (1.06)	−0.02
Family integration (range: 0–4)	2.73 (1.15)	−0.14*
Children’s activities engagement and preference
Engagement (range: 0–55)	34.8 (6.0)	−0.15**
Frequency (range: 0–110)	41.0 (8.0)	−0.07
Home food inventory
Availability of food (range: 0–190)	60.3 (15.9)	0.03
Total access to food (range: 0–36)	13.6 (5.0)	0.12*

Abbreviations: ENR_HOME, enriched home environment; HOME, Home Observation for Measurement of the Environment; RRV_FOOD, relative reinforcement of food.

^aFederal assistance at enrollment including Supplemental Nutrition Assistance Program, Special Supplemental Nutrition Program for Women, Infants, and Children, Temporary Assistance for Needy Families, Home Energy Assistance Program, housing vouchers, the Emergency Assistance Food Program, child care and development fund, and low-income home energy assistance.

^bPercent over the 85th BMI percentile.

^cn = 289.

^*p < 0.05;

^**p < 0.0148;

^***p < 0.001.

Eligible children were required to report a moderate liking (6/10 Likert-type scale) of the foods and activities offered in the study and to spend at least 65% of their time in the participating parent’s household. Exclusion criteria included the following: 1) dietary restrictions (food allergies, religious/ethnic practices) or medical conditions that could modify nutritional status and food absorption; 2) medical/physical activity restrictions; 3) psychopathology (e.g., attention-deficit/hyperactivity disorder), developmental disabilities, and/or taking medications that could affect activity or appetite levels (e.g., methylphenidate, daily corticosteroids); and (4) body mass index (BMI) z score > ±3.0. Siblings of enrolled participants were not eligible. To ensure variability in home environments and alternative activity access, 41.9% of the sample included families who were classified in a low education/income strata, as defined by household education (less than a college degree) or eligibility for government assistance (online Supporting Information Methods).

Measures

Anthropometrics

The participating parent and child had weight and body fat measured using a Tanita bioelectrical impedance digital scale (Tanita Corporation, Inc., Arlington Heights, Illinois) and height using a digital stadiometer (QuickMedical, Issaquah, Washington). Child BMI z score was calculated as a standardized z score of BMI (kilograms per meters squared) using Centers for Disease Control and Prevention (CDC) growth charts [15]. Percent overweight was calculated as percent over the 85th BMI percentile [16] using the following formula: (BMICHILD - 85th BMI percentile)/(85th BMI percentile) _ 100. Scores greater than zero indicate that the child is in the overweight category (BMI ≥ 85th percentile), whereas negative numbers would indicate that the child is under the 85th BMI percentile.

RRV_FOOD

The behavioral measure of RRV_FOOD, which followed previous procedures [5, 17], has been validated in children [5]. Briefly, this task provides discrete food and activity reinforcers that can be earned by completing button presses for each reinforcer desired. The reinforcement schedule is a progressive ratio schedule in which the response requirements (button presses) double for each subsequent reinforcer (e.g., 20 presses for the first reinforcer, 40 for the second, 80 for the third, etc.). This allows for a measure of individual differences in work completed for access to each reinforcer. Children had the option of earning 8-g portions of food (e.g., chocolate-chip cookies, chocolate wafer crème cookies, candy rope, cheese crackers, potato chips, flavored tortilla chips, pretzels, fruit snacks, chocolate candies, sugar candies, pepperoni, peanut butter crackers, cheese puffs) and 2 minutes of access to their activity alternative (e.g., jump rope, hula hoop, skipping toy, balance board, active video games, coloring, interlocking blocks, puzzles, dolls, toy cars, crafts, activity books, beanbag toss) during the task. The procedures consisted of two computer stations and two reinforcer stations between which children were able to freely choose. Each reinforcer was associated with one computer station for earning access and one reinforcer station for consumption (eating or playing). Food portions and cards indicating 2 minutes of activity were placed on the respective stations as soon as the child earned a reinforcer. Children were instructed that earning points and consuming food/activity time was their choice and that they could end the task when they did not want to earn additional reinforcers.

The reinforcing value breakpoint, i.e., the highest schedule completed, for food and alternative reinforcers was determined by averaging across both sessions. The reinforcing value breakpoint measures were significantly skewed and were log-transformed prior to analysis. RRV_FOOD was calculated using the following formula: (Breakpoint_FOOD)/([Breakpoint_FOOD + Breakpoint_ALTERNATIVE]).

Child engagement and frequency of activities

The Children’s Assessment of Participation and Enjoyment questionnaire (CAPE; Pearson, New York, New York), validated in children 6 to 21 years old [18], was used to assess children’s participation and enjoyment of alternative activities over the previous 4 months [19]. The questionnaire includes 55 activities in the following categories: recreational, physical, social, skill based, and self-improvement (e.g., puzzles, team sports, visiting, music lessons, writing a story, respectively). Parents completed an engagement dimension indicating whether the child participated in the activity over the previous 4 months (α = 0.80) and frequency of engagement (1–7 scale, 1x-1x per day or more; α = 0.80). Children completed enjoyment scales (1–5 scale, “not at all” to “love it”) for activities in which they participated. For each measurement dimension, summed scores were calculated for total activities [19].

Home Observation for Measurement of the Environment Inventory

The middle-childhood Home Observation for Measurement of the Environment (HOME) Inventory [20, 21] was used to measure cognitive and emotional enrichment in children’s homes (Home Inventory LLC, Madison, Wisconsin). The HOME Inventory consists of 59 questions and observations (scored 0/1 for negative/positive characteristics) made during a semi-structured interview conducted by a research assistant with the parent and child in their home. The items are divided into eight different subscales based on the content: learning materials and opportunities (e.g., availability of 10+ books); enrichment (ENR_HOME, e.g., parent encourages hobbies); family companionship (e.g., parent helped child learn gross motor skills); responsivity (e.g., parent answers child’s questions during visit); encouragement of maturity (e.g., parent requires child to complete simple household tasks); parental warmth (e.g., parent uses term of endearment for child during visit); family integration (e.g., child has contact with second parent/parental figure); and physical environment (e.g., house is reasonably clean; details in online Supporting Information Methods).

Home food inventory

The home food inventory [22] is a checklist of 190 typical foods used to assess the home availability of a range of foods. A research assistant and parent reviewed the list of foods, and a research assistant confirmed via cupboard survey. Food availability was defined as the food being present in the household, whereas accessibility was defined as the food being prepared/ready to eat and/or stored in a location accessible to children. Both food availability and accessibility were calculated for the total foods in the household.

Demographics and medical history

The MacArthur Questionnaire was used to assess household income levels, household size, and years of education for both parents [23, 24]. Household education was coded in years for the parent with the higher education level. Parents self-identified sex (0/1, male/female individuals), race, and ethnicity for themselves and their child. Minority status was coded as White/non-Hispanic and non-White or Hispanic (0/1).

Physical activity

The Physical Activity Questionnaire [25, 26] was used to assess children’s level of physical activity. Information was collected about physical activity level during a normal school day, weeknight, and weekend day, including the number of days children did different physical activities. Parents completed questions about recent physical activity, whereas children answered questions about physical activity at school and preferences for physical activity during leisure time. A physical activity score was calculated ranging from one (low physical activity) to five (high physical activity).

Procedures

Baseline appointments 1 through 3 and home visit

Eligible families were scheduled for three 90-minute laboratory visits and one 60-minute home visit for baseline measures (Table 1). Children were asked to refrain from eating 3 hours prior to laboratory appointments to standardize hunger. During the initial visit, parents and children signed consent and assent forms in which their responsibilities and study methods were explained, and parents completed a comprehension check about the basic study responsibilities and commitments. Parents and children had height, weight, and body fat measured, and children had the opportunity to sample all of the provided physical and sedentary activities. Children rated and ranked the foods and activities provided for the RRV task. To determine which foods and activities were presented as reinforcers in the behavioral task, children completed an experimenter-administered RRV questionnaire for their top five foods, sedentary activities, and physical activities offered (additional details in online Supporting Information Methods) [27]. Two snack foods were chosen for measurement in the behavioral RRV task, i.e., the most reinforcing and moderately reinforcing snack food, as well as one activity with the highest reinforcing value.

In laboratory sessions 2 and 3, children completed the behavioral laboratory measure of RRV_FOOD [5], with the order of favorite/moderate value foods counterbalanced. After completing hunger/fullness and food-liking scales, children were provided with a preload snack to ensure the behavioral task was not unduly influenced by hunger. Children completed a second set of hunger/fullness and food liking scales before starting the RRV_FOOD task. Children were provided instructions on how to earn and consume the foods and activities within the RRV_FOOD task and they completed a practice session. Children were allowed to choose how many reinforcers they earned and when to consume them. At the end of each session, final hunger/fullness and food-liking scales were collected, along with a same-day food recall, completed with the help of the parent, as a manipulation check for being 3 hours post prandial.

Parents completed demographic questionnaires, child medical history, parent food reinforcement questionnaires, child activity engagement [19], and physical activity questionnaires for their child in sessions 1 through 3. At the end of sessions 2 and 3, children completed the Physical Activity Questionnaire with the help of their parent [28] and enjoyment scales for their recent activities (CAPE).

The final data collection session for baseline was the home visit, attended by two research assistants. At the home visit, the research assistant conducted an open-ended interview using the HOME Inventory to assess and observe the home environment [20]. The parent and research assistant completed the home food inventory checklist to record the foods and drinks that the families had in their homes. After completion of appointments, participating families were compensated $100 ($70/$30 for child/parent) through checks or gift cards. Each family was put in a raffle (1-in-10 winners) for a chance to win an additional $100 check after completing baseline appointments.

Follow-up appointments

Children and one parent completed height and weight measures, in addition to several questionnaires (online Supporting Information Methods) at 1- and 2-year follow-up. Starting in March 2020, due to COVID-19, appointments were completed remotely. Children who completed remote follow-up appointments had portable height and weight equipment delivered to their home (Supporting Information Methods; see Supporting Information Tables S5 and S6 for validity and reliability of remote height and weight measures).

All procedures were conducted in accordance with the guidelines for the ethical conduct of human research and with approval of the University at Buffalo Institutional Review Board (identifier: 030–385420).

Analytic plan

ANOVA and χ2 tests were used to examine potential demographic differences in completers versus non-completers of follow-up. Pearson product moment correlations were used to establish associations between predictor variables and child percent overweight. False discovery rate (10%) was examined for multiple testing correction for the correlations [29]. Percent overweight was chosen as the main dependent measure of obesity because it is sensitive to changes in weight over time [16, 30].

Linear mixed-effect models were used to examine predictors of baseline percent overweight and weight change over 2 years. Associations with baseline were examined by setting the intercept to baseline. Covariates were mean-centered and included baseline child age, sex, minority status, and physical activity score, as well as years of household education and a variable indicating months of COVID-19 exposure. The best-fit model included both intercept and year as random effects using −2 log likelihood, as well as a visual examination of graphical representations (Q-Q plot and scatterplot) of the marginal and conditional residuals for normality and homoscedasticity. Covariates were added in the first step and examined for potential subgroup analyses if there was a significant effect of the covariate on percent overweight intercept. Household education was also examined for significant effects on slope to test the secondary aim.

To test the primary aim predicting percent overweight change, RRV_FOOD, ENR_HOME, and home food access were simultaneously entered into a model predicting intercept and slope. To examine moderation effects, the two- and three-way interactions were entered. Exploratory analyses were conducted on other independent variables that were significantly correlated with percent overweight, specifically family integration, and they are presented in online Supporting Information. Predictor variables were mean-centered, and coefficients presented are unstandardized betas (B). Simple slopes were used to examine the main effects of significant interactions by re-centering independent variables at ±1 standard deviation (SD).

Research examining RRV_FOOD as a predictor of 1-year relative weight changes in 7- to 10-year-old children showed an effect size of 0.185 [4]. This effect can be detected (α = 0.05, power = 80) with a sample size of 230 children, with an additional 60 children anticipating 25% attrition after 2 years.

RESULTS

No differences in sex, age, baseline percent overweight, or household education between completing families and non-completers for 1-year follow-up (n = 7) or 2-year follow-up (n = 10) were observed. Child sex was the only significant difference between families completing remote versus laboratory follow-up appointments (X² = 8.1, p = 0.004), with a higher percentage of boys completing remote measures.

Associations with baseline percent overweight

Bivariate correlations showed that several variables were significantly correlated with percent overweight at baseline (Table 2), including positive relationships among weight measures and parent BMI, child RRV_FOOD, and total child home food access. Parent education, HOME cognitive enrichment (ENR_HOME), HOME family integration, and total activity engagement were negatively related to percent overweight. When accounting for multiple comparisons, false discovery rate cutoffs (p < 0.0148) excluded the correlations for percent overweight with RRV_FOOD and home food access. RRV_FOOD and home environment were hypothesized as having a relationship with percent overweight changes and were examined in the mixed-model analyses.

Percent overweight was negatively skewed, and regression analyses were conducted with the log-transformed variable. The first model included covariates and the effect of household education on intercept and slope (represented by year). There was not a significant effect of household education on weight change. Given this, mediation analyses were not conducted, and subsequent models did not include household education x year.

ENR_HOME (F[1, 279] = 4.25), RRV_FOOD (F[1, 279] = 4.54), and home food access (F[1, 279] = 7.14) were significantly related to baseline percent overweight (Table 3), but no interactions were observed. Based on the multiple regression, the additive effects of RRV_FOOD, ENR_HOME, and access to food on child percent overweight are shown in Figure 1. Examining models with total activity engagement as the ENR_HOME variable showed the same pattern of significant additive baseline effects (total activity engagement: B = −0.004, p = 0.048; Supporting Information Table S1) in models with home food access and RRV_FOOD.

TABLE 3.

Predictors of baseline and change in percent overweight in children

	Baseline percent overweight (intercept)		Change in percent overweight (slope)
	Estimate (95% CI)	p value	Estimate (95% CI)	p value
Covariates
Intercept	4.386 (4.358 to 4.413)	<0.0001	N/A	N/A
Year	0.008 (0.003 to 0.014)	0.003	N/A	N/A
Child age at baseline	0.012 (−0.030 to 0.006)	0.175	N/A	N/A
Child sex	−0.035 (−0.079 to 0.008)	0.111	N/A	N/A
Child minority status	0.046 (_0.001 to 0.093)	0.055	N/A	N/A
Household educationa	−0.013 (−0.022 to −0.004)	0.007	−0.002 (−0.004 to 0.0002)	0.073
Physical activity score	0.001 (_0.039 to 0.040)	0.978	N/A	N/A
COVID exposure (mo)	−0.004 (−0.009 to 0.002)	0.197	N/A	N/A
Model 1
RRVFOOD	0.110 (0.008 to 0.211)	0.034	0.017 (−0.008 to 0.042)	0.172
ENRHOME	−0.019 (−0.036 to −0.001)	0.040	−0.003 (−0.007 to 0.001)	0.115
HFA	0.006 (0.002 to 0.011)	0.008	0.0002 (−0.001 to 0.001)	0.742
Model 2
RRVFOOD x ENRHOME	0.014 (_0.061 to 0.089)	0.718	0.018 (0.0002 to 0.037)	0.048
RRVFOOD x HFA	−0.003 (−0.024 to 0.019)	0.806	0.003 (−0.003 to 0.008)	0.340
ENRHOME x HFA	0.001 (−0.002 to 0.004)	0.542	−0.001 (−0.001 to 0.0002)	0.110
Model 3
RRVFOOD x ENRHOME x HFA	0.013 (−0.003 to 0.029)	0.108	0.001 (−0.003 to 0.005)	0.555

Note: Bolded numbers indicate p < 0.05.

Abbreviations: ENR_HOME, enriched home environment; HFA, home food access; N/A, not applicable; RRV_FOOD, relative reinforcing value of food.

^aHousehold education x year was not included in subsequent models.

Figure 1.

The additive effects of ENR_HOME, food access, and RRV_FOOD (±1 SD) on percent overweight, based on the regression model. Each independent variable was calculated at +/− 1 SD for each combination of risk factors. The zero line represents the 85th BMI percentile, indicating the definition of childhood obesity. ENR_HOME, enriched home environment; RRV_FOOD, relative reinforcing value of food

Predicting change in percent overweight

There was a significant RRV_FOOD × ENR_HOME × year interaction (F[1, 279] = 3.96, p = 0.048; Table 3), showing that children with low RRV_FOOD and high ENR_HOME (both hypothesized protective factors) were more likely to remain stable or decrease percent overweight across 2 years (Figure 2). Children with both risk factors, high RRV_FOOD and low ENR_HOME, were more likely to increase percent overweight over 2 years and to develop obesity. Simple slopes analysis (Supporting Information Table S4) showed, for children with high ENR_HOME, that there was a significant RRV_FOOD x year effect on percent overweight change (p = 0.017) that was not seen with low ENR_HOME (p = 0.76). Children with low RRV_FOOD had a significant ENR_HOME x year effect (p = 0.011), but not children with high RRV_FOOD (p = 0.80). Children with both protective factors (high ENR_HOME and low RRV_FOOD) showed a nonsignificant effect of year (B = −0.006, p = 0.29), suggesting that their percent overweight did not significantly increase over 2 years.

Figure 2.

The interaction between ENR_HOME and RRV_FOOD on 2-year percent overweight change in children. The zero line indicates the 85th BMI percentile, indicating the definition of childhood obesity. ENR_HOME, enriched home environment; RRV_FOOD, relative reinforcing value of food.

Children with at least one risk factor showed a significant increase in percent overweight across 2 years, with similar slopes. There were no significant main effects (by year) of RRV_FOOD, ENR_HOME, home food access (Table 3), total activity engagement (Supporting Information Table S1), or family integration (Supporting Information Tables S2 and S3) on percent overweight change.

DISCUSSION

Our primary aim was to examine cross-sectional and prospective relationships among RRV_FOOD, ENR_HOME, and child percent overweight. High RRV_FOOD, low ENR_HOME, and high home food access were independently correlated with baseline child percent overweight. The baseline results support previous research on the relationship between high RRV_FOOD and child weight [4, 5]; cognitively enriched home environment and child weight outcomes [9, 10]; and home food availability and weight status [31]. Previous research has not studied these factors together. Home food availability and access often have been studied in the context of the availability of physical and sedentary activities, particularly complements to eating such as screen availability [31–34]. Alternative substitute reinforcers, including cognitively enriching activities such as reading, have not been considered in the same context.

Research has shown effects of both RRV_FOOD and ENR_HOME on weight change over time in children. In this sample, there was a significant interaction between RRV_FOOD and ENR_HOME. Children with both protective factors, i.e., low RRV_FOOD and high ENR_HOME, showed stable percent overweight, being an average 10% under the 85^th BMI percentile, whereas children having one or two risk factors (high RRV_FOOD, low ENR_HOME) showed increases in percent overweight across 2 years, crossing over the 85th BMI percentile. Previous research examining the moderation between RRV_FOOD and ENR_HOME has focused on the protective effects of ENR_HOME for children in a weight-loss treatment program [35]. Children with low RRV_FOOD benefited from high ENR_HOME, losing more weight over 16 weeks than those with lower ENR_HOME.

Behavioral economic theory posits that accessibility influences choice: if two choices are equally reinforcing, people will choose the more-accessible option [36]. When reinforcers are not equally valued or accessible, choice depends on a variety of factors, including the reinforcer quality, access, and availability of alternatives, which contribute to the substitutability among reinforcers. If one finds food highly reinforcing, then non-food alternatives will need to be high quality, be more accessible, or include more variety to encourage switching among reinforcers. A home with ENR_HOME can be considered one in which a variety of high-quality non-food reinforcers are available. This provides the opportunity for children to engage in behaviors other than eating or behavioral substitutes for eating. Our results suggest that the presence of multiple risk factors may increase the likelihood of developing obesity because RRV_FOOD, ENR_HOME, and home food access have additive effects on child percent overweight. Children with greater RRV_FOOD are at risk for increased consumption and they may be doubly so when they are offered fewer enriching alternatives or offered complementary activities, rather than substitutes, to eating.

One approach to decreasing child energy intake is identifying non-food reinforcers to compete with eating to reduce snack consumption or provide motivation to limit large meals. Laboratory research has shown that non-food alternatives can substitute for food in children [17, 37]. Differential access to food can moderate reinforcing value of snack food, as prior access to a healthy snack was associated with lower reinforcing value of a less-healthy snack [38]. In clinical studies, greater reinforcing value of non-food alternatives has been related to better weight-loss outcomes in both child [39] and adult obesity treatment programs [40], and ENR_HOME also led to better weight-loss outcomes for children [35]. In our sample, RRV_FOOD was related to baseline percent overweight, suggesting that treatments changing either RRV_FOOD or alternatives may improve relative weight gain in children. These studies also indicate that alternative reinforcers to food are an important aspect of obesity risk and that they may provide new avenues for developing effective childhood obesity treatment and prevention programs.

The present study focused on the number of cognitively enriching activities, with the assumption that children with a large number of potential activities will find at least one behavior that can substitute or compete with food. A better approach might be to assess the reinforcing value of alternatives and to selectively arrange the environment to improve access to the most reinforcing, high-quality alternative reinforcers. Research with infants showed a brief, 6-week program to strengthen alternatives to food can reduce RRV_FOOD [41]. These data suggest that building alternative reinforcers may be a way to modify obesity risk, but experimental research is needed with older children over longer time periods to understand the optimal dose needed to alter relative body weight.

This study included children from a range of socioeconomic backgrounds and had strong participant retention, but there were some limitations. Limited families from the lowest levels of education and income were included. These were also younger children, and effects of RRV_FOOD and enriched environment may be different for infants/toddlers or for adolescents. The activities offered during the measurement of RRV_FOOD may not have been representative of the child’s preferred activities and did not include social activities. Adding a social element to reinforcers may increase their value, and a child with many friends or siblings may have available alternatives that we were not able to assess against RRV_FOOD. Although we observed a correlation among baseline percent overweight, RRV_FOOD, and ENR_HOME, as well as an interaction of RRV_FOOD with ENR_HOME on relative weight change, neither had a main effect on weight change. It is important to recognize that this is an observational study, and these ideas should be assessed in a randomized trial to test whether manipulating alternatives and environment would alter obesity development.

Examining ways to improve children’s weight trajectories may include building alternative reinforcers in an enriched home environment to reduce RRV_FOOD. Future research should examine ways to achieve these goals simultaneously to improve treatment or prevent childhood obesity. This study provides a framework, based on behavioral economic theory, for a novel approach to preventing and treating childhood obesity.

Study Importance

What is already known?

• Relative reinforcing value of food (RRV_FOOD) is indexed by the amount of effort one engages in for access to food and depends on the reinforcing value of food and alternative behaviors.

• RRV_FOOD is associated with the development of obesity in children.

• The enriched home environment, i.e., the cognitive and emotional stimulation available at home to a child, is also associated with risk of developing obesity in children.

What does this study add?

• RRV_FOOD and an enriched home environment interacted to predict changes in percent overweight in children over 2 years.

• Children with lower RRV_FOOD and a more enriched home environment maintained percent overweight, whereas children with one risk factor increased percent overweight. Having two risk factors was associated with the development of obesity by year 2.

How might these results change the direction of research or the focus of clinical practice?

• Although RRV_FOOD is a risk factor for obesity, having access to non-food alternatives that a child is motivated to engage in may be protective against weight gain.

• A novel obesity prevention or treatment for children should be tested in which alternatives to food that are substitute reinforcers for food are provided in a child’s environment, whereas less-healthy foods are made less accessible within the home.

DATA AVAILABILITY STATEMENT

The individual deidentified data and data dictionary that support the findings of this study will be made available from the corresponding author, upon reasonable request and approval.