Associations of Less Healthy Snack Food Consumption with Infant Weight-for-Length Z-Score Trajectories: Findings from the Nurture Cohort Study

Little is known about the impact of less healthy snack foods on weight trajectories during infancy. This secondary analysis of data from the Nurture cohort explored prospective associations of less healthy snack foods with infant weight trajectories. Pregnant women were recruited and, upon delivery of a single live infant, 666 mothers agreed to participate. Mothers completed sociodemographic and infant feeding questionnaires, and infant anthropometrics were collected during home visits at 3, 6, 9, and 12 months. Less healthy snack food consumption was assessed by asking how frequently baby snacks and sweets were consumed each day during the previous three months. Multilevel growth curve models explored associations of baby snacks and sweets with infant weight-for-length (WFL) z-scores. On average, mothers were 27 years old, 71.5% were non-Hispanic Black, and 55.4% had household incomes of ≤$20,000/year. Consumption of less healthy snack foods increased during infancy with a median intake of 3.0 baby snacks/day and 0.7 sweets/day between 10 and 12 months. Growth curve models showed that infants who consumed sweets >2 times/day had significantly higher WFL z-scores during the second half of infancy compared to infants who never consumed sweets. Less healthy snacks may contribute to the risk of obesity during infancy and promoting healthy snack food choices during this critical time is important.


Introduction
Food preferences and dietary patterns that impact weight trajectories emerge during infancy [1]. Recent national data suggest that 8% of infants and toddlers are at risk for obesity (weight-for-length (WFL) ≥95th percentile) with non-Hispanic Black infants and toddlers at greater risk compared to non-Hispanic Whites [2]. Foods and beverages consumed during infancy influence food preferences and subsequent dietary patterns [3]. For example, higher consumption of fruits and vegetables [4], sweet desserts [3], and sugar-sweetened beverages (SSB) [5] are associated with higher consumption in later childhood. Less healthy dietary patterns, which include foods high in added sugars, sodium, and saturated fats, are associated with an increased risk for obesity during infancy [6] and later

Study Design and Participants
This was a secondary analysis of data from the Nurture study, a prospective observational birth cohort of predominantly non-Hispanic Black mothers and their infants residing in the Southeastern USA [28]. The Nurture study was designed to explore longitudinal associations between various infant caregivers and infant adiposity during the first year of life. Women between 20 and 36 weeks' gestation were recruited from a county health department prenatal clinic and a private prenatal clinic in Durham, North Carolina from 2013 to 2015. Recruited mothers were ≥18 years of age with a singleton pregnancy with no known congenital abnormalities. After delivery, mothers confirmed continued interest in participating or were excluded if: no longer interested, their infants were born before 37 weeks gestation or were unable to take breastmilk or formula by mouth at hospital discharge. A total of 666 mother-infant dyads were enrolled in the study. Details on the study design have been provided elsewhere [28]. Mothers provided written informed consent and parental permission for their infants. All procedures were approved by Duke University Medical Center Institutional Review Board (human subjects committee, Pro 0036242).

Measures
Data collection occurred from 2013 to 2016. Trained data collectors conducted four home visits when infants were 3, 6, 9, and 12 months of age. Mothers completed sociodemographic and infant feeding questionnaires, and infant heights and weights were measured during each home visit. Mothers reported how frequently infants consumed foods and beverages each day during the previous three months using items from the Infant Feeding Practices Study II (IFPS II) [29] and the Feeding Infants and Toddlers Study (FITS) [30]. Items included how frequently infants consumed baby snacks (teething biscuits, puffs, and melts), sweets (cookies, cakes, or candy), SSBs, fruits (not including fruit juice), vegetables (not including vegetable juice), dairy (yogurt and cheese), protein (meat, fish, and eggs), grains (breakfast cereals, crackers, bread, pasta, and rice), breastmilk, and infant formula each day. Based on AAP recommendations for the introduction of solid foods [14], any foods and beverages (other than breastmilk and infant formula) consumed between birth and 3 months of age were further categorized as early introduction to solid foods.

Sociodemographics
Sociodemographic characteristics were collected at recruitment and during each home visit. Maternal variables of interest included age, pre-pregnancy body mass index (BMI), race (black, white, or other), education (≤high school diploma, some college, college graduate, or graduate degree), household income (≤$20,000, $20,001-$40,000, ≥$40,001), Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) status, and total number of weeks of any breastfeeding between birth and 12 months. Infant variables of interest included gender, birth weight for gestational age (WGA) z-scores, and WFL z-scores.

Exposure Variable (Categorical)-Baby Snacks and Sweets
Less healthy snack food consumption during the first year of life was assessed using two items: 1) "How often was he or she (their infant) fed baby snacks (teething biscuits, puffs, or melts) during the month" and 2) "How often was he or she [their infant] fed sweets (cookies, cakes, or candy) during the month". Response options for both items were: 0 = never, 1 = just to try, 2 = sometimes but less than once/day, and from 3 = 1 time/day to 7 = 5 or more times/day. To capture important transitions in infant feeding, monthly responses for baby snacks and sweets were averaged across three months (4-6 months, 7-9 months, and 10-12 months) to create an average score for each time point. To reflect AAP recommendations for snacking frequency, averaged scores for baby snacks were further categorized as: 0 = never (never, just to try, or less than once/day), 1 = sometimes (1-3 times/day), or 2 = often (>3 times/day). Given the narrow distribution of scores for sweets, averaged scores were categorized as: 0 = never (never, just to try, or less than once/day), 1 = sometimes (1-2 times/day), or 2 = often (>2 times/day).

Outcome Variable (Continuous)-Infant WFL z-Scores
Standardized measurements of infant recumbent length (ShorrBoard Portable Length Board, Issaquah, WA) and weight (Seca Infant Scale, Hanover, MD) were collected in triplicate by trained staff during the four home visits. An average of the three measurements was used to calculate ageand sex-specific WFL z-scores at 3, 6, 9, and 12 months using World Health Organization reference standards [31].

Covariates
Several mother and infant sociodemographic characteristics were examined as possible covariates based on previous findings of an association with infant weight status (pre-pregnancy BMI, birth WGA z-scores, total number of weeks any breastfeeding (non-exclusive), and early introduction of solid foods) [17]. Other possible covariates were examined based on research suggesting an association with infant feeding and infant weight status (mother's age, race, education, household income, and infant gender) [29]. Covariates were included in adjusted models when the magnitude of the association between baby snacks or sweets consumed and WFL z-scores changed by approximately 10% when added separately to the model [32].

Statistical Analyses
Descriptive statistics including means and standard deviations (SD) or medians and interquartile ranges (IQR) for continuous variables and frequencies and percentages for categorical variables were used to summarize sociodemographic characteristics and infant feeding. Spearman's rho correlation coefficients were computed to examine the associations between foods and SSBs (as continuous variables) consumed averaged across three time points (i.e., 4-12 months). Multilevel growth curve models were used to explore the prospective associations of baby snacks and sweets consumed between 4-6 months, 7-9 months, and 10-12 months with infant WFL z-scores at 6 months, 9 months, and 12 months. These models are appropriate for longitudinal data with repeated measures, and were built following published guidelines [33], with infant age as the measure of time point (level 1) nested within infants (level 2). Baby snacks and sweets were modeled as fixed effects with individual infants' slopes and intercepts modeled as random effects. Models were estimated using restricted maximum likelihood estimation (REML) [33], and a change in Bayesian Information Criteria (BIC) of >10 was used to indicate significant improvement in model fit [34]. All participants who completed at least one home visit were included in models. First, an unconditional means model (model 1) with no predictors was estimated and used to calculate the intraclass correlation coefficient (ICC). A larger ICC indicates more between-infant variation and a smaller ICC indicates more within-infant variation in WFL z-scores. Second, an unconditional (unadjusted) growth model (model 2) was estimated to examine the impact of time point as a fixed effect on infant WFL z-scores. Time point was centered at 4-6 months to reflect recommendations for solid food introduction, such that the intercept represented mean WFL z-scores at 6 months, and the slope represented change in mean WFL z-scores per time point. Next, conditional (adjusted) growth models were estimated to examine the effects of level 2 categorical predictors, baby snacks (model 3) and sweets (model 4). Baby snacks and sweets were modeled as never (reference), sometimes, and often to examine differences in less healthy snack food frequency and timing on WFL z-scores. Both models were adjusted to control for the potential confounding effects of infant birth WGA z-scores and total number of weeks of any breastfeeding (i.e., the only covariates changing the magnitude of the association by approximately 10%). Interactions between baby snacks and sweets and maternal covariates (race, pre-pregnancy BMI, and income) were explored and were not significant at p < 0.05. All analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC, USA), and a p value of < 0.05 was used to determine statistical significance. Table 1 shows sociodemographic characteristics of mother-infant dyads participating in the Nurture study. Mothers were on average 27.1 (SD = 5.8) years of age with a mean pre-pregnancy BMI of 29.9 (SD = 9.3). Mothers were predominantly non-Hispanic Black (71.5%) with over half reporting a household income of ≤$20,000/year (55.4%). Nearly half of infants were female (48.8%), and birth WGA z-scores (M = −0.3, SD = 0.9) and WFL z-scores (across all time points) were within the normal range. Infants were non-exclusively breastfed an average of 14.7 weeks (SD = 18.2) with 82.9% of infants consuming breastmilk at least one time per day between birth-3 months of age, 54.7% between 4-6 months, 39.3% between 7-9 months, and 35.2% between 10-12 months. One-third of mothers (30.3%) reported introducing any solid foods before 4 months of age.

Consumption of Baby Snacks and Sweets during the First Year of Life
Very few mothers introduced baby snacks (n = 7) or sweets (n = 3) between birth and 3 months. Overall, 25.8% of infants consumed at least one baby snack per day between 4-6 months, 82.2% between 7-9 months, and 87.6% between 10-12 months. Similarly, 7.1% of infants consumed at least one sweet per day between 4-6 months, 28.7% between 7-9 months, and 47.2% between 10-12 months. Table 2 shows medians (IQR) for baby snacks and sweets consumed per day across the first year of life. Baby snacks consumption increased across the first year of life reaching a median of 3.0 (IQR = 2.0-4.0) times/day between 10-12 months. Sweets consumption also increased reaching a median of 0.7 (IQR = 0.0-1.7) times/day between 10-12 months.
a Includes teething biscuits, puffs, and melts. b Includes cakes, cookies, and candies. c SSB (sugar-sweetened beverages) includes juice drinks, soda, and sweetened tea.
Spearman's rho correlations for less healthy snacks and other foods consumed from 4 to 12 months are shown in Table S1. There was a weak negative correlation between baby snacks (ρ = −0.16) and sweets (ρ = −0.12) with breastmilk, and a weak positive correlation between baby snacks (ρ = 0.03) and sweets (ρ = 0.06) and infant formula. Baby snacks were positively correlated with all other foods and SSBs with values ranging from 0.17 to 0.47. Similarly, sweets were positively correlated with all other foods and SSBs with values ranging from 0.22 to 0.49.

Discussion
The goal of this analysis was to explore the association between less healthy snack foods (baby snacks and sweets) and infant WFL z-score trajectories during the first year of life. In this sample of predominantly low-income, non-Hispanic Black mothers and their infants, 25% of infants consumed baby snacks and 7% consumed sweets between 4-6 months of age. There was an increasing trend across the first year of life, where 87% of infants consumed baby snacks and 47% consumed sweets by 10−12 months of age. Our results suggest that consuming sweets impacts infant weight trajectories. At 7−9 and 10−12 months of age, infants consuming sweets often (>2 times /day) had higher WFL zscores compared to infants who never had sweets. Continued efforts to reduce less healthy foods during infancy, especially sweets (e.g., cookies, cakes, and candies), may be critical to the development of healthy food preferences, dietary patterns, and weight trajectories that begin to emerge during this early developmental period.
The findings that mothers introduced less healthy snack foods during the first year of life are consistent with previous research [19,20]. Data from NHANES found increasing trends for less healthy snack foods, where 5% of young infants (birth-5 months) and 50% of older infants (6−11 months) consumed a sweet or salty snack each day, which is consistent with the IFPS II and our study. The AAP recommends three nutrient-dense small meals and two or three small snacks per day [14], which leaves little room for discretionary calories from nutrient-poor foods. Notably, the FITS found that self-reported snack foods provided about one-fifth of an infant's daily energy needs [35]. Although not all snacks were nutrient-poor (e.g., 48% consumed fruits and 9% consumed vegetables), over 20% of infants consumed sweets, SSBs, or desserts. Given the prevalence and energy contribution of less healthy snack foods during the first year of life, it is critical to examine if these snacking patterns contribute to early weight gain and to investigate factors that influence parents and other caregivers to offer less healthy snack foods during infancy.
This study adds to the literature as one of the first to explore the association between less healthy snack foods with infant WFL z-score trajectories. In line with our a priori hypothesis, sweet consumption had a significant impact on weight trajectories between 4-12 months of age. Of interest is our finding that infants who consumed sweets more often (>2 times/day) during early complementary feeding (4−6 months) had lower WFL z-scores. Although this was a significant finding, results should be interpreted with caution given the very small sample size in the sweets category. In contrast, between 7-9 and 10-12 months, infants who consumed sweets often had higher WFL z-scores compared to infants who never consumed sweets. This is also in contrast to other studies finding no association or a protective association between snack foods and weight status in older children [25,26,36]. However, our study targeted less healthy snacking during infancy and results may reflect that infants have little room for discretionary or "empty" calories. In contrast to our a priori hypothesis, baby snacks did not have a significant impact on infant weight trajectories at

Discussion
The goal of this analysis was to explore the association between less healthy snack foods (baby snacks and sweets) and infant WFL z-score trajectories during the first year of life. In this sample of predominantly low-income, non-Hispanic Black mothers and their infants, 25% of infants consumed baby snacks and 7% consumed sweets between 4-6 months of age. There was an increasing trend across the first year of life, where 87% of infants consumed baby snacks and 47% consumed sweets by 10−12 months of age. Our results suggest that consuming sweets impacts infant weight trajectories. At 7−9 and 10−12 months of age, infants consuming sweets often (>2 times /day) had higher WFL z-scores compared to infants who never had sweets. Continued efforts to reduce less healthy foods during infancy, especially sweets (e.g., cookies, cakes, and candies), may be critical to the development of healthy food preferences, dietary patterns, and weight trajectories that begin to emerge during this early developmental period.
The findings that mothers introduced less healthy snack foods during the first year of life are consistent with previous research [19,20]. Data from NHANES found increasing trends for less healthy snack foods, where 5% of young infants (birth-5 months) and 50% of older infants (6−11 months) consumed a sweet or salty snack each day, which is consistent with the IFPS II and our study. The AAP recommends three nutrient-dense small meals and two or three small snacks per day [14], which leaves little room for discretionary calories from nutrient-poor foods. Notably, the FITS found that self-reported snack foods provided about one-fifth of an infant's daily energy needs [35]. Although not all snacks were nutrient-poor (e.g., 48% consumed fruits and 9% consumed vegetables), over 20% of infants consumed sweets, SSBs, or desserts. Given the prevalence and energy contribution of less healthy snack foods during the first year of life, it is critical to examine if these snacking patterns contribute to early weight gain and to investigate factors that influence parents and other caregivers to offer less healthy snack foods during infancy.
This study adds to the literature as one of the first to explore the association between less healthy snack foods with infant WFL z-score trajectories. In line with our a priori hypothesis, sweet consumption had a significant impact on weight trajectories between 4-12 months of age. Of interest is our finding that infants who consumed sweets more often (>2 times/day) during early complementary feeding (4−6 months) had lower WFL z-scores. Although this was a significant finding, results should be interpreted with caution given the very small sample size in the sweets category. In contrast, between 7-9 and 10-12 months, infants who consumed sweets often had higher WFL z-scores compared to infants who never consumed sweets. This is also in contrast to other studies finding no association or a protective association between snack foods and weight status in older children [25,26,36]. However, our study targeted less healthy snacking during infancy and results may reflect that infants have little room for discretionary or "empty" calories. In contrast to our a priori hypothesis, baby snacks did not have a significant impact on infant weight trajectories at any category or time point. This finding suggests that commercially available baby snacks may not contribute sufficient discretionary calories that may place infants at risk for obesity.
To our knowledge, this was one of the first studies to explore the impact of commercially available baby snacks (teething biscuits, puffs, and melts), which are snacks marketed specifically to the parents of infants, on infant weight trajectories. A recent study found that many commercially available infant and toddler foods contain added sugars and salt [37], which are not recommended for this age group [14] and may also contribute excess calories. Although baby snacks did not have a significant impact on infant weight trajectories in our study this area warrants further exploration. Particularly given that food and beverages offered during early infancy influence food preferences, dietary patterns, and weight trajectories that often persist into later childhood, further exploring the impact of commercially available baby snacks on infant weight status and promoting healthy snacking is important.
The primary limitation of this study was the use of self-report infant dietary questionnaires, which have been shown to be biased by under-and overreporting, to examine less healthy snack food consumption. However, the dietary questionnaire has been used in other cohort studies [29,30], and the validity is supported by similar findings of less healthy snack food consumption in a large national sample [19]. The dietary questionnaire also did not include serving sizes, so the energy contribution of baby snacks and sweets is unknown. Future studies should consider including 24 hour recalls to collect data on serving size in order to assess the energy contribution of snack foods. To assess less healthy snacking behaviors, our study examined baby snacks (teething biscuits, puffs, and melts) and sweets (cookies, cakes, and candies); however, it is unknown if mothers would describe these foods as snacks or if there are other foods that mothers would describe as snacks that were not included in our analysis. However, given recent increases in less healthy snack food consumption during early childhood [22], these snack foods warrant examination as independent predictors of infant weight trajectories. Future studies may consider qualitative research with mothers of young infants to understand how mothers define snacking and to explore factors that influence snacking during infancy. Although social desirability bias may contribute to underreporting of less healthy snack food consumption, the prevalence of less healthy snack foods in our sample were in line with other studies assessing snack food consumption during infancy [19,20]. In addition, very few infants in the Nurture study had WFL z-scores placing them at risk for obesity, therefore we were unable to examine the impact of less healthy snacks by weight status (i.e., normal weight compared to infants at risk for obesity).
This study adds to the literature on snack food consumption during infancy and has a number of strengths that warrant mention. This analysis included primarily low-income, non-Hispanic Black mothers and their infants who are underrepresented in the research literature. Including underrepresented groups in research is a public health priority and is a vital component of reducing health disparities [38]. In addition, this study examined less healthy snack food consumption across important transitions in infant feeding (i.e., from a milk-based diet to solid foods). Given that less healthy dietary patterns that begin during infancy often track into later childhood and have been associated with increased weight status, examining the impact less healthy snacks on weight during infancy is essential. In addition, future studies may consider examining snack food consumption and weight status during the toddler years to understand how snacking and weight trajectories track across the first two years of life. This study also highlights the need for recommendations around healthy snack foods, particularly in this young age group who are learning to eat and developing food preferences.

Conclusions
This analysis of the Nurture study, a cohort of predominantly low-income, non-Hispanic Black mothers and their infants, found that mothers introduce less healthy snacks during the first year of life. Our results suggest that less healthy sweets (cookies, cakes, and candies) are associated with increased weight trajectories during later infancy, making these snack foods important targets for childhood obesity prevention efforts. Given that less healthy snacks are offered during early infancy and may contribute to the risk of adiposity, promoting healthy snack food choices during this critical window is important. Future studies should explore drivers of snacking during infancy to help inform the evidence-base for healthy snacking recommendation during infancy. In addition, understanding the drivers of infant snacking would help to inform developmentally appropriate infant feeding interventions and help to address factors in the food environment that influence less healthy snacking during infancy.