Young children are dependent on parents and caregivers for food, making parents' choices about feeding key determinants of children's eating experiences. These choices include when eating will occur, the extent to which feeding occurs in response to children's indication of hunger or distress, the contexts within which eating will occur, the foods and portions sizes that will be made available to children, and which feeding practices will be used to promote or discourage children's eating. All of these choices have the potential to influence children's early learning about food and eating. Thus, parents influence children by shaping their eating environments, but this influence is bidirectional, as parenting is, in part, a reaction to child characteristics. This reality must be kept in mind, especially interpreting evidence from observational, cross-sectional research. Holistically understanding the factors that influence parenting and feeding is key to success in attempts to positively impact children's eating and weight outcomes. In the interest of effectively preventing childhood obesity, the current evidence shows there are many modifiable risk factors for childhood obesity that reside in young children's family environments.
The preponderance of evidence examining associations between parenting, child eating and child weight provides support for associations between parenting practices and child weight, however this evidence alone is not sufficient to answer the question "does parenting influence child weight?" As shown in Table , the majority of the studies that address the association between parenting and child weight are cross-sectional, do not include measures of child eating, and cannot provide evidence for direction. The direction of influence is an especially pertinent issue within this literature, as parent-child interactions are characterized by bidirectional influences [6
]. In the absence of solid evidence that parenting causally influences child weight via an influence on child eating, evidence from the general parenting literature, showing that parenting is responsive to and influenced by child characteristics and behaviors, suggests any direct association seen between parenting and child weight is in the other direction, with child weight influencing parenting [6
With the ultimate goal of developing a comprehensive understanding of how parents affect children's eating and weight status outcomes, as well as how children's eating and weight status affects parenting, research is needed to empirically test the meditational model presented in Figure . Research designs should reflect the inclusion of all three pathways, use validated measures of all three constructs, use experimental designs to assess causality and include assessment of covariates (for example, SES or maternal weight status) to rule out spurious associations. If all of these considerations are in place, then mediation can be tested by the traditional four step method proposed by Baron and Kenny [74
]: (1) show that parenting is correlated with child weight status (Pathway 1), (2) show that parenting is correlated with child eating behaviors (Pathway 2), (3) show that child eating behaviors are correlated with child weight status (Pathway 3), and (4) show that the association between parenting and child weight (Pathway 1) is no longer significant when child eating is added to the model.
of the 67 studies reviewed above met all criteria needed to test this full conceptual mediational model. Additionally, only 4 studies included all three pathways (see Table 3 [see Additional file 1
]); these studies provided some support for the influence of parenting practices (pressure, restriction) on child eating (eating more, eating in the absence of hunger), which then predicted higher weight and greater weight change across childhood [31
]. However, the generalizability of these and other studies was limited. First, many authors maintain a unidirectional focus in their interpretation of findings, asserting that parenting is influencing child eating and weight; the issue of how child eating and weight influences parenting has not been sufficiently addressed in the literature. Another limitation is that many studies fail to also assess covariates representing characteristics of families that may influence parenting, child eating and child weight, thus spurious associations (e.g., associations between family income or parent weight status and both parenting practices and dietary choices) cannot be ruled out. Future research utilizing experimental manipulation or appropriately designed longitudinal studies needs to encompass and appreciate the constant bidirectional influences that occur between parents and children with respect to parenting, child eating and child weight, as well as the potential confounders that are also influential within this research realm.
Across all pathways and study designs, there is a lack of consistency with respect to the definition of constructs and the validity of measures. This problem is particularly severe in the case of defining parenting styles, which makes comparability across these studies very difficult. Additionally, variability in construct definition and measure validity across studies can result in inconsistencies in findings that are not due to the true nature of the variables, rather due to discrepancies in how the variables were measured. Finally, most research in this field has been conducted with white, middle to higher SES children, which limits our ability to generalize findings. Research is needed with diverse samples, as there is evidence for racial/ethnic variation and for differences between daughters and sons, in both the parenting practices used and the relations between parenting and child outcomes [75
Where does the field go from here? It is time to start providing evidence for causal pathways, which will provide a stronger evidence-base for building obesity prevention and intervention programs targeting young children and their parents. The above review revealed that this field has an oversupply of evidence for associations between parenting on child eating and weight. Experimental and appropriately designed longitudinal research is needed to provide solid, causal evidence for the direction of these associations. Only two studies assessed whether parenting could be modified and whether this modification resulted in significant changes in child eating and weight [52
]; such evidence can provide important insights for potential intervention components that will have a high likelihood of success. Future research is needed to examine other relatively unexplored issues, including how child attributes influence parenting choices and how other characteristics, such as ethnicity, SES, and parent weight status, moderate the association between parenting and child eating and weight.
One promising framework for taking these next steps is the Multiphase Optimization Strategy (MOST) developed by Collins and colleagues [79
]. Implementation of this framework allows for the design and evaluation of optimized interventions through a three phase approach: 1) a screening phase
where candidate intervention components are selected or rejected depending on causality established through randomized experimentation; 2) a refining phase
where closer examination of component dosage and tailoring is assessed; and 3) a confirming phase
where an intervention is finally built from the candidates identified in the first two phases. The key to this approach is that an intervention is not developed until a sufficient evidence-base is formed, allowing for a higher likelihood of intervention success and a better understanding of the independent and interactive effects all intervention components.
In summary, the dramatic increase in studies examining links between parenting, child eating and child weight is one indicator of the progress this field has made over the past three decades. Although our review reveals some well established associations between aspects of parenting and child eating and weight, the evidence for the influence of parenting and feeding practices on children's eating and weight status is limited. Additional research addressing these limitations is essential to informing the design of early interventions intended to modify the early feeding context and to influence children's eating and weight. Further research is needed to assess: (1) causal influences of parenting on child eating and weight; (2) whether child behavior mediates the influence of parenting on child weight; and (3) how these relations may be moderated by demographic and individual factors such as family ethnicity, income, education and parental weight status. Finally, we suggest a multiphase research strategy [79
], which provides an approach for identifying, selecting, and evaluating aspects of parenting and child feeding practices as components in optimized preventive interventions for childhood obesity.