Food parenting and child snacking: a systematic review

Background While the role of parenting in children’s eating behaviors has been studied extensively, less attention has been given to its potential association with children’s snacking habits. To address this gap, we conducted a systematic review to describe associations between food parenting and child snacking, or consuming energy dense foods/foods in between meals. Methods Six electronic databases were searched using standardized language to identify quantitative studies describing associations of general and feeding-specific parenting styles as well as food parenting practices with snacking behaviors of children aged 2–18 years. Eligible peer-reviewed journal articles published between 1980 and 2017 were included. Data were extracted using a standard protocol by three coders; all items were double coded to ensure consistency. Results Forty-seven studies met inclusion criteria. Few studies focused on general feeding (n = 3) or parenting styles (n = 10). Most studies focused on controlling food parenting practices (n = 39) that were not specific to snacking. Parental restriction of food was positively associated with child snack intake in 13/23 studies, while pressure to eat and monitoring yielded inconsistent results. Home availability of unhealthy foods was positively associated with snack intake in 10/11 studies. Findings related to positive parent behaviors (e.g. role modeling) were limited and yielded mixed results (n = 9). Snacking was often assessed using food frequency items and defined post-hoc based on nutritional characteristics (e.g. energy-dense, sugary foods, unhealthy, etc.). Timing was rarely included in the definition of a snack (i.e. chips eaten between meals vs. with lunch). Conclusions Restrictive feeding and home access to unhealthy foods were most consistently associated with snacking among young children. Research is needed to identify positive parenting behaviors around child snacking that may be used as targets for health promotion. Detailed definitions of snacking that address food type, context, and purpose are needed to advance findings within the field. We provide suggested standardized terminology for future research. Electronic supplementary material The online version of this article (10.1186/s12966-017-0593-9) contains supplementary material, which is available to authorized users.


Background
Childhood overweight and obesity persist as significant health risks for children globally [1,2]. Given that excessive energy intake is a primary driver for inappropriate weight gain among children, it is not surprising that child snacking has consistently increased in recent decades [3,4]. Snacking has been defined interchangeably in the literature as foods consumed between meals and/or consuming "snack foods", typically identified as energydense and nutrient-poor (i.e. candy, chips, cookies, sugary drinks). Individual study participants may also selfdefine snacking occasions. The inconsistency across definitions is problematic and limits the generalizability of findings. Snacking in between meals currently contributes an estimated one third of children's daily energy intakes in the United States [5] and a quarter of daily energy for youth in some European nations [6]. Though data on snacking and obesity in children are limited and equivocal, there is evidence that children who snack frequently consume greater energy, have poorer quality diets, and exhibit other risk factors for excessive weight gain [7,8].
Although parental influence on children's overall eating behaviors and weight status has been studied extensively [9,10], less attention has been given to how food parenting might affect the snacking behaviors of children. Food parenting includes both parent feeding practices, the specific behaviors or strategies that parents use to feed their children (i.e. pressuring a child to eat), and feeding styles, the generalized patterns of these practices. General parenting styles (e.g. uninvolved, authoritarian) approximate how caregivers engage with their children through interaction and disciplinary strategies and may also be informative in the context of child snacking, as different styles have been associated with a variety of childhood dietary and weight-related outcomes [11]. Current literature suggests that in order to promote healthy eating habits, parents must strike a balance between setting reasonable limits, providing healthful foods and structured eating occasions, and supporting children's unique food preferences and regulation of appetite [12,13].
A recent theoretically guided conceptual model of snack-specific food parenting practices [14] identified four domains specific to snack feeding, which included Coercive Control, Permissiveness, Structure, and Autonomy Support. Coercive Control practices, such as restricting food or rewarding children with food, have been linked with increased energy intake, lower diet quality, and increased weight in children [15,16]. It is surmised that this domain may be particularly important in the context of snacking, as qualitative work suggests parents of young children, often use snack foods as tools to manage children's behaviors [17,18]. Permissive practices, such as feeding children to provide comfort, or having few rules or limits on snack intake, have been associated with excessive energy intake and elevated body mass index in children [19]. Given the low cost and portability of many processed snack foods, unrestricted access in the home may be especially problematic [20]. Conversely, it has been proposed that positive food parenting that provides Structure (e.g. routines, making healthy foods available) and Autonomy Support (e.g. role modeling, praise) is more likely to encourage children to establish healthy eating habits [21]. However, there are limited findings that describe such practices, and it is not clear what impact they may have on snacking intake among children [14]. Despite limited data, it is likely that overall parenting practices, whether positive or negative, have a differential impact on the quality of snack foods consumed by children.
To provide an overview of prominent findings in the literature, we conducted a systematic review to describe quantitative studies between 1980 and 2017 that have evaluated associations of parenting styles and food parenting practices with child snacking. Given the inconsistency in definitions, we describe all studies utilizing the word(s) snack/snacking, and provide distinctions between how they are measured and defined. We define snacking as consuming foods or beverages between meals, and snack foods are defined as energy-dense, nutrient poor foods/ beverages. Snacking behaviors refer to any behaviors related to snacking/consuming snack foods. To our knowledge, this is the first systematic review that assesses food parenting specifically in the context of child snacking. We are aware of one review that assessed the influence of two specific food parenting practices (e.g. parental pressure to eat and restriction) on children's dietary intake [22], but this review did not include a range of parenting behaviors and did not focus specifically on snacking.
The aims of this review were to: 1) present characteristics of studies on parenting and child snacking, including study design, setting, participant demographics, and measures used to assess food parenting, 2) present the frequency with which food parenting practices were characterized in the literature, 3) summarize associations between food parenting practices and child snack intake, 4) describe characteristics of measures of child snacking, and 5) identify recommendations for future research.

Search criteria
To ensure consistency in data collection and presentation, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist to conduct our search [23] (Additional file 1) and registered our review with PROSPERO (Registration number: CRD42017062520). To standardize abstract review, we employed a protocol containing inclusion and exclusion criteria, along with an electronic search strategy for the study (Additional file 2).
We searched for English-language articles published in peer-reviewed journals in the following electronic databases: CAB Abstracts, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Embase, PsycINFO, PubMed, and Web of Science. Key search terms were used to search titles, abstracts, and Medical Subject Headings and included text related to parents/caregivers (e.g. mother, father, parent), parenting style (e.g. parenting, parent-child relations, child rearing), food parenting (e.g. child feeding, control, restriction, pressure), and child snacking (e.g. snacks). Abstract files were downloaded, screened, assessed for eligibility, and organized by inclusion or exclusion in EndNote X7 by RB and AK. Full-texts of articles were assessed if they met all inclusion criteria.

Eligibility criteria
We included studies published between January 1980 and January 2017 in order to provide a scope of modern literature over the past four decades that reflects current parenting practices as well as those corresponding to increases in obesity prevalence in children over time [1].
Articles were included if they met the following criteria: 1) Measured snacking or snack-related behaviors of children aged 2 -18 years and 2) Measured the general parenting, feeding style, and/or food parenting practices of the child's parent or primary caregiver in the context of child snacking. We focused on children aged 2 and older to remove studies of infant breastfeeding and/or complementary feeding. We included studies with samples that included children younger than 2 only if solid food snacks were assessed (e.g. sample of toddlers and preschoolers aged 18 months -5 years), but excluded studies with samples comprised of only children under 2 years.
Experimental studies that assessed children eating in the absence of hunger (EAH), following meals were included. Their protocols were developed to evaluate dimensions of satiety in children, but we believed the general paradigm was relevant because it focused on eating outside of meals [24]. More specifically, these studies evaluated the extent to which a meal suppressed subsequent intake of snack foods.
We excluded studies that did not directly assess primary caregivers (e.g. child care workers, laboratory feeding studies where parent was not present/assessed). We also excluded studies that did not appreciably measure food parenting, such as those solely assessing frequency of family meals or home availability of food (e.g. pantry audit), as these are often markers of other factors such as socioeconomic status.
We also excluded conference abstracts or dissertations because we sought to describe peer-reviewed journal articles. Qualitative studies and reviews were not included because they are not appropriate for drawing inferences about association. Articles were also excluded if their scope was outside the field of child/family nutrition (e.g. focus on oral health and dental caries) or only studied children with special healthcare needs (e.g. eating disorders, developmental delays) due to lack of applicability to the general population.

Data extraction and analysis
To ensure consistency all full-text articles were extracted and double coded by researchers (AK, RK, RB); 25% were triple coded using the constant comparative method [25] to identify discrepancies in protocol interpretation and to reach a consensus when clarifying questions. Fewer than 5% of data items entered were in disagreement, and thus the protocol and data extraction tool were deemed appropriate for use.
Data extraction of full-texts occurred using a predefined list of items to be coded (Additional file 3) that were collected using Survey Gizmo for ease of data entry and summarization. After data extraction was complete, two researchers (AK and RK) also assessed study quality using existing tools: the National Institutes of Health Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies [26] and the Quality Assessment Tool for Quantitative Studies designed to assess experimental studies [27]; 25% of studies were double coded to ensure study quality tool consistency and no disagreement was found. We used Stata/SE 12.1 (Stata Corporation, College Station, Texas, USA) to obtain frequencies for categorical variables and mean values and standard deviations for continuous variables.

Study characteristics
We documented general study information such as publication date, country, journal name, and study design. To describe study samples, we assessed age of target children, populations recruited (e.g. low-income, minority), and sample sizes of caregivers/children. To describe participant demographics, we examined caregiver race/ethnicity, gender (i.e. mothers vs. fathers), and level of education.
We described the extent to which studies reported on important demographic information associated with child feeding (e.g. parent education, race/ethnicity, inclusion of male caregivers) as well as instrument quality to see how often validated tools were used in their intended way (e.g. all items vs. select subscales vs. individual items), and the level of dietary assessment (e.g. 24-h recall vs. food frequency questionnaire) [9,28]. Additionally, we described whether or not child snacking outcomes were predefined by the researchers before the outset of data collection, or defined post-hoc during analysis. We also examined the sample sizes and journals of publication to provide a general discussion about the diversity in publication. Finally, we described quality ratings for cohort and crosssectional studies (Range: Good, Fair, Poor) and experimental studies (Range: Strong, Moderate, Weak) using existing tools [26,27].

Measures of food parenting
We collected data on whether general parenting style vs. specific practices were assessed and whether measures were snacking-specific. We also identified the type of practices studied using a pre-determined list of specific snack-feeding practices (e.g. role modeling, rewarding behavior) based upon a recently published conceptual model of food parenting practices specific to child snacking [14]. Practices were organized by four higher dimensions from the conceptual model: Coercive Control, Structure, Autonomy Support, and Permissiveness.

Association between food parenting and child snacking
We summarized study results on the association between food parenting and child snacking outcomes. We postcoded these result summaries as positive, negative, null, or mixed in order to summarize trends in association. Since both the exposure (food parenting) and outcome (child snacking) were measured in myriad ways and not generalizable quantitatively, we opted to conduct a narrative summary of our findings using tables and figures at the level of each individual study.

Measures of child snacking
We examined the types of measures used to assess child snacking, and collected data on the source (i.e. parent vs. child), use of validated tools, how "snack" was defined in both the tools and in the analysis post-hoc, and what types of contextual information was presented about child snacking (e.g. timing, nutrient profile, frequency).

Study characteristics
Our search yielded 2846 articles, of which 84 duplicates were identified and removed (Fig. 1). After reviewing 2762 abstracts based upon inclusion and exclusion criteria (Table 1), 2696 were excluded and 66 were included for full-text assessment. Of full texts reviewed, 47 were included for analysis [13,18,. The primary reason for exclusions was that parenting/feeding practices were not assessed.
We present a brief narrative description of each study, the measures used, and study quality in Table 2 and a summary of overall study characteristics in Table 3. Nearly half of all studies (n = 31) were published within the past 5 years. More than 90% of all studies occurred in four Western nations: the United States (n = 14, 29.8%), the Netherlands (n = 12, 25.5%), Australia (n = 8, 17.0%), and the United Kingdom (n = 8, 17.0%). With the exception of Appetite, which published 36% of eligible articles, studies were published in a variety of journals (n = 25), with most journals publishing 1-2 studies each. There was significant diversity in authorship as well, with no author contributing more than 3 studies to the literature.
Overall, quality was high across cross-sectional and observational articles, with 39/40 receiving a Good quality rating (Range: Good, Fair, Poor) ( Table 2). Among experimental studies (n = 6) quality was weaker due to a lack of reporting study participation rates (Range: Strong, Moderate, Weak); most experimental studies scored as moderate (n = 4) compared with weak (n = 1) or strong (n = 1).
Most studies defined the gender of caregivers (n = 29, 55.3%) who were predominantly female. Forty percent of studies exclusively contained mothers/female caregivers (n = 12); when included, males made up 11% of caregiver samples on average. Although these studies distinguished between male and female caregivers, only about one third (n = 10) explicitly mentioned the word "father" or defined the number of fathers in their sample. Most studies reported caregiver level of education (n = 34, 72.3%), with two studies reporting that their samples contained at least 40% of caregivers with a low level of education.

Measures of food parenting
The most commonly used tool adapted to measure food parenting practices was the Child Feeding Questionnaire (n = 16, 34.0%) [74], followed by the Comprehensive Feeding Practices Questionnaire (n = 3) [75]. General feeding styles (n = 3) or parenting styles (n = 10) were examined in fewer studies than specific food parenting practices (n = 42), and often focused their findings on specific practices within styles; few studies evaluated parenting specific to child snacking (n = 10).
Using a theoretically-driven conceptual framework [14], we summarized the frequency with which specific food parenting practices were described across four dimensions of snack feeding in Fig. 2. The practices are presented across the four key dimensions (Coercive Control, Structure, Autonomy Support, and Permissiveness), to indicate how many studies provided data about each practice. Studies appeared to focus on more negative aspects of food parenting, with a strong focus on the dimension of coercive control (n = 39, 90.0%) in the context of child snacking. Within this dimension, specific behaviors related to restriction (n = 32) and pressure to eat (n = 20) were most often described. Within the dimension of structure (n = 32, 68.0%), most studies measured home availability of healthy foods (n = 25) and monitoring of food intake (n = 17), compared with fewer studies examining planning and routines (n = 8) and home availability of healthy foods/snacks (n = 12). Fewer studies described practices within the dimension of autonomy support (n = 20, 42.5%) and permissiveness (n = 15, 31.9%), where home availability of unhealthy food (n = 12) was assessed most frequently.

Association between food parenting on child snacking
We summarize associations of the most commonly studied aspects of parenting with child snacking in Fig. 3. No noticeable differences in trends based on feeding practices versus feeding or parenting styles were observed. Parental restriction of food was positively associated with child snack intake in 13/23 studies (n = 2 experimental, n = 2 longitudinal, n = 9 cross-sectional), while pressure to eat and monitoring yielded inconsistent results. Home availability of unhealthy foods was positively associated with snack intake in 10/11 studies (n = 8 cross-sectional, 2 = experimental). Instrumental feeding was described in 7 studies and was typically a combination of coercive controlling practices (e.g. restriction and rewarding with food). Findings related to positive parent behaviors (e.g. role modeling, reasonable rules about eating) were limited to less than a fifth of all studies (n = 9). Four of seven studies found parent food rules were negatively associated  . Using preferred reporting items for systematic reviews and meta-analyses (PRISMA), diagram illustrates studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage with snack intake. Based on the small sample sizes, it is not possible to identify trends by study design (e.g. experimental vs. cross-sectional).

Measures of child snacking
We summarize characteristics of measures used to assess child snacking in Table 4. A wide variety of measures were used, with little consistency across the literature. The vast majority of studies used self-report to assess child snacking behaviors (n = 39), with caregivers frequently reporting on their child's intake (n = 20). Nearly half the time, a food frequency questionnaire (FFQ) was used to assess snacking (n = 22, 46.8%), with survey tools used less frequently (n = 14). Open-ended tools (e.g. 24-h recalls) were rarely used.
Most studies adapted an existing tool (n = 33, 70.2%); fewer reported the use of a validated tool to assess their particular age group (n = 10, 21.3%). The definition of snacking or snack intake varied greatly across measures. Since FFQs were employed often, it is not surprising that many studies defined individual food items as "snacks." However, snacks were also defined categorically based on healthy or nutritional characteristics (e.g. "junk food", "sweets", "dessert", "unhealthy", "energy dense"), or in other ways (e.g. "excessive snacking: eating between meals and at night" [63]).
Although snacks were typically measured as individual food items, they were often grouped together in a variety of ways post-hoc and then defined as snacks during analysis (n = 26, 55.3%). For example, a FFQ might assess child consumption of cookies, chips, and soda as separate food items, but during subsequent analysis, the author(s) would group them together and label them as "energydense snacks." Studies of EAH (n = 6) were often laboratory-based and presented children with a specific set of foods, sometimes described as palatable snack foods, to evaluate children's satiety [40,43,48,52,57,62]. Three studies did not provide any definition of snacks and left it to the caregiver or child to determine what this word meant (e.g. "How often do you give your child snacks…").
Timing was not consistently assessed as a factor used to define a snack (i.e. chips eaten between meals vs. with lunch) during measurement or analysis. More than half the time (n = 26, 55.3%) beverages would be included in the definition of snack (e.g. soda and chips combined together as "unhealthy snacks"), but only 2 studies distinguished between beverages consumed during or between meal times. Consequently, a soda consumed with lunch could not be distinguished from a soda consumed with chips during a snack.
Frequency of snacking was the factor most often assessed (n = 38, 80.9%), but some studies also evaluated total energy intake from snacks or child snack preferences. In rare cases, fat intake was estimated. No studies reported on snack context (e.g. where or precisely when snacking occurred) and only one described parent rationale/purpose for providing snacks.

Discussion
The aim of this systematic review was to describe how food parenting behaviors were described in the context of child snacking in quantitative studies published between 1980 and 2017. We also sought to identify how child snacking was operationalized in studies that examined food parenting and describe the demographic characteristics of study participants present in this field of research. Using evidence-based, replicable methods, we found that most studies were of good quality and reported cross-sectional findings utilizing samples that contained mostly white, college educated, female caregivers who self-reported their food parenting behaviors and their children's snack behaviors. Dietary assessment was selfreported in 3 out of 4 studies, typically using abbreviated food frequency questionnaires or brief survey items. No Table 1 Inclusion and Exclusion Criteria Inclusion Criteria Exclusion Criteria 1. Measured snacking or snack-related behaviors of children aged 2 years to 18 years through either objective (e.g., meal observations) or subjective (i.e., self-report) methods. This could include nutrient intake, snack foods, frequency, quality, or context. 2. Measured the feeding style, feeding practices, and/or parenting style of the child's parent or primary caregiver through self-report of caregiver, child, or direct observation (e.g. observed snack time) in the context of child snacking.
1. Not in English 2. Published prior to 1980 3. Not in a peer-reviewed journal (e.g. TIME magazine) 4. Not a research article (e.g. published in Pediatrics but is an interest piece or compilation of abstracts) 5. Scope of article is outside of child/family nutrition or weight status (e.g. focus on oral health, a particular foodstuff, etc.) 6. Population studied was not children ages 2-18. As long as child was under 18 at baseline, we can use the study. 7. Exclude studies of nursing 8. Population focused on children with special healthcare needs (e.g. feeding disorders, diabetes, eating disorders 9. Child snacks or snacking not assessed 10. Parenting/parent feeding and child snacking not examined together a,b 11. Review paper 12. Qualitative paper a Did not include family meals or parent diet as a marker of food parenting b Excluded if primary caregivers were not assessed at all (e.g. a study of the feeding patterns of child care workers)   or "no" for consuming a snack).
• Adolescent version of the CFQ was used and assessed three parent feeding subscales: restriction, concern, and monitoring.
Weighed food intake of snack items consumed in the laboratory over 2 days was used to assess snack EAH.

Good
Parental restriction and concern about child eating were positively associated with observed adolescent snack EAH. Parental monitoring was not associated with snack intake. Cross-sectional study of n = 943 children aged 7-12 years in the Netherlands.

Rhee
• A children's version of the CFQ, using two subscales: restriction and pressure to eat.
Child report using food frequency items assessed consumption of sweet and/or savory snacks. Source of measure undefined.

Good
Perceived maternal restriction to eat was negatively associated with snack intake; pressure to eat was not associated with child snack intake.
• Restriction, pressure to eat, and monitoring subscales of CFQ.
Parent report using food frequency items assessing child intake of "high calorie snacks". Source of measure undefined.

Good
Restriction and monitoring mediated the relationship between maternal education and child snack intake; restriction was positively     Cross-sectional study of n = 739 female adolescents aged 12-15 years in Iran.
• Items developed for the study assessed perceived parental control over junk food consumption (e.g. "My parents tell me how much junk food I may consume").
Child report using modified Iranian FFQ (Mirmiran et al., 2007) assessing snacking behaviors over a 1-week period and classified into healthy and unhealthy snacks, or "junk food".

Good
Adolescents who perceived stricter parental control reported less frequent consumption of "junk food", or unhealthy snacks. The relationship was partially mediated by the child's perceived own behavioral control over snack consumption.
• New measure of parental overt control (detectable by child) and covert control (undetectable by child) of child eating adapted from previous tool (Brown & Ogden, 2004).
Parent report of child intake of healthy (e.g. grapes, yogurt, toast) and unhealthy (e.g. sweets, crisps) snacks using existing tool (Brown & Ogden, 2004) Good Greater covert control was associated with less child intake of unhealthy snacks. Greater overt control associated with greater child intake of healthy snacks. Good Home availability of snacks at baseline was associated with increased energy-dense snack intake after 2 years; family support for healthy eating was inversely associated. Maternal modeling of healthy eating was not associated with a change in snack intake.

Pearson
van Ansem, 2015, [67] Cross-sectional study of n = 1203 parent-child pairs, • One binary item assessed presence of snack consumption Child report using validated FFQ assessed energy-dense Good Home availability of snacks was positively associated with Cross-sectional study of n = 502 parent-child pairs, with children aged 12-14 years in the Netherlands.
• Three binary items developed for the study assessed the presence of parent-imposed snack rules about number of snacks, timing of snacks, and which snacks child may eat.
Child report using items adapted from validated FFQs assessing sweet and savory snack consumption. Multiple measures cited.

Good
Presence of rules regarding the quantity and timing of child snack consumption was positively associated with the child's snack intake, based upon child self-report. Verstraeten,
• Two items developed for study based on qualitative data assessed child report of parental permissiveness (e.g. fast food/ snacks allowed any time).
Child report using 2 days of 24-h recalls with "unhealthy snacks" identified as foods high in sodium, fat, or sugar.

Good
Parental permissiveness (e.g. no limits) was not associated with unhealthy snacking among adolescents.
• Parenting style was assessed using two constructs: parental warmth (e.g. affectionate behaviors) and parental hostility towards child (e.g. irritable and angry behaviors). Source of measure undefined. noticeable differences in trends based on feeding practices versus feeding or parenting styles were observed. There was a notable range in the measurement of types of food parenting practices and in the definition of child snacking, thus creating opportunities for improvement in future exploration of these topics. Restrictive feeding and access to unhealthy foods were most consistently associated with increases in children's snack intake, though the frequency of cross-sectional study designs limits the ability to determine causality. Few studies described autonomysupporting (e.g. praise, encouragement) or permissive (e.g. feeding to comfort) food parenting behaviors.

Inconsistent definition of snacks
Describing child snack intake presents several challenges.
First, there appears to be no consensus on a universally accepted definition of child snacking in the literature we examined. Snacks were described both as a food type and as foods consumed in between meals. In most studies the word "snack" was a catch-all phrase to describe energydense, nutrient poor food types similar to "junk food"; few studies distinguished between unhealthy (e.g. chips, cookies) and healthy snacks (e.g. fruits and vegetables) [31,58,59]. Additionally, multiple dimensions were included in the definitions: half of the studies included beverages as snacks, while one third specified the timing when a snack food was consumed (e.g. between meals). Another measurement challenge is that many studies defined "snacks" post-hoc, meaning the definition of snacks was often developed after data were collected, introducing possible bias depending on how or why certain foods were grouped together (e.g. relevance in the diet, statistical viability). There was great variation regarding which unhealthy foods were included or excluded across studies of similar populations. Additionally, beverages, though likely consumed alongside snack foods, often received their own separate category for analysis since timing of their intake was not routinely assessed.
Our findings that snacking definitions vary within food parenting literature are reflected elsewhere. A 2010 review of general snacking definitions concluded that studying the impact of snacking on various dietary and health outcomes was limited by the variation in definitions [76]. In another review of child snacking patterns, authors reported limited evidence of association between snacking behaviors and weight status, but emphasized that methodological limitations in the measurement of snacking might have severely limited their ability to conduct the analysis [7].

Relationship between food parenting and child snacking
Despite a doubling in the number of studies describing food parenting and child snacking over the previous decade, the lack of consistency in methodology limits generalizability of findings across studies. On one hand, some of our findings appear consistent with existing literature on food parenting and general dietary intake. We found that restriction was positively associated with child snack intake in a majority of studies, which included experimental and cross-sectional designs. In other studies of food parenting, restriction of food has been linked with both increased caloric intake and elevated body mass index in children [11,77]. The underlying basis for this association is likely bidirectional, complex, and mediated by multiple factors such as a child's weight status (e.g. parents may restrict out of concern if a child is overweight). Additionally, how parents restrict (i.e. with warmth and supportive structure versus with hostility and coercive controlling practices), which may lead children to more disinhibited eating and interest in high-calorie, or "off limits" foods [10,77]. We also found that home availability of unhealthy foods was positively associated with snack intake in 10 out of 11 studies. The home food environment has been discussed as an important risk factor for childhood obesity. However, it is not clear if this is explicitly due to the presence of the food or represents a proxy, such as role modeling or that fact that parental food and beverage intake strongly predicts that of their children [78,79]. Our review did not yield enough studies of parental role modeling using consistent methods (n = 2) to determine what impact it might have on child snacking. Mixed findings were obtained regarding associations of pressure to eat and snacking. In the wider literature on child feeding, parental pressure to eat has been associated with both lower energy intake and body mass index in children in some studies, and increased energy intake in others, possibly because parents may be trying to encourage underweight or picky children to eat [22,80,81]. It is also possible that this construct is less The total number of studies that described specific food parenting practices related to child snacking. Practices are arranged within 4 dimensions of child snack feeding derived from a theoretically guided conceptual model of food parenting around child snacking [14] Fig. 3 Summary of Commonly Described Food Parenting Practices and Their Association with Child Snack Intake (n = 33). Number of studies describing positive, negative, or null associations between specific food parenting practices and child snack intake utilized in the context of child snacking, as parents may be more likely to pressure children to eat foods deemed "healthy." This is consistent with a qualitative conceptual study of food parenting around child snacking that found very few low-income parents identified pressure as part of their schemas around snacking [10,14].
We also found monitoring food intake bore null findings in a majority of studies [81]. One possible reason for this may be that monitoring can be characterized as controlling when paired with other behaviors (e.g. restriction) and may be positive if it is paired with structure-supporting behaviors (e.g. reasonable limits, offering healthy foods) [14]. Additionally, few studies employed measures that focused specifically on snack food parenting, which may reduce their relevance for some food parenting practices.
Although a number of validated tools exist to assess food parenting practices [9], few studies in our review utilized complete measures, and instead took specific items or partial subscales from tools like the Child Feeding Questionnaire [74] to assess specific controlling feeding practices (e.g. restriction). Measurement of food parenting presents a challenge, as many child feeding tools have numerous items and subscales, which affects participant burden. However, adaptation presents a threat to validity, as psychometric properties of validated scales do not necessarily apply when subsets of items are administered. It is possible such adaptations contributed to mixed findings when we examined associations between food parenting practices and child snacking.
Recommendations for future research Recommendation #1: Investigate parenting specific to child snacking In general, the literature presents negative food parenting practices like restriction and pressure to eat, compared with role modeling, healthy limit-setting, or encouragement. Therefore, it would be beneficial for future studies to include positive parenting behaviors to identify how these can be supported and translated into public health interventions. At present, there are a limited number of tools that exist to measure food parenting specific to snacking. The Toddler Snack Food Feeding Questionnaire [36] assesses both negative and positive food parenting dimensions and is validated for use with caregivers of children aged 1-2 years. The Parent Mealtime Action Scale [49] measures overall parent mealtime behavior, but does present two dimensions that are specifically positive and snack focused (e.g. snack limits and snack modeling); this tool was validated with caregivers of children in 1st-4th grade (aged 6-9 years). In the future, it would be beneficial to expand these measures or create a new tool to assess the full spectrum of food parenting practices around snacking.

Recommendation #2: Increase diversity in caregiver perspectives
Our review found that mothers almost exclusively represented caregivers of interest with respect to food parenting around child snacking. We noted that a vast majority of studies either did not mention fathers or male caregivers (e.g. stepfather, live-in partner of mother), and if mentioned, they comprised 10% or less of samples. Increasingly, men are playing a greater role in child rearing, and their absence in studies of food parenting [28] and childhood obesity-related risk factors [82,83] presents a major gap in the literature. Thus, it is important to intentionally recruit men in studies of snack food parenting and examine whether their practices conflict with or support that of female partners, or female caregivers as a whole. Future studies should define a parent or caregiver, and clearly convey the number of female and male caregivers included in the sample. Additionally, there is evidence that other informal caregivers, such as grandparents, may play an increasingly important role in the provision of snacks to children [84,85]. Caregivers in the studies reviewed were typically white and highly educated, consistent with other literature exploring parenting and obesity-related risk factors in children [11,86]. In light of the health disparities that lowincome children from racial/ethnic minority groups face with respect to food quality, healthy food availability, and childhood obesity [87,88], an intentional approach towards recruiting diverse families is warranted. Additionally, recent qualitative work suggests that low-income parents may use snack foods specifically as an affordable way to comfort children or provide treats in the absence of other costly pleasures (e.g. vacations, movies) [17,89,90]. Therefore, more quantitative studies are also needed to identify differences in food parenting intentions and practices based upon such sociodemographic factors.

Recommendation #3: Describe child snacking contexts and purposes
The context in which child snacking occurs is poorly defined in the literature. Although most quantitative studies described the number of snacks children consume, only one described the purpose, or parent rationale for providing snacks (e.g. reward, to promote health) [18]. No studies in our review described the physical context or timing in which snacking occurred. There is reason to believe that timing may also be an important factor, as a recent review of American children's snacking patterns found that afternoon snacks might be more energy dense and nutrition-poor that morning snacks [91].
One qualitative study of low-income multi-ethnic caregivers of 2-5-year-old children provides additional insight, revealed that snacking timing and location were important parts of their definition of a snack [92]. Parents reported that children were often fed in response to environmental stimuli (e.g. ice cream truck, while grocery shopping) or that physical context dictated their child's snacking habits (e.g. whenever the TV was turned on) [14,89]. Another analysis from the same study found that nutritional quality of snacks varied greatly based upon self-reported purposes; children received healthier snacks when parents were addressing their hunger and less healthy snacks when they were being rewarded [93]. Therefore, understanding both context and the underlying purpose of snack feeding is critical to developing effective public health messages for parents and may also help to identify environmental triggers for food parenting practices that are most obesogenic.
Recommendation #4: Move toward more consistent terminology and detailed definitions around child snacking The current heterogeneity in definitions of child snacking limits the field in progressing towards greater understanding of snacking behaviors. Given that measurement of snacking varies based upon populations, research aims, and methodologies, it is not likely feasible to provide one universal definition of child snack foods. However, we propose the use of consistent terminology and dimensions of snacking (Table 5).
Primarily, we suggest that snack foods be defined as foods or beverages consumed between meals in order to standardize language across studies. Within this definition, nutrient-rich items like fruits, vegetables, and whole grains consumed between meals may also qualify as snacks, thus leading the field towards including more healthful eating behaviors in research. If items are defined as "unhealthy" snack foods, we recommend providing explicit details about all food/beverages assessed and the specific rationale for such categorization. Nutrientpoor foods assessed without the context of the timing (e.g. junk food or soda consumed at any time of day) would not be considered snack foods within this proposed definition.
Some studies may use qualitative research to define snacking within a population in order to identify the full range of foods consumed between meals as "snacks". For example, one caregiver-defined definition of snacking Snack foods (and beverages, if applicable) Foods and/or beverages that are consumed by children between meals. Researchers may provide their own specific qualifiers (e.g. "energy-dense snack foods", "sugary snack foods") along with explicit criteria for these classifications. Terminology may be shortened to "snack" or "snacks" after it has been defined.

Snacking occasions
The number of between-meal eating episodes in a given day.

Snacking purposes
Reasons that parents offer foods between meals (e.g. child request, reward, special occasion, routine).

Snacking contexts
Places where between-meal eating occurs (e.g. at home, in the car, at church).
among preschool-aged children that was recently presented by Younginer et al. [92] is, "A small portion of food that is given in-between meals, frequently with an intention of reducing or preventing hunger until the next mealtime." When parents in this population were asked about why or when they give their children "snacks", this definition is useful to properly interpret the findings. Measuring all dimensions of snacking certainly has implications for participant burden and is not likely to be feasible in most studies. A smaller-scale study that utilizes high-burden measures to validate a lower burden questionnaire-based assessment of various snacking dimensions would be a promising strategy to enable largescale assessment of associations with food parenting and other factors in the future.

Strengths and limitations
Our review presents several strengths. First, we provide transparent and replicable methods using PRISMA guidelines. We provide our search protocol, detailed search strategy, and data extraction tool with our findings. We also utilized double coding of all data extracted, including screening and full-text analysis in order to increase validity of our results. Additionally, we built our review upon a theoretically guided conceptual model of food parenting around child snacking so that our findings could be presented in the context of the current momentum within the literature. We use the same terminology and definitions of food parenting practices presented in the model in order to maximize construct operationalization.
Our review also has limitations. Due to the vast number of studies requiring screening, we did not review the bibliographies of full-texts to identify additional articles. We also did not include grey literature in our search, which could have increased the number of possible publications. The cross-sectional design of most studies we present also limits our ability to assess causality or temporality of the relationship between food parenting and child snacking. Due to the lack of standardization across measures of food parenting and child snacking, our review is limited to a descriptive, narrative summary of the state of the research, rather than a meta-analysis. However, our hope is that providing recommendations to improve future methodology will allow for such analysis in the future.

Conclusions
Snacking among children is nearly universal and significantly contributes to children's intake of energy and other nutrients. Parents play an important role in shaping children's dietary behaviors, including snacking. This study is the first to systematically describe food parenting specifically in the context of child snacking.
Restrictive feeding and child access to unhealthy foods have been most consistently associated with increases in children's snack intake. Pressure to eat and monitoring have yielded mixed and null findings. With mounting attention paid to the role of child snacking on obesity risk in recent years, a universal definition of snacking that addresses both food type and timing is needed to maximize generalizability across studies and advance findings within the field. Future research should include positive food parenting behaviors around child snacking that may be used as targets for health promotion.