Predictors of Age at Juice Introduction and Associations with Subsequent Beverage Intake in Early and Middle Childhood

ABSTRACT

Background

The American Academy of Pediatrics recommends that if parents choose to introduce juice, they wait until ≥12 months, citing concerns of obesity and dental caries.

Objectives

We sought to identify correlates of early juice introduction (<6 months) and determine whether early introduction establishes a pattern of sugary beverage intake in childhood.

Methods

Upstate KIDS is a prospective birth cohort study with follow-up through 7 years (n = 4989). The age of juice introduction was assessed from responses on periodic questionnaires from 4–18 months and categorized as <6,  6 to <12, and ≥12 months. Sociodemographic information was reported using vital records or maternal questionnaires. At 24, 30, and 36 months and 7 years, mothers reported their child's regular juice, soda, water, and milk intakes. The analysis was restricted to singletons and 1 randomly selected twin from each pair with information on juice introduction (n = 4067). We assessed associations of sociodemographic correlates with juice introduction using Cox proportional hazard models. The relations of juice introduction with beverage intake were evaluated using Poisson or logistic regression for adjusted risk ratios (aRR) or ORs, adjusting for sociodemographic covariates and total beverage intake.

Results

Of the mothers, 25% and 74% introduced juice prior to 6 and 12 months, respectively. Younger maternal age; black or Hispanic race/ethnicity; lower educational attainment; Special Supplemental Nutrition Program for Women, Infants, and Children participation (yes); smoking during pregnancy; a higher pre-pregnancy BMI; a lower household income; and living in a townhouse/condominium or mobile home were associated with earlier juice introduction. Earlier juice introduction was related to a higher childhood juice intake, any soda intake, and lower water intake, holding total beverage intake constant [aRR, 1.5 (95% CI: 1.3–1.7; P-trend < 0.0001); adjusted OR 1.6 (95% CI: 1.0–2.4; P-trend = 0.01); aRR 0.9 (95% CI: 0.8–0.9; P-trend < 0.0001), respectively].

Conclusions

Markers of lower socioeconomic status are strongly associated with earlier juice introduction, which, in turn, relates to sugary beverage intake in childhood, potentially replacing water.

Introduction

Whether 100% fruit juice intake is a healthful component of young children's diets remains controversial. Intake of 100% fruit juice in children has been cross-sectionally associated with more dietary diversity and higher intake of key micronutrients (1–5). Alternatively, the lack of fiber and the high sugar content in 100% fruit juice raises concerns of excess consumption; indeed, high 100% fruit juice consumption in early childhood has been related to failure to thrive (6, 7), weight gain in childhood (8), and dental caries (9). As such, in 2017 the American Academy of Pediatrics (AAP) recommended that, if introduced at all, 100% fruit juice not be introduced into infants’ diets prior to 12 months (10), revising a previous guideline recommending that 100% fruit juice not be introduced prior to 6 months (11).

Diet in infancy is hypothesized to “program” dietary preferences (12–14). Specifically, earlier introduction of sweet foods has been associated with a preference for sweet tastes, a major predictor of intake in younger ages (13–15). Intake of juice at 12 months has been associated with more 100% fruit juice and sugar-sweetened beverage (SSB) intake and less water intake in middle childhood (16). However, the age of juice introduction has not been examined in relation to subsequent beverage intake in early or middle childhood. If related to subsequent beverage intake, delaying 100% fruit juice introduction may represent an easier intervention for parents than restricting intake once a child has already demonstrated preferences for 100% fruit juice.

Upwards of 90% of pediatricians indicate that they discuss limiting juice and SSB intakes in children under 2 years of age (17); thus, identifying factors associated with early juice introduction would allow providers to better target at-risk groups. Previously identified correlates of 100% fruit juice intake prior to 6 months include lower parental educational attainment (18–21); younger maternal age (21); lower income (20, 21); participation in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) (22, 23); less breastfeeding (21, 24); higher birthweight (19); and non-Hispanic black and Hispanic race/ethnicity (21, 25–27). However, the majority of studies on correlates of 100% fruit juice introduction have been cross-sectional (19, 20, 22–26) and thus do not prospectively evaluate juice introduction, which may introduce recall bias or right truncation.

The aims of this study were 2-fold: first, to examine correlates of juice introduction timing within the context of a longitudinal cohort with prospectively collected information; and second, to evaluate whether earlier juice introduction is related to beverage intake in early and middle childhood. Of note, this study is limited in that we cannot distinguish introduction of 100% fruit juice from that of fruit drinks with added sugars. As such, the term “juice” is used throughout the paper to indicate 100% fruit juice as well as juice products, which may additionally contain added sugars. As the cohort was established prior to the AAP's 2017 guidelines, juice introduction was evaluated based on the old (<6 months) and new (<12 months) recommendations.

Methods

Study population

The Upstate KIDS study is a population-based birth cohort designed to evaluate the associations of fertility treatment with child health and development (28). At approximately 4 months postpartum, mothers of children born between July 2008 and May 2010 in New York State, excluding New York City, were recruited (5034 mothers and 6171 children). Mothers of all infants conceived using fertility treatments and multiple births were invited to participate. Three singletons not conceived by fertility treatment were sampled for every child conceived using treatment, with frequency matched by the region of birth. The fertility treatment status was not related to any significant differences in early childhood development (29).

The New York State Department of Health and the University of Albany Institutional Review Board (IRB) approved of the study and served as the IRB designated by the NIH for this study under a reliance agreement. Parents provided written informed consent prior to enrollment.

Measurements

Potential correlates of age at juice introduction

At 4 months postpartum, mothers completed a baseline questionnaire, from which we obtained information on maternal race/ethnicity, education, marital status, smoking and alcohol use during pregnancy, height, weight, daily vitamin use during pregnancy, fertility treatment use, residence type during pregnancy (single family home, apartment, townhouse/condominium, or mobile home), household income, and any breastfeeding at 4 months, as well as father's age, height, and weight. Maternal age, private health insurance status, height, weight, parity, and WIC participation (yes compared with no), and infant sex, plurality, birthweight, and gestational age were obtained from vital records (i.e., birth certificates). Race/ethnicity was primarily self-reported by mothers on the questionnaire at 4 months. For race, mothers were asked to choose from a list that included “white/Caucasian,” “black or African American,” “Asian Indian (South Asian),” “Chinese,” “Filipino,” “Japanese,” “Korean,” “Vietnamese,” “Native Hawaiian,” “Guamanian or Chamorro,” “Samoan,” “American Indian or Alaska Native,” “Other Asian,” “Other Pacific Islander,” and “Other.” With regards to ethnicity, mothers were asked to choose from a list which included “Not Hispanic or Latino,” “Mexican, Mexican American, Chicano,” “Puerto Rican,” “Cuban,” and “Other Central or South American.” Pre-pregnancy BMI was calculated as kg/m² from the pre-pregnancy height and weight provided in vital records or, if missing, from self-report on the baseline questionnaire, which inquired about maternal weight before this pregnancy. Gestational hypertension and gestational diabetes were identified by combining information from the maternal report, vital records, and linkage with in- and outpatient hospital data. The Townsend index, a measure of economic deprivation, was calculated using census information, with increasing scores denoting greater neighborhood deprivation (30, 31). The language spoken at home was assessed by maternal report when the child was 36 months old. Details on the breastfeeding assessment have been previously reported (32). Household income was reported at 12 months. Questionnaires when the child was 4, 8, 12, and 18 months were used to assess the timing of solid food introduction and any daycare use prior to 18 months.

Timing of juice introduction

Mothers completed questionnaires when their child was 4, 8, 12, and 18 months old. On each of these questionnaires, mothers were asked whether they had introduced juice into their child's diet and, if so, to indicate the age when they first gave juice to their child. Unfortunately, we are unable to distinguish whether mothers introduced 100% fruit juice or other types of juice (e.g., juice sweetened with sugar). If mothers indicated that they had introduced juice on multiple questionnaires, the response on the first chronological questionnaire was prioritized to minimize recall bias. When mothers indicated that juice had been introduced but did not include the age of juice introduction, the midpoint between the child's age at the time of the current questionnaire and the last questionnaire on which they had indicated that their child had not yet begun to drink juice was used as the time at which juice was introduced. The timing of juice introduction was then categorized as <6,  6 to <12, and ≥12 months, in line with AAP guidelines (10, 11).

Beverage intake

At 24, 30, and 36 months, mothers reported the number of 8-ounce (236.6 mL) cups of “juice,” “soda or pop,” “water,” and “milk” that their child drank daily. As the term “juice” was used on the questionnaires, we cannot distinguish between 100% fruit juice and juice with added sugars. As our questionnaire specified “soda or pop” intake at 24–36 months, our measure of SSB intake at this time likely does not include that of other SSBs, such as fruit drinks including Tang, Hi-C, fruit punch, Kool Aid, and so forth; sports drinks; or sweetened tea/coffee. As such, this measurement will be referred to as “soda” throughout. Soda intake at 24–36 months was classified as any or none, because drinking soda was rare (∼5% reported any intake). Milk intake at this time does not specify the type of milk consumed or whether it was flavored milk.

At 7 years, mothers completed a 27-item FFQ, where they indicated how many times per day, times per week, or times per month their child ate or drank each item within the past month. This FFQ was adapted from the 28-item FFQ used in the Infant Feeding Practices Study II (IFPS II) (33–35). Food items pertinent to this analysis as listed on the FFQ include “100% pure fruit juice or 100% pure vegetable juice,” “regular soda or pop that contains sugar or corn syrup,” “water: include tap, bottled, or unflavored sparkling water,” “milk: all types to drink or in cereal,” and “sweetened drinks: Kool-Aid, lemonade, sweet tea, Hi-C, cranberry cocktail, Gatorade, etc.” In this questionnaire, 100% fruit juice was distinguished from sweetened drinks. For consistency with beverage intake in early childhood, juice, milk, and water intakes were transformed into times per day and rounded to the nearest whole number; sensitivity analyses that were conducted with values rounded up instead of to the nearest whole number produced similar results (data not shown). In the main analysis, soda intake only included that of regular soda or pop at 24, 30, and 36 months and 7 years. In sensitivity analyses, SSB intake at 7 years additionally included that of “sweetened drinks: Kool-Aid, lemonade, sweet tea, Hi-C, cranberry cocktail, Gatorade, etc.” Soda and SSB intake was classified as any or none. Given that the FFQ asks mothers to report on all types of milk consumption, this likely includes flavored milk; however, only 7 (0.49%) mothers indicated that their child primarily drank flavored or sweetened milk at this time. At each time point, total beverage intake was calculated by summing the daily report of juice, soda, water, and milk intake; at 7 years, this was done prior to rounding.

Statistical analyses

Analyses were restricted to singletons and 1 randomly selected twin from each twin pair (n = 4989) due to a high correlation in juice introduction timing between twin pairs (Spearman's r = 0.9). Information on the timing of juice introduction was available for 4067 children (81.5%; Supplemental Figure 1).

Correlates analysis

The distributions of covariates were compared across categories of age at juice introduction. P values were calculated from a Cox proportional hazards model where time to introduction of juice (months) was the outcome. Children whose mothers did not report juice introduction were censored at the last questionnaire they completed. We then assessed which correlates were robust to multivariable adjustment by including covariates related to the timing of juice introduction at 2-sided P values < 0.05 and child's sex into a Cox proportional hazard model. Reference values were chosen to represent the majority of the population (e.g., non-Hispanic white race/ethnicity, educational attainment of college or advanced degree). Only covariates which remained significant at P values < 0.05 were retained in the final model and are presented in Table 1. Kaplan-Meier curves were plotted for select covariates.

TABLE 1.

Multivariable adjusted correlates of timing of juice introduction among singletons and 1 randomly selected twin from each twin pair in the Upstate KIDS cohort

	HR (95% CI)	P value1
Maternal characteristics
Age, year	0.97 (0.96–0.98)	<0.0001
Maternal race/ethnicity
Non-Hispanic white	Reference	—
Non-Hispanic black	1.4 (1.2–1.6)	0.0001
Non-Hispanic Asian	1.1 (0.9–1.3)	0.52
Hispanic	1.2 (1.1–1.4)	0.0001
Mixed race/other	1.1 (1.0–1.4)	0.13
Education: college or advanced degree vs. less then high school,high school or GED equivalent, or some college (reference)	1.2 (1.1–1.3)	<0.0001
WIC participation	1.1 (1.0–1.2)	0.03
Smoking during pregnancy	1.2 (1.1–1.3)	<0.0001
Pre-pregnancy BMI, kg/m2	1.01 (1.00–1.01)	0.006
Gestational hypertension	1.2 (1.1–1.4)	0.002
Residential type
Single family home	Reference	—
Apartment	1.0 (0.9–1.1)	0.47
Townhouse or condominium	1.2 (1.0–1.4)	0.009
Mobile home	1.3 (1.1–1.5)	0.006
Household income2	0.94 (0.90–0.98)	0.001
Child characteristics
Breastfed at 4 months	0.8 (0.7–0.8)	<0.0001
Timing of solid food introduction, months3	0.91 (0.89–0.93)	<0.0001

WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

¹Wald chi-squared test from a proportional hazards model where introduction of juice (months) was the outcome and all covariates included in the table were introduced into the model. Twenty imputed data sets were used; imputed covariates included maternal insurance status (n = 4), WIC participation (n = 69), smoking during pregnancy (n = 2), pre-pregnancy BMI (n = 11), child breastfeeding (n = 981), and timing of solid food introduction (n = 343).

²Categorical variable representing household incomes of $10,000–$24,999, $25,000–$49,999, $50,000–$74,999, $75,000–$99,999, and ≥$100,000, which was entered as a continuous covariate.

³Midpoint between age at last response to a survey and age when the mother responded “yes” to introducing a food item or food group. The minimum age of food introduction was considered 3 months: that is, the midpoint for a positive response at 4 months would be 3.5 months.

Beverage intake analysis

We compared the distributions of juice, soda, water, milk, and total beverage intakes at each time point across categories of age at juice introduction. Other SSBs were examined at 7 years. P-trend values were calculated using a Wald chi-squared test, where the amount of beverage intake was the outcome and the 3-category variable for juice introduction timing was introduced as a continuous covariate.

We then used generalized linear mixed-effect models to estimate associations between the timing of juice introduction and subsequent beverage intakes. For juice, water, and milk intakes, a Poisson regression model was used; a logistic regression model was used for soda and other SSBs due to low intake of soda and other SSBs in early childhood (<6%). In all models, a child-level random intercept was included to account for repeated measures of beverage intake. We applied sampling weights and inverse probability weights (IPW), which respectively accounted for the oversampling of infants conceived using fertility treatment and nonresponses to follow-up at each time point. Time was modeled as a categorical variable (i.e., 24, 30, 36, and 84 months). An unadjusted analysis was first conducted (Model 0). Then, covariates which had previously been identified as correlates of juice introduction were added in Model 1. In Model 2, the total beverage intake at each time point was additionally included. Results from this model reflect the association of timing of juice introduction with future beverage intake, holding the total amount of beverages consumed constant.

For all multivariable models, covariates with missing information were imputed using the chained equations method with fully conditional specification. Twenty multiply imputed data sets were created. All analyses were conducted with SAS version 9.4 (SAS Institute Inc.).

Results

Among those with responses to the juice introduction question by 18 months (n = 4067), 25% (n = 1012) of mothers reported introducing juice into their child's diet at <6 months of age, 49% (n = 1995) at 6 to <12 months, and 26% (n = 1060) at ≥12 months. Thus, 75% and 26% of parents were in line with the AAP 2001 (≥6 months) and 2017 (≥12 months) guidelines for juice introduction, respectively.

Correlates

Clear socioeconomic divides are seen with regards to the age of juice introduction, with earlier juice introduction being associated with markers of lower socioeconomic status (Supplemental Figure 2; Supplemental Table 1). In a multivariable analysis, maternal non-Hispanic black and Hispanic race/ethnicity were positively associated with early juice introduction [adjusted HRs, 1.4 (95% CI: 1.2–1.6) and 1.2 (95% CI: 1.1–1.4), respectively], compared to non-Hispanic white race/ethnicity. Likewise, maternal educational attainment of less than high school, high school or GED equivalent, or some college; WIC participation (yes); smoking during pregnancy; gestational hypertension; higher BMI; and residing in a townhouse/condominium or mobile home were associated with earlier juice introduction. Conversely, higher maternal age, higher household income, being breastfed at 4 months, and later introduction of solid foods were inversely associated with early juice introduction (Table 1).

Associations with subsequent beverage intake

In an unadjusted analysis, an earlier age of juice introduction was associated with more juice intake at 24, 30, 36 months, and 7 years. Likewise, an earlier age of juice introduction was related to drinking any soda at each time point and drinking nonsoda SSBs at 7 years. An earlier age of juice introduction was also associated with less water intake at 7 years (Table 2).

TABLE 2.

Type of beverage intake at 24, 30, and 36 months and 7 years by timing of juice introduction among singletons and 1 randomly selected twin from each twin pair, the Upstate KIDS cohort¹

			Age at juice introduction
	n	All	<6 months	6 to <12 months	≥12 months	P value2
Juice intake, mean ± SD
24 months, 8 oz cups (236.6 mL)	2170	1.4 ± 1.3	1.6 ± 1.2	1.6 ± 1.2	0.8 ± 1.1	<0.0001
30 months, 8 oz cups (236.6 mL)	2152	1.4 ± 1.2	1.8 ± 1.2	1.6 ± 1.1	0.9 ± 1.0	<0.0001
36 months, 8 oz cups (236.6 mL)	2115	1.3 ± 1.1	1.6 ± 1.2	1.4 ± 1.1	0.9 ± 0.9	<0.0001
7 years, times per day	1215	0.8 ± 0.9	0.9 ± 0.9	0.8 ± 1.0	0.6 ± 0.8	<0.0001
Sugar-sweetened beverages, n (%)
24 months, any soda intake	2211	48 (2.2)	13 (3.2)	30 (2.7)	5 (0.8)	0.004
30 months, any soda intake	2170	117 (5.4)	63 (14.8)	44 (4.0)	10 (1.6)	<0.0001
36 months, any soda intake	2115	104 (4.9)	40 (10.1)	53 (4.9)	11 (1.7)	<0.0001
7 years, any soda intake	1225	403 (32.9)	89 (40.5)	224 (34.7)	90 (25.1)	<0.0001
7 years, nonsoda SSB intake	1218	702 (57.6)	135 (61.1)	390 (60.7)	177 (50.0)	0.003
7 years, any SSB intake	1216	794 (65.3)	153 (69.5)	441 (68.7)	200 (56.5)	0.0003
Water intake, mean ± SD
24 months, 8 oz cups (236.6 mL)	2101	2.0 ± 1.2	2.1 ± 1.3	1.9 ± 1.2	2.1 ± 1.2	0.88
30 months, 8 oz cups (236.6 mL)	2126	2.0 ± 1.2	2.1 ± 1.2	1.9 ± 1.2	2.1 ± 1.2	0.29
36 months, 8 oz cups (236.6 mL)	2074	2.1 ± 1.2	2.0 ± 1.2	2.0 ± 1.2	2.1 ± 1.2	0.26
7 years, times per day	1209	3.1 ± 1.9	3.0 ± 2.0	2.9 ± 1.9	3.3 ± 2.0	0.01
Milk intake, mean ± SD
24 months, 8 oz cups (236.6 mL)	2176	2.3 ± 1.2	2.4 ± 1.3	2.3 ± 1.2	2.4 ± 1.2	0.96
30 months, 8 oz cups (236.6 mL)	2123	2.3 ± 1.2	2.4 ± 1.2	2.2 ± 1.2	2.3 ± 1.1	0.24
36 months, 8 oz cups (236.6 mL)	2060	2.1 ± 1.1	2.2 ± 1.1	2.1 ± 1.1	2.2 ± 1.1	0.69
7 years, times per day	1207	1.6 ± 1.2	1.6 ± 1.0	1.6 ± 1.2	1.7 ± 1.3	0.33
Total beverage intake, mean ± SD
24 months, 8 oz cups (236.6 mL)	2236	5.5 ± 2.2	6.0 ± 2.4	5.6 ± 2.1	5.1 ± 2.1	<0.0001
30 months, 8 oz cups (236.6 mL)	2187	5.7 ± 2.0	6.3 ± 2.4	5.7 ± 2.0	5.2 ± 1.7	<0.0001
36 months, 8 oz cups (236.6 mL)	2115	5.5 ± 2.0	5.8 ± 2.2	5.5 ± 2.0	5.2 ± 1.8	<0.0001
7 years, times per day	1227	5.7 ± 2.6	5.9 ± 2.9	5.7 ± 2.7	5.6 ± 2.4	0.33

SSB, sugar-sweetened beverage.

¹The percentage of children with juice introduced at <6,  6 to <12, and ≥12 months was 18.69, 50.74, and 30.58 at 24 months, respectively; 19.57, 50.71, and 29.72 at 30 months, respectively; 18.68, 50.50, and 30.83 at 36 months, respectively; and 17.95, 52.74, and 29.32 at 7 years, respectively.

²Wald chi-squared test for beverage intake from a Poisson or logistic regression model where the amount of beverage intake (cups or times per day; juice, water, milk, total beverage intake) or any SSB intake were the outcomes, respectively, and a 3-category variable for age at juice introduction was introduced in the model as a continuous predictor.

In longitudinal analyses, juice introduction at <6 and 6 to <12 months related to higher subsequent juice intake compared to juice introduction at ≥12 months following adjustment for sociodemographic correlates of age at juice introduction. Juice introduction at <6 and 6 to <12 months was associated with 1.6 (95% CI: 1.4–1.8) and 1.7 (95% CI: 1.5–1.8) times more subsequent juice intake, respectively (P-trend, <0.0001). An additional adjustment for total beverage intake modestly attenuated the associations. Juice introduction at <6 and 6 to <12 months were positively but not statistically significantly related to higher odds of drinking soda in childhood [adjusted ORs (aOR), 1.5 (95% CI: 1.0–2.3) and 1.2 (95% CI: 0.8–1.7); P-trend, 0.02] compared with juice introduction at ≥12 months. Results were attenuated when drinking any SSB was included in the outcome at 7 years [aOR, <6 compared with ≥12 months: 1.3 (95% CI: 1.0–1.7); aOR, 6 to <12 compared with ≥12 months: 1.1 (95% CI: 0.9–1.4); P-trend, 0.04]. Earlier juice introduction (<6 or 6 to <12 compared with ≥12 months) was associated with 10% less water intake (95% CIs, 0.9–1.0 and 0.9–1.0, respectively; P-trend: 0.005). This association was exaggerated upon adjustment for total beverage intake. An earlier age of juice introduction was inversely associated with milk intake only after adjustment for total beverage intake (Table 3).

TABLE 3.

Associations of timing of juice introduction with beverage intake at 24, 30, and 36 months and 7 years among singletons and 1 randomly selected twin from each twin pair, the Upstate KIDS cohort

		Age at juice introduction
	n	<6 months	6 to <12 months	≥12 months	P value1
Juice intake, RR (95% CI)2	3233
Unadjusted		2.0 (1.8–2.3)	1.9 (1.7–2.1)	Reference	<0.0001
Model 1		1.6 (1.4–1.8)	1.7 (1.5–1.8)	Reference	<0.0001
Model 2		1.5 (1.3–1.7)	1.6 (1.4–1.7)	Reference	<0.0001
Soda intake, OR (95% CI)3	3234
Unadjusted		17.2 (7.2–41.0)	3.0 (1.6–5.5)	Reference	<0.0001
Model 1		1.5 (1.0–2.3)	1.2 (0.8–1.7)	Reference	0.02
Model 2		1.6 (1.0–2.4)	1.2 (0.8–1.7)	Reference	0.01
Water intake, RR (95% CI)2	3199
Unadjusted		1.0 (0.9–1.0)	0.9 (0.9–1.0)	Reference	0.09
Model 1		0.9 (0.9–1.0)	0.9 (0.9–1.0)	Reference	0.005
Model 2		0.9 (0.8–0.9)	0.9 (0.8–0.9)	Reference	<0.0001
Milk intake, RR (95% CI)2	3213
Unadjusted		1.0 (1.0–1.1)	1.0 (0.9–1.0)	Reference	0.60
Model 1		1.0 (0.9–1.0)	1.0 (0.9–1.0)	Reference	0.52
Model 2		0.9 (0.9–1.0)	0.9 (0.9–1.0)	Reference	0.04

WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

¹Model 1 is adjusted for maternal age, BMI, race/ethnicity, education, smoking during pregnancy, WIC participation, gestational hypertension, type of residence, and household income, as well as child's age in years, age at solid food introduction (midpoint), and any breastfeeding at 4 months. Model 2 is additionally adjusted for total beverage intake at the corresponding time point. The P value is from a model where the categorical variable of age at juice introduction was entered as a continuous covariate. Twenty imputed data sets were used; imputed covariates included maternal pre-pregnancy BMI (n = 11), smoking during pregnancy (n = 2), WIC participation (n = 69), type of residence (n = 191), household income (n = 2079), child breastfeeding (n = 981), and timing of solid food introduction (n = 343).

²Values are RRs from a Poisson regression model where the amount of beverage consumed daily was the outcome.

³Values are ORs from a logistic regression model where the any consumption of soda at 24, 30, and 36 months and 7 years was the dichotomous outcome.

Discussion

In this prospective cohort study, the majority of mothers (75%) reported introducing juice to their child's diet at ≥6 months of age, in line with AAP guidelines at the time of the study. An earlier age of juice introduction was positively associated with indicators of lower socioeconomic status, which included lower maternal age, black or Hispanic race/ethnicity, lower educational attainment, not having private health insurance, and WIC participation (yes). Further, earlier juice introduction was related to increased juice and soda intake, as well as less water intake, in early and middle childhood. These associations with beverage intakes were independent of potential confounding socioeconomic indicators.

The estimates of the prevalence of juice introduction prior to 6 and 12 months reported here are similar to those reported in IFPS II, a nationally distributed longitudinal cohort. In IFPS II, 22.9% and 78.7% of mothers reported introducing juice or juice drinks prior to 6 and 12 months, respectively (n = 1,653,691 at 6 and 12 months) (18, 36). In longitudinal studies conducted in specific regions of the United States (i.e., Iowa, the Southeast, Northern California), estimates of the prevalence of juice introduction prior to 6 and 12 months vary from 13%–55% and 75%–89%, respectively (20, 21, 37). Of note, only 1 of these longitudinal studies was able to distinguish between 100% fruit juice and other types of juice drinks (21). Differing definitions of juice introduction, as well as regional and temporal variations in juice consumption in the United States (38, 39), may explain the varying estimates.

In this study, an earlier age of juice introduction was positively associated with markers of lower socioeconomic status: namely, lower maternal age, lower educational attainment, WIC participation (yes), residing in a townhouse/condominium or mobile home, lower household income, and non-Hispanic black and Hispanic race/ethnicity. Although most other studies which examined the association between maternal age and juice introduction did not find an association (19, 21, 24), other factors identified here are similar to previously identified correlates. A lower income and WIC participation (yes) have been related to earlier juice introduction (19–22). Additionally, all but 1 other study (19) found non-Hispanic black or Hispanic race/ethnicity to relate to early juice consumption compared with non-Hispanic whites (21, 23, 25–27). Thus, we have identified 1 additional potential correlate of early juice introduction (i.e., younger maternal age), and corroborated other sociodemographic correlates.

Maternal behaviors and health status markers generally associated with lower socioeconomic status were also positively related to earlier juice introduction. These included maternal smoking during pregnancy, higher pre-pregnancy BMI, having gestational hypertension, lack of breastfeeding in the first 4 months, and earlier solid food introduction. Of these, only lack of breastfeeding has been consistently related to early juice introduction in previous studies (19, 21, 24, 40–42). Earlier introduction of solid foods has been associated with SSB intakes in infancy, including fruit drink intake (43), though no other studies have examined the associations with age at juice introduction. Maternal obesity was inversely related to juice introduction in 1 study (24) and not associated in another (21). Both studies sampled black mothers in the Southeastern United States, whereas our population is from New York state and predominately non-Hispanic white; thus, results may not be generalizable across such discrepant populations. No other study examined the associations with smoking during pregnancy and gestational hypertension.

We also found that earlier juice introduction was positively associated with juice and soda intakes and inversely related to water intake in early and middle childhood. The age at juice introduction was not consistently related to milk intake. Similar to our findings, the amount of juice children drank at 1 year was positively related to 100% fruit juice and SSB intakes at 3 and 8 years in the Project Viva cohort (16). Cross-sectional associations examining simultaneous consumption of juice intake and SSBs have been mixed, with 2 studies finding an inverse association (44, 45) and 2 others finding no association (46, 47) in early childhood. Notably, these cross-sectional studies are small (<200 participants) and examined juice intake in childhood rather than the age of introduction. In the Project Viva cohort, high juice consumption at 1 year was inversely associated with water intake in middle childhood (16).

Taken together, our results suggest that a younger age at juice introduction is associated with higher juice and soda intakes, which replace water intake in childhood. Infants demonstrate an innate preference for sweet tastes (12), which can be augmented with earlier introduction of sweet foods. For instance, when sweetened and unsweetened water were given to infants at birth and again at 6 months, only those who had received sweetened water at birth continued to exhibit the same degree of preference for sweet tastes as in infancy (13). In children, food preferences often drive intake since children tend to eat what they prefer and leave the rest of their food untouched (14). Moreover, food preferences in early childhood are relatively stable across middle childhood (15). Thus, early introduction of juice in infancy may serve to “program” a preference for sweet tastes in children. While an inverse association with early juice introduction and future water intake has not been previously reported, it is plausible that children's preference for sweet tastes may establish a pattern of beverage intake that does not include water.

The ability to assess juice, soda, water, and milk intakes longitudinally across early and middle childhood is one of several strengths of this study. Others include regularly asking mothers about the age of juice introduction from 4 to 18 months, which minimizes recall bias, and measurements on many different familial socioeconomic and health indicators. There are limitations as well. Primarily, we are not able to distinguish between introduction of 100% fruit juice and introduction of juice with added sugars. Although our data are comparable to those of most other longitudinal studies in this regard (16, 18, 20, 36, 37), the lack of specification in our questionnaire may introduce measurement error. This would be problematic if 100% fruit juice and juice with added sugars differentially impact future beverage intake, a question which future research should address. Another potential source of measurement error is that mothers reported the number of 8-ounce cups (236.6 mL) their child drank, which could lead to an overestimation of beverage intake if the child was drinking less than 8 ounces (236.6 mL) per sitting or if they diluted juice with water. However, the total amount of beverage intake in early childhood is similar to that which has been previously reported in national surveys (48, 49). At 24–36 months, our questionnaires did not capture intakes of SSBs other than soda or pop (e.g., Tang, Hi-C, fruit punch, Kool Aid, sports drinks, and sweetened tea and coffee drinks), nor were we able to distinguish between unflavored and flavored milks. Although we were able to control for many potential confounders, we did not have information on costs of beverages or family purchasing decisions, which may be an important confounder influencing the age of 100% fruit juice introduction and subsequent beverage intakes. Similarly, we were not able to control for maternal diet or additional indicators of socioeconomic status that may predict the age of juice introduction, and residual confounding may bias our effect estimates. There was also nonresponse to follow-up questionnaires, which we accounted for by using IPW to mitigate potential selection bias. Lastly, our cohort is predominantly non-Hispanic white, so generalizability may be limited.

In conclusion, this study shows that behaviors that exacerbate health disparities begin early. Markers of lower socioeconomic status were strongly associated with earlier juice introduction. In turn, earlier juice introduction was associated with increased juice and SSB consumption in early and middle childhood, potentially replacing water intake. Given the consistent associations with juice intake in infancy and markers of lower socioeconomic status, these findings suggest that further research is needed to identify drivers of this association. Further, future studies should examine whether introduction of 100% fruit juice and introduction of juice drinks with added sugars have differing effects on beverage preferences in childhood. Finally, identifying pathways that mediate the association between the age of juice introduction and beverage intakes in middle childhood, such as the child's physical activity, overall diet quality, social interactions, and TV screen time, could help identify interventions in early and middle childhood.

Notes

This work was supported by the Intramural Research Program of theEunice Kennedy Shriver National Institute of Child Health and Human Development (contracts #HHSN275201200005C, #HHSN267200700019C, #HHSN275201400013C, and #HHSN275201300026I/27500004).

Author disclosures: The authors report no conflicts of interest.

Supplemental Figures 1 and 2 and Supplemental Table 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn.

Abbreviations used: AAP, American Academy of Pediatrics; aOR, adjusted OR; IFPS II, Infant Feeding Practices Study II; IPW, inverse probability weights; IRB, Institutional Review Board; SSB, sugar-sweetened beverage; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.

Data Availability

Data described in the manuscript, code book, and analytic code will be made available upon request.