Eating in the absence of hunger (EAH), or the extent to which a child continues to eat despite experiencing satiety [1], is an important behavioral phenotype indicating elevated risk of obesity. EAH, which is typically measured as the amount of highly-palatable snack foods that a child consumes after a satiating meal, has been associated with higher weight, weight gain, and obesity among children [2–7]. Although there is evidence that EAH is in part genetically determined [8], intervention studies suggest that EAH is modifiable [9, 10] and thus there is great interest in how sociocultural factors influence the development of EAH among children.

Much attention has been paid to how parenting practices may impact children’s obesogenic eating behavior, particularly EAH [11, 12]. Parents’ use of restrictive feeding practices has been hypothesized to promote EAH by undermining children’s ability to self-regulate their eating and increasing children’s desire for restricted foods [13, 14]. However, existing research regarding the association between restrictive feeding and EAH among children has produced mixed results. Early research among a primarily white, moderate socioeconomic status sample of girls found that between ages 5 and 7, the greatest increases in EAH were observed among overweight girls whose mothers reported higher levels of restrictive feeding [13]. Two additional cross-sectional studies among preschool-age children also found positive associations between maternal restriction and EAH among girls [1, 3]. Together, this body of work contributed to the current US practice guidelines for preventing and treating obesity among children, which encourage parents to avoid overly restricting children’s intake or restricting access to specific foods [15, 16]. Other studies of preschool-aged children however, have not observed any associations between restrictive feeding practices and EAH [17, 18]. Further, longitudinal data increasingly suggest that parents’ use of restrictive feeding practices is a response to, not a cause of, increasing child weight [19, 20]. In sum, the relations between restrictive feeding and child EAH are unclear.

Given the uncertainty regarding whether restrictive feeding contributes to EAH among children, the objective of the current study is to utilize longitudinal data from a contemporary cohort of toddlers to examine the bidirectional associations between maternal restriction and child EAH. All prior research on maternal restriction and child EAH has been conducted among children preschool age and older; no studies have been conducted among toddlers. However in the US, clinical recommendations for the prevention of childhood obesity advise parents of children 12 years and younger to avoid overly restrictive feeding [15], therefore it is essential to understand the potential impact of restrictive feeding among all ages of children in this range. Among the cohort of toddlers observed in the current study, EAH at 27 months predicted greater body mass index z-score at 33 months, demonstrating that EAH is a concerning behavior that emerges early in life and may establish risk of obesity throughout childhood [7]. These findings reinforce the urgency of identifying factors that contribute to EAH during toddlerhood.

As depicted in Fig. 1, the results of the cross-lagged analysis demonstrate the simultaneously-adjusted longitudinal associations between the same measures over time, concurrent correlations between maternal restriction and child EAH at each age point, and cross-lagged associations estimating the effect of maternal restriction at an earlier age with child EAH at a later age, and vice versa. The fit of both cross-lag models (Model 1: Restriction with regard to food amount and Model 2: Restriction with regard to food quality) was good with the ppp value for each model well within the recommended 0.05 to 0.95 range.

In Model 2, mothers’ restriction with regard to food quality was strongly correlated between 21 and 27, and 27 and 33 months (b ₁  = 0.68 and b ₂  = 0.59, respectively, p < .05). Similar to Model 1, in Model 2 adjusting for concurrent and longitudinal associations with maternal restriction with regard to food quality, a non-significant correlation between measures of child EAH was observed between 21 and 27 months (b ₃  = 0.14, p > .05) while a significant correlation was observed between 27 and 33 months (b ₄  = 0.47, p < .05).

Maternal restriction with regard to food quality was not concurrently associated with child EAH at any time point (b ₅  = −0.03 at 21 months, b₆ = −0.09 at 27 months, and b₇ = −0.08 at 33 months, all p > .05) and did not prospectively predict child EAH at either of the future age points (b₁₀ = 0.05 and b₁₁ = 0.08, p > .05).

Child EAH did not prospectively predict either type of maternal restriction at the future age points (Model 1: b₈ = −0.07 and b₉ = 0.01, p > .05; Model 2: b₈ = 0.01 and b₉ = 0.13, p > .05).

The objective of this study was to examine the bidirectional associations between maternal restrictive feeding and EAH among toddlers between ages 21 and 33 months. Cross-lagged modeling of longitudinal data, as conducted in this and other recent studies [20, 30], can provide important clarity to the bidirectional relations inherent in child feeding. Contrary to research conducted primarily among preschool and school age children that suggests that restrictive feeding increases EAH among children [1, 3, 13], we observed an inverse association between restriction with regard to food amount at 21 months and EAH at 27 months. Within this cohort, EAH at 27 months was predictive of higher BMI z-score at 33 months [7], suggesting that lower restriction with regard to food amount in early toddlerhood may contribute to obesogenic eating during a critical developmental period. Restriction with regard to food amount at 27 months was unrelated to EAH at 33 months and restriction with regard to food quality did not prospectively predict EAH. These findings extend previous cross-sectional research that observed no associations between maternal restriction and EAH during the preschool period [17, 18]. While at 21 months, mothers of children with high EAH were more likely to report restriction with regard to food amount, prospectively, child EAH did not predict changes in mothers’ restrictive practices. This suggests that mothers of young children may not consistently recognize eating in the absence of hunger or be concerned enough about their children’s over eating at this young age to modify their feeding practices in response [31]. Mothers’ beliefs about the amount of food toddlers should eat or the types of food toddlers should eat may also be more strongly influenced by external sources, such as the media or clinicians, as opposed to characteristics of their child. Together, these and other findings [19, 20] suggest that current practice guidelines that pediatric providers should counsel parents to avoid overly restrictive feeding practices [15, 16] may be unfounded. While as a whole, maternal restriction appears to have little impact on children’s EAH, mothers’ efforts to limit the quantity of food that young toddlers eat may protect against obesogenic eating.

Differences between the current study and seminal studies suggesting that restrictive feeding increases EAH [1, 3, 13] may serve to explain our contrary findings. First, all previous research measured restrictive feeding practices used either the Child Feeding Questionnaire (CFQ) [32] or less commonly, the Comprehensive Feeding Practices Questionnaire (CFPQ) [33], while the IFSQ was used in the current study. The IFSQ may capture sufficiently different approaches to or perspectives on restriction as compared to other measures. Further, the IFSQ distinguishes between restriction with regard to food quality and restriction with regard to food amount. Most items in the CFQ, the most commonly-used measure of maternal restriction, ask about restriction of children’s “favorite foods”, junk foods, and sweets. Therefore, the CFQ restriction scale is more similar to the IFSQ’s restriction with regard to food quality subscale than the restriction with regard to food amount subscale. The lack of items specifically tapping into restriction with regard to food amount in the CFQ may serve to explain why previous studies using the CFQ have not observed that restriction may be protective against EAH.

Children in our sample were also younger than those in early studies of restriction [1, 3, 13]. While maternal restriction among older children may be detrimental, restrictive feeding does not appear to be detrimental to young children’s intake. This discrepancy by age of study population may be due to differences in the cognitive development of toddlers versus older children, or differences in the ways mothers restrict children of different ages. For example, mothers of older children may use more overt approaches to restriction [34] and/or older children may be more aware than younger children that their intake is being restricted.

Finally, sociodemographic and secular differences in study populations may explain differences between the current study’s findings and those of early studies. Original research on restrictive feeding was conducted over 15 years ago among primarily normal-weight samples, often exclusively girls from moderate-income families [1, 3, 13]. Our contemporary sample consisted exclusively of low-income children, many of whom were already overweight or obese by age 2. In the US, 40% of children live in low-income households [35]. These children are at substantially higher risk for overweight and obesity in early childhood [36], therefore identifying factors that contribute to the early emergence of obesogenic eating among this population is a priority. However, low-income children may be differently influenced by maternal restriction than higher income children. For example, low-income children are more likely than higher income children to live in environments where large portions of highly-palatable food are frequently available [37, 38]. In such environments, maternal restriction of food amount may be helpful to set children’s expectations of appropriate intake. In comparison, for children from higher income families where unhealthy food is less commonly available in their environment, maternal restriction may be experienced as more severe or limiting, resulting in increased desire for those foods. Given this potential for a differential impact of restrictive feeding on child eating by socioeconomic status, further research is needed with diverse study samples to inform clinical guidelines.

The lack of increase in EAH among children exposed to higher maternal restriction with regard to food amount has important implications for the development of obesity prevention interventions and clinical guidance targeting families. Serving portion sizes that lead to appropriate growth patterns may help to entrain children’s recognition of and response to satiety cues linked with these portion sizes. This perspective supports current obesity prevention practice guidelines that encourage parents to limit children’s portion sizes [15]. Previous research has also demonstrated that serving age-appropriate portion sizes limits the energy consumed by children during eating occasions when parents are present [39]. Our findings extend this research by suggesting that these portion size limits may also positively impact consumption during future eating occasions when children’s portion sizes are not under the control of parents. The current study additionally suggests that restricting children’s consumption of specific types of food, in particular foods of low nutritional quality that are high in added sugar and/or solid fats (e.g., candy, cookies, potato chips), does not increase children’s consumption of these types of foods when children are provided unrestricted access. This finding further supports current obesity prevention guidelines that recommend limiting children’s intake of energy dense, low nutritional quality foods [15], and provides evidence that restriction may not increase young children’s future consumption of these highly-palatable foods.

There are several limitations to this study to consider. The longitudinal design is a strength, but due to the high-risk nature of the study cohort, attrition was high and there were missing data, which affected study power. Missing data were handled using the recommended method—a Bayesian approach using appropriate posterior distribution to account for the missing data. This approach, versus listwise deletion of participants who were missing data at any time point, protects against inducing selection bias [40]. All the data available at each time period were used in the path analysis. That is, if a dyad only participated at one time point, their data only contributed to the estimations of the mean and variance at that time point. If a dyad had longitudinal data available, their data contributed to the estimation of a longitudinal path. Specifically, 167 dyads contributed information at 21 months, 186 dyads at 27 months, and 161 dyads at 33 months. With regard to the longitudinal paths, 125 dyads contributed information to the paths between 21 and 27 months and 149 dyads contributed to the paths between 27 and 33 months. Our examination of attrition supports the use of this strategy as there were few differences in families who participated in a single versus multiple data collections. Another study limitation is that the EAH protocol used differed slightly from the more commonly-used EAH protocols published in the literature. For example, in this study, data collection was conducted in dyads’ homes and children were not provided a standardized meal before having free access to the snack foods. However, there is similar variation across other studies of EAH in factors such as location of protocol administration and types of foods served as the preload meal [41]. Standardizing the EAH protocol for use across varying study populations may increase consistency and reproducibility within the scientific literature. Finally, similar to other measures of restrictive feeding, the IFSQ captures both mothers’ restrictive practices and mothers’ beliefs with regard to restriction. Future longitudinal research is needed to identify whether specific restrictive practices increase children’s risk for, or protect against, EAH.

The current study identified that among low-income toddlers, restrictive feeding was not a risk factor for EAH. This suggests that interventions to improve children’s diet quality through limiting non-nutritive, energy-dense foods, as is recommended by current pediatric obesity prevention guidelines, are not detrimental to future eating behavior. Restriction with regard to food amount in particular may be protective against increasing EAH during early toddlerhood. Limiting toddlers’ intake to age-appropriate portions may help children appropriately recognize hunger and satiety signals as they age, and promote self-regulation of eating in situations where parents are not in control of children’s intake. Our results call into question the appropriateness of discouraging restrictive feeding, particularly among young children at high risk for obesity. Findings suggest that there may be appropriate approaches to restrictive feeding at specific points in children’s development. Future work is needed to identify the most effective parental feeding approaches that promote healthy weight throughout children’s development.