Influence of infant feeding practices on childhood dietary patterns in Upstate KIDS

Abstract

Background:

Earlier feeding practices may influence dietary preference. We evaluated if age of introduction to select complementary foods shape intake and diet quality as measured by the Youth Healthy Eating Index (YHEI) in childhood.

Methods:

Parents from the Upstate KIDS cohort reported complementary food introduction of 4–12-month-old infants on food questionnaires. Children with information on infant feeding and diet at 30–36m (n=2826) and 7–9 years of age (n=1449) were included. Associations of age of complementary food introduction with intake in childhood were modeled with Poisson regression and diet quality score with linear models, adjusting for sociodemographic factors.

Results:

Approximately 84% (n=2383) of mothers were non-Hispanic White and about 19% (n=526) of children were twins. At 2–3 years, compared to introducing fruits and vegetables between 5–8 months, introducing later was associated with 13% lower daily intake of fruits and vegetables (aRR, 0.87; 95%CI: 0.79, 0.95); while dairy and grains were associated with a 10% and 17% lower intake, respectively. Later introduction of protein was associated with 6% (aRR, 0.94; 95%CI: 0.90, 0.98) lower intake. For diet quality, introducing fruits and vegetables later (adjusted B: −4.01; 95%CI: −7.42, −0.60) was associated with lower diet quality relative to 5–8m. Later (adjusted B: −1.98; 95%CI: −3.21, −0.74) introduction to dairy was associated with lower diet quality.

Conclusion:

Timing of select complementary foods was associated with lower subsequent intake and lower diet quality in childhood. Further research is needed to evaluate feeding practices that may affect food preferences during infancy as a way to impact healthy dietary patterns and diet quality.

1. Introduction

Behavioral and dietary patterns associated with childhood obesity are known to begin during infancy and continue across the life span (1–3). Early infant feeding practices during infancy not only impact short- and long-term health but are also known to influence the formation of food preferences in childhood (4–6). (7). Evidence suggests that between 4 and 6 months of age is a critical time window during which infants more readily accept a wide range of tastes (8). This critical window is one of many reasons the American Academy of Pediatrics (AAP) and the World Health Organization (WHO) recommend beginning introduction of complementary foods around 6 months of age (9, 10). Based on the timing of introduction of complementary foods, differing dietary patterns may begin to emerge (11).

Around the age of 2, the window for acceptance of new tastes and textures becomes small, and children’s dietary preferences become more established (11–13). Delaying introduction to new tastes and textures during infancy may lead to food aversion and being less receptive to trying new foods (13, 14). Little is known about how differences in dietary preferences, characterized by exposure to select complementary foods introduced, may impact dietary patterns in late childhood. Few studies have explored the timing of fruit and vegetable introduction on later fruit and vegetable intake in childhood. Cooke et al. found in 572 children, timing of introduction to fruits and vegetables was inversely correlated with frequency of fruit intake in early childhood (p<0.005), whereas others observed no associations with intake during early childhood (15–17). However, there is even more limited evidence on timing of other complementary foods (i.e., dairy, grains, and proteins) introduced on dietary patterns in childhood, while other research has focused on examining the effects between longer duration of breastfeeding and subsequent intake and diet diversity in childhood (18–20).

To our knowledge, few studies have examined the timing of more than one complementary food introduced on dietary patterns. A study conducted by Switkowski et al. in 1162 children found different associations between early and late introduction of fruit juices and sweets on select foods (e.g., fish, eggs, and peanut butter) based on their Youth Healthy Eating Index (YHEI) component score at 3 years of age (11). In a randomized clinical controlled trail (RCT), Saunders et al. found early introduction of various complementary foods among 2059 infants improved the subsequent intake of select foods at 9 and 12 months of age (p<0.05), although the study did not follow the infants to childhood to see if there is lasting differences (21).

Evidence remains scarce on the associations between the timing of introduction of select complementary foods and diet in childhood as there are no prospective studies that have observed timing of complementary foods on later childhood intake. There is also little research in U.S. populations where introduction to complementary foods may differ from other countries. Therefore, the aim of the present study is to address whether the timing of introducing select complementary foods between 4–12 months, is associated with differences in both subsequent intake and overall diet quality measured by Youth Healthy Eating Index (YHEI) scores at ages 30–36 months and 7–9 years. We hypothesize that at ages 2–3 and 7–9 years, children will be more likely to consume foods that were introduced within AAP recommendations and contribute to a higher diet quality. For instance, we expect infants introduced fruits and vegetables within recommendations will be more likely to consume fruits and vegetables at 2–3 and 7–9 years of age.

2. Methods

2.1. Study population

Upstate KIDS was a prospective cohort study that enrolled 5,034 mothers and 6,171 infants born between 2008–2010 in New York State (excluding New York City) (22). The study was designed to determine the effects of infertility treatment on childhood development and growth. As previously described, singleton infants conceived by infertility treatment were identified based on birth certificate data and were frequency matched (3:1) to infants without treatment by perinatal region of birth (22). For this analysis, a subset of singletons and twins with information on early infant feeding practices at 4–12 months and diet at 30–36 (n=2826) at age 7–9 (n=1449) years of age were included (Figure 1). The New York State Department of Health and the University at Albany (State University of New York) Institutional Review Boards (IRBs) served as IRBs designated by the National Institutes of Health under reliance agreements. Parents provided written informed consent prior to enrollment.

Figure 1.

Study participant flow diagram, 2008–2010.

2.2. Timing of introduction of complementary foods

Complementary foods were assessed from questionnaires at 4, 8, and 12 months. At each of these 3 time points parents reported using checkbox items which solid foods and/or beverages were introduced to their child which were grouped into the following categories: grains (cereal in a bottle, rice and other cereals, and biscuits); fruits or vegetables (pureed or solid table food); dairy (cow’s milk, goat milk, soy milk, unpasteurized milk, cheese, and other dairy foods); and protein (meats and eggs). Timing of introduction was based on dates of completion on the questionnaires, and for fruits and vegetables, grains, and dairy were categorized as <5 months, between 5–8 months, and between 9–12 months of age as some foods have been introduced around 4 months of age or were introduced later around 8 and 12 months of age. Timing of protein was dichotomized as infants introducing between 5–8 months of age and between 9–12 months age as our study did not ask about the introduction of protein separately on the 4-month questionnaire.

2.3. Dietary patterns in early and middle childhood

Parents completed questionnaires on foods consumed at 30 and 36 months of age. At each time point, parents reported the number of servings per day of the following foods: Grains (rice, pasta, or bread); dairy (milk, yogurt, or cheese); potatoes; legumes (beans, peas, or lentils); vegetables (other than potatoes and legumes); peanut butter; eggs; soy food (such as tofu, soy cheese); meat (beef, poultry, or pork); fish or seafood; fruits; and sweets (such as candy, cakes). Frequency of fruit and vegetable intake was combined to stay consistent with timing of fruits and vegetables during infancy. If the child was provided fast food, parents were asked to specify the frequency throughout the week.

At 7–9 years, mothers completed a 27-item Food Frequency Questionnaire (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) (16, 23). Food items pertinent to this analysis as listed on the FFQ include: milk (all types to drink or in cereal); cheese (all types); ice cream or other frozen dairy desserts (frozen yogurt and sherbet); other dairy products (pudding or corn syrup); fried potatoes (French fries, home fries, or hash browns); beans (refried beans, baked beans, beans in soup, pork and beans, or any other cooked dried beans, not including green beans); rice; pasta; pizza (frozen pizza, fast food pizza, homemade pizza, or other pizza); tomato sauces (Mexican-type salsa, spaghetti or noodles with tomato sauce, or mixed into foods such as lasagna); processed meat (bacon, ham, lunch meats, hot dogs, spam, nuggets, etc.); Non-processed meat (chicken, turkey, pork, beef, lamb); fish or shellfish; peanut butter or peanuts; bread (toast, rolls, bagels, cornbread, tortillas, in sandwiches, pancakes, waffles, etc.); popcorn; and snacks (potato chips, corn chips, pretzels, or crackers). Consistent with previous studies (18, 19) and to stay consistent with our timing of fruit and vegetable exposure, frequency of fruit and vegetable intake was calculated as the sum of two variables: fruits (fresh, frozen or canned, not including juice); green leafy or lettuce salad (with or without other vegetables).

2.4. Youth Healthy Eating Index (YHEI)

We calculated the YHEI (24) total and component scores using maternal report of child’s daily intake of different food items (servings/day) at 30 and 36 months and via FFQ completed by the mothers at 7 and 9 years of age to report frequency of each food item. Consistent to previous literature (11, 25), the number of times a mother reported a child’s intake of each food item, it was assumed to be equivalent to the number of servings/day (e.g., one time equals one serving). A detailed overview of how the YHEI was calculated is presented in Supplemental Table 1. We had data available to calculate 10/13 components (e.g., dairy, whole fruit, vegetables, snack foods, meat ratio, eating breakfast, fast-food outside home, multi-vitamin use, soda and drinks, and eating dinner with family) of the YHEI in early (2–3 years) and middle (7–9 years) childhood. For component 4, the YHEI calculates a ratio for the consumption of meat and other protein sources. Based on the data available at both 2–3 and 7–9 years of age, all low-fat meat and proteins (e.g., fish, peanuts, legumes) were included in the numerator while all high-fat meats and proteins were included in the denominator (e.g., non-processed and processed meats) to calculate the meat ratio. The three components that were not available are consumption of visible animal fat, margarine butter, and whole grains. Though the questionnaire reported information on frequency of grains during both the 30- and 36-month questionnaire, no additional information was requested to distinguish whether the grains consumed were whole wheat, therefore this component was removed. YHEI components 1–6 are based on a minimum score of 0 and a maximum score of 10 while 7–10 are based on a minimum score of 0 and a maximum score of 5. For all components, middle scores between the minimum and maximum standards were scored proportionately consistent with the Healthy Eating Index (HEI) scoring method (26–28). Thus, our total YHEI score included 10 components with a possible minimum score of 0 to a maximum score of 80 with a higher score indicating higher diet quality.

2.5. Covariates

Information on covariates was obtained primarily from birth certificates and maternal questionnaires. Birth certificates provided maternal and paternal age, private health insurance status, participation in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), maternal pre-pregnancy height and weight, parity, infant sex, birthweight, and gestational age. At the time of enrollment, maternal race/ethnicity (non-Hispanic white, non-Hispanic black, Asian, Hispanic, other), education (less than high school, high school or GED, some college, college, advanced degree), marital status (married or living as married), and fertility treatment use were obtained from the baseline questionnaire at 4 months postpartum. As we previously observed, a range of child and maternal factors has been associated with child dietary patterns, hence the above factors were considered in the current analyses (11, 29–31).

2.6. Statistical Analyses

Descriptive statistics of maternal and child characteristics were calculated for the analytical sample and presented in Table 1. All singletons and one randomly selected multiple from each family were included. We used Poisson regression to model the associations between timing of introduction to select complementary foods and subsequent intake of select foods at ages 30–36 months and 7–9 years. We assessed the model fitness using Pearson chi-square goodness-of-fit test (X²/DF). Generalized linear models were used to examine the associations between age of introduction to select complementary foods and diet quality. Those introducing complementary foods between 5–8 months of age were used as the reference group based on the AAP’s recommendations (10) to introduce complementary foods around 6 months of age. Those introducing complementary foods prior to the recommendations and after were categorized as <5 months and between 9–12 months, respectively. All models were adjusted for mother’s age, race/ethnicity, education, insurance status, pre-pregnancy BMI, child’s gestational age, parity, and WIC participation. A false discovery rate (FDR) correction based on the Benjamini-Hochberg method was applied to account for multiple testing (32). Because infants conceived by fertility treatment and twins were oversampled in this study, we used sampling weights in each model to account for this study design feature (22). To account for missing exposure and covariate data, 20 data sets were created using multiple imputations by the chained equations method (33). Sampling and inverse probability of censoring weights were multiplied for analyses to account for nonresponses to follow-up at 30–36 months and at age 7–9 years, respectively.

Table 1.

Maternal and infant characteristics (n=2826), Upstate KIDS study

Overall, n, %
Maternal characteristics
Age, y, mean ± SD	31.2 ± 5.9
Race/ethnicity, n (%)
Non-Hispanic White	2383 (84.3)
Non-Hispanic Black	78 (2.8)
Asian	88 (3.1)
Hispanic	206 (7.3)
Other	71 (2.5)
Education, n (%)
Less than high school	103 (3.6)
High school or GED equivalent	269 (9.5)
Some college	778 (27.5)
College	704 (24.9)
Advanced degree	972 (34.4)
Private insurance, n (%)	2290 (81.0)
WIC recipient, n (%)	590 (20.9)
Married or living as married, n (%)	2568 (90.9)
Height, cm, mean ± SD	164 ± 7.3
Weight, kg, mean ± SD	74.1 ± 18.5
Pre-pregnancy BMI, kg/m2, mean ± SD	26.7 ± 6.7
Gestational diabetes	271 (9.6)
Gestational hypertension	290 (10.3)
Fertility treatment use	945 (33.4)
Child characteristics
Child sex, male	1475 (52.2)
Birth weight, g, mean ± SD	3215 ±682
Plurality, twin	526 (18.6)
Gestational age, weeks, mean ± SD	38.2 ± 2.4
Congenital malformations, n (%)	169 (6.0)
Breastfeeding duration, n (%)
Formula-fed only	553 (19.6)
<6 months	1132 (40.1)
6-<12 months	577 (20.4)
≥ 12 months	564 (20.0)

Abbreviations: GED, general educational diploma; BMI, body mass index; WIC, Women, Infants, and Children.

^*Values are n (%) unless otherwise noted.

In sensitivity analyses, we additionally adjusted for breastfeeding duration and frequency of fast-food intake to further examine potential associations for subsequent intake and overall diet quality based on the age of select complementary food introduction, as previous literature has shown these feeding behaviors to influence later dietary patterns (11, 34, 35). In addition, children with registered congenital malformations (e.g., cleft palate and/or heart defects), were removed (n=169) and all models re-run. This sensitivity analysis was conducted to determine whether the results remained consistent as it has been previously reported that children with congenital malformations often influence feeding and developmental problems (36, 37). Analyses were conducted using SAS version 9.4 (SAS Institute Inc.).

3. Results

Descriptive statistics are presented in Table 1. Based on the entire analytical sample, approximately 84% (n=2383) of mothers were non-Hispanic White, 34% (n=972) obtained an advanced degree, 81% (n=2290) had private insurance, 91% (n=2568) were married or living as married, and mothers had a mean pre-pregnancy BMI of 26.7 kg/m². About 19% (n=526) of children were twins.

3.1. Intake in early and middle childhood

The distributions of the timing and type of complementary foods are provided in Supplemental Table 2. Approximately 60.5% of infants were introduced to grains at <5 months, while 61.0% and 48.8% were introduced to fruits and vegetables and protein between 5 and 8 months, respectively. Few infants were introduced to these foods past 9 months of age, although 65.4% and 51.2% of infants were introduced to dairy and protein during this latter part of infancy.

Earlier introduction of complementary foods before 5 months of age was generally not associated with their respective intake at 2–3 and 7–9 years of age, particularly after adjustment of covariates (Table 2). This was evident as the unadjusted models revealed that children introduced to fruits and vegetables prior to 5 months was associated with an 8% (RR, 0.92; 95% CI: 0.88, 0.97; FDR p =0.001) and 12% (RR, 0.88; 95% CI: 0.82, 0.95; FDR p=0.003) lower daily subsequent intake of fruits and vegetables at 2–3 and 7–9 years of age, respectively. Adjustment for sociodemographic factors including pre-pregnancy BMI reduced these associations to null. Associations with dairy, grains, and protein followed a similar pattern.

Table 2.

Association with intake of select foods based on timing of complementary food introduction at 2–3 (n=2797) and 7–9 (n=1443) years of age, Upstate KIDS Study^†

	Age of complementary food introduction
Model	<5 vs 5–8 months	FDR	9–12 vs 5–8 months	FDR	X2/DF
Outcome at 2–3 y	Risk Ratio (95% CI)
Fruit and vegetable intake	Fruit and vegetable introduction
Unadjusted	0.92 (0.88, 0.97)	0.001	0.82 (0.75, 0.90)	<.0001	0.965
Model 1	0.99 (0.93, 1.05)	0.680	0.87 (0.79, 0.95)	0.002	0.929
Model 2	0.99 (0.93, 1.04)	0.618	0.93 (0.79, 1.09)	0.378	0.926
Dairy intake	Dairy introduction
Unadjusted	0.85 (0.79, 0.93)	0.0003	0.89 (0.85, 0.94)	<.0001	0.765
Model 1	0.98 (0.87, 1.09)	0.765	0.90 (0.86, 0.94)	<.0001	0.749
Model 2	0.98 (0.89, 1.09)	0.765	0.95 (0.85, 1.06)	0.540	0.749
Grain intake	Grain introduction
Unadjusted	0.86 (0.81, 0.90)	<.0001	0.79 (0.69, 0.91)	0.002	0.701
Model 1	0.98 (0.90, 1.07)	0.674	0.83 (0.72, 0.95)	0.012	0.645
Model 2	0.98 (0.90, 1.07)	0.606	0.90 (0.72, 1.14)	0.381	0.638
Protein intake	Protein introduction
Unadjusted	1		0.93 (0.89, 0.98)	0.007	0.831
Model 1	1		0.94 (0.90, 0.98)	0.007	0.796
Model 2	1		0.94 (0.90, 0.98)	0.007	0.795
Outcome at 7–9 y	Risk Ratio (95% CI)
Fruit and vegetable intake	Fruit and vegetable introduction
Unadjusted	0.88 (0.82, 0.95)	0.003	0.74 (0.62, 0.87)	0.002	1.142
Model 1	0.98 (0.91, 1.07)	0.690	0.80 (0.68, 0.95)	0.011	1.117
Model 2	0.97 (0.82, 1.13)	0.681	0.89 (0.69, 1.16)	0.392	1.052
Dairy intake	Dairy introduction
Unadjusted	0.78 (0.68, 0.89)	0.001	0.88 (0.82, 0.95)	0.004	1.084
Model 1	0.96 (0.80, 1.16)	0.679	0.89 (0.83, 0.96)	0.006	1.082
Model 2	0.96 (0.81, 1.13)	0.678	0.95 (0.84, 1.07)	0.577	1.076
Grains intake	Grain introduction
Unadjusted	0.72 (0.64, 0.82)	<.0001	0.78 (0.56, 1.10)	0.412	0.774
Model 1	0.97 (0.83, 1.13)	0.683	0.80 (0.57, 1.13)	0.412	0.761
Model 2	0.95 (0.77, 1.18)	0.656	0.87 (0.60, 1.26)	0.462	0.743
Protein intake	Protein introduction
Unadjusted	1		1.02 (0.92, 1.13)	0.725	0.914
Model 1	1		1.03 (0.93, 1.14)	0.725	0.893
Model 2	1		1.02 (0.92, 1.13)	0.725	0.895

Abbreviations: FDR, False Discovery Rate; X²/DF, Pearson Chi-Square Goodness of Fit Test.

^†Model 1s were partially adjusted for mother’s age, race/ethnicity, education, insurance status, pre-pregnancy BMI, child’s gestational age, parity, Women Infants and Children (WIC) participation; Model 2s were fully adjusted for frequency of fast- food intake and breastfeeding duration.

On the other hand, delaying the introduction of fruits and vegetables and other select complementary foods was associated with lower intake in early childhood even after adjustment of sociodemographic factors and maternal pre-pregnancy BMI. At 2–3 years of age, compared to those introducing fruits and vegetables between 5–8 months, those introducing later was associated with a 13% lower daily subsequent intake of fruits and vegetables (adjusted RR [aRR], 0.87; 95% CI: 0.79, 0.95; FDR p=0.002). Introduction to dairy and grains between 9–12 compared to 5–8 months was associated with a 10% and 17% lower daily subsequent intake, respectively. Lastly, introducing protein between 9–12 months of age was also associated with a 6% (aRR, 0.94; 95% CI: 0.90, 0.98; FDR p=0.007) lower subsequent intake relative to introduction between 5–8 months. When additionally adjusting for other feeding practices such as breastfeeding duration and frequency of concomitant fast-food intake, associations did not persist in the adjusted analysis for fruits and vegetables, dairy, and grains (Table 2). However, it did persist for protein with an 8% (aRR, 0.94; 95% CI: 0.90, 0.98; FDR p=0.007) lower daily subsequent intake.

Associations for intake reported in middle childhood at 7–9 years were examined. Compared to introducing around AAP recommendations, delaying introduction of fruits and vegetables and dairy was associated with a 20% (aRR, 0.80; 95% CI: 0.68, 0.95; FDR p=0.011) and 11% (aRR, 0.89; 95% CI: 0.83, 0.96; FDR p=0.006) lower daily subsequent intake of the select foods, respectively. However, the timing of introducing grains (aRR, 0.80; 95% CI: 0.57, 1.13; FDR p=0.412) and protein (aRR, 1.03; 95% CI: 0.93, 1.14; FDR p=0.725) was no longer associated with later intake at 7–9 years of age after adjusting for covariates. Findings were attenuated after additionally adjusting for breastfeeding duration and frequency of fast-food intake (Table 2).

3.2. Diet Quality

Mean score for each YHEI component at 2–3 and 7–9 years of age is presented in Supplemental Table 3. Mean YHEI [SD] was 39.05 ± 14.10 at 2–3 years of age and 36.38 ± 14.81 at 7–9 years of age.

Apart from protein, U-shaped associations between timing of introduction to select complementary foods and diet quality were observed for diet reported in early and in middle childhood (Table 3). In other words, both earlier and later than AAP recommended introduction to foods were associated with lower YHEI scores. For instance, at 2–3 years of age, compared to children introducing fruits and vegetables between 5–8 months of age, introducing prior to 5 months of age (adjusted B: −3.90; 95% CI: −5.10, −2.70; FDR p=0.0002) and later (adjusted B: −3.24; 95% CI: −5.64, −0.85; FDR p=0.008) introduction of fruits and vegetables were associated with lower diet quality. Results of other select complementary foods were congruent as earlier (adjusted B: −3.54; 95% CI: −4.72, −2.36; FDR p=0.0002) and later (adjusted B: −4.22; 95% CI: −6.99, −1.45; FDR p=0.003) introduction to grains and earlier (adjusted B: −3.23; 95% CI: −5.20, −1.27; FDR p=0.003) and later (adjusted B: −1.98; 95% CI: −3.21, −0.74; FDR p=0.003) introduction to dairy were associated with lower diet quality. Later introduction to protein was not associated with lower diet quality (adjusted B: −0.38; 95% CI: −1.55, 0.80; FDR p=0.549). When adjusting for breastfeeding duration and frequency of fast-food intake, associations remained for fruits and vegetables, grains, and dairy with diet quality (Table 3).

Table 3.

Association with diet quality at 2–3 (n=2826) and 7–9 (n=1449) years of age based on timing of complementary food introduction, Upstate KIDS Study^†

	2–3 years old						7–9 years
Exposure	Unadjusted* Beta (95% CI)	FDR	Model 1† Beta (95% CI)	FDR	Model 2† Beta (95% CI)	FDR	Unadjusted* Beta (95% CI)	FDR	Model 1† Beta (95% CI)	FDR	Model 2† Beta (95% CI)	FDR
Fruits and vegetables
<5 months	−4.94 (−6.20, −3.69)	0.0002	−3.90 (−5.10, −2.70)	0.0002	−3.41 (−4.64, −2.18)	0.0002	−5.07 (−6.76, −3.37)	0.0002	−3.67 (−5.36, −1.98)	0.0002	−2.78 (−4.46, −1.10)	0.002
5–8 months	Ref		Ref		Ref		Ref		Ref		Ref
9–12 months	−5.88 (−8.36, −3.41)	0.0002	−3.24 (−5.64, −0.85)	0.008	−3.48 (−5.86, −1.10)	0.005	−7.21 (−10.57, −3.85)	0.0002	−4.01 (−7.43, −0.60)	0.021	−4.18 (−7.54, −0.82)	0.018
Dairy
<5 months	−4.90 (−6.92, −2.87)	0.0003	−3.23 (−5.20, −1.27)	0.003	−2.63 (−4.60, −0.66)	0.009	−4.36 (−7.29, −1.43)	0.022	−2.69 (−5.56, 0.19)	0.134	−1.42 (−4.29, 1.46)	0.335
5–8 months	Ref		Ref		Ref		Ref		Ref		Ref
9–12 months	−2.72 (−3.99,−1.44)	0.0003	−1.98 (−3.21, −0.74)	0.003	−1.90 (−3.13, −0.67)	0.003	−1.89 (−3.70, −0.08)	0.123	−1.42 (−3.19, 0.35)	0.140	−1.25 (−2.98, 0.48)	0.188
Grains
<5 months	−4.88 (−6.08, −3.67)	0.0002	−3.54 (−4.72, −2.36)	0.0002	−2.87 (−4.10, −1.64)	0.0002	−5.07 (−6.70, −3.43)	0.003	−3.33 (−4.99, −1.66)	0.0003	−2.08 (−3.78, −0.37)	0.034
5–8 months	Ref		Ref		Ref		Ref		Ref		Ref
9–12 months	−5.51 (−8.46, −2.55)	0.0004	−4.22 (−6.99, −1.45)	0.003	−4.68 (−7.44, −1.93)	0.0008	−1.81 (−6.04, 2.42)	0.514	−1.11 (−5.27, 3.05)	0.600	−2.05 (−6.16, 2.05)	0.490
Protein
5–8 months	Ref		Ref		Ref		Ref		Ref		Ref
9–12 months	−0.62 (−1.94, 0.70)	0.549	−0.38 (−1.55, 0.80)	0.549	−0.44 (−1.60, 0.72)	0.549	0.31 (−1.52, 2.14)	0.738	0.59 (−1.11, 2.29)	0.738	0.50 (−1.20, 2.20)	0.738

Abbreviations: FDR, False Discovery Rate.

^†Model 1s were partially adjusted for mother’s age, race/ethnicity, education, insurance status, pre-pregnancy BMI, child’s gestational age, parity, Women Infants and Children (WIC) participation; Model 2s were fully adjusted for frequency of fast- food intake and breastfeeding duration.

At 7–9 years of age, associations persisted for introduction to fruits and vegetables and grains but not dairy. Introduction to fruits and vegetables prior to 5 months (adjusted B: −3.67; 95% CI: −5.36, −1.98; FDR p=0.0002) and after 9 months of age (adjusted B: −4.01; 95% CI: −7.43, −0.60; FDR p=0.021) was associated with lower diet quality. Earlier (adjusted B: −3.33; 95% CI: −4.99, −1.66; FDR p=0.0003) introduction to grains was also associated with lower diet quality. Dairy followed a similar trend for introduction prior to 5 months (adjusted B: −2.69; 95%: −5.56, 0.19) and after 9 months (adjusted B: −1.42; 95% CI: −3.19, 0.19) albeit attenuated after adjustment of sociodemographic factors and maternal pre-pregnancy BMI. Lastly, later introduction to protein was again not associated with diet quality (adjusted B: 0.59; −1.11, 2.29; FDR p=0.738). Associations persisted for fruits and vegetables and for grains for those introduced prior to 5 months compared to 5–8 months when adjusting for breastfeeding duration and frequency of fast-food intake.

3.3. Sensitivity analyses

In a sensitivity analysis excluding those with congenital malformations, results were consistent with our overall findings and did not change in magnitude for later intake (Table 4) and overall diet quality (Table 5) in both early and middle childhood.

Table 4.

Association with intake of select foods based on timing of complementary food introduction at 2–3 (n=2657) and 7–9 (n=1369) years of age, excluding those with diagnosed congenital malformations^†

	Age of complementary food introduction
Model	<5 vs 5–8 months	FDR	9–12 vs 5–8 months	FDR	X2/DF
Outcome at 2–3 y	Risk Ratio (95% CI)
Fruit and vegetable intake	Fruit and vegetable introduction
Unadjusted	0.93 (0.89, 0.97)	0.003	0.83 (0.76, 0.91)	0.0001	0.981
Adjusted	0.99 (0.94, 1.04)	0.683	0.88 (0.80, 0.97)	0.018	0.943
Dairy intake	Dairy introduction
Unadjusted	0.87 (0.80, 0.95)	0.003	0.90 (0.86, 0.95)	0.0002	0.764
Adjusted	0.98 (0.89, 1.08)	0.764	0.91 (0.86, 0.96)	0.0006	0.747
Grains intake	Grain introduction
Unadjusted	0.86 (0.82, 0.91)	<.0001	0.80 (0.69, 0.92)	0.004	0.710
Adjusted	0.98 (0.91, 1.06)	0.676	0.83 (0.73, 0.96)	0.021	0.650
Protein intake	Protein introduction
Unadjusted	1		0.93 (0.89, 0.97)	0.006	0.831
Adjusted	1		0.94 (0.89, 0.98)	0.006	0.796
Outcome at 7–9 y	Risk Ratio (95% CI)
Fruit and vegetable intake	Fruit and vegetable introduction
Unadjusted	0.89 (0.82, 0.96)	0.007	0.73 (0.61, 0.87)	0.002	1.156
Adjusted	0.98 (0.91, 1.06)	0.694	0.79 (0.67, 0.95)	0.020	1.121
Dairy intake	Dairy introduction
Unadjusted	0.78 (0.68, 0.89)	0.002	0.89 (0.83, 0.96)	0.010	1.104
Adjusted	0.96 (0.80, 1.16)	0.679	0.90 (0.83, 0.97)	0.015	1.100
Grains intake	Grain introduction
Unadjusted	0.73 (0.65, 0.83)	<.0001	0.78 (0.56, 1.09)	0.409	0.774
Adjusted	0.97 (0.83, 1.13)	0.684	0.81 (0.58, 1.13)	0.409	0.760
Protein intake	Protein introduction
Unadjusted	1		1.02 (0.92, 1.13)	0.687	0.856
Adjusted	1		1.03 (0.93, 1.14)	0.687	0.826

Abbreviations: FDR, False Discovery Rate; X²/DF, Pearson Chi-Square Goodness of Fit Test.

^†Models were adjusted for mother’s age, race/ethnicity, education, insurance status, pre-pregnancy BMI, child’s gestational age, parity, Women Infants and Children (WIC) participation.

Table 5.

Association with diet quality at 2–3 (n=2632) and 7–9 (n=1364) years of age based on timing of complementary food introduction, excluding those with diagnosed congenital malformations^†

	2–3 years old				7–9 years old
Exposure	Unadjusted* Beta (95% CI)	FDR	Adjusted Model † Beta (95% CI)	FDR	Unadjusted* Beta (95% CI)	FDR	Adjusted Model† Beta (95% CI)	FDR
Fruits and vegetables
<5 months	−4.78 (−6.06, −3.49)	0.0001	−3.75 (−4.98, −2.52)	0.0001	−4.94 (−6.71, −3.16)	0.0003	−3.52 (−5.31, −1.74)	0.0001
5–8 months	Ref		Ref		Ref		Ref
9–12 months	−5.68 (−8.24, −3.11)	0.0001	−3.05 (−5.54, −0.56)	0.016	−7.02 (−10.47, −3.57)	0.0003	−3.84 (−7.36, −0.33)	0.032
Dairy
<5 months	−4.79 (−6.89, −2.69)	0.0004	−3.04 (−5.09, −1.00)	0.005	−4.67 (−7.72, −1.62)	0.011	−2.80 (−5.81, 0.20)	0.094
5–8 months	Ref		Ref		Ref		Ref
9–12 months	−2.52 (−3.85, −1.2)	0.0004	−1.78 (−3.06, −0.51)	0.006	−1.73 (−3.60, 0.14)	0.094	−1.22 (−3.05, 0.61)	0.193
Grains
<5 months	−4.58 (−5.81, −3.36)	0.0002	−3.29 (−4.50, −2.08)	0.0002	−5.10 (−6.78, −3.42)	<.0001	−3.51 (−5.22, −1.79)	0.0002
5–8 months	Ref		Ref		Ref		Ref
9–12 months	−5.23 (−8.26, −2.2)	0.001	−3.98 (−6.85, −1.11)	0.007	−1.70 (−5.89, 2.50)	0.571	−0.92 (−5.09, 3.24)	0.663
Protein
5–8 months	Ref		Ref		Ref		Ref
9–12 months	−0.8 (−2.16, 0.57)	0.351	−0.58 (−1.81, 0.65)	0.351	0.38 (−1.47, 2.24)	0.684	0.67 (−1.07, 2.41)	0.684

Abbreviations: FDR, False Discovery Rate.

^†Models were adjusted for mother’s age, race/ethnicity, education, insurance status, pre-pregnancy BMI, child’s gestational age, parity, Women Infants and Children (WIC) participation.

4. Discussion

In our population based, longitudinal cohort, we observed that later introduction of complementary foods, such as fruits and vegetables, dairy, and protein, was associated with lower subsequent intake in early and middle childhood. Furthermore, we found associations between introduction of fruits and vegetables and dairy prior to 5 months and after 9 months of age associated with lower diet quality even after adjustment for other feeding behaviors at 2–3 years of age such as breastfeeding and fast-food consumption. These associations persisted for age of introduction for fruits and vegetables and only prior to 5 months for grains with lower diet quality at 7–9 years of age. Lastly, findings remained consistent when accounting for congenital malformations that may influence feeding behaviors.

Although we found significant associations between later timing of select complementary food introduction and subsequent intake, there are few studies comparable to our findings. While we previously observed that earlier introduction to juice increased the subsequent intake of sugar-containing beverages defined as juice and sugar-sweetened beverages in the Upstate KIDS cohort (38), the study did not examine how the timing of select complementary foods may influence intake of select foods and children’s diet quality. One study, however, found no association between the age of introduction of fruits and vegetables prior to <5 months of age and intake at age 6 years, although the study was underpowered in their age at introduction groups (16). Another study conducted by Rose et al. in 1203 children found fruit and vegetable intake at 9 months of age was associated with higher intake of fruit and vegetables at 6 years of age (p<0.0001), although it did not distinguish whether this was influenced by the age in which fruit and vegetables were introduced during infancy (18). Moss et al. found in 4891 children, that those who were fed fruits or vegetables as their first complementary food ate fruit (adjusted β: 0.19; 95% CI: 0.08, 0.31) and vegetables (adjusted β: 0.13; 95% CI: 0.04, 0.22) more frequently at age 7, respectively, relative to children introduced to cereal first (39).

Currently, there are limited studies that have evaluated the effects of age of introduction to select complementary foods and diet quality in later childhood (11, 21). The prior studies have focused on effects of the age of complementary foods overall or only on select complementary foods introduced during infancy and diet quality, although not consistent with our study. One study conducted by Weinfield et al. in 1223 children found no association between the age at which complementary foods overall were introduced and diet quality measured by the Healthy Eating Index (HEI)-2015 (adjusted β: 0.05; 95% CI: −0.01, 0.12) at 3 years of age. Another study by Switkowski et al. found later introduction of both sweets and/or fruit juice resulted in higher total YHEI score (adjusted β: 4.5; 95% CI: 1.0, 7.4), although in children fully or partially breastfed at 6 months of age not on the entire analytical sample (25). Therefore, limited evidence remains on whether the timing of select complementary foods influences overall diet quality in later childhood.

Although few studies have found an association between complementary food introduction and diet quality, previous studies have found an association between breastfeeding duration and later diet quality, therefore we accounted for this in our fully adjusted analysis. The hypothesized biological mechanism between breastfeeding and children’s later diet quality is that breastfeeding may influence the acceptability of certain foods mainly by the introduction of a diverse range of flavors as infants are exposed to their mother’s breastmilk. The repeated exposure to these flavors through maternal diet can help to facilitate infants’ acceptance of new foods (40, 41), therefore increasing the programming of healthier dietary preferences and consumption of select foods in childhood (42). A systematic review by Eslami et al. included 11 studies examining the association between breastfeeding exposure and duration and diet quality scores over 1 year of age and 9 studies examining overall dietary patterns (34). Three studies found duration of any breastfeeding was positively associated with diet variety between 2–4 years of age (43–46), while 4 studies found longer duration of breastfeeding was significantly associated with diet quality in children aged 1–5 years (47, 48). The remaining studies found no association between 7–10 years of age for breastfeeding duration and diet quality scores (49, 50).

Though the YHEI calculates for fast-food outside home, previous literature has shown in children and adults that the daily frequency of fast-food intake is associated with intake of food groups (51–53) and overall diet quality (35, 54, 55), therefore we further weighed any association adjusting for the frequency of fast-food intake in the model to examine if associations would hold due to its influence on dietary patterns. When adjusting for frequency of fast-food, we observed a significant association between introduction of select complementary foods and diet quality in early and middle childhood. This is consistent with previous evidence suggesting that other behaviors during the complementary food period are responsible for shaping later dietary preferences (42, 56).

4.1. Strengths and limitations

There are several limitations. Reporting error has been documented when parents complete FFQs about their children’s diets (57). The study did not incorporate food models and images to overcome this limitation to help parents standardize the size of each serving consumed. In relation to questions on the timing of complementary feeding, the questionnaire did not obtain data on the frequency or quantity of the solid foods and beverages consumed prior to 30 months. Certain foods may have been introduced more than once or never introduced again, and our study was unable to grasp these differing feeding practices. The specific questions for reporting food items, either in infancy for classifying complementary food introduction or for the amounts consumed as toddlers, were not previously validated. In addition, due to unequal group sizes for the timing of complementary foods, the models had variable power to detect differences across food groups. However, one strength is the ability to assess the infant’s introduction to separate types of complementary foods consumed longitudinally across early childhood versus asking mothers to recall the age of introduction after the event has occurred from 4 to 12 months (38). Lastly, the present sample is predominantly non-Hispanic White and of higher socioeconomic status which may limit the generalizability of the study findings and our ability to see differences between racial/ethnic groups. However, our study remains novel as there are limited prospective studies examining timing of complementary food introduction and dietary patterns in both early and middle childhood.

5. Conclusion

In conclusion, in terms of intake of certain food items, later introduction was associated with lower subsequent intake of that complementary food item in preschool and middle childhood, but not after accounting for breastfeeding duration and fast-food intake. When examining diet quality, both earlier and later introduction were associated with lower diet quality even when adjusting for additional feeding behaviors such as breastfeeding duration and frequency of fast-food exposure. Our research builds on previous research on this topic supporting observed associations on the influence early infant feeding practices have on later diet quality and behaviors in childhood. Overall, this study highlights the need for further research to evaluate feeding practices that may affect food preferences during infancy and their impact on the child’s diet quality, which is key in implementing healthy dietary patterns.

Abbreviations

AAP

American Academy of Pediatrics

WHO

World Health Organization

YHEI

Youth Healthy Eating Index

HEI

Healthy Eating Index

RCT

Randomized Clinical Control Trial

FFQ

Food Frequency Questionnaire

IFPSII

Infant Feeding Practices Study II

WIC

Women, Infants, and Children

BMI

Body Mass Index

Data and code availability

Data described in the manuscript, code book, and analytic code will be made available upon request pending internal review. Requests for data and other information can be sent to the corresponding author, Dr. Edwina Yeung, Division of Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health.