Gender expression and sexual orientation differences in diet quality and eating habits from adolescence to young adulthood

Abstract

Background:

Diet and eating habits during youth have implications on diet and eating habits during adulthood; however, little longitudinal research has examined sexual orientation and gender expression differences in diet.

Objective:

To examine sexual orientation and gender expression differences in diet quality and eating habits from adolescence to young adulthood.

Design:

Data across multiple time-points from the longitudinal Growing Up Today Study cohorts (1997–2011) were used.

Participants/setting:

Participants (n=12,880; ages 10–23 years) are the children of women from the Nurses’ Health Study II cohort.

Main outcome measures:

Diet quality scores were assessed using the Alternative Healthy Eating Index-2010. Additionally, breakfast consumption (≥5 days/week) and family dinners (≥5 days/week) were assessed.

Statistical analyses performed:

Multivariable generalized estimating equations regression models were fit to estimate sexual orientation and gender expression differences in diet quality scores, breakfast consumption, and family dinners, stratified by sex assigned at birth over available repeated measures.

Results:

“Gender nonconforming” males had significantly higher diet quality scores than “very gender conforming” males (p<0.05). Diet quality scores did not differ by gender expression among females. “Mostly heterosexual” females and gay males had higher diet quality scores than their same-sex completely heterosexual counterparts (p<0.05). Adjustment for mother’s diet quality scores attenuated effects, except for gay males (p<0.05). “Gender nonconforming” females were less likely to consume breakfast than “very gender conforming” (p<0.05). Similar results were found for “mostly heterosexual” and bisexual compared to completely heterosexual females. There were no gender expression or sexual orientation differences in family dinners among males and females.

Conclusions:

Sexual orientation and gender expression have independent effects on diet quality scores and eating habits for both males and females. Very gender conforming and completely heterosexual males had the lowest diet quality scores compared to other gender expression and sexual orientation groups. Additional research to explore the effects of sexual orientation and gender expression on diet-related health is needed to build upon these findings.

INTRODUCTION

Poor diet quality and eating habits during adolescence and young adulthood have implications on the development of chronic conditions, including type 2 diabetes and cardiovascular disease.¹ Most notably, some less favorable eating habits developed in childhood and adolescence may persist into adulthood,² thus, increasing risk for the development of chronic diseases throughout the life course. Poor eating habits during childhood and adolescence, such as breakfast skipping^3,4 and fewer family meals,^5,6 has been associated with lower diet quality. One longitudinal study found that greater frequency of family meals during adolescence was associated with better diet quality and healthier eating habits during young adulthood.⁶ Further, breakfast consumption and frequent family meals have numerous social benefits including associations with better academic achievement,^7–9 making these behaviors particularly salient eating habits to examine in young populations.

Gender expression refers to the ways an individual is socially and culturally perceived as masculine or feminine.¹⁰ While a large body of research has examined gender or sex (i.e., male/man/boy and female/woman/girl) differences in body image and other weight-related health indicators, few studies have examined differences based on gender expression. Studies using longitudinal cohort data from the Growing Up Today Study (GUTS) have explored associations of gender expression with physical activity and body mass index (BMI) based on gender expression.^11,12 Findings indicated that the most gender conforming males had significantly higher physical activity, while the most gender conforming females had significantly lower physical activity compared to gender nonconforming same-sex peers.¹² More recently, researchers found that compared to very gender conforming females, mostly conforming and gender nonconforming females had higher BMIs, while the opposite trend was found for males.¹¹ These findings are consistent with Gender Intensification Theory,¹³ which posits that social pressure to conform to traditional gender norms may intensify adolescents’ identification with gender stereotyped interests and behaviors, thus resulting in some of the observed gender differences in health. To our knowledge, studies examining gender expression differences in diet quality and eating behaviors do not exist.

Research on diet quality and eating habits among sexual minority (e.g., gay, lesbian, bisexual, those with same-sex attractions) young people is particularly lacking, despite evidence suggesting they may be at greater risk for a variety of diet-related risk factors for chronic diseases. Population-based data from the Youth Risk Behavior Surveillance System (YRBSS) suggest that among both male and female students, sexual minority youth may be more likely to be overweight or obese and engage in disordered eating behaviors.^14–16 Findings for YRBSS dietary behaviors, such as fruit, vegetable, milk, and soda consumption were generally mixed.^14,15,17 Recent YRBSS data suggest that sexual minority youth may be less likely to eat breakfast, however, findings were not stratified by sex.¹⁷ Other studies examining sexual orientation differences among youth and young adults have generally found higher BMI or more overweight or obesity among most sexual minority groups (except gay males)^18–20 and poorer eating habits, including less breakfast consumption and more eating away from home as well as more disordered eating behaviors such as purging and binge eating, among sexual minority young people.^19,21–23 The minority stress model posits that stressful social conditions, including explicit or covert discrimination, stigmatization, or prejudice based on one’s sexual orientation, may contribute to disparities in mental and physical health.^24,25 These minority stress experiences can influence diet quality and eating habits, where coping strategies, such as stress-related eating or disordered eating behaviors may be used to help regulate reactions to negative experiences. Although little research has examined diet quality and eating habits in relation to minority stress experiences, a recent study of adult lesbian women found that minority stress was associated with binge eating.²⁶ Thus, sexual minority stress may contribute to disparities in diet and eating-related outcomes.

Regardless of sexual orientation, every person has a gender expression. However, evidence suggests that sexual minority people are more likely than their heterosexual counterparts to exhibit gender nonconformity.^27–29 Few studies have examined the simultaneous effect of gender expression and sexual orientation on nutritional or weight-related behaviors. In the same GUTS study described earlier, researchers also explored the role of gender expression in the relationship between sexual orientation and physical activity.¹² Findings indicated that among males only, gender nonconforming expression accounted for lower physical activity among gay and bisexual males compared to completely heterosexual males. However, gender expression did not explain sexual orientation differences in physical activity among females.

The body of research on sexual orientation, gender expression, and weight-related health suggests a complicated relationship among these factors, particularly based on sex. Building on previous research on differences in BMI and physical activity by sexual orientation, gender expression, and assigned sex, this study examined the sexual orientation and gender expression differences in diet quality and eating habits among young people 10–23 years old. In line with Gender Intensification Theory, gender conformity among males and gender nonconformity among females is likely associated with poorer diet quality and eating habits. Extending Gender Intensification Theory to incorporate sexual orientation, gender nonconformity is likely associated with poor diet quality and eating habits among sexual minority males, while it would have the opposite effect for sexual minority females.

MATERIALS AND METHODS

Study Population.

Participants were from GUTS, a longitudinal cohort study of the children of women participating in the Nurses’ Health Study II. With consent from their mothers, the children were mailed questionnaires and those who returned completed questionnaires were considered to be assenting to the study and were enrolled in the GUTS cohort. The original GUTS cohort began in 1996 with participants 9–15 years old (N=16,882). GUTS was expanded in 2004 to include a new group of children of Nurses’ Health Study II participants (N=10,923). For the cohort which began in 1996, there have been twelve follow-up questionnaires; for the cohort which began in 2004, there have been six follow-up questionnaires. The study is ongoing and additional information is available online.³⁰ Data collection for this study was approved by the Brigham and Women’s Hospital Institutional Review Board.

Gender expression.

Gender expression was assessed in 2010–2011 with the following validated items:³¹ “A person’s appearance, style, or dress may affect the way people think of them. On average, how do you think people would describe your appearance, style, or dress?” A similar question was used to assess mannerisms, such as the way they walk or talk. Response options ranged from “Very masculine” to “Very feminine.” Consistent with previous GUTS research,^11,32 a gender conformity score was created by taking the mean response (range 1–7) of both items. Then, a 3-category variable was created and used in analyses. A mean score of <1.5 was categorized as “very gender conforming,” a score of 1.5–3 was categorized as “mostly gender conforming,” and a score >3 was categorized as “gender nonconforming.”

Sexual orientation.

Sexual orientation was assessed beginning in 1999 with the following item, “Which of the following best describes your feelings?” Response options included, “completely heterosexual (attracted to persons of the opposite sex),” “mostly heterosexual,” “bisexual (equally attracted to men and women),” “mostly homosexual,” “completely homosexual (gay/lesbian, attracted to persons of the same sex),” and “not sure.” Due to small sample sizes, “mostly homosexual” and “completely homosexual” were combined into a “gay or lesbian” category. Respondents were treated as missing in each wave they reported “not sure” due to the small number of respondents reporting “not sure.” Sexual orientation was back-assigned from the next available wave for years when sexual orientation and diet measures were not simultaneously assessed. As a result, data from only 1997–2011 were included in these analyses (age range 10–23 years).

Alternative Healthy Eating Index.

A robust food frequency questionnaire (FFQ)^33,34 assessed dietary intake of participants during seven times from baseline to 2011 (1997, 1998, 2001, 2004, 2006, 2008, 2011). The Alternative Healthy Eating Index-2010 (AHEI-2010) evaluates diet quality based on intake of 11 items: vegetables, fruit, whole grains, nuts and legumes, long-chain (n-3) fats, polyunsaturated fatty acids, sugar-sweetened beverages and fruit juice, red and processed meat, trans fat, sodium, and alcohol.³⁵ Among adults, lower AHEI-2010 scores are associated with excess weight gain and chronic conditions such as type 2 diabetes and cardiovascular disease.³⁵ The AHEI-2010 was adapted to an adolescent population by removing alcohol as an indicator, which is not included as a recommendation for a healthy diet for this age group, and reducing the cut-point for whole grains consumption for males under the age of 13 years to 75g, based on national data on estimated energy.³⁶ The resulting 10-item index was scored from 0 (unhealthy) to 10 (healthiest), and the total score ranged from 0 to 100.

Breakfast consumption and family dinners.

Consuming breakfast and having family dinners during youth is associated with better diet quality^3–6 and healthier eating habits⁶ during young adulthood, as well as better academic achievement.^7–9 Measures of breakfast consumption and family dinners were included in the same years as AHEI assessment, except for 2001, when breakfast consumption and family dinners were not measured. Participants were asked, “How many times each week (including weekdays and weekends) do you eat breakfast?” and “How often do you sit down with other members of your family to eat dinner or supper?” Response options for both items included, “Never or almost never,” “1–2 times per week,” “3–4 times per week,” and “5 or more times per week.” Consistent with previous research examining breakfast consumption among young people³⁷ and to address small sample sizes for family dinners, both breakfast consumption and family dinners were dichotomized as “≥5 times/week” and “<5 times/week.”

Covariates.

Covariates were selected based on previous studies that found significant sexual orientation differences in age, race/ethnicity, and region of residence.³⁸ Therefore, covariates included age group (10–12, 13–14, 15–16, 17–18, 19–23 years), race/ethnicity [non-white (includes black, Hispanic, Asian or Pacific Islander, American Indian or Alaskan Native, and Other) vs. non-Hispanic white], and region of residence (west, midwest, south, and northeast). Due to small sample sizes of participants of color (94% are non-Hispanic white), these racial and ethnic categories were collapsed into a single category. Further, analyses were stratified based on sex assigned at birth (male/female), as reported by their mothers during study recruitment. Transgender participants (n=3) were not included in these analyses because gender expression questionnaire items in 2010/2011 was missing for these participants. For models examining AHEI-2010 scores, energy intake (kilocalories/day) and alcohol intake (drinks/day) were estimated from the FFQ and included as covariates. Further, diet quality information from mothers participating in the Nurses’ Health Study II were calculated, as described previously,³⁵ and included as a covariate for models examining AHEI-2010 scores.

Statistical Analyses.

Participants who were missing sexual orientation data for all waves of GUTS and those who were missing gender expression information were excluded; this yielded an analytic sample of 12,880 participants (n_observations=49686) with exposure information. The overall distribution of observations by age group were as follows: n_observations 10–12 years old=8,293; n_observations 13–14 years old=9,100; n_observations 15–16 years old=10,510; n_observations 17–18 years old=7,646; n_observations 19–23 years old=14,137. Further, for family dinners, the sample was restricted to participants 18 years and younger to account for the fact that many participants may not be living at home with their parents. Individual models contained varying number of observations over the repeated measures due to missing diet data (n_observations AHEI-2010=30,959; n_observations breakfast consumption=25,907; n_observations family dinners=21,152). Sex-stratified generalized estimating equations (GEE) using linear regression were fit to estimate average AHEI-2010 scores over available repeated measures based on respondent age. To examine sexual orientation and gender expression differences in breakfast consumption and family dinners, sex-stratified GEE models using modified Poisson regression were fit to estimate the odds of eating breakfast and having family dinners ≥5 times/week over available repeated measures. For AHEI-2010 scores and eating habits, β and odds ratios (OR) are presented, respectively, which is consistent with the types of models that were fit for each outcome type.

For all outcomes, a series of three models were fit. The first model included age group only as a covariate and fit gender expression and sexual orientation in separate models. The second model also fit gender expression and sexual orientation in separate models and both included age group, race/ethnicity, and region as covariates. Further, for AHEI-2010 scores the second model also included energy intake and alcohol consumption as covariates. The interaction between sexual orientation and age as well as gender expression and age were also included in the second model to examine potential effect modification based on age. The third model that was fit included gender expression and sexual orientation in the same model and adjusted for all covariates that were included in the second model. Very gender conforming and heterosexual youth were the reference groups for gender expression and sexual orientation, respectively. For AHEI-2010 scores, an additional two models were fit. One model included mom’s AHEI-2010 score as a covariate and the other model examined the associations between gender expression and AHEI-2010, including mom’s AHEI-2010 score, but was restricted to participants 18 years and younger only. P-values <0.05 were considered statistically significant for all analyses. The data analyses were generated using SAS software.³⁹

RESULTS

Tables 1 and 2 contain sex-stratified age-standardized distributions of study participant characteristics by gender expression and sexual orientation, respectively. The majority of participants were mostly gender conforming, completely heterosexual, and non-Hispanic white. Energy intake was highest among very gender conforming participants and lowest among gender nonconforming participants. Overall, participants reported low levels of alcohol consumption (0.1–0.2 drinks/day). In addition, the majority of participants reported eating breakfast ≥5 days/week, while less than half reported having family dinners ≥5 days/week.

Table 1.

Age-standardized distribution^a of study variables across all available measures from ages 10–23 years by gender expression^b and sex, GUTS 1 & 2, 1997–2011

	MALES
	Very Gender Conforming	Mostly Gender Conforming	Gender Nonconforming
Sample size, n	1,796	2,607	172
Observations, n	7,086	10,244	704
Sexual orientationc
Completely heterosexual	98.4%	90.3%	64.1%
Mostly heterosexual	1.4%	6.8%	14.2%
Bisexual	0.1%	0.7%	6.0%
Gay or Lesbian	0.1%	2.2%	15.7%
Region
Northeast	30.9%	29.0%	29.8%
Midwest	33.2%	33.2%	34.6%
South	14.9%	13.5%	10.6%
West	14.2%	18.2%	18.5%
Race/Ethnicity (non-Hispanic White)	94.3%	93.6%	92.1%
Energy intake, [kcal/day; mean(SD)]	2374.6 (823.6)	2258.2 (793.4)	2087.5 (748.7)
Alcohol [drinks/day; mean(SD)]	0.2 (0.6)	0.2 (0.5)	0.2 (0.5)
Mother’s AHEI-2010d Score [mean(SD)]	53.5 (12.6)	54.1 (13.0)	55.6 (13.9)
Breakfast consumption (≥5 days/week)	72.9%	71.5%	68.2%
Family Dinner (≥5 days/week)e	49.0%	46.3%	45.7%
	FEMALES
	Very Gender Conforming	Mostly Gender Conforming	Gender Nonconforming
Sample size, n	2,282	5353	670
Observations, n	8,559	20,428	2,665
Sexual orientationc
Completely heterosexual	95.3%	88.5%	76.8%
Mostly heterosexual	4.1%	9.6%	15.2%
Bisexual	0.5%	1.6%	5.0%
Gay or Lesbian	0.2%	0.4%	3.1%
Region
Northeast	32.9%	31.4%	31.6%
Midwest	32.4%	34.9%	33.2%
South	15.8%	13.0%	13.4%
West	14.0%	15.5%	16.1%
Race/Ethnicity (non-Hispanic White)	94.3%	93.4%	91.5%
Energy intake, [kcal/day; mean(SD)]	1979.5 (706.6)	1953.4 (702.2)	1928.2 (717.7)
Alcohol [drinks/day; mean(SD)]	0.1 (0.5)	0.1 (0.4)	0.1 (0.5)
Mother’s AHEI-2010d Score [mean(SD)]	53.0 (12.6)	53.0 (12.4)	53.5 (12.8)
Breakfast consumption (≥5 days/week)	70.5%	70.4%	64.5%
Family Dinner (≥5 days/week)e	42.2%	41.2%	40.9%

^aData are mean(SD) or percentage unless otherwise indicated

^bassessed in the 2010/2011 wave of data collection

^ccross-tabulation is based on the distribution of gender expression in the denominator. Therefore, values represent the percent of individuals within each gender expression category.

^dAHEI-2010: Alternative Healthy Eating Index-2010

^eincludes only participants up until 18 years of age

Table 2.

Age-standardized distribution of study variables^a across all available measures from ages 10–23 years by sexual orientation^b and sex, GUTS 1 & 2, 1997–2011

	MALES
	Completely Heterosexual	Mostly Heterosexual	Bisexual	Gay
Observations, n	12,589	672	88	253
Gender expressionc
Very gender conforming	42.5%	12.0%	3.9%	2.6%
Mostly gender conforming	55.0%	77.0%	63.3%	65.4%
Gender nonconforming	2.5%	11.0%	32.8%	32.0%
Region
Northeast	31.4%	32.8%	58.3%	30.0%
Midwest	36.3%	38.9%	13.2%	38.1%
South	14.7%	10.6%	11.0%	15.8%
West	17.4%	16.5%	17.5%	16.1%
Race/Ethnicity (non-Hispanic White)	94.2%	93.8%	98.2%	91.0%
Energy intake, [kcal/day; mean(SD)]	2316.1 (811.6)	2143.8 (829.1)	2179.2 (735.6)	2072.1 (899.1)
Alcohol [drinks/day; mean(SD)]	0.2 (0.6)	0.2 (0.8)	0.2 (0.4)	0.2 (0.6)
Mother’s AHEI-2010d Score [mean(SD)]	54.2 (12.9)	56.1 (14.0)	55.8 (11.3)	56.2 (13.3)
Breakfast consumption (≥5 days/week)	71.4%	71.7%	71.4%	62.2%
Family Dinner (≥5 days/week)e	46.7%	43.0%	50.6%	47.8%
	FEMALES
	Completely Heterosexual	Mostly Heterosexual	Bisexual	Lesbian
Observations, n	22,613	2,161	390	143
Gender expressionc
Very gender conforming	29.1%	13.4%	8.5%	7.7%
Mostly gender conforming	64.1%	72.6%	66.0%	48.6%
Gender nonconforming	6.8%	14.0%	25.5%	43.8%
Region
Northeast	32.9%	35.9%	40.1%	34.3%
Midwest	36.8%	31.6%	26.5%	28.2%
South	14.7%	11.5%	16.2%	11.2%
West	15.5%	20.6%	17.0%	26.4%
Race/Ethnicity (non-Hispanic White)	93.8%	91.6%	93.4%	81.9%
Energy intake, [kcal/day; mean(SD)]	1970.7 (707.0)	1906.2 (718.8)	1817.0 (733.8)	1984.8 (802.6)
Alcohol [drinks/day; mean(SD)]	0.2 (0.5)	0.2 (0.6)	0.2 (0.5)	0.2 (0.7)
Mother’s AHEI-2010d Score [mean(SD)]	53.2 (12.5)	55.2 (13.4)	52.6 (12.9)	58.1 (12.8)
Breakfast consumption (≥5 days/week)	71.1%	64.5%	52.8%	63.6%
Family Dinner (≥5 days/week)e	41.8%	36.7%	39.6%	46.3%

^aData are mean(SD) or percentage unless otherwise indicated

^bassessed every wave

^ccross-tabulation is based on the distribution of sexual orientation over multiple waves in the denominator. Therefore, values represent the percent of individuals within each sexual orientation category.

^dAHEI-2010: Alternative Healthy Eating Index-2010

^eincludes only participants up until 18 years of age

Table 3 contains sex-stratified age-standardized distributions of AHEI-2010 scores and scoring components by gender expression and sexual orientation. Overall, diet quality scores were generally higher among female participants than male participants. Across gender expression, gender nonconforming participants had the highest AHEI-2010 scores. Across sexual orientation, gay or lesbian participants had the highest AHEI-2010 scores. Completely heterosexual and bisexual males had similarly low AHEI-2010 scores among male participants, while completely heterosexual females had the lowest AHEI-2010 scores among female participants.

Table 3.

Age-standardized distribution^a of AHEI-2010^b score components across all available measures from ages 10–23 years by gender expression^c, sexual orientation, and sex, GUTS 1 & 2, 1997–2011

	Very Gender Conforming	Mostly Gender Conforming	Gender Nonconforming	Completely Heterosexual	Mostly Heterosexual	Bisexual	Gay or Lesbian
	MALES
AHE1–2010b Total Score	37.6 (10.2)	38.3 (10.3)	40.2 (10.9)	38.4 (10.4)	40.0 (10.6)	38.3 (10.9)	41.7 (11.9)
Red/processed meat (servings/d)	5.0 (2.8)	5.4 (2.8)	6.0 (2.9)	5.2 (2.8)	6.0 (2.8)	5.8 (3.0)	6.6 (2.9)
Nuts and legumes (servings/d)	2.6 (2.4)	2.6 (2.5)	2.6 (2.7)	2.6 (2.5)	2.4 (2.7)	2.3 (2.4)	2.7 (3.0)
Sugar-sweetened beverages and fruit juice (servings/d)	1.6 (2.8)	1.8 (3.0)	2.3 (3.4)	1.8 (2.9)	2.0 (3.1)	2.0 (3.0)	2.1 (3.2)
Vegetables (servings/d)	2.7 (2.0)	2.7 (2.1)	2.7 (2.2)	2.7 (2.1)	2.9 (2.2)	2.8 (2.3)	3.3 (2.9)
Fruit (servings/d)	2.4 (2.0)	2.3 (1.9)	2.3 (2.0)	2.4 (2.0)	2.3 (1.9)	1.9 (1.9)	2.6 (2.3)
Polyunsaturated fat (% energy)	5.0 (1.7)	4.9 (1.7)	4.9 (1.9)	5.0 (1.7)	4.9 (1.8)	4.5 (1.9)	5.0 (1.8)
Trans fat (% energy)	2.5 (2.5)	2.5 (2.5)	2.3 (2.4)	2.5 (2.5)	2.4 (2.6)	2.1 (2.5)	2.2 (2.5)
Long-chain (n-3) fats (mg/d)	28.7 (106.4)	29.0 (147.4)	26.7 (101.7)	32.1 (142.6)	32.5 (109.6)	22.9 (66.6)	28.5 (102.1)
Whole grains (g/d)	2.5 (1.8)	2.6 (1.9)	2.7 (2.1)	2.6 (1.9)	2.8 (2.2)	2.3 (2.0)	2.5 (2.0)
Sodium (mg/d)	4.7 (3.1)	5.1 (3.1)	5.8 (3.2)	4.9 (3.1)	5.6 (3.3)	5.6 (3.3)	6.1 (3.4)
	FEMALES
AHE1–2010b Total Score	40.5 (11.2)	40.5 (10.9)	40.7 (11.1)	40.8 (11.2)	42.1 (11.4)	41.6 (11.4)	43.6 (11.2)
Red/processed meat (servings/d)	6.8 (2.6)	6.7 (2.6)	6.8 (2.6)	6.7 (2.5)	7.2 (2.6)	7.2 (2.7)	7.6 (2.4)
Nuts and legumes (servings/d)	2.1 (2.3)	2.2 (2.3)	2.3 (2.5)	2.3 (2.4)	2.7 (2.9)	2.7 (3.0)	3.0 (2.8)
Sugar-sweetened beverages and fruit juice (servings/d)	2.4 (3.4)	2.5 (3.4)	2.6 (3.4)	2.7 (3.5)	2.8 (3.6)	3.2 (3.7)	2.9 (3.8)
Vegetables (servings/d)	3.1 (2.3)	3.1 (2.2)	3.1 (2.3)	3.1 (2.2)	3.2 (2.4)	3.0 (2.3)	3.6 (2.3)
Fruit (servings/d)	2.6 (2.1)	2.5 (2.0)	2.4 (2.0)	2.6 (2.1)	2.5 (2.1)	2.2 (2.0)	2.7 (2.2)
Polyunsaturated fat (% energy)	5.1 (1.7)	5.0 (1.7)	4.9 (1.8)	5.0 (1.7)	4.9 (1.8)	4.9 (2.0)	5.0 (1.8)
Trans fat (% energy)	2.1 (2.2)	2.0 (2.2)	2.0 (2.2)	2.1 (2.2)	2.2 (2.4)	1.8 (2.3)	2.2 (2.6)
Long-chain (n-3) fats (mg/d)	22.1 (84.3)	22.5 (86.9)	22.3 (82.4)	23.8 (89.8)	26.1 (93.6)	24.0 (87.1)	29.9 (105.7)
Whole grains (g/d)	2.6 (1.9)	2.6 (1.9)	2.6 (1.9)	2.6 (1.9)	2.7 (2.0)	2.5 (1.9)	2.9 (2.4)
Sodium (mg/d)	4.9 (3.1)	5.0 (3.2)	5.1 (3.3)	4.9 (3.1)	5.2 (3.3)	5.7 (3.4)	4.7 (3.6)

^aData are mean(SD)

^bAHEI-2010: Alternative Healthy Eating Index-2010

^cassessed in the 2010/2011 wave of data collection

Table 4 displays results of linear regression models using GEE for AHEI-2010 scores. In models 1–3, very gender conforming males had the lowest diet quality scores; gender nonconforming males had statistically significantly higher diet quality scores compared to very gender conforming males. While mostly gender conforming males had statistically significantly higher diet quality scores in Models 1 and 2, when sexual orientation was included in the model (Model 3), the association between mostly gender conforming gender expression and AHEI-2010 scores were slightly attenuated and no longer statistically significant. Inclusion of mother’s diet quality scores in Model 4 did not substantially alter results, although estimates were attenuated slightly (β=1.6 in Model 3; β=1.1 in Model 4) such that there was no longer a statistically significant difference between gender nonconforming and very gender conforming males. Age-restriction to those 18 years and younger (Model 5) did not substantially change results, although mostly gender conforming males had statistically significantly higher diet quality scores than very gender conforming males. There were no statistically significant differences in diet quality score by gender expression among female participants.

Table 4.

Results from generalized estimating equations linear regression examining associations with AHEI-2010^a score from ages 10–23 years, stratified by sex, GUTS 1 & 2, 1997–2011

		β (95% CI)
		MALES
	N obs.	Model 1b	Model 2c	Model 3d	Model 4e	Model 5f
Gender Expression
Intercept		33.1 (32.7,33.5)	41.2 (40.0,42.3)	41.3 (40.0,42.5)	30.2 (28.7,31.7)	31.4 (29.9,32.9)
Very gender conforming	4,217	ref	ref	ref	ref	ref
Mostly gender conforming	5,967	0.9 (0.4,1.4)***	0.6 (0.1,1. 0)**	0.4 (−0.0,0.9)	0.4 (−0.1,0.8)	0.5 (0.0,0.9)*
Gender nonconforming	382	2.8 (1.6,4.0)***	2.0 (0.9,3.1)***	1.6 (0.2,3.0)*	1.1 (−0.2,2.4)	1.3 (−0.0,2.6)
Sexual Orientation
Intercept		33.8 (33.4,34.1)	41.6 (40.3,42.9)	g	g	g
Completely Heterosexual	9,850	ref	ref	ref	ref	ref
Mostly Heterosexual	471	1.7 (0.4, 3.0)**	1.1 (0.1,2.4)*	1.0 (−0.1,21)	0.8 (−0.2,1.9)	0.8 (−0.3,19)
Bisexual	64	0.9 (−2.3,4.1)	0.0 (−2.8,2.9)	−0.3 (−3.3,2.6)	−0.3 (−3.3,2.7)	−2.3 (−5.0,0.3)
Gay	181	3.9 (1.7,6.0)***	2.6 (0.7,4.6)**	2.2 (0.2,4.2)*	2.3 (0.3,4.2)*	1.0 (−1.2,3.2)
		FEMALES
	N obs.	Model 1b	Model 2c	Model 3d	Model 4e	Model 5f
Gender Expression
Intercept		34.4 (34.0,34.8)	41.9 (41.1,42.8)	41.8 (40.9,42.7)	29.0 (28.0,30.0)	30.1 (29.1,31.2)
Very gender conforming	5,583	ref	ref	ref	ref	ref
Mostly gender conforming	13,232	0.2 (−0.2,0.6)	0.0 (−0.3,0.4)	−0.1 (−0.4,0.3)	−0.1 (−0.4,0.3)	0.0 (−0.3,0.4)
Gender nonconforming	1,578	0.3 (−0.5,1.0)	0.1 (−0.5,0.8)	0.1 (−0.6,0.7)	−0.1 (−0.7,0.6)	0.1 (−0.5,0.7)
Sexual Orientation
Intercept		34.5 (34.2,34.7)	41.9 (41.0,42.7)	g	g	g
Completely Heterosexual	18,428	ref	ref	ref	ref	ref
Mostly Heterosexual	1,579	1.0 (0.3,1.6)**	0.7 (0.2,1.3)*	0.8 (0.3,1.4)**	0.5 (−0.1,1.0)	0.7 (0.1,1.3)*
Bisexual	288	0.5 (−1.0,2.0)	0.3 (−1.0,16)	0.4 (−0.9,1.6)	0.5 (−0.7,1.6)	0.3 (−0.9,15)
Lesbian	98	0.4 (−2.0,2.7)	1.2 (−1.1,3.5)	1.4 (−0.9,3.6)	1.0 (−1.1,3.2)	2.0 (−0.7,4.7)

^aAHEI-2010: Alternative Healthy Eating Index-2010

^bModel 1: age group-adjusted only; (sexual orientation and gender expression are in separate models)

^cModel 2: adjusted for age group, race/ethnicity [non-white (includes black, Hispanic, Asian or Pacific Islander, American Indian or Alaskan Native, and Other) vs. non-Hispanic white], region, energy intake, and alcohol consumption; (sexual orientation and gender expression are in separate models)

^dModel 3: includes gender expression and sexual orientation as independent variables and adjusted for age group, race/ethnicity [non-white (includes black, Hispanic, Asian or Pacific Islander, American Indian or Alaskan Native, and Other) vs. non-Hispanic white], region, energy intake, and alcohol consumption

^eModel 4: Model 3 + Mother’s AHEI-2010 score as a covariate

^fModel 5: Model 4, but age restricted to ≤ 18 years old

^gNo independent intercept because gender expression and sexual orientation are included in the same model

^*p<0.05,

^**p<0.01,

^***p<0.001

In models 1–3, completely heterosexual males had statistically significantly lower diet quality scores than gay males, on average across repeated measures. Mostly heterosexual males and females had statistically significantly higher diet quality scores than completely heterosexual males and females, respectively. After inclusion of gender expression in the model (Model 3), there was no longer a statistical difference in diet quality scores between mostly and completely heterosexual males. There were no statistically significant differences between bisexual and completely heterosexual males and females in diet quality score. Among females, there was no statistically significant difference in diet quality scores between lesbian and completely heterosexual females. Inclusion of mother’s diet quality did not substantially change results for the association between sexual orientation and diet quality scores among males (Model 4). However, effects were slightly attenuated for mostly heterosexual females, such that results were no longer statistically significant (β=0.8 in Model 3; β=0.5 in Model 4). Restricting to participants age 18 and younger (Model 5) attenuated effects for gay males, but not for mostly heterosexual females.

There were no statistically significant gender expression-by-age interactions for either males or females. Sexual orientation-by-age interactions were generally not significant (results not shown), except for mostly heterosexual males and females in the 13–14 years old age group and males in the 17–18 years old age group. Interaction findings suggest that differences in diet quality scores between mostly heterosexual and completely heterosexual males and females were wider at 13–14 years of age than at 10–12 years of age. Similarly, differences in diet quality scores were wider at 17–18 years of age than at 10–12 year of age between mostly heterosexual and completely heterosexual males.

Results for breakfast consumption and family dinners ≥5 days/week are presented in Table 5. Among males, there were no statistically significant associations between gender expression, sexual orientation, breakfast consumption and family dinners. In all models, gender nonconforming females had significantly lower risk of breakfast consumption than very gender conforming females. In all models, mostly heterosexual and bisexual females had statistically significantly lower risk of breakfast consumption than completely heterosexual women. There was no statistically significant difference in breakfast consumption between lesbian and completely heterosexual females. Further, there were no statistically significant differences in family dinners by gender expression or sexual orientation among females.

Table 5.

Results from generalized estimating equations modified Poisson regression examining associations with breakfast consumption and family dinners from ages 10–23 years, stratified by sex, GUTS 1 & 2, 1997–2011

		RR (95% CI)
		MALES				FEMALES
	N obs.	Model 1a	Model 2b	Model 3c	N obs.	Model 1a	Model 2b	Model 3c
			Breakfast Consumption (≥5 times/week)
Gender Expression
Very gender conforming	3,695	ref	ref	ref	4,546	ref	ref	ref
Mostly gender conforming	5,129	1.0 (1.0,10)	1.0 (1.0, 1.0)	1.0 (1.0,1.0)	10,863	1.0 (1.0,1.0)	1.0 (1.0,1.0)	1.0 (1.0,10)
Gender nonconforming	340	1.0 (0.9,1.0)	1.0 (0.9,1.0)	1.0 (0.9,1.1)	1,334	0.9 (0.9,1.0)***	0.9 (0.9,1.0)***	0.9 (0.9,1.0)*
Sexual Orientation
Completely Heterosexual	8,566	ref	ref	ref	15,218	ref	ref	ref
Mostly Heterosexual	390	1.0 (0.9, 1.1)	0.9 (0.9, 1.0)	1.0 (0.9,1.1)	1,219	0.9 (0.8,0.9)***	0.9 (0.8,1.0)***	0.9 (0.9,01.0)***
Bisexual	55	1.0 (0.9,1.2)	1.0 (0.9,1.2)	1.0 (0.9,1.2)	220	0.7 (0.6,0.8)***	0.7 (0.6,0.8)***	0.7 (0.6,0.8)***
Gay or Lesbian	153	0.9 (0.7,1.0)	0.9 (0.7,1,0)	0.9 (0.7,1.0)	86	0.8 (0.7,11)	0.8 (0.7,1.1)	0.9 (0.7,1.1)
		Family Dinners (≥5 times/week)d
Gender Expression
Very gender conforming	3,026	ref	ref	ref	3,779	ref	ref	ref
Mostly gender conforming	4,128	1.0 (0.9,1.0)	1.0 (0.9, 1.0)	0.9 (0.9, 1.0)	8,893	1.0 (0.9,1. 0)	0.9 (0.8,1. 00	1.0 (0.9,1.0)
Gender nonconforming	265	1.0 (0.9,1.1)	1.0 (0.9,1.1)	0.9 (0.8,1.0)	1,061	1.0 (0.9,11)	1.0 (0.9,1.1)	1.0 (0.9,1.1)
Sexual Orientation
Completely Heterosexual	7,006	ref	ref	ref	12,657	ref	ref	ref
Mostly Heterosexual	272	1.0 (0.8,1.1)	1.0 (0.8,1.1)	1.0 (0.8,1.1)	866	0.9 (0.8,1. 0)	0.9 (0.8,1.0)	0.9 (0.8,10)
Bisexual	44	1.0 (0.7,15)	1.0 (0.7,15)	1.1 (0.7,15)	162	1.1 (0.8,13)	1.0 (0.8,1.3)	1.0 (0.8,13)
Gay or Lesbian	97	1.1 (0.8,13)	1.1 (0.8,13)	1.1 (0.9,14)	48	1.1 (0.8,16)	1.1 (0.8,1.6)	1.1 (0.8,16)

^aModel 1: age group-adjusted only; (sexual orientation and gender expression are in separate models)

^bModel 2: adjusted for age group, race/ethnicity [non-white (includes black, Hispanic, Asian or Pacific Islander, American Indian or Alaskan Native, and Other) vs. non-Hispanic white], and region; (sexual orientation and gender expression are in separate models)

^cModel 3: includes gender expression and sexual orientation as independent variables and adjusted for age group, race/ethnicity [non-white (includes black, Hispanic, Asian or Pacific Islander, American Indian or Alaskan Native, and Other) vs. non-Hispanic white], and region; no independent intercept because gender expression and sexual orientation are included in the same model

^drestricted to participants ≤18 years old

^*p<0.05,

^**p<0.01,

^***p<0.001

There were no significant gender expression-by-age interactions among females for either breakfast consumption or family dinners. For gender expression-by-age interactions among males (results not shown), the difference in breakfast consumption between mostly gender conforming and very gender conforming males was statistically wider at 19–23 years old age than at 10–12 years old. Sexual orientation-by-age interactions for breakfast consumption were statistically significant for bisexual males and mostly heterosexual females in the 13–14 years old age group (results not shown). These findings indicated that the difference in breakfast consumption was wider for bisexual males and mostly heterosexual females compared to the same-sex completely heterosexual counterparts at age 13–14 years old compared to at age 10–12 years old. There were no statistically significant sexual orientation-by-age interactions for family dinners among both males and females.

Model 3 presented in Tables 4 and 5 included both gender expression and sexual orientation as independent predictors. Compared to the crude and adjusted models, which examined sexual orientation and gender expression in separate models, findings did not change substantially. This suggests that gender expression and sexual orientation each have an independent effect on diet quality and eating habits. It should be noted that a small change to the confidence intervals altered statistical significance in the association between gender expression, sexual orientation, and AHEI-2010 scores for mostly gender conforming and mostly heterosexual males.

DISCUSSION

Findings indicate that the relationship between sexual orientation, gender expression, and diet quality, breakfast consumption, and family dinners is complex. Most notably, completely heterosexual males and very gender conforming males tended to have poorer diet quality than sexual minority males and males in other gender expression categories, respectively. Among females, there were significant relationships between sexual orientation and diet outcomes, however, gender expression was generally not associated with diet outcomes. While mostly heterosexual females had slightly higher diet quality scores than completely heterosexual females, mostly heterosexual and bisexual women were less likely to consume breakfast. In addition, gender nonconforming expression among females was associated with less breakfast consumption, but associations with other diet outcomes were not observed.

The relationship between gender expression and diet quality was more consistent among males than females. As hypothesized, very gender conforming males had the lowest diet quality scores of all the groups. Some research has examined the social perceptions of “masculinity” on various aspects of eating habits and diets; these studies highlight how values associated with masculinity may negatively impact health by construing “healthy eating” as an inherently “feminine” behavior.^40–42 Consistent with Gender Intensification Theory, very gender conforming males may exhibit gender expression behaviors that reinforce the constructs associated with masculinity and thus likely contributing to disparities in dietary quality. Further, in this study, regression results suggest that completely heterosexual males had lower diet quality scores than mostly heterosexual and gay males. Previous work exploring the relationship between gender and sexual orientation have noted the importance of “heterosexuality” in the embodiment of “masculine” gender norms.^43,44 It is likely that the social forces underpinning what it means to be “masculine” or “completely heterosexual” among males negatively influences their relationship to healthy eating, as evidenced by recent work.^45,46 Given the importance of dietary quality on future risk of numerous chronic conditions, include cardiovascular disease, type 2 diabetes, and cancer,³⁵ addressing dietary quality among very gender conforming and completely heterosexual males may have a significant public health impact on our population’s health. For example, diet interventions targeted to men, may want to consider incorporating intervention elements that challenge perceptions of healthy eating as inherently feminine or particular food items as reinforcing masculine ideals. Additional work in the development of effective diet interventions targeted to completely heterosexual as well as very gender conforming males continues to be needed. Overall, in order to address dietary quality effectively, it is vital to shift the negative social, gender, and sexuality norms that link masculinity and heterosexuality with less healthy eating.

Among females, gender expression was not consistently associated with diet-related outcomes, which is contrary to what we expected. Using Gender Intensification Theory to inform potential explanations for this lack of association, it is possible that gender nonconformity among young females, and the adaptation of more masculine forms of gender expression does not include eating habits. Previous research found that gender nonconformity was associated with less physical activity among males, but more physical activity among females,¹² so it was surprising that a similar effect was not found for dietary quality or eating habits. Future research should explore if there may be other diet-related outcomes (such as disordered eating behaviors) that may be more strongly associated with gender expression among females.

Previous research on diet quality by sexual orientation is limited to one study examining women in the Nurses’ Health Study II cohort, which includes the mothers of the GUTS participants. This study found that lesbian and bisexual women had significantly higher diet quality scores than heterosexual women.³⁸ This finding differs from current study findings, which found small non-significant associations between sexual orientation and diet quality scores among lesbian and bisexual females. Examination of diet quality components did not yield substantive information on specific factors that could explain this lack of difference. One potential reason could be that youth in general tend to have lower diet quality scores than adults⁴⁷ and therefore, there is less variability in their diets until adulthood. Further, although this discrepancy could not be directly assessed, cohort effects could also contribute to this difference. Since the participants in GUTS are younger and grew up with greater obesity in their environments⁴⁸ compared to their mothers, this may modify the relationship between sexual orientation and dietary quality. Additional research on change over time between birth cohorts is needed to better understand how environmental shifts may affect diet-related health.

Related, inclusion of mother’s diet quality scores did not have a substantial effect on results. While statistically significant, but small, effects were attenuated slightly and thus, no longer statistically significant, mother’s diet quality score did not seem to contribute in clear ways to gender expression or sexual orientation differences in diet quality scores. While previous research has linked mother’s diet quality to their children’s diet quality,⁴⁹ it is unlikely that mother’s diet quality is associated with their child’s gender expression or sexual orientation. This may explain why mother’s diet quality scores did not have a bigger effect on the associations of gender expression and sexual orientation with diet quality scores.

Findings regarding breakfast consumption and sexual orientation indicating that sexual minority females were less likely than completely heterosexual females to eat breakfast ≥5 days/week aligns with recent 2015 data from YRBS,¹⁷ however, is not consistent with college student data from Minnesota, which found only bisexual females were less likely to eat breakfast than heterosexual females.¹⁹ Among males, previous studies generally found no significant sexual orientation differences in breakfast consumption,^17,19 which is consistent with findings from the current study. Only one previous study, which also used GUTS data, examined family dinners as a 4-level variable (“some to never”, “most to some”, “most”, and “everyday) across sexual orientation and found no significant difference among males, but found significant differences among females. However, the study did not examine which specific sexual orientation groups differed at what level of family dinners.⁵⁰ Additional research using other data sources is needed to further determine whether the finding from the current study that there are no differences in frequency of family dinners by sexual orientation is replicable. Research on sexual orientation differences in eating habits is particularly lacking and more work is needed to understand why differences exist as well as its potential influence on health outcomes that negatively affect sexual minorities, such as obesity. Tentative evidence from this study as well as others suggests that breakfast consumption may be an important intervention point, especially for sexual minority females.

Despite previous GUTS research showing that sexual minority youth are also more likely to be gender nonconforming,⁵¹ for diet quality scores and eating habits, sexual orientation and gender expression seem to not confound each other. In other words, sexual orientation and gender expression each had independent effects on diet quality and eating habits. These findings differ from some previous research, which has found that gender expression may partially mediate the relationship between sexual orientation and physical activity among males¹² and the relationship between sexual orientation and childhood abuse as well as posttraumatic stress disorder.⁵² Another study found that including sexual orientation as a covariate only slightly attenuated the statistically significant positive association between gender expression and BMI.¹¹ Findings from these studies suggest that gender expression influences vary depending on the health-related outcome being examined. Further, this evidence suggests that Gender Intensification Theory and sexual minority stress may have related, but separate contributions to diet quality and eating habits. For example, while socialized gender norms may affect overall diet quality, sexual minority stress may affect specific eating behaviors, such as stress-related eating or other eating-related coping mechanisms.

Several limitations of our study are important to consider. First, there is little racial/ethnic or socioeconomic diversity in the GUTS cohort due to the participants being the children of nurses, therefore generalizability of findings are limited and may not represent the experiences of people of color or people in low or high socioeconomic positions. However, participants were not recruited based on either sexual orientation or gender expression, thus improving generalizability of findings. In addition, gender development (including self-identity as well as understanding of social gender roles and behaviors) is a dynamic and complex process that begins at a very young age.⁵³ In this study, gender expression was assessed at only one point in time, which could have resulted misclassification of some participants, particularly those who may have experienced less gender expression stability throughout adolescence. It is unclear how stable gender expression may have been over time for these study participants. If participants were more likely to report gender conformity during adolescence, estimates in the current study were likely biased toward the null, which could explain why there were no gender expression differences for some diet-related outcomes. Future longitudinal research should incorporate multiple time-point measurements of gender expression in order to assess important shifts in diet outcomes in relation to changes in gender expression. For some gender expression and sexual orientation subgroups, the sample sizes are relatively small, which may contribute to insufficient statistical power to detect differences in diet outcomes. Future studies with larger sample sizes for these subgroups (such as bisexual males and very gender conforming gay males) are needed to better understand the relationship between gender expression, sexual orientation and diet outcomes. Finally, although validated measures were used, sexual orientation, gender expression, and diet outcomes were assessed using self-reports, which could result in bias of findings due to misclassification and measurement error.

CONCLUSION

The relationships of sexual orientation and gender expression with diet quality and breakfast consumption are complex. Gender conformity and heterosexuality among males may be a critical social factor to consider when developing interventions to improve dietary quality and prevent the development of future chronic conditions. Overall, findings indicate that gender, sex, and sexual orientation all play a role in diet-related health. However, as this continues to be an underexplored area, more research is needed to establish a stronger evidence base to inform future interventions for youth.

RESEARCH SNAPSHOT.

Research Question:

Are there differences in diet quality and eating habits based on gender expression as well as sexual orientation among males and females 10–23 years old?

Key Findings:

On average, from ages 10–23 year old, very gender conforming males tended to have lower diet quality scores than less conforming males. Further, completely heterosexual males also had lower diet quality scores than some sexual minority males. Sexual minority females were less likely to consume breakfast most days of the week compared to completely heterosexual females.