Diet quality comparisons in Hispanic/Latino siblings: Results from the Hispanic Community Children’s Health Study/Study of Latino Youth (SOL Youth)

Abstract

The objective of this study was to determine how well Hispanic/Latino siblings’ diet quality correlate with each other and whether social and environmental factors explained potential differences. Hispanic/Latino 8–16-year-olds from the cross-sectional Hispanic Community Children’s Health Study/Study of Latino Youth (SOL Youth) with at least one sibling enrolled in the study were examined (n=740). Diet quality was assessed with the Healthy Eating Index 2010 (HEI-2010), calculated from two 24-hour recalls. Mixed effects models were used with HEI-2010 score as the outcome, and correlations in siblings’ diet quality were assessed with intraclass correlation coefficients (ICCs). All models were examined stratified by age and sex. Diet-related social and environmental measures were added as fixed effects in a secondary analysis. Mean (standard deviation) overall HEI-2010 score was 53.8 (13.0). The ICC for siblings’ HEI-2010 score was 0.31 (95% CI: 0.25, 0.38). Siblings who were born <3 vs. ≥3 years apart had stronger correlations in overall diet quality (0.47 [95% CI: 0.37, 0.58] vs. 0.21 [95% CI: 0.13, 0.30]), but no differences were observed in overall HEI-2010 score according to sex. Greater peer support for fruit and vegetable intake (β=1.42 [95% CI: 0.62, 2.21]) and greater away-from-home food consumption (β=−1.24 [95% CI: −2.15, −0.32]) were associated with differences in siblings’ diet quality. Overall diet quality scores of Hispanic/Latino siblings in this study were slightly correlated, with stronger correlations among siblings closer in age. Differences in peer support and foods consumed outside the home may explain differences in siblings’ diet quality. Future research should investigate additional determinants of differences in siblings’ diets.

1. INTRODUCTION

Among 2- to 19-year-olds in the United States (US), the prevalence of obesity is significantly higher for Hispanic/Latino youth compared to non-Hispanic White youth (26% vs. 14%).¹ Improvements in diet quality and increases in physical activity reduce risk for childhood obesity in the short- and long-term,² but maintaining these healthy behaviors is largely contingent on family support, encouragement, and participation.³ While the parent-child relationship is the most frequently researched component of family determinants of obesity, siblings may also play a critical role.^4-6 This is particularly true in Hispanic/Latino families whose cultural values are distinctive for stressing close and supportive relationships among extended and immediate family, including siblings.^7-9

Siblings act as both companions/peers and caregivers (e.g., in the absence of parental supervision), resulting in siblings having multiple spheres of influence on one another’s health behaviors.^10-12 Siblings may promote healthy diets for one another¹³ by role-modeling healthy food choices (which in turn can shape siblings’ food selection and dietary preferences),¹⁴ maintaining healthy diets themselves,¹⁵ or encouraging siblings to follow healthy diets¹⁵. They may also promote unhealthy diets through teasing one another about healthy food intake or by role modeling unhealthy dietary intake.^14,16 Despite the potential importance of siblings for shaping dietary behaviors and intake in youth, few studies have examined the role of siblings.¹³ As such, there has been a call for additional research to clarify the impact sibling have on one another’s diet in youth.¹³

Research on the association between siblings’ dietary intake during childhood or adolescence has primarily been conducted in twins, and findings have been inconsistent.^17,18 Associations in diet have also been observed in non-twin siblings, but the significance and degree of correlation has been shown to vary depending on the component of diet examined (e.g., fruit and vegetable vs. fast food consumption¹⁷) and on whether siblings are close in age.^19,20 It has also been suggested that sex concordance among siblings may determine the degree of correlation in siblings’ diet,¹³ but data to support this notion is currently limited.^19,20 To our knowledge, only two previous studies of siblings’ diets have been conducted in primarily Hispanic/Latino samples, both of which examined Mexican American families over 25 years ago.^21,22 Given that children’s dietary intake has changed from the 1980s to 2010²³ and that these studies limited their examinations to Mexican American youth, there is a need to examine how Hispanic/Latino siblings’ diets correlate using more recent data and with individuals from diverse Hispanic/Latino backgrounds. Findings of such research can inform future obesity interventions and clinical dietary counseling by providing insight on whether including siblings in these practices may help create or maintain healthy dietary change.²⁴

Further, there is a need to identify potential modifiable determinants of differences in overall diet quality in siblings, including social and environmental determinants.²⁵ While siblings generally share their home and neighborhood environments, non-shared environments can impact the degree to which siblings’ diets resemble one another.²⁶ One of the most promising non-shared social determinants of health behavior in youth is peers,¹³ with previous studies citing peer support for healthy or unhealthy dietary intake as a key determinant of dietary intake.^15,27,28 Additionally, non-shared food environments of siblings, including away-from-home food establishments and the school food environment, have been identified as determinants of youth’s diet and may explain variance in siblings’ healthy and unhealthy dietary intake.^29-31 By identifying these modifiable determinants of differences in siblings’ diet quality, future family-based obesity interventions can target these measures to create successful behavior change in siblings who spend more of their time outside of shared environments.

The objective of this study is thus to examine the correlation of Hispanic/Latino siblings’ diet quality, as measured using Healthy Eating Index 2010 (HEI-2010) scores,³² and to examine social and environmental determinants of differences in siblings' diet quality to inform future family-based obesity interventions and dietary counseling. We selected diet quality as our outcome due to its importance regarding childhood obesity prevention³³ and because no previous study, to our knowledge, has examined sibling correlations in this diet measure. We expected that HEI-2010 scores would be positively correlated in siblings, with stronger correlations among siblings who were closer in age. Further, we hypothesized that individuals with greater peer support for fruit and vegetable intake, lower away-from-home food consumption, and a healthier school food environment than their siblings would have better diet quality.

2. METHODS

2.1. Study population.

The Hispanic Community Children’s Health Study/Study of Latino Youth (SOL Youth) is a cross-sectional, ancillary study of the Hispanic Community Health Study/Study of Latinos (HCHS/SOL).³⁴ HCHS/SOL is a prospective, community-based cohort study of 16,415 self-identified Hispanic/Latino adults (ages 18-74 years) who were selected using a stratified, two-stage probability sampling design across four US communities (Bronx, New York; Chicago, Illinois; Miami, Florida; San Diego, California), supported by a Coordinating Center at the University of North Carolina at Chapel Hill.^34,35 Between 2012 and 2014, SOL Youth ³⁶ enrolled 1466 youth aged 8 to 16 years living in the household of a parent/caregiver (henceforth referred to as the parent) who completed the HCHS/SOL baseline examination (2008-2011). All siblings living in the same household were eligible for participation in SOL Youth. SOL Youth study participation included three components: 1) an initial clinical examination at the field center, 2) seven days of wearing a physical activity monitor, and 3) a second 24-hour dietary recall via telephone within a month of the initial clinic visit. Protocols for HCHS/SOL and SOL Youth were approved by the institutional review boards of each institution participating in the study and are published elsewhere.^34-36 Written informed consent and assent were obtained from parents and youth, respectively.

2.2. Diet assessment.

Youth’s diet was assessed using two interviewer-administered 24-hour dietary recalls in the participant’s preferred language (Spanish or English). The first interview was conducted in person at the clinical examination, and the second interview was conducted via telephone between five and thirty days later. All interviews employed the multiple pass procedure using the Nutrition Data System for Research (NDSR) software from the Nutrition Coordinating Center at the University of Minnesota. NDSR versions 10-12 (2010-2012) were used to collect data, and all raw files were processed using version 13 (2013).^37,38 To aid in recalling portion sizes, participants were provided with food models (first recall) or a Food Amounts Booklet (second recall).

Adherence to the 2010 Dietary Guidelines for Americans was assessed using HEI-2010 score.³² HEI-2010 is comprised of 12 component scores: nine adequacy components (total fruit, whole fruit, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant proteins, and fatty acids) and three moderation components (refined grains, sodium, and empty calories).³² Higher scores for the adequacy components indicate higher consumption, while higher scores for the moderation components indicate lower consumption.³² The twelve component scores are summed to determine overall HEI-2010 score (range: 0-100), with higher scores indicating better diet quality.³² Averaged dietary intake data from all available 24-hour dietary recalls was used to calculated HEI-2010 score.

2.3. Social and environmental determinants.

Interviewers administered the following questionnaires in English or Spanish to each youth participant. All questionnaires used in the present study were validated in similar youth populations and pilot-tested in age-appropriate samples before their inclusion in SOL Youth. Data from the interviews were entered into a standardized, web-based data management system developed by the Coordinating Center. Additional quality control checks of the interview process included directly observing selected procedures, audio-taping interviews, and conducting repeatability studies.³⁶

The present analysis uses only select items from each questionnaire. Thus, to confirm that the constructs of interest were being assessed by the selected items, an exploratory factor analysis (EFA) was run on candidate items from each questionnaire. For all EFAs, factors were retained by the proportion criterion and were rotated using varimax rotation. Items that loaded <0.30 on all factors were removed and the EFA was re-run.³⁹

Composite scores for each questionnaire were created based on the items that remained after the EFA. Composite scores were set to missing if individuals were missing responses for more than one item. Details on the identification of candidate items and the creation of composite scores can be found below.

2.3.1. Peer support for fruit and vegetable intake.

Peer support for healthy dietary intake was assessed using three items from a 16-item questionnaire on family and friend support for diet and physical activity.^40,41 The diet-specific items of the SOL Youth questionnaire were originally developed in a sample of college students in the US.⁴¹ Results of the original study showed the diet-specific social support scale had acceptable internal consistency (Cronbach’s alpha=0.82) and test-retest reliability (intraclass correlation coefficient [ICC]=0.68).⁴¹

In the present study, individuals indicated, using a 5-point Likert-type response scale (1=never to 5=everyday), the number of days 1) a friend encouraged them to eat fruits and vegetables, 2) a friend ate fruits and vegetables with them, or 3) other kids teased them for eating fruits and vegetables. Based on the EFA, the item on teasing was dropped, and the resulting 2-item scale had acceptable internal consistency (Cronbach’s α=0.66). Individuals received a mean score (range: 1-5) based on responses to these two items.

2.3.2. Away-from-home food consumption.

Foods consumed away from home were measured using a 6-item questionnaire.⁴² These items were adapted from another questionnaire designed to assess away-from-home food intake in elementary school students living in Southern California.⁴² For each item, participants reported the number of days that they ate ready-to-eat food from relatives/friends’ homes, fast food restaurants, other restaurants, grocery stores, school cafeterias, and other outlets (e.g., street vendors) in the past week. Based on results of the EFA, the item on school cafeterias was dropped, and the resulting 5-item scale had acceptable internal consistency (Cronbach’s α=0.62). Individuals received a mean score (range: 0-7) representing the mean number of days that individuals reported they ate away-from-home foods across these five locations.

2.3.3. School food environment.

The school food environment was assessed using eight items about the presence of healthy and unhealthy vendors at school and the frequency with which the participant used them.⁴³ The items were selected from a larger questionnaire developed to assess the physical environment related to chronic diseases in school-aged youth in the US (test-retest reliability of these items: ICCs ranged from 0.13 to 0.79).⁴³ Results of the EFA showed that only the four items on use of school vending machines (1. Food vending machines, 2. Drink vending machines, 3. Healthy food vending machines, and 4. Healthy drink vending machines) had factor loadings >0.30. Although all four items loaded on the same factor, the Cronbach’s α for the four items was poor (Cronbach's α=−0.30; items for healthy vending machines were reverse-coded). As such, the items for healthy vending machines were dropped, and the resulting 2-item scale showed acceptable internal consistency (Cronbach’s α=0.75). Individuals received a mean score (range: 0-5) representing the number of days they used the school’s food and drink vending machines.

2.4. Covariates.

Age, sex, Hispanic/Latino background, and nativity (born in the US 50 states/DC vs. US territory or foreign country) were self-reported by youth. The parent reported annual household income and their own educational attainment. These covariates are consistently used in analyses of correlations in siblings’ diet^17,19,44 and are recommended to account for the known differences in diet according to age, sex, ethnicity, and socioeconomic status.⁶

2.5. Statistical Analysis.

Of the 1466 youth in SOL Youth, 792 had at least one other sibling enrolled in the study and were thus eligible for this analysis. Of these 792 youth, we excluded seven individuals whose dietary recalls were missing (n=4) or all completed recalls were considered implausible (n=3)⁴⁵ and an additional 36 individuals for missing sociodemographic information. We further excluded 9 individuals who did not have at least one sibling in the analytic dataset after the exclusion criteria were applied. The final analytic sample size was 740 youth.

The below analytic plan and all hypotheses were pre-specified during the manuscript proposal process. Data for these analyses are not openly available due to privacy concerns of clinical data; data can be made available upon request and at the approval of the Coordinating Center. We used linear mixed-effects models to estimate correlations in siblings’ diet quality. Models used overall HEI-2010 score and each HEI-2010 score component as the outcome. A random intercept representing the random effect of family was included in all models, allowing for examination of correlations in sibling sets of sizes ranging from dyads to multiples. Youth’s age, sex, Hispanic/Latino background, and nativity; parent’s education; and household income were fixed effects. ICCs were calculated and were used to estimate correlations in siblings’ diet quality.^46,47 Confidence intervals were constructed for all ICCs using a cluster bootstrap procedure (with “family” as the cluster), where selection occurred with equal probability and with replacement. Bootstrapping was conducted for 5000 iterations and 95% CIs were estimated based on the standard deviation of the bootstrapped distribution. ICCs were considered significant if the 95% CI did not include the null hypothesis (ICC=0.00). ICCs were interpreted according to the guidelines by Shrout⁴⁸: virtually no similarity=0.00–0.10; slight=0.11–0.40; fair=0.41–0.60; moderate=0.61–0.80; and substantial similarity=0.81–1.00.

To examine whether correlations in diet quality varied according to age, age differences were assigned to each sibling in a family relative to the youngest sibling in the analytic sample. Analyses were stratified by whether all siblings were <3 years apart or ≥3 years apart (the median “maximum age difference in sibling sets”, similar to the approach of a previous study¹⁹). Exploratory analyses examined whether correlations differed for siblings who were all 8-11 years old, 12-16 years old, or of mixed ages. To determine whether correlations in diet quality differed according to sex, analyses were stratified according to if all siblings in the analytic sample were brothers, sisters, or a mix of brothers and sisters. The same bootstrap approach described above was used to calculate 95% CIs for each ICC. Significant differences in sibling diet quality correlations according to age difference or sex concordance were based on whether the 95% CIs overlapped for the stratified ICCs. To explore whether potential differences in siblings’ diet quality were explained by differences in siblings’ peer support for fruit and vegetable intake, away-from-home food consumption, and the school food environment, these variables were added into the models together as fixed effects. All analyses were conducted in SAS software version 9.4 (SAS Institute). Significance was set at p <0.05 for all analyses.

3. RESULTS

An overview of the analytic sample from SOL Youth is provided in Table 1. In this sample of 740 individuals, 66.2% had one sibling, 27.2% had two siblings, and 6.6% had at least three siblings enrolled in the study. Approximately half of the sample was 12 years of age or older (54.2%) and female (50.1%). Youth were predominately born in the US (78.8%), of Mexican heritage (47.0%), and from low socioeconomic status households (54.0% of parents with household income ≤$20,000). Mean overall HEI-2010 score was 53.8, and mean BMI percentile was 71.2.

Table 1.

Characteristics of the analytic sample of siblings from SOL Youth (n=740)

	Mean (SD) or n (%)
Age group, years (n [%])
8-11	339 (45.8)
12-16	401 (54.2)
Sex (n [%])
Female	371 (50.1)
Male	369 (49.9)
Hispanic/Latino background (n [%])
Dominican	80 (10.8)
Central American	44 (5.9)
Cuban	48 (6.5)
Mexican	348 (47.0)
Puerto Rican	72 (9.7)
South American	27 (3.6)
Mixed/Other	121 (16.4)
Nativity (n [%])
Born outside of United States	157 (21.2)
Born in United States	583 (78.8)
Parental education (n [%])*
<HS or GED	133 (41.0)
HS or GED	98 (30.2)
>HS or GED	93 (28.7)
Household income (n [%])*
≤$20,000	175 (54.0)
>$20,000 to <$40,000	103 (31.8)
≥$40,000	46 (14.2)
BMI, kg/m2 (mean [SD])	22.1 (6.4)
BMI percentile (mean [SD])	71.2 (28.4)
HEI-2010 overall score (mean [SD])	53.8 (13.0)
Peer support for F&V intake (mean [SD])†	2.0 (1.1)
Away-from-home food consumption, # days/week (mean [SD])	1.3 (1.0)
School vending machine use, #days use/week (mean [SD])	0.6 (1.1)

HS, High School; F&V, Fruit and Vegetable

^*Parent-reported (n=324 parents)

^†Measured as frequency of peer support on a 5-point Likert scale

Table 2 shows the correlations in siblings’ HEI-2010 overall and component scores. The ICC for siblings’ overall HEI-2010 score was 0.31 (95% CI: 0.25, 0.38), indicating “slight” correlation⁴⁸ in siblings’ overall diet quality. Examination of the twelve HEI-2010 components showed that siblings’ intake of empty calories and sodium had the highest correlations (0.31 [95% CI: 0.24, 0.38]; 0.32 [95% CI: 0.25, 0.38], respectively), while intake of total protein foods had the lowest correlation (0.13 [95% CI: 0.04, 0.21]).

Table 2.

Intraclass correlation coefficients (ICC [95% CI]) for association of HEI-2010 overall and component scores in siblings in SOL Youth (n=740)

	Overall
HEI-2010, overall	0.31 (0.25, 0.38) *
Total fruit†	0.20 (0.13, 0.27)*
Whole fruit‡	0.20 (0.13, 0.27)*
Total vegetables§	0.19 (0.11, 0.26)*
Greens and beans§	0.22 (0.14, 0.30)*
Whole grains	0.20 (0.12, 0.27)*
Dairy¶	0.18 (0.10, 0.25)*
Total protein foods#	0.13 (0.04, 0.21)*
Seafood and plant proteins**	0.29 (0.22, 0.36)*
Fatty acids††	0.21 (0.14, 0.28)*
Refined grains	0.20 (0.13, 0.28)*
Sodium	0.32 (0.25, 0.38)*
Empty calories‡‡	0.31 (0.24, 0.38)*

^*Significant correlation based on not containing the null hypothesis (ICC=0.00)

All ICCs are from mixed models adjusted for youth’s age, youth’s sex, youth’s Hispanic/Latino background, youth’s nativity, parent’s education, and household income as fixed effects and family as a random intercept

^†Includes fruit juice

^‡Includes all forms except juice

^§Includes any beans and peas not counted as total protein foods

^¶Includes all milk products (e.g., fluid milk, cheese) and fortified soy beverages

^#Beans and peas are included here when the total protein foods standard is otherwise not met

^**Includes seafood, nuts, seeds, and soy products (other than beverages), as well as beans and peas counted as total protein foods

^††Ratio of polyunsaturated fatty acids (PUFA) and monounsaturated fatty acids (MUFA) to saturated fatty acids (SFA)

^‡‡Calories from solid fats, alcohol, and added sugars; threshold for counting alcohol is >28 grams/day

Table 3 shows the correlations in siblings’ HEI-2010 overall and component scores, stratified by age difference. Siblings who were <3 years apart had “fair” correlation in overall diet quality (0.47 [95% CI: 0.37, 0.58]) compared to “slight” correlation (0.21 [95% CI: 0.13, 0.30]) among siblings ≥3 years apart. Examination of the HEI-2010 components showed that only correlations for total protein foods significantly differed for siblings <3 years apart (0.32 [95% CI: 0.18, 0.46]) vs. ≥3 years apart (0.02 [95% CI: −0.08, 0.13]). An exploration of the combined role of age difference and age groups on siblings’ correlations showed that this difference in total protein food correlations for siblings <3 vs. ≥3 years apart may be driven by sibling sets who are closer in age and younger (Supplementary Table 1). Specifically, sibling sets who were all between the ages of 8-11 years old and thus close in age had more similar total protein and seafood and plant protein scores compared to sibling sets who were further apart in age and thus spanned both age groups examined (8-11 and 12-16 years).

Table 3.

Intraclass correlation coefficients (ICC [95% CI]) for association of HEI-2010 overall and component scores in siblings in SOL Youth, stratified by age difference (n=740)

	Age difference
	<3 years (n=271)	≥3 years (n=469)
HEI-2010, overall (Mean[SD])	52.7 (13.9)	54.4 (12.5)
HEI-2010, overall	0.47 (0.37, 0.58) *	0.21 (0.13, 0.30) *
Total fruit†	0.29 (0.16, 0.42)	0.16 (0.06, 0.26)
Whole fruit‡	0.29 (0.17, 0.42)	0.15 (0.06, 0.25)
Total vegetables§	0.20 (0.06, 0.34)	0.19 (0.09, 0.28)
Greens and beans§	0.29 (0.15, 0.42)	0.19 (0.09, 0.29)
Whole grains	0.19 (0.06, 0.32)	0.19 (0.10, 0.28)
Dairy¶	0.22 (0.08, 0.36)	0.16 (0.06, 0.26)
Total protein foods#	0.32 (0.18, 0.46)*	0.02 (−0.08, 0.13)*
Seafood and plant proteins**	0.34 (0.23, 0.46)	0.29 (0.20, 0.38)
Fatty acids††	0.31 (0.19, 0.43)	0.17 (0.08, 0.26)
Refined grains	0.22 (0.09, 0.34)	0.20 (0.11, 0.30)
Sodium	0.29 (0.16, 0.42)	0.33 (0.24, 0.41)
Empty calories‡‡	0.40 (0.28, 0.51)	0.27 (0.18, 0.35)

^*Significantly different based on non-overlapping 95% CIs

All ICCs are from mixed models adjusted for youth’s age, youth’s sex, youth’s Hispanic/Latino background, youth’s nativity, parent’s education, and household income as fixed effects and family as a random intercept

^†Includes fruit juice

^‡Includes all forms except juice

^§Includes any beans and peas not counted as total protein foods

^¶Includes all milk products (e.g., fluid milk, cheese) and fortified soy beverages

^#Beans and peas are included here when the total protein foods standard is otherwise not met

^**Includes seafood, nuts, seeds, and soy products (other than beverages), as well as beans and peas counted as total protein foods

^††Ratio of polyunsaturated fatty acids (PUFA) and monounsaturated fatty acids (MUFA) to saturated fatty acids (SFA)

^‡‡Calories from solid fats, alcohol, and added sugars; threshold for counting alcohol is >28 grams/day

Table 4 provides correlations according to sex concordance of siblings. Overall HEI-2010 scores did not significantly differ across the three sib-types. However, examination of the HEI-2010 components showed that correlations for greens and beans (0.48 [95% CI: 0.30, 0.67] vs. 0.18 [95% CI: 0.08, 0.29]) and whole grains (0.43 [95% CI: 0.27, 0.59] vs. 0.14 [95% CI: 0.05, 0.23]) were significantly greater for siblings who were all brothers vs. a mix of sisters and brothers.

Table 4.

Intraclass correlation coefficients (ICC [95% CI]) for association of HEI-2010 overall and component scores in siblings in SOL Youth, stratified by sex concordance (n=740)

	Sisters (n=141)	Brothers (n=146)	Sisters-brothers (n=453)
HEI-2010, overall (Mean[SD])	55.4 (13.2)	52.9 (12.4)	53.6 (13.2)
HEI-2010, overall	0.40 (0.23, 0.56)	0.52 (0.39, 0.65)	0.27 (0.18, 0.36)
Total fruit†	0.38 (0.20, 0.55)	0.27 (0.09, 0.46)	0.15 (0.05, 0.25)
Whole fruit‡	0.33 (0.16, 0.50)	0.27 (0.08, 0.47)	0.17 (0.07, 0.26)
Total vegetables§	0.24 (0.05, 0.43)	0.16 (−0.05, 0.36)	0.18 (0.08, 0.29)
Greens and beans§	0.16 (−0.03, 0.36)	0.48 (0.30, 0.67)*	0.18 (0.08, 0.29)*
Whole grains	0.20 (−0.02, 0.42)	0.43 (0.27, 0.59)*	0.14 (0.05, 0.23)*
Dairy¶	0.24 (0.03, 0.46)	0.11 (−0.10, 0.32)	0.19 (0.09, 0.29)
Total protein foods#	0.06 (−0.12, 0.24)	0.28 (0.03, 0.54)	0.10 (−0.02, 0.21)
Seafood and plant proteins**	0.13 (−0.05, 0.32)	0.33 (0.15, 0.50)	0.35 (0.26, 0.44)
Fatty acids††	0.11 (−0.08, 0.30)	0.23 (0.01, 0.45)	0.23 (0.14, 0.32)
Refined grains	0.05 (−0.15, 0.26)	0.29 (0.10, 0.48)	0.21 (0.11, 0.30)
Sodium	0.27 (0.09, 0.45)	0.27 (0.08, 0.47)	0.36 (0.27, 0.44)
Empty calories‡‡	0.35 (0.17, 0.53)	0.37 (0.20, 0.55)	0.30 (0.21, 0.39)

^*Significantly different based on non-overlapping 95% CIs

All ICCs are from mixed models adjusted for youth’s age, youth’s sex (sisters-brothers model only), youth’s Hispanic/Latino background, youth’s nativity, parent’s education, and household income as fixed effects and family as a random intercept

^†Includes fruit juice

^‡Includes all forms except juice

^§Includes any beans and peas not counted as total protein foods

^¶Includes all milk products (e.g., fluid milk, cheese) and fortified soy beverages

^#Beans and peas are included here when the total protein foods standard is otherwise not met

^**Includes seafood, nuts, seeds, and soy products (other than beverages), as well as beans and peas counted as total protein foods

^††Ratio of polyunsaturated fatty acids (PUFA) and monounsaturated fatty acids (MUFA) to saturated fatty acids (SFA)

^‡‡Calories from solid fats, alcohol, and added sugars; threshold for counting alcohol is >28 grams/day

Table 5 shows the association of differences in siblings’ social and environmental determinants with differences in siblings’ diet quality. Individuals who reported greater peer support for fruit and vegetable intake than their siblings had higher overall HEI-2010 scores (β=1.42 [95% CI: 0.62, 2.21]). Individuals who reported greater away-from-home food consumption than their siblings had lower overall HEI-2010 scores (β=−1.24 [95% CI: −2.15, - 0.32]). School vending machine use did not explain differences in siblings’ diet quality. Significant associations of differences in siblings’ social and environmental determinants with differences in siblings’ individual HEI-2010 components were limited (Table 5). However, the direction of the associations were the same as those reported for overall HEI-2010 score. Specifically, individuals who reported greater peer support for fruit and vegetable intake than their siblings had higher scores for total fruit (β=0.19 [95% CI: 0.07, 0.31]), whole fruit (β=0.23 [95% CI: 0.09, 0.36]), whole grains (β=0.33 [95% CI: 0.09, 0.58]), dairy (β=0.19 [95% CI: 0.01, 0.37]), and refined grains (β=0.24 [95% CI: 0.02, 0.46]). Individuals who reported greater away-from-home food consumption than their siblings reported lower scores for whole grains (β=−0.28 [95% CI: −0.56, −0.01]) and empty calories (β=−0.39 [95% CI: −0.71, −0.07]).

Table 5.

Beta coefficients (95% CI) for association of differences in siblings’ social and environmental determinants with siblings’ diet quality in SOL Youth (n=720)

	Peer support for F&V intake	Away-from-home food consumption	School vending machine use
HEI-2010, overall	1.42 (0.62, 2.21) **	−1.24 (−2.15, −0.32) **	−0.05 (−0.89, 0.80)
Total fruit†	0.19 (0.07, 0.31)**	−0.14 (−0.28, 0.00)	0.04 (−0.09, 0.17)
Whole fruit‡	0.23 (0.09, 0.36)**	−0.09 (−0.25, 0.06)	−0.03 (−0.17, 0.11)
Total vegetables§	0.02 (−0.07, 0.12)	0.05 (−0.06, 0.15)	−0.02 (−0.12, 0.08)
Greens and beans§	0.01 (−0.11, 0.13)	0.06 (−0.07, 0.20)	0.01 (−0.11, 0.14)
Whole grains	0.33 (0.09, 0.58)**	−0.28 (−0.56, −0.01)*	0.08 (−0.17, 0.34)
Dairy¶	0.19 (0.01, 0.37)*	−0.18 (−0.39, 0.02)	0.02 (−0.18, 0.21)
Total protein foods#	−0.01 (−0.09, 0.07)	0.01 (−0.08, 0.10)	−0.02 (−0.11, 0.07)
Seafood and plant proteins††	0.06 (−0.08, 0.19)	−0.02 (−0.18, 0.14)	−0.06 (−0.20, 0.09)
Fatty acids‡‡	−0.14 (−0.36, 0.08)	−0.10 (−0.35, 0.15)	0.05 (−0.19, 0.28)
Refined grains	0.24 (0.02, 0.46)*	−0.09 (−0.34, 0.16)	−0.12 (−0.35, 0.11)
Sodium	0.07 (−0.14, 0.28)	−0.02 (−0.26, 0.22)	−0.12 (−0.34, 0.10)
Empty calories§§	0.23 (−0.05, 0.51)	−0.39 (−0.71, −0.07)*	0.10 (−0.19, 0.40)

F&V, Fruit and Vegetable

Sample size for these analyses was 720 instead of 740 due to missing data on peer support for fruit and vegetable intake, away-from-home food consumption, and the school food environment

All coefficients are from mixed models adjusted for youth’s age, youth’s sex, youth’s Hispanic/Latino background, youth’s nativity, parent’s education, household income, peer support for F&V intake, away-from-home food consumption, and school vending machine use as fixed effects and family as a random intercept

^*p<0.05

^**p<0.01

^†Includes fruit juice

^‡Includes all forms except juice

^§Includes any beans and peas not counted as total protein foods

^¶Includes all milk products (e.g., fluid milk, cheese) and fortified soy beverages

^#Beans and peas are included here when the total protein foods standard is otherwise not met

^††Includes seafood, nuts, seeds, and soy products (other than beverages), as well as beans and peas counted as total protein foods

^‡‡Ratio of polyunsaturated fatty acids (PUFA) and monounsaturated fatty acids (MUFA) to saturated fatty acids (SFA)

^§§Calories from solid fats, alcohol, and added sugars; threshold for counting alcohol is >28 grams/day

4. DISCUSSION

In this sample of youth from diverse Hispanic/Latino backgrounds across four US communities, overall diet quality scores among siblings were slightly correlated, with the strongest correlations observed for sodium and empty calorie intake. Correlations in diet quality were stronger among siblings closer in age, particularly for total protein intake. Additionally, though correlations in overall diet quality did not differ according to siblings’ sex concordance, correlations in greens and beans and whole grain intake were stronger for siblings who were all brothers compared to a mix of brothers and sisters. Differences in peer support for fruit and vegetable intake and away-from-home food consumption were associated with differences in siblings’ diet quality.

We expected that diet quality of siblings in the age range under study (8-16 years) would resemble one another, given that siblings raised in the same household share the same home and neighborhood food environment, plus various other unmeasured genetic and familial factors.^46,49,50 Further, we expected correlations in this population given Hispanic/Latino culture’s emphasis on close, supportive family relationships.^7-9 Existing studies of correlations in siblings’ diets in both Hispanic/Latino and non-Hispanic/Latino populations have examined aspects of diet including breakfast frequency,¹⁷ fast food consumption,^17,19 total energy intake,^{20,21,44,51,52}, macronutrient intake,^21,44,52 and specific healthy and unhealthy food intake (e.g., fruit and vegetable intake)^17,20. However, this is the first study, to our knowledge, to examine correlations in siblings’ diet quality in any ethnicity.

We found evidence of a slight correlation among siblings’ diets, with the strongest correlations for unhealthy dietary components—empty calories (calories from solid fats, alcohol [>99.7% of this population consumed <0.1 mean servings/day], and added sugars) and sodium. This is largely consistent with findings from the two previous studies of siblings’ diets conducted in Hispanic/Latino youth.^21,22 Specifically, the study conducted by Patterson et al.²¹ found that both non-Hispanic White siblings and Mexican American siblings (4^th and 5^th graders) had similar total energy intakes (r=0.38), with Mexican American siblings uniquely having similar total fat (r=0.42), saturated fat (r=0.26), and sodium (r=0.30) intake. The second study of Hispanic/Latino siblings’ diets by Mitchell et al.²² also examined macronutrient intake among Mexican American siblings (aged >16 years) and found the strongest correlations were observed for saturated fat and protein intake (both r=0.10). Together with the present study, these findings indicate slight to fair correlations in Hispanic/Latino siblings’ diets, with the magnitudes of the associations being similar across studies.

While correlations in the present study were strongest for the unhealthy dietary components, correlations were also significant, albeit smaller, for healthy components such as fruit and vegetable intake. Previous studies have compared the strength of correlations for siblings’ healthy vs. unhealthy dietary components with mixed findings. One previous study of 58 racially/ethnically diverse (19% Hispanic; mean age=14.5 years) sibling pairs in the US reported positive correlations for fast food consumption but not for fruit and vegetable consumption.¹⁷ Conversely, studies of ~750 sibling pairs across Europe (aged 2-19 years) reported positive correlations in siblings’ intake of both fast food and fruit and vegetables, along with other healthy and unhealthy food groups.^19,20 Given these inconsistent findings, there is a need for future work to investigate whether similarities in siblings’ diets are truly stronger for unhealthy vs. healthy dietary components.^5,6,13,14

In the present study, we also found that correlations in overall diet quality were stronger for siblings close in age, consistent with previous research.^19,20 Notably, differences in siblings’ correlations in total protein intake (0.32 vs. 0.02 for siblings <3 vs. ≥3 years apart) seemed to be driving this difference in the present study, with exploratory analyses of age groups indicating that siblings who were both close in age and younger seeming to underpin the observed difference. We also observed that siblings who were all brothers compared to a mix of brothers and sisters had stronger correlations in greens and beans and whole grains, which is somewhat inconsistent with previous research.^19,20 Specifically, previous studies have reported no differences in fast food consumption for siblings of the same vs. different sex¹⁹ or for healthy and unhealthy dietary components (e.g., fruits and vegetables, sugar and sweets) across the three sib-types.²⁰ It is possible that the previous studies did not observe differences in correlations according to sex concordance as we did due to their focus on different dietary measures and due to combining sister-sister and brother-brother dyads into a single category in one study.¹⁹ However, given the novelty of these findings for age difference and sex concordance, future research should aim to replicate these observations.

Finally, to examine potential explanations for differences in siblings’ diet quality, we looked at whether differences in siblings’ social and built environments could explain differences in siblings’ diet quality. Existing research on non-shared determinants of childhood obesity in siblings has largely focused on differential parental treatment (e.g., food parenting practices), but there is comparatively little research examining how the siblings’ individual social and environmental determinants may explain discrepancies in weight-related behaviors.²⁵

In line with our hypothesis, we found that greater peer support for fruit and vegetable intake was associated with greater overall diet quality amongst siblings, with higher diet quality scores primarily being driven by greater intake of the foods that were encouraged (i.e., total and whole fruit), along with greater intake of whole grains and dairy and lower intake of refined grains (note: higher refined grain scores indicated lower consumption of this item because it is a moderation component). Peers have been shown to play an important role in youth’s dietary intake, with peer similarities in fast food consumption surpassing that of siblings’ by the time they are 9-10 years of age.¹⁹ Further, studies of adolescents have shown that peers are primarily a negative influence on dietary intake, encouraging unhealthy food consumption^27,28,53,54. However, evidence of a significant association between peer support for healthful dietary intake and adolescents’ consumption of healthful items is limited.^13,55 This study contributes to existing literature by indicating that greater peer support for fruit and vegetable intake is associated with more healthful overall diet quality in siblings.

Our study also provides evidence that greater away-from-home food consumption is associated with poorer overall diet quality in siblings. Specifically, we found that siblings reporting greater away-from-home food consumption had lower overall diet quality scores, driven by lower intake of whole grains and greater intake of empty calories (note: lower empty calorie scores indicated higher consumption of this item because it is a moderation component). Our findings support previous research documenting that away-from-home food consumption is linked with poor overall diet quality^56,57—likely due to unhealthy away-from-home vendors selling highly processed items containing elevated amounts of fat, sodium, and added sugar.⁵⁷ Additionally, our study indicated that use of school vending machines was not related to diet quality in siblings, in contrast with our hypothesis. It is possible this is due to school and vending machines combined providing less than 10% of total calories for youth aged 7-18 years old and thus minimally affecting diet quality scores.⁵⁶

4.1. Strengths and Limitations

SOL Youth provided a unique opportunity to examine a large sample of diverse Hispanic/Latino siblings living across the US. Data were collected by trained interviewers and examiners using established quality control and assurance procedures, ensuring high quality data. Use of multiple 24-hour dietary recalls allowed us to capture youth’s usual intake and calculate diet quality scores. Further, the wide range of questionnaires administered in SOL Youth allowed us to examine novel social and environmental determinants of siblings’ diet quality.

Despite the study’s strengths, there are some limitations worth noting. Siblings in this study lived in the same household, but we do not know if they were biologically related. As such, we could not parse out the role genetics played in correlations in siblings’ diet quality or identify if twins were included in this analysis. We also did not exclude individuals with severe food allergies or adjust for birth order (e.g., first-born, second-born) due to not collecting this data in SOL Youth. Birth order could not be determined in SOL Youth using youth’s ages alone due to all youth in the household not necessarily being enrolled in SOL Youth as a result of interest/feasibility or the age-based exclusion criterion. We aimed to identify non-shared social and environmental determinants of siblings’ diets, but it is possible that siblings in each household were attending the same school, for example, and thus had similar environments. Regarding the diet measure, HEI-2010 was developed to reflect adherence to US dietary guidelines rather than to assess individual dietary components as reported here.³² Additionally, while diet quality reflects the diversity of dietary intake, it does not fully capture quantity. Further, there is the possibility of misreporting dietary intake.⁵⁸

4.2. Conclusions

In this study of Hispanic/Latino US youth, overall diet quality scores were slightly correlated amongst siblings, with stronger correlations among siblings closer in age. Differences in peer support for fruit and vegetable intake and frequency of away-from-home food consumption may explain differences in siblings’ diet quality and may need to be incorporated in future family-based obesity interventions. Overall, the significant correlations in Hispanic/Latino siblings’ dietary intake suggest that including siblings in family obesity interventions targeting dietary change may be beneficial. Although the mechanisms through which siblings influence each other’s’ dietary intake warrants further investigation to inform the design of such interventions, the significant correlations in siblings diet quality suggest that siblings may have a reciprocal influence on each other’s dietary intake. The findings also suggest that clinicians should consider the role of siblings, including age difference and sex concordance among siblings in a family, when counseling families on healthy diets. Future studies should consider examining additional non-shared environmental determinants of differences in siblings’ diets and should aim to account for measures such as birth order and dietary behaviors to better understand correlations in siblings’ diet.

Abbreviations:

HCHS/SOL

Hispanic Community Health Study/Study of Latinos

SOL Youth

Hispanic Community Children’s Health Study/Study of Latino Youth

HEI-2010

Healthy Eating Index 2010

ICC

Intraclass Correlation Coefficient

CI

Confidence Intervals

NDSR

Nutrition Data System for Research