Infant Appetite Traits, Feeding Practices, and Child Obesity in Low-Income Hispanic Families

Abstract

Background:

Appetite traits and feeding practices are important determinants of child weight and obesity.

Objectives:

This study examined whether: 1) infant appetite traits were associated with feeding practices; and 2) feeding practices mediated the link between appetite traits and weight-for-age z-scores at age 3 years.

Methods:

We conducted a secondary data analysis from the “Starting Early Program” of low-income, Hispanic mother-child pairs. Appetite traits were assessed using the Baby Eating Behavior Questionnaire. Infant feeding practices were collected using 24-hour dietary recalls and surveys: 1) breastfeeding exclusivity, intensity, and duration; 2) early introduction to complementary foods/liquids; and 3) any 100% fruit juice consumption at age 10 months. Regression and mediation analyses were used to explore associations between appetite, feeding, and weight.

Results:

Higher infant Slowness in Eating scores were associated with greater breastfeeding exclusivity, intensity, and duration, compared to lower Slowness in Eating. Infants with higher Slowness in Eating and Satiety Responsiveness had lower odds of early introduction to complementary foods/liquids. Infants with higher Enjoyment of Food had greater odds of 100% juice consumption. Breastfeeding duration mediated the relationship between higher infant Slowness in Eating and lower weight-for-age z-scores.

Conclusions:

Appetite traits represent potential targets for early life infant feeding interventions.

1. INTRODUCTION

The increasing prevalence of obesity among young children, aged 2 to 5 years old,¹ especially among Hispanic children, has created a serious public health concern in the United States (U.S.) and highlights the need to identify modifiable causes of early obesity. There is a growing evidence that links child appetite traits, which describe characteristics of eating behavior, to factors related to weight gain, such as quantity and frequency of energy intake,² and eating in the absence of hunger.³ These traits can be classified as food approach/obesogenic (Food Responsiveness and Enjoyment of Food) or food avoidance/obesity protective (Slowness in Eating and Satiety Responsiveness) appetite traits.⁴ These traits are found to be related to excess weight and obesity in older children,^5–8 more recently in infants and young children,^9–13 and may elucidate why some children are more susceptible to obesity early in the life course.¹⁴

Infant feeding practices, including limited breastfeeding, the early introduction of complementary foods/liquids, and the provision of sugar-sweetened beverages and juice are recognized as potential determinants of early excess weight gain and obesity. Although the American Academy of Pediatrics (AAP) recommends feeding infants with breast milk exclusively for at least six months, and continuing to breastfeed along with providing complementary foods thereafter, most infants do not receive optimal feeding.¹⁵ According to the Centers for Disease Control and Prevention (CDC) breastfeeding report card, only 26% of infants in U.S. were exclusively breastfed for 6 months in 2018, with the lowest rates of exclusive breastfeeding among Hispanic (23%) and non-Hispanic Black (20%) racial and ethnic minority groups.¹⁶ Studies have also focused on the association of 100% fruit juice intake during early childhood with later obesity.¹³ Although the AAP recommends avoidance of juice and sugar-sweetened beverage consumption during the first year of life,¹⁷ according to the national Feeding Infants and Toddlers Study (FITS) survey, about 27% and 9% of 6-12 months old infants consumed 100% fruit juice and sugar-sweetened beverages in 2016, respectively.¹⁸ The high prevalence of obesity-related infant feeding practices, particularly among Hispanic families,¹⁸ underscore the need to identify related factors. Appetite traits are one possible characteristic that can influence parent feeding practices.¹⁹ Parents of infants with higher levels of food approach/obesogenic appetite traits might be more likely to provide juice or early introduction to solids to satisfy perceived infant hunger.²⁰

While a number of studies have found independent associations between appetite traits and weight in older children,^5–8 less research has studied these associations during infancy.^9–12 To our knowledge, limited studies have explored the associations between infant appetite traits and infant feeding practices,^9,21–25 and none have looked at their potential mediation role in the pathways affecting child weight in low-income and racial/ethnic U.S. minority families. Therefore, this study aimed to assess: 1) associations between infant appetite traits and feeding practices; and 2) whether feeding practices mediate associations between infant appetite traits and later child weight in low-income Hispanic families.

2. Methods

2.1. Study Design

We conducted a secondary analysis of data from the Starting Early Program (StEP), a randomized controlled trial of low-income Hispanic mother-child pairs with the goal of reducing early child obesity.^26,27 The study design has been described elsewhere.^26,27 StEP enrolled women in the third trimester of pregnancy and continued follow-up until child age 3 years. Enrollment and prenatal baseline assessments occurred between August 2012 and December 2014 in a large public hospital in New York City. The institutional review boards of New York University Grossman School of Medicine, and New York City Health+Hospitals approved this study. The StEP trial is registered on clinicaltrials.gov (NCT01541761).

2.2. Study Sample

Women were eligible to participate in the StEP trial if they: 1) were ≥18 years old; 2) had a singleton uncomplicated pregnancy; 3) self-identified as Hispanic/Latina; 4) planned to receive prenatal and pediatric care at the study sites; and 5) were fluent in English or Spanish. Women who had significant medical or psychiatric illness, homelessness, substance use disorder, or severe fetal anomalies were excluded. Participation was offered during a 3^rd trimester prenatal visit and those who chose to participate signed informed consent and completed baseline assessments. Trained research assistants conducted surveys in English or Spanish at a baseline prenatal visit, with follow-up at child ages 3 months, 10 months, 19 months, 2 years, and 3 years. The current analytic sample included mother-child pairs with baseline maternal data, BEBQ measures of infant appetite traits, collected concurrently and retrospectively, infant feeding practices, and child anthropometric data. Figure 1 provides an overview of the present study measures and includes the frequency of intervention and follow-up visits.

Figure 1.

Study Design of the Analytic Sample (n=332)

2.3. Measures

Appetite Trait Measures:

Appetite traits were measured using the Baby Eating Behavior Questionnaire (BEBQ), a scale that has been validated in English and Spanish to measure infant appetite traits both concurrently and retrospectively.²⁸ Mothers completed the BEBQ at the 3-month assessment (n=216 [40.6%]) or retrospectively for 258 subjects at either the 10-month (n=118 [24.9%]), 19-month (n=95 [20%]), 2-year (n=29 [6.1%]), or 3-year (n=16 [3.4%]) assessment time-points related to BEBQ measures at infant age 3 months old. Since BEBQ appetite traits have high internal consistency (Cronbach’s α estimates of 0.73 to 0.81) and were not associated with the infant age at assessment,^12,28 the present analysis combined prospectively and retrospectively collected BEBQ data. In addition, sensitivity analyses were conducted using both 1) models stratified by age of the infants at BEBQ completion (concurrently versus retrospectively) and found similar associations between appetite traits and feeding practices in both samples (results not shown); 2) models with interaction terms for the appetite traits and age of the infants at BEBQ completion (no significant interaction terms were found). The BEBQ was used to measure four infant appetite traits during the milk feeding period, including two food approach/obesogenic appetite traits (Food Responsiveness and Enjoyment of Food), and two food avoidance/obesity protective (Slowness in Eating and Satiety Responsiveness) appetite traits.⁴ The appetite traits, sample questions for each trait, and mean scores are displayed in Table 1.

Table 1.

Baby Eating Behavior Questionnaires factors and sample questions¹

Appetitive Traits	Items (n)	Sample Question	Mean Scores (SD) 2
BEBQ
Food Responsiveness	5	My baby is always demanding a feed My baby frequently wanted more milk than I provided	2.2 (1.0)
Enjoyment of Food	4	My baby loves milk My baby seemed contented while feeding	4.7 (0.4)
Slowness in Eating	4	My baby took >30 minutes to finish feeding My baby sucked more and more slowly during the course of a feed	2.5 (0.8)
Satiety Responsiveness	5	My baby gets full up easily My baby found it difficult to manage a complete feed	2.0 (0.9)

¹(Llewellyn et al., 2011);

²1= never, 2= rarely, 3= sometimes, 4= often, 5= always

Infant Feeding Practices Measures:

Breastfeeding exclusivity and intensity were measured via a 24-hour diet recall at infant age 3 months. Exclusive breastfeeding (EBF) was defined as feeding the infant nothing besides breast milk, except medicine or vitamins.²⁶ We dichotomized EBF as “EBF at 3 months” vs. “No breastfeeding or EBF<3 months.” Breastfeeding intensity was defined as the percentage of all feedings in the past 24 hours that were breast milk. A continuous measure of breastfeeding intensity was defined as the total breast milk proportion of all feedings in the past 24 hours. Breastfeeding intensity score was categorized as “low” for <20%, “medium” for 20%-80%, and “high” for >80% of all feeds from breast milk.²⁹

Breastfeeding duration was assessed via questions adapted from the Infant Feeding Practices Study II.³⁰ Mothers were asked at all survey time points if they had ever fed breast milk and at what age they completely stopped breastfeeding. We used both continuous breastfeeding duration in months (a single variable generated from data collected at several times during the first 24 months of a child’s life) and breastfeeding duration categories: 1) “Any breastfeeding at 3 months” vs. “No breastfeeding or breastfeeding <3 months,” and 2) “Any breastfeeding ≥6 months” vs. “No breastfeeding or breastfeeding <6 months.”

Mothers reported the date of introduction and average frequency of providing any complementary foods (cereal, fruit, and vegetables) or liquids (juice, tea, or water) via survey questions. Introduction of complementary foods/liquids was categorized as early if it was done prior to infant age 4 months vs. ≥4 months old. Mothers were also asked if they have ever given 100% fruit juice at the time of the 10-month assessment.³¹ We dichotomized juice intake as “Any 100% fruit juice given at infant age 10 months” vs. “No 100% fruit juice at 10 months.”

Anthropometric Measures (Outcome Variable):

Medical record review was used to collect child weight at age 3 years measured by medical assistants at clinical visits.³² While the original aim for the StEP trial was to use BMI-for-age and weight-for-length/height z-scores, weight-for-age z-scores (WFAz) were used because biologically implausible variability in clinically measured lengths/heights was identified^27,33 and WFA percentiles correspond well with the BMI cutoff for obesity in children.³⁴ Child WFAz at age 3 years was calculated based on the CDC 2000 growth charts.³⁵

Baseline Characteristics:

Maternal age, parity, marital status, country of birth, education, depressive symptoms (using the Patient Health Questionnaire-9),³⁶ household food insecurity (using the Core Food Security Module from the U.S. Department of Agriculture),³⁷ and participation in supplemental food assistance programs were obtained at baseline. Maternal pre-pregnancy BMI and gestational weight gain were calculated using weight and height from medical record review.³⁸ Child sex, gestational age, birthweight, and delivery mode were also obtained from medical record review.

2.4. Statistical Analysis

First, the mean scores for the four infant appetite traits were calculated (Table 1). The normality test was also performed for each appetite trait using the Shapiro-Wilk test, which showed that the data were not normally distributed. Therefore, the scores were log-transformed and all analyses were performed with the log-transformed data. Second, we used unadjusted simple linear regression to test associations between infant appetite traits and continuous feeding practice measures (breastfeeding intensity and duration). Third, we used unadjusted logistic regressions to assess associations between infant appetite traits and categorized feeding practices. All linear and logistic regression models were subsequently adjusted for potential covariates that were selected a priori based on their known associations with child feeding practices and weight from prior literature. These included child sex, birthweight, maternal weight-related factors (pre-pregnancy BMI, gestational weight gain), socio-demographic factors (age, education, marital status, country of birth), psychosocial factors (depression, household food insecurity, WIC participation), mode of delivery, age of the infants at BEBQ completion, and intervention group status. All regression models were analyzed with and without the interaction term between each appetite trait and the intervention group status.

Next, to test whether feeding practices mediated the relationships between infant appetite traits and child WFAz at 3 years, we used separate simple linear regressions to assess direct associations between infant appetite traits, feeding practices, and WFAz at 3 years, followed by bootstrapping (5000 bootstrap replicates) for significance testing. Mediation analyses were regression-based, as described in Hayes (2013).³⁹ The SPSS Macro PROCESS , developed by Hayes, estimates separate regression models for the mediator and the outcome to estimate the strength of each pathway in the mediation model (both direct and indirect components). If significantly related to child WFAz, each feeding practice, as a mediator in the relationship between each appetite trait and child WFAz, was tested using the PROCESS macro for SPSS version 25 (“Model 4”). Indirect effects were assessed using bootstrapping. Statistical analyses were performed using SPSS version 25.0 and Stata version 15.0.

3. RESULTS

3.1. Sample Characteristics

The StEP trial randomized 533 pregnant women. Of these, 474 completed the BEBQ, 439 participants had complete data on infant appetite traits and breastfeeding exclusivity, intensity, and duration and early introduction of complementary foods/liquids, and 411 participants had complete data on infant appetite traits and juice intake. Since log-transformation did not change any of the findings, therefore we reported the results from the non-transformed data. Variables describing the baseline socio-demographic, psychosocial, and weight-related characteristics for those who were included in this analysis (n=439) are listed in Table 2. Only 3% of the sample (n=12) were born preterm (<37 wk. gestational age); however, the majority of the preterm babies were born at 36 wk. gestation age and because excluding them did not change the results, they were included in the analyses. A total of 332 participants who had complete data on WFAz at 3 years, infant appetite traits, and feeding practices, were included in the mediation analysis. (Figure 2)

Table 2.

Baseline maternal and child characteristics of StEP participants

Characteristicsa	Analytic Sample (n=439)
Child
Male sex	215 (49)
Birth weight (kg, mean ±SD)	3.4 ± 0.5
Premature <37 wk. gestational age	12 (3)
LGA	33 (8)
Intervention group status	218 (49.8)
Expectant Mother Socio-demographic Factors
Age, (mean ±SD)	28.6 ± 5.8
Primiparous	154 (35)
Married or living as married	326 (74)
Born outside of US	355 (81)
Completed high school	198 (45)
Expectant Mother Psychosocial Factors
Depressive symptoms	145 (33)
Household food insecurity	133 (30)
WIC participation	332 (76)
SNAP participation	160 (36)
Expectant Mother Weight-Related Factors
Gestational weight gain (kg, mean ±SD)	9.8 ± 5.2
Pre-pregnancy BMI (mean ±SD)	27.6 ± 5.7
Expectant Mother Mode of Delivery
Vaginal delivery	352 (80)

^aValues are n (%) unless otherwise stated; LGA, large for gestational age; SNAP, Supplemental Nutrition Assistance Program; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children

Figure 2:

Path Analysis – Breastfeeding Duration Mediates the Relationship between the Infant Appetite Trait of Slowness in Eating and Child Weight-for-Age Z-Scores at 3 Years^a

^aValues shown are unstandardized regression coefficients (standard error)

^b Unadjusted, p<0.001

^c Adjusted for Slowness in Eating at 3 months, p=0.01

^d Unadjusted, p=0.036

^e Adjusted for breastfeeding duration, p=0.15

WFAz: Weight-for-Age Z-Scores

3.2. Associations between Infant Appetite Traits and Infant Feeding Practices

Associations between infant appetite traits and continuous measures of breastfeeding intensity and duration are shown in Table 3. In the unadjusted model, Slowness in Eating was associated with higher breastfeeding intensity scores at age 3 months (B=12.48, 95%CI=8.1-16.9, p<0.001) and longer breastfeeding duration in months (B=2.4, 95%CI=1.5-3.3, p<0.001). These associations remained significant after adjusting for covariates, with higher breastfeeding intensity (B=12.73, 95%CI=8.4-17.1, p<0.001) and longer duration (B=2.43, 95%CI=1.5-3.3, p<0.001). No other appetite traits were associated with breastfeeding intensity or duration.

Table 3.

Associations between infant appetite traits and BF intensity and duration^a

	Unadjusted Model				Adjusted Modelb
Appetite Traits	B	SE	95% C.I.c	P	B	SE	95% C.I.c	P
	BF Intensity Score d
Food Responsiveness	−0.16	1.95	(−3.98, 3.66)	0.93	−0.10	1.96	(−3.94, 3.79)	0.96
Enjoyment of Food	5.12	4.65	(−4.03, 14.26)	0.27	4.20	4.67	(−4.93, 13.34)	0.36
Slowness in Eating	12.48	2.23	(8.10, 16.87)	<0.001	12.73	2.22	(8.37, 17.08)	<0.001
Satiety Responsiveness	2.82	2.23	(−1.55, 7.19)	0.21	3.54	2.24	(−0.85, 7.93)	0.11
Total BF Duration, months
Food Responsiveness	−0.29	0.39	(−1.06, 0.48)	0.46	−0.32	0.39	(−1.10, 0.45)	0.41
Enjoyment of Food	0.27	0.95	(−1.61, 2.14)	0.78	0.06	0.95	(−1.83, 1.89)	0.97
Slowness in Eating	2.40	0.46	(1.50, 3.30)	<0.001	2.43	0.45	(1.54, 3.31)	<0.001
Satiety Responsiveness	0.35	0.45	(−0.54, 1.23)	0.45	0.46	0.45	(−0.33, 1.45)	0.31

^aLinear regression analyses were used to assess the associations between each appetite trait and the continuous infant feeding practices.

^bCovariates: child sex, birthweight, maternal pre-pregnancy BMI, gestational weight gain, age, education, marriage status, non US vs. US born, depression, household food insecurity, WIC participation, mode of delivery, age of the infants at BEBQ completion, and intervention group status.

^c95.0% Confidence Interval; Dependent variables: BF duration and BF intensity. Significant P-values (<0.05) are bolded.

^dBF intensity: the percentage (ranging from 0%-100%) of all feedings in the past 24 hours that were breast milk.

Associations between appetite traits and other feeding practices are shown in Table 4. In both unadjusted and adjusted models, Slowness in Eating was associated with higher odds of EBF at 3 months (aOR=1.7, 95%CI=1.3-2.2, p<0.001), any breastfeeding at 3 months (aOR=2.0, 95%CI=1.4-2.8, p<0.001), and total breastfeeding duration ≥6 months (aOR=2.1, 95%CI=1.5-2.8, p<0.001). Similarly, Slowness in Eating was associated with higher odds of medium and high breastfeeding intensity scores at 3 months compared to the reference group (low breastfeeding intensity) (aOR=1.8, 95%CI=1.2-2.7, p<0.01 [Medium intensity]; aOR=2.5, 95%CI=1.7-3.6, p<0.001 [High intensity]). Infants with higher Slowness in Eating and higher Satiety Responsiveness had lower odds of receiving early introduction of complementary foods/liquids in both unadjusted models and adjusted models (aOR=0.6, 95%CI=0.4-0.9, p=0.04 [Slowness in Eating]; aOR=0.5, 95%CI=0.3-0.8, p<0.01 [Satiety Responsiveness]). Infants with higher Enjoyment of Food had greater odds of any 100% fruit juice intake at age 10 months after adjusting for covariates (aOR=1.8, 95%CI=1.1-3.0 p=0.02). The other appetite traits were not significantly related to these feeding practices. The interaction term between each appetite trait and the intervention group status was not significant in any of the models and therefore there were no significant interactions between appetite traits at infant age 3 months and the intervention group status.

Table 4.

Associations between infant appetite traits and categorized infant feeding practices^a

	Unadjusted Model		Adjusted Modelb
Appetite Traits	OR (95% C.I.)c	P	aORc (95% C.I.)c	P
Exclusive BF (EBF) at 3 Months
Not EBF at 3 months	1.00	Reference	1.00	Reference
Food Responsiveness	0.98 (0.81, 1.20)	0.86	0.99 (0.81, 1.21)	0.89
Enjoyment of Food	1.21 (0.75, 1.96)	0.44	1.22 (0.74, 2.00)	0.43
Slowness in Eating	1.64 (1.29, 2.10)	<0.001	1.69 (1.32, 2.17)	<0.001
Satiety Responsiveness	1.07 (0.86, 1.35)	0.53	1.14 (0.91, 1.44)	0.26
Any BF at 3 Months
No BF at 3 months	1.00	Reference	1.00	Reference
Food Responsiveness	0.97 (0.76, 1.23)	0.98	1.01 (0.78, 1.27)	0.98
Enjoyment of Food	0.90 (0.50, 1.62)	0.73	0.84 (0.46, 1.53)	0.56
Slowness in Eating	1.91 (1.37, 2.65)	<0.001	1.98 (1.41, 2.78)	<0.001
Satiety Responsiveness	1.14 (0.85, 1.52)	0.38	1.21 (0.90, 1.63)	0.20
Medium BF Intensity Score at 3 Months (≥20% - ≤80)
Low BF intensity (<20%)	1.00	Reference	1.00	Reference
Food Responsiveness	0.94 (0.72, 1.24)	0.68	0.95 (0.71, 1.27)	0.73
Enjoyment of Food	1.46 (0.77, 2.76)	0.25	1.26 (0.65, 2.44)	0.49
Slowness in Eating	1.76 (1.22, 2.55)	<0.01	1.83 (1.24, 2.70)	<0.01
Satiety Responsiveness	1.12 (0.82,1.55)	0.47	1.15 (0.82,1.61)	0.41
High BF Intensity Score (>80%)
Low BF intensity (<20%)	1.00	Reference	1.00	Reference
Food Responsiveness	0.97 (0.75, 1.24)	0.81	0.96 (0.74, 1.25)	0.76
Enjoyment of Food	1.37 (0.78, 2.42)	0.27	1.28 (0.71, 2.29)	0.41
Slowness in Eating	2.41 (1.70, 3.42)	<0.001	2.51 (1.74, 3.62)	<0.001
Satiety Responsiveness	1.16 (0.86, 1.55)	0.34	1.20 (0.88, 1.64)	0.24
BF Duration≥6 Months
BF duration<6 months	1.00	Reference	1.00	Reference
Food Responsiveness	0.95 (0.78, 1.16)	0.60	0.95 (0.77, 1.16)	0.62
Enjoyment of Food	1.04 (0.65, 1.67)	0.87	0.96 (0.58, 1.61)	0.85
Slowness in Eating	2.17 (1.64, 2.88)	<0.001	2.05 (1.52, 2.76)	<0.001
Satiety Responsiveness	0.97 (0.77, 1.22)	0.81	1.02 (0.81, 1.30)	0.86
Introduction of Complementary Foods/Liquids
Early introduction to complementary foods/liquids	1.00	Reference	1.00	Reference
Food Responsiveness	1.12 (0.48, 2.51)	0.79	1.29 (0.53, 3.16)	0.57
Enjoyment of Food	1.06 (0.76, 1.48)	0.74	0.99 (0.69, 1.44)	0.95
Slowness in Eating	0.62 (0.39, 0.97)	0.036	0.60 (0.37, 0.98)	0.04
Satiety Responsiveness	0.52 (0.32, 0.84)	<0.01	0.50 (0.30, 0.84)	<0.01
Any 100% Fruit Juice Given at Infant Age 10 Months
No 100% Fruit Juice at infant age 10 months	1.00	Reference	1.00	Reference
Food Responsiveness	1.20 (0.96, 1.50)	0.12	1.16 (0.92, 1.45)	0.21
Enjoyment of Food	1.53 (0.95, 2.48)	0.08	1.70 (1.04, 2.77)	0.03
Slowness in Eating	0.91 (0.70, 1.17)	0.47	0.89 (0.69, 1.16)	0.39
Satiety Responsiveness	1.09 (0.85, 1.42)	0.46	1.06 (0.82, 1.38)	0.65

^aBinomial logistic regression was used to assess the associations between each appetite trait and the categorized infant feeding practices.

^bCovariates: child sex, birthweight, maternal pre-pregnancy BMI, gestational weight gain, age, education, marriage status, non US vs. US born, depression, household food insecurity, WIC participation, mode of delivery, age of the infants at BEBQ completion, and intervention group status.

^caOR: Adjusted odds ratio; Dependent variable: infant feeding practices; Significant P-values (<0.05) are bolded.

3.3. Appetite Traits, Feeding Practices, and Child Weight

In our prior study, Slowness in Eating was the only infant appetite trait that had direct associations with child WFAz at 3 years.¹³ Therefore, the current analysis only explored mediation for Slowness in Eating. Path analysis was used to determine whether breastfeeding duration (months) mediated associations between Slowness in Eating and WFAz at 3 years. Four criteria for mediation were found (Figure 2): 1) Slowness in Eating was directly associated with lower child WFAz at 3 years (B=−0.16, 95%CI=−0.31 to −0.01; p<0.05,); 2) Slowness in Eating was associated with longer breastfeeding duration (B=2.11, 95%CI=1.04-3.17; p<0.001); 3) Longer breastfeeding duration was associated with lower child WFAz at 3 years (B=−0.02, 95%CI=−0.04 to −0.01; p=0.01); and 4) the association between Slowness in Eating and child WFAz at 3 years decreased after adjusting for breastfeeding duration (B=−0.12, 95%CI=−0.28 to 0.03; p=0.15 [direct effect]). The overall indirect effect (mediation pathway) was significant (B=−0.04, 95%CI=−0.09 to −0.01; p<0.001), documenting mediation. Mediation was not found for the other feeding practices associated with Slowness in Eating (breastfeeding exclusivity, intensity, introduction of complementary foods/liquids, and any 100% fruit juice).

4. DISCUSSION

In this randomized controlled trial of low-income Hispanic mother-child pairs, we found that infant appetite traits were associated with feeding practices related to child weight gain and obesity risk, such as breastfeeding exclusivity, intensity, and duration, and early introduction of complementary foods/liquids and juice. Slowness in Eating was associated with breastfeeding (greater exclusivity, higher intensity, and longer duration) and lower child WFAz at age 3 years; and the relationship between Slowness in Eating and WFAz was mediated by breastfeeding duration. Although many studies have reported associations between breastfeeding and child weight,⁴⁰ few have explored associations between infant appetite traits and breastfeeding^21–24,41 and to our knowledge only one study reported similar associations between Slowness in Eating, breastfeeding duration, and child weight.²⁴ However, the study did not assess the possible mediation effect of breastfeeding on the link between Slowness of Eating and child weight. The Growing Up in Singapore towards Healthy Outcomes (GUSTO) cohort of 386 Singaporean children showed that children who had overweight or obesity demonstrated faster eating rates and higher energy intake associated with larger bite size and shorter oral exposure time per bite.⁴² However, this study did not include appetite traits collected with the Child Eating Behavior Questionnaire and did not report on associations with infant appetite traits and feeding practices. Another study from the GUSTO trial (n=210) also found that infant appetite traits at 3 months but not 12 months were associated with BMI and weight gain over the first 2 years of life,¹⁰ but did not specifically report on associations with feeding mode and practices during infancy. While the UPBEAT trial of 353 mothers and infant pairs in the United Kingdom, which assessed the relationship between feeding mode, infant growth and appetite traits in 6-month-old infants of women with obesity, found that formula fed infants gained more weight than those who were exclusively breastfed, this association was not related to the food avoidant trait of Slowness in Eating, and it did not report later child weight status.⁹ The BASELINE trial found that 1) appetite traits assessed at age 2 years (n=1,189) were associated with 2 year old child weight status, 2) food avoidant traits of Slowness in Eating and Satiety Responsiveness were associated with underweight at 2 years and 3) food approach traits of Enjoyment of Food and Food Responsiveness were associated with overweight/obesity.⁶ Overall, while there are similarities in these analyses, our study advanced this work by exploring how infant appetite traits relate to several aspects of infant feeding including milk type, juice and introduction to solids/liquids, in a low-income U.S. Hispanic population. This research suggests that one mechanism by which infant appetite traits affect weight trajectories may be through influencing obesity promoting parent feeding practices during infancy and early childhood.

Several mechanisms are proposed to explain the link between breastfeeding and obesity risk, including breast milk composition, appetite-regulating hormones in breast milk,⁴³ and infant self-regulation of milk intake.⁴⁴ The mediation effect of breastfeeding on the link between Slowness in Eating and child weight may represent an additional factor to explain how breastfeeding and child weight are related. Differing breast and formula milk compositions may explain how breastfeeding duration mediates associations between appetite traits and child weight. Compared to formula-fed infants, breastfed infants have lower levels of leptin and ghrelin,⁴⁵ which may promote response to eating and reduce hunger during infancy and early childhood.⁴⁶ Breastfed infants have higher Glucagon-like peptide-1 and peptide YY levels than their formula-fed counterparts, which inhibit gastric acid secretion and emptying and suppress energy intake through the sensation of satiety.^45,47 It is shown that both breastfeeding duration and Slowness in Eating increase response to satiety²⁴ and given the fact that breastfed infants have to work harder to suckle and self-regulate milk than those fed from bottle,⁴⁴ infants with higher Slowness in Eating who are breastfed for longer duration may have lower weight gain and obesity risk. Additional research is warranted to examine in more depth the pathways between appetite traits, breastfeeding and child weight.

Higher Slowness in Eating and Satiety Responsiveness were associated with lower odds of early introduction of complementary foods/liquids. Infant feeding guidelines from the AAP and the USDA recommend complementary food introduction between ages 4 and 6 months.^48,49 Our study findings support that those with higher levels of the obesity-protective appetite traits (Slowness in Eating and Satiety Responsiveness) have greater odds of receiving optimal feeding after 4 months of age. However, our findings are not consistent with findings of the PEAS Study 217 mother and infant pairs in the U.S. The PEAS trial reported a positive association between Enjoyment of Food, an obesogenic appetite trait, at age 6 months old with age at introduction to solids. However, this study did not report any relationships related to the food avoidant traits of Slowness in Eating and Satiety Responsiveness and age at introduction to solids, and it did not report any associations between appetite traits, feeding mode, and later child weight status.²⁵

These associations between infant appetite traits on obesity-related feeding practices,^50,51 provide a potential mechanism to explain how these infant traits could affect child weight. Parents who formula feed have greater odds of introducing complementary foods before age 4 months.⁵² Infants with higher Slowness in Eating and Satiety Responsiveness may suckle less milk and respond to satiety faster, which may consequently lower risk of obesity.

Higher Enjoyment of Food was associated with greater odds of juice intake at age 10 months. Given that the AAP also recommends no fruit juice for infants prior to 12 months,¹⁷ our findings support that higher levels of obesogenic traits (e.g., Enjoyment of Food) may lead to practices that potentially increase obesity risk. The few studies that have looked at the associations of appetite traits and other feeding practices were mainly in older children and have reported inconsistent findings.^2,5,43,53 The PEAS study reported an inverse association between Slowness in Eating and sweet food/beverage intake frequency at age 12 months old.²⁵ However, this study did not specifically report on associations with 100% fruit juice intake and later child weight. A study of 187 low-income Hispanic preschool-aged children reported a positive association between Enjoyment of Food and snacking frequency, and consequent energy and sugar intake in children with overweight or obesity. However, the opposite relationship was seen for healthy weight children.² Another study of 1,211 Australian and British preschool-aged children showed an inverse relationship between Enjoyment of Food and snacking frequency.⁵³ A few studies have shown a positive association between Enjoyment of Food with preferences for nutrient-poor food.⁵³ Manifestation of self-regulation of food intake in late infancy and toddlerhood may explain how early introduction of solids and sugary beverages are related to child weight.^43,54 One reason parents may be more likely to introduce juice to infants with higher Enjoyment of Food is because they may be happy to foster this enjoyment and may have difficulty setting healthy limits. This may also be related to the higher sugar concentration in juices than other non-juice drinks.

This study has some limitations. Child appetite traits were reported by mothers, therefore, the findings might be affected by maternal beliefs, perceptions, and experiences about eating behaviors, practices, and the weight of the child.^55,56 However, the validity of maternal reports of child eating behaviors by independent observations has been previously reported.⁵⁷ Another limitation was the likelihood of recall bias because of inclusion of the retrospective BEBQ data. However, in the present study, appetite trait scores were not associated with the age of the infants at BEBQ completion and a sensitivity analysis, stratified by prospective versus retrospective data, found similar associations in both samples. To avoid bias from retrospectively collected data, future studies should query mothers about their infant’s appetitive traits currently. In addition, types of 100% fruit juice were not assessed in this analysis. Another limitation was obtaining anthropometrics data from medical records, and the inability to utilize BMI due to biologically implausible heights. Finally, the participants were primarily low-income Hispanic mother-child pairs; therefore, results may not be generalizable to other age groups and populations.

In summary, we found that infants with higher obesity protective appetite traits had healthier infant feeding practices. Slowness in Eating was associated with greater breastfeeding exclusivity, intensity, and duration, and breastfeeding duration mediated the link between this trait and child weight. In addition, Slowness in Eating and Satiety Responsiveness had lower risk of early introduction to complementary foods/liquids. Those with higher levels of the obesogenic appetite trait Enjoyment of Food had greater odds of 100% fruit juice intake at 10 months. Our findings have implications for the design of early obesity prevention interventions. Infant appetite traits may influence infant feeding practices, which consequently lead to excess child weight gain and obesity risk. To better understand the underlying mechanisms linking appetite traits to child diet and weight, larger scale prospective population-based longitudinal studies in diverse populations are warranted.