A Systematic Review of Responsive Feeding and Child Obesity in High-Income Countries

Abstract

Child overweight/obesity continues to be a serious public health problem in high-income countries. The current review had 3 goals: 1) to summarize the associations between responsive feeding and child weight status in high-income countries; 2) to describe existing responsive feeding measures; and 3) to generate suggestions for future research. Articles were obtained from PubMed and PsycInfo using specified search criteria. The majority (24/31) of articles reported significant associations between nonresponsive feeding and child weight-for-height Z-score, BMI Z-score, overweight/obesity, or adiposity. Most studies identified were conducted exclusively in the United States (n = 22), were cross-sectional (n = 25), and used self-report feeding questionnaires (n = 28). A recent trend exists toward conducting research among younger children (i.e. infants and toddlers) and low-income and/or minority populations. Although current evidence suggests that nonresponsive feeding is associated with child BMI or overweight/obesity, more research is needed to understand causality, the reliability and validity between and within existing feeding measures, and to test the efficacy of responsive feeding interventions in the prevention and treatment of child overweight/obesity in high-income countries.

Introduction

Childhood overweight/obesity is a serious public health problem that can lead to long-lasting health and developmental consequences. Over the past 30 y, childhood overweight/obesity, often beginning in infancy and toddlerhood, has reached epidemic proportions, particularly among minority and/or low-income populations in high-income countries (1). Overweight/obesity poses a major risk for chronic diseases, including type 2 diabetes, cardiovascular disease, hypertension and stroke, and certain forms of cancer (1). Increased consumption of more energy-dense, nutrient-poor foods has led to overweight/obesity rates that have risen 3-fold or more since 1980 in some areas of North America, the United Kingdom, Eastern Europe, the Middle East, the Pacific Islands, Australia, and China (2, 3). Of particular concern is the increasing prevalence in young children. Dietary patterns track over time, making the first few years of life an ideal period to help children establish healthy eating behaviors and avoid overweight/obesity (4). Although the role of the family in children’s early growth is well recognized, most research and interventions to prevent child overweight/obesity have focused on advice regarding children’s diet. Only recently has attention been directed toward responsive feeding. Responsive feeding is characterized by caregiver guidance and recognition of the child’s cues of hunger and satiety. Nonresponsive feeding is dominated by a lack of reciprocity between the parent and child, with the caregiver taking excessive control of the feeding situation (forcing/pressuring or restricting food intake), the child completely controlling the feeding situation (indulgent feeding), or the caregiver being completely uninvolved during meals (uninvolved feeding) (5, 6).

Although reports on the relationships between responsive feeding and child weight status in high-income counties have been inconsistent, the most consistent finding has been a positive association between nonresponsive feeding (i.e. restriction) and child weight status (7). This review had 3 goals: 1) to summarize the evidence for associations between responsive feeding and child weight status in high-income countries; 2) to describe responsive feeding measures; and 3) to generate suggestions for future research.

Methods

Selection criteria.

The following criteria were used to select articles for the review: 1) empirical research on children with an age range from 0 to 60 mo; 2) research published or available electronically in peer-reviewed English-language journals between 1990 and 2009; and 3) study locations in high-income countries. Theoretical articles, case studies, and articles focusing on eating disorders were excluded.

Data sources and review procedures.

A graduate student obtained articles from PubMed and PsycInfo using the following search terms: feeding; feeding patterns; feeding styles; permissive, indulgent, uninvolved, responsive, controlling, restrictive, and forceful feeding; and eating patterns crossed with childhood obesity; childhood overweight; child weight status; and child growth patterns. References of the selected articles were searched to identify additional articles that met the selection criteria. Articles were reviewed to determine the following components of each study: setting, study design and sample, ethnicity and socioeconomic status of study participants, measure(s) used to assess feeding, measure(s) used to assess child growth, and main findings relating feeding to child weight/growth and adiposity.

Results

Study characteristics

Table 1 summarizes characteristics and findings from the 31 articles that met the search criteria. The articles were categorized according to 3 age ranges: predominantly infants/young toddlers (n = 4), predominantly toddlers and preschoolers (n = 20), and predominantly preschoolers and children in early elementary school (n = 7).

TABLE 1.

Characteristics and significant findings of included studies, presented by child age range

Authors (reference), setting	Study design/sample	Ethnicity/SES1	Feeding measure (subscale)2/child weight3	Main findings
Infants and toddlers
Farrow and Blissett (13)UK, Birmingham	Longitudinal69 mother-child (0–12 mo)	Ethnicity and SES not reported	FIS at 6 mo (observation)Weight at birth and 6 and 12 mo	Infant weight gain between 6 and 12 mo was predicted by an interaction between early infant weight gain and maternal control during feeding at 6 mo
Farrow and Blissett (20)UK, Birmingham	Longitudinal62 mother-child (prenatal–2 y)	Ethnicity not reportedmixed SES	CFQ (questionnaire) at 1 y monitoring; restriction; pressure weight SDS at 1 and 2 y	Pressure and restriction at 1 y significantly predicted lower child weight at 2 y
Thompson et al. (10)US, North Carolina	Cross-sectional 150 mother-infant (3–18 mo)	100% B low-income	IFSQ (questionnaire)laissez-faire; pressuring; restrictive, responsive; indulgent WH (length) Z	Responsive, pressuring and indulgent feeding styles had significant, negative association with child WH (length) Z
Worobey et al. (14)US, New Jersey (WIC)	Longitudinal 96 mother-infant dyads (0–12 mo)	24% B; 76% H low-income	MFA (questionnaire)NCAST (observation)maternal sensitivity to infant cues at 3 and 6mo weight gain at 3, 6, and 12 mo	Lower maternal sensitivity to infant satiety cues associated with more weight gain from 6–12 mo, but not between 0–3 mo or 3–6 mo
Toddlers and preschool
Baughcum et al. (41) USKentucky (WIC) Ohio (Pediatric practices)	Cross-sectional453 mother-child (11–24 mo)634 mother-child (23–60 mo)	Predominantly Cmixed SES	IFQ (11–24 mo) (questionnaire)PFQ (23–60 mo) (questionnaire)WH Z and percentiles	No significant relationship between feeding and child overweight.
Blissett and Haycraft (26)UK (Midlands/Cambridge)	Cross-sectional48 parent-child dyads (24–59 mo)	Predominantly Cmiddle-class	CFQ (questionnaire)BMI Z	Higher child BMI was predicted by lower paternal (but not maternal) pressure to eat
Carnell and Wardle (27)England, London	Cross-sectional439 parent-child (mean = 4.4 y; SD = 0.6 y)	Predominantly Cmixed SES	CFQ (questionnaire)pressure; restriction; monitoring PFQ (questionnaire)pushing child to eatPFSQ (questionnaire)prompting; emotional feeding instrumental feeding BMI Z	Scores on CFQ (pressure to eat) and PFQ (pushing to eat) had significant negative association with BMI Z. No significant relationships were found between PFSQ (prompting to eat), CFQ (restriction) and BMI Z
Drucker et al. (21)US, California (laboratory)	Cross-sectional77 mother-child dyads (mean = 3.5 y)	Predominantly C; 80% mothers with college degree	BATMAN (observation) BMI	Child BMI had positive association with number of discouraging remarks (e.g., “that's too much food!”) during feeding.
Faith et al. (23)US, New York	Cross-sectional and prospective cohort971 parent-child dyads (1–5 y; mean = 30.2 mo)	35.8% C;20.5% B 34.5% H; 9.2% Olow-income (WIC)	4 feeding items (questionnaire)BMI Z and percentiles	Restriction is greater among parents with children initially overweight/obese compared with normal weight children
Faith et al. (25)US National Survey	Cross-sectional 1790 children (35–73 mo)	18% H; 30.1% C 51.9% NHNAAmixed-income	Mother-allotted child food choice (questionnaire)BMI Z	Children allowed no food choice by mother had lowest BMI Z but it was not associated with child overweight status
Francis et al. (15)US, Pennsylvania	Cross-sectional197 mother-daughter dyads (5 y)	100% C mixed-income	CFQ (questionnaire)perceived child OWconcerns about child OWrestriction and pressure BMI /triceps and subscapular	Mothers used more restriction when daughters were heavier Mothers used more pressure when daughters were thinner
Galloway et al. (28)US, Pennsylvania	Repeated-measures, within-group experiment27 children(3–5 y)	Predominantly CParents highly educated	CFQ (questionnaire)pressure BMI percentiles	Children who were pressured to eat at home had lower BMI percentile scores
Haycraft and Blissett (42)UK, Cambridgeshire and West Midlands	Cross-sectional23 parent-child dyads (18–67 mo)	Ethnicity not reportedMost parents professionals	CFQ (questionnaire)FMCS (observation)BMI Z	No significant relationship between feeding and child weight
Hoerr et al. (29)US, Texas and Alabama	Cross-sectional715 parent-child dyads (3–5 y)	43% B; 29% H28% Clow-income (Head Start)	CFSQ (questionnaire)authoritative, authoritarian, indulgent, uninvolved BMI Z	BMI Z scores were higher in children of indulgent parents and lower in children of authoritarian parents (high demandingness and low responsiveness)
Hughes et al. (24)US, Texas	Cross-sectional231 parent-child dyads (3–5 y)	45% B 55% Hlow-income (Head Start)	CFSQ (questionnaire)authoritative, authoritarian, indulgent, and uninvolved feeding stylesBMI Z	Negative association between the parent-centered/high control and BMI Z for H girls/boys Negative association between parent-centered/contingency management, child-centered strategies and BMI Z for B boys
Hughes et al. (9)US, Texas	Cross-sectional231 parent-child dyads (3–5 y)	43.7% B56.3% Hlow-income (Head Start)	CFSQ (questionnaire)authoritative, authoritarian, indulgent, and uninvolved BMI Z	Children with indulgent parents were higher in weight compared with children with authoritarian parents
Hughes et al. (30)US, Texas and Alabama	Cross-sectional718 parent-child dyads (3–5 y)	43% B; 29.1% H 27.9% Clow-income (Head Start)	CFSQ (questionnaire)authoritative, authoritarian, indulgent, and uninvolved BMI Z	Indulgent feeding style associated with higher child BMI Z
Johannsen et al. (16)US, South Dakota	Cross-sectional211 parent-child dyads (3–5 y)	96.6% CParents: = 15.1 y school	CFQ (questionnaire) BMI DXA scan	Girls whose fathers were more controlling had a significantly higher percentage fat
Kasemsup and Reicks (43)US, Minnesota	Cross-sectional80 mother-child dyads (3–5 y)	100% Hmonglow-income (WIC)	CFQ (questionnaire) BMI Z	No significant association between maternal child-feeding practices and child BMIHowever, child's BMI Z was positively related to maternal perception of child's weight, and perception of child's weight was related to maternal restriction of child's food intake
Montgomery et al. (44)Scotland, Glasgow	Cross-sectional and longitudinal117 parent-child dyads (3–5 y)	Ethnicity not reportedmixed-income	CFQ (questionnaire)BMI SDS	No significant association between pressure, restriction, monitoring and BMI SDS
O'Connor et al. (45) US Houston, Texas, Northern Alabama	Cross-sectional755 parent-child dyads (3–5 y)	43.8% B 29.8% H 26.4% Clow-income (Head Start)	Feeding practices facilitating child FV intake (questionnaire)Cluster 1 (low-involved) Cluster 2 (non-directive) Cluster 3 (indiscriminate) BMI Z	No significant child BMI Z score differences were found across feeding clusters
Powers et al. (22)US, Ohio	Cross-sectional296 mother-child dyads (24–59 mo)	100% B low-income (WIC)	CFQ (questionnaire)restriction, pressure PFSQ (questionnaire)control CEBQ (questionnaire)desire to drink food responsiveness BMI Z	Maternal pressure to eat negatively associated with child BMI ZMaternal restriction and control was positively associated with BMI Z in children with obese but not nonobese mothers
Sherman et al. (18)US, Arizona	Cross-sectional377 mother-child dyads (3–5 y)	50% MA; 50% Clow-income (WIC)	Maternal Feeding Practices (questionnaire)WH Z and skinfolds	In both ethnic groups, use of bottle to comfort child was associated with child obesity
Sherman et al. (17)US, Arizona	Cross-sectional375 mother-child dyads (mean = 4.3 y)	49.9% H 50.1% Clow-income (WIC)	Maternal feeding practices (questionnaire)bottle given to stop infant cryingWH Z and skinfolds	Bottle given to child to stop crying was associated with child overweight
Preschool and early elementary
Birch and Fisher (36)US, Pennsylvania	Cross-sectional197 mother-daughter dyads (4.6–6.4 y)	100% Cmixed SES	CFQ (questionnaire) WH Z	Mothers’ perception of daughters’ risk of overweight was positively associated with restrictive feeding and in turn daughters diet and relative weight
Brown et al. (46)England, Southern	Cross-sectional518 parent-child dyads (4–7 y)	Predominantly C51% of parents had college degree	Controlling feeding practices (questionnaire)snack overt control; snack covert control; meal overt control; meal covert control; pressure to eatBMI SDS	Controlling feeding was not significantly associated with Child BMI SDS
Fisher and Birch (32)US, Pennsylvania	Cross-sectionalExperiment 1: 31 children (3–5 y)Experiment 2: 37 children (3–6 y)	Experiment 1: ethnicity not reported Experiment 2: 80%C, 15% A 4% B, 1% O	Maternal restriction of child access to snack (questionnaire)WH percentiles	Experiment 2: Restriction was significantly, positively associated with child WH
Joyce and Zimmer-Gembeck (12) Australia	Cross-sectional230 caregiver-child dyads (4–8y)	94% C 30% of caregivers had university degree	CEBQ (questionnaire)food responsiveness; emotional overeating CFQ (questionnaire)restriction PFDQ (questionnaire) BMI Z	Restriction positively associated with child BMI Z
Keller et al. (33)US, New York City	Cross-sectional15 mother-sibling pair dyads (3–7 y)	46.7% B 33.3% H ;20% CSES not reported	CFQ (questionnaire) BMI Z	Mothers reported less pressure and marginally more restriction (P = 0.17) toward heavier than thinner children
Lee et al. (19)US, Central Pennsylvania	Cross-sectional 192 mother-daughter dyads (5–7 y)	100% C Mixed SES	CFQ (questionnaire)BMI and skinfold thickness	Child BMI was negatively associated with pressure; positively associated with restriction
Musher-Eizenman et al. (31)France and US	Cross-sectional97 US parent-child dyads (3.7–6.8 y)122 French parent-child dyads (4.0–6.8)	US sample: 97% Cmiddle-incomeFrench sample: middle-income	CFPQ (questionnaire) BMI	In the US, higher child BMI was related to greater use of food as rewardIn France, higher child BMI was related to more restriction and lower use of food as reward

¹Race/ethnicity: B, black; C, white; H, Hispanic; NHNAA, non-Hispanic/non-African American; MA, Mexican-American; A, Asian; O, other.

²BATMAN, Bob and Tom's Method for Assessing Nutrition (47); CEBQ, Child Eating Behavior Questionnaire (22); CFPQ, Comprehensive Feeding Practices Questionnaire (11); CFSQ, Caregiver's Feeding Styles Questionnaire (9, 29); FIS, Feeding Interaction Scale (48); IFQ, Infant Feeding Questionnaire (41); IFSQ, Infant Feeding Style Questionnaire (10); MFA, Maternal Feeding Attitudes (49); NCAST, Nursing Child Assessment Satellite Training Feeding Scale (50); PFDQ, Parent Feeding Dimensions Questionnaire (12); PFQ, Preschooler Feeding Questionnaire (41); PFSQ, Parental Feeding Style Questionnaire (51).

³WH, weight-for-height; OW, overweight; DXA, dual energy X-ray absorptiometry.

Study setting, design, and sampling.

Most (22/31) articles in the current review present findings from the United States. Beginning in 2006, 7 came from the United Kingdom, 1 from Australia, and 1 from a cross-national comparison of France and the United States. The majority of articles present data derived from cross-sectional (n = 25) rather than longitudinal (n = 3) or repeated-measures (n = 1) designs. Two studies used both a cross-sectional and longitudinal design. The articles varied across the ethnicity or ethnicities of the samples. Many (39%; 12/31) of the studies examined predominantly white samples. One sample consisted entirely of Hmong participants, 2 entirely of black participants, and 11 studies consisted of a bi- or multi-ethnic sample. Five studies reported no information on race/ethnicity.

Measures used to assess feeding and child growth

Feeding measures.

The 31 identified articles used 16 different feeding measures (12 different self-report questionnaires and 4 different feeding observations). Eight articles used subscales from existing feeding measures; 4 used a single subscale. With 14 articles using subscales from the Child Feeding Questionnaire (CFQ) (8), it was the most commonly used feeding measure. The CFQ consists of 4 scales measuring parental perception (perceived parent weight, perceived child weight, parental concern about child weight, and parental responsibility) and 3 scales measuring parental behavior (restriction, pressure, and monitoring). Support for validity of the CFQ has been provided in 3 samples (2 white, 1 Hispanic) in children 5–11 y of age (8). Many studies that used the CFQ included predominantly white samples; however, some measured feeding perceptions and behaviors in black, Hmong, or other multi-ethnic groups. The CFQ was developed to measure children 2–11 y of age, with a focus primarily on feeding restriction and pressure. Since the creation of the CFQ, researchers have continued to develop and validate self-report feeding questionnaires that extend beyond feeding restriction and pressure and can be used in multiple race/ethnic and child age populations (9–12).

Growth measures.

To assess children’s weight status, most studies used weight-for-length percentiles/Z-scores (<2 y) or BMI percentiles/Z-scores (≥2 y). Two studies examined growth velocity during infancy (13, 14) and 5 examined percentage body fat (15–19).

Findings relating feeding to child weight/growth

Infancy/early toddlerhood (n = 4).

All articles examining participants in the infancy/early toddlerhood period found significant relationships between feeding and child weight; however, the direction of the associations was mixed. Using feeding observations, 1 longitudinal study conducted in the UK found that maternal control during feeding interfered with the infant’s ability to regulate weight gain over time (i.e. infants with early rapid weight gain continued to gain weight too quickly and vice versa) (13). Similarly, another study found that infants whose mothers were less sensitive to children’s satiety cues gained significantly more weight from 6 to 12 mo than infants of sensitive mothers (14). Conversely, based on self-report feeding questionnaires, 2 studies (1 in the US and 1 in the UK) found that responsive and/or nonresponsive feeding strategies (i.e. controlling and indulgent) were negatively associated with child weight-for-length Z-scores (10) and weight SD scores (20).

Toddler/preschool period (n = 20).

Seventy-five percent (15/20) of articles presenting data from the toddler/preschool period found significant relationships between feeding and child weight status or adiposity. One-half (n = 10; 9 via self-report questionnaires, 1 via observation) found significant relations with controlling feeding strategies: 4 found positive associations between restrictive feeding and BMI (15, 21, 22) and obesity (23), 1 found parental feeding control in general was negatively associated to BMI Z-scores (24) and positively associated with child adiposity (16), 1 found that children allowed no food choice had the lowest BMI Z-scores (25), and 5 found negative associations between pressuring and BMI (15, 22, 26–28). Five studies (all using self-report questionnaires) found positive associations between indulgent feeding and BMI (9, 29, 30) or overweight/obesity status (17, 18).

Preschool/early elementary school period (n = 7).

Five of 7 articles from the preschool/early elementary school period found significant relationships between feeding and child weight status. Five found positive relationships between restrictive feeding and BMI (12, 19, 31–33) and 2 found negative relationships between pressuring during feeding and BMI (19, 33). One study found a positive relationship between indulgent feeding and BMI (31).

Discussion

The majority (24/31) of the articles reported significant associations between parental feeding and child BMI Z-score, weight status, or adiposity. The most frequent finding (16/31; across the 3 age ranges) was an association between parental feeding control and child weight gain/status (i.e. restriction was related to higher BMI and/or overweight/obesity; pressure during feeding was related to lower BMI/weight gain). Although these findings suggest that controlling feeding strategies may have counterproductive effects on children’s eating behaviors (32), leading to either increased (in the case of restriction) or decreased (in the case of pressure) energy consumption, the cross-sectional design of most studies makes it difficult to interpret the direction of causality. An alternative explanation is that caregivers may respond to concerns regarding their children’s weight status or feeding behavior with controlling feeding practices.

The other feeding domain examined was indulgence. Except for 1 study that included only infants and toddlers (10), findings suggest a positive relationship between indulgent feeding and BMI and/or overweight/obesity (9, 17, 18, 29–31). These findings are consistent with other studies that have found positive relationships among perceived child fussiness and indulgent feeding (34). One possible explanation for these associations, particularly during the toddler/preschool period, is parents’ use of feeding as a strategy to calm and/or regulate their child’s behavior.

Some studies have examined the potential mechanisms linking feeding to child weight status. For example, Hoerr et al. (29) found a negative association between indulgent feeding and children’s intake of fruits and vegetables. These findings are consistent with others who found intake of soda and sweets was positively associated with indulgent feeding (35). Two studies found that children’s selection of food may increase with exposure to restrictive feeding, thereby potentially increasing their risk of overweight/obesity (12, 32).

Responsive feeding measures

Our review found diverse methods of measuring responsive feeding. Although studies differed with respect to study design, setting, sample characteristics, and feeding measurements, several similarities are evident.

First, most studies used cross-sectional (n = 25) rather than longitudinal (n = 3) or repeated-measures (n = 1) designs. Two studies used both a cross-sectional and longitudinal design. Thus, because the direction of effects cannot be determined in cross-sectional studies and given the inconsistency in results and study design from the 3 longitudinal studies, reverse causality may be possible. At least 1 longitudinal study showed that restriction is higher in parents whose children were initially overweight or obese compared with children who were not initially overweight or obese (23). Similarly, others have shown that birth weight influences later feeding behaviors. For example, Farrow and Blissett (20) found that women with lighter-born infants exerted greater pressure to eat at 1 y, while Hurley et al. (34) found that lower birth weight was related to indulgent feeding and higher birth weight was related to restrictive feeding during infancy.

Second, although most studies (n = 22) were conducted in the United States, a few (n = 8) were conducted in other high-income areas (i.e. Europe and Australia). In addition, 1 study included a cross-national sample (France and US). Similar findings were reported across countries, with the positive association between feeding restriction and BMI or overweight/obesity the most common. However, the cross-cultural comparison found that restrictive feeding behaviors were more prevalent in France, whereas indulgent behaviors (i.e. parental use of foods for non-nutritive purposes) were more prevalent in the US (31).

In the sample of French participants, associations between feeding restriction and parents’ perceived responsibility for child’s eating, perception of increased child’s body weight, and parental restrained eating partially explained the association between restriction and child weight (31). These factors, along with parental weight status and psychosocial characteristics, have also been associated with restrictive feeding behaviors in the US (9, 15, 16, 34, 36).

In the U.S. sample, the association between indulgence and parental uncontrolled or emotional eating explained the relationship between indulgence and increased child weight. The latter finding is consistent with another U.S. study that found that congruence of parent-child emotional characteristics (i.e. lack of negative affect) was associated with indulgent parent feeding behaviors (30).

Third, whereas most studies examined children in the middle to older age ranges, there is a recent trend toward including infants and toddlers. The increasing prevalence of overweight/obesity among infants and toddlers (1) and the recent focus on early obesity prevention and development of healthy eating behaviors (37) may explain this trend.

Fourth, race/ethnicity and socioeconomic information varied widely across studies, with many of the earlier studies examining predominantly white, middle-income U.S. populations. Beginning in 2005, the focus largely shifted to low-income, minority populations (e.g. African American and Hispanic participants). This trend coincides with the increased risk of child overweight/obesity among minority and/or low-income populations and their associated early-life risk factors (38). Although the understanding of how the relationships between feeding and weight status vary by demographic characteristics remains limited, some studies suggest that Hispanic and African American populations are more likely to report nonresponsive feeding styles (i.e. controlling, indulgent, and uninvolved) than whites, thereby possibly increasing their risk of overweight/obesity (9, 34).

Finally, 16 different feeding measures were identified in the current review. Most (n = 12) were self-report questionnaires and the remaining (n = 4) were observational techniques. Although the findings appeared consistent regardless of the method (self-report questionnaire vs. observation), few studies have examined the reliability between the 2 methods (39). The theoretical understanding behind feeding behaviors and the validated tools needed to measure them across different child ages and race/ethnic populations have shifted in recent years from a primary focus on control during feeding (e.g. restriction and pressure) to one that includes responsive and indulgent feeding (9, 10).

In the last decade, scientific and theoretical underpinnings based in anthropology, nutrition, and child development research have greatly increased our ability to measure how child-feeding strategies are related to child underweight and obesity in low-, middle- and high-income countries (5). In many low- and middle-income countries, where growth faltering and stunting continue to be highly prevalent in infants and young children, feeding strategies, the mechanisms linking feeding to child weight/growth, and the methods used to measure and intervene efficiently may differ between low- and middle-income and high-income countries (40).

Future directions

Additional research is needed to understand the relationships between responsive feeding and child weight status in high-income countries. Most studies are cross-sectional, making it difficult to determine the direction of effect. For example, does nonresponsive feeding cause child obesity, or do parents of obese children react to concerns about their child’s obesity by using nonresponsive feeding strategies? Three types of research needed include longitudinal studies to better understand causality and the stability of feeding behaviors over time, methodological studies that examine the reliability and validity between and within existing self-report feeding questionnaires and feeding observations along with their age and race/ethnic specificity, and research examining mechanisms (e.g. diet and self-regulation) linking nonresponsive feeding to child weight gain/status. Further, randomized controlled trials are needed to test the efficacy of responsive feeding interventions to treat and to prevent childhood obesity.