Differences in Dietary Quality by Sexual Orientation and Sex in the US: NHANES 2011–2016

Abstract

Background:

There are persistent disparities in weight- and diet-related diseases by sexual orientation. Lesbian and bisexual females have a higher risk of obesity and cardiovascular disease compared to heterosexual females. Gay and bisexual males have a higher risk of diabetes and cardiovascular disease compared to heterosexual males. However, it remains unknown how sexual orientation groups differ in their dietary quality.

Objective:

This study aimed to determine whether dietary quality differs by sexual orientation and sex among US adults.

Design:

This was a cross-sectional study of 24-h dietary recall data from a nationally representative sample of adults aged 20–65 participating in the 2011–2016 National Health and Nutrition Examination Survey (NHANES).

Participants/setting:

Study participants were adults (n=8,851) with complete information on dietary intake, sexual orientation, and sex.

Main outcome measures:

The main outcome measures were daily energy intake from 20 specific food and beverage groups and Healthy Eating Index-2015 (HEI-2015) scores for sexual orientation groups (heterosexual vs. gay/lesbian/bisexual (GLB)).

Statistical analyses performed:

Ordinary least squares regressions were used to calculate adjusted means for each food and beverage group and HEI-2015, stratified by sex and controlling for covariates (e.g., age, race/ethnicity) and survey cycles (2011–2012; 2013–2014; 2015–2016).

Results:

Among males, red and processed meat/poultry/seafood (p=0.01) and sandwiches (p=0.02) were smaller contributors to energy intake for gay/bisexual males compared to heterosexual males. Among females, cereals (p=0.04) and mixed dishes (p=0.02) were smaller contributors to energy intake for lesbian/bisexual females compared to heterosexual females. Gay/bisexual males had significantly higher total HEI-2015 scores than heterosexual males (53.40 ± 1.36 vs. 49.29 ± 0.32, difference=4.14, p=0.004). Lesbian/bisexual females did not differ in total or component HEI-2015 scores from heterosexual females.

Conclusions:

While GLB groups were similar for a variety of dietary outcomes compared to heterosexual groups, gay and bisexual males displayed healthier dietary quality for processed meat (by consuming smaller amounts) and overall dietary quality (according to HEI-2015) compared to heterosexual males.

Background

Persistent disparities exist in chronic diseases by sexual orientation.^1–4 Gay/lesbian/bisexual (GLB) adults of both sexes are more likely to experience a co-occurance of at least 2 chronic diseases (i.e., diabetes, coronary heart disease, cancer)⁵ and are also more likely to die from preventable chronic diseases (i.e., chronic liver disease, hypertensive heart disease, smoking-related cancers) compared to heterosexual adults.⁶ Additionally, GLB adults may have an elevated cancer risk compared to heterosexual adults.⁷

Disparities among sexual orientations may also be stratified by sex, though evidence is conflicting. Gay males have a lower risk of being overweight or obese compared to heterosexual males,^8–10 but show a stronger positive relationship between body weight and chronic disease risk than heterosexual males.¹¹ One study found that gay and bisexual males are at higher risk for a number of chronic diseases compared to heterosexual males, including diabetes, high blood pressure, high cholesterol, and cardiovascular disease.² However, another recent study found no difference in cardiovascular disease risk facors between gay and heterosexual males, yet found an elevated risk for bisexual males.¹² Lesbian and bisexual females have a higher risk of being overweight and obese than heterosexual females,^2,8,13 as well as a higher risk of cardiovascular disease¹⁴ and type 2 diabetes,¹⁵ compared to heterosexual females. However, one systematic review found that lesbian and bisexual females did not experience weight-related physical disorders at a higher prevalence than heterosexual females, despite having higher BMIs.¹³

Diet is a major contributor to the development of obesity and many chronic diseases experienced by GLB adults, such as cardiovascular disease, cancer, and type 2 diabetes.^16–18 Given this, diet may be a contributing factor for differences in chronic disease risk by sexual orientation. For example, existing research has found that the elevated cardiovascular disease risk for lesbian and bisexual females compared to heterosexual females is due in part to higher BMI and greater alcohol consumption, in addition to illicit drug and tobacco use and poor mental health.¹⁴ However, the same review found that gay and bisexual males had a higher risk of cardiovascular disease compared to heterosexual males due to illicit drug and tobacco use and poor mental health, not diet-related factors.¹⁴ This conflicting evidence suggests a need for continued research on dietary quality differences between sexual orientation groups. Research also shows differences in diet-related behaviors which may contribute to elevated chronic disease risk. For example, gay males consume more meals away from home than heterosexual males,¹⁹ and lesbian and bisexual females are more likely to consume meals away from home compared to heterosexual females.²⁰ Higher consumption of food away from home is associated with elevated risk of obesity,^21,22 diabetes,^23–25 chronic disease biomarkers,^21,26 and all-cause mortality.²⁷ GLB adults are also more likely to experience disordered eating,^28–30 which can lead to a variety of negative health consequences,³¹ such as cardiovascular,^32,33 endocrine,^34,35 and bone^36–38 issues, as well as an elevated risk of mortality.^38–42

Yet, despite these persistent disparities in diet-related diseases and diet-related behaviors, limited research has characterized in detail how diet quality differs between GLB and heterosexual individuals. Existing research has primarily examined intake of only a few specific nutrients or food groups, such as energy intake, fruits and vegetables, desserts, and SSBs,^{9,19,20,43–46} but these categories do not capture overall dietary quality. Additionally, these are not the only food groups associated with obesity and other diet-related disease; other food groups of interest include red meats and refined grains.⁴⁷ Researchers and practitioners need a more detailed understanding of differences in dietary quality by sexual orientation, including understanding consumption of food groups (beyond fruits and vegetables, desserts, and SSBs) and of overall dietary quality. This information may elucidate whether dietary behaviors of GLB and heterosexual individuals differ in ways that are meaningful for non-communicable disease risk. Understanding these relationships could help develop targeted interventions to reduce health disparities between and within GLB and heterosexual populations. The aim of this study was to address this gap by describing differences in dietary quality between GLB and heterosexual individuals, including whether differences exist by sex.

Methods

Participants

This study used data from NHANES, a cross-sectional survey providing a nationally representative sample of civilian, non-institutionalized individuals that live in the United States. NHANES is continuous, multistage probability survey that includes at-home health interviews and objective health measurements collected in a Mobile Examination Center.⁴⁸ It is the most comprehensive nationally-representative dataset for assessing dietary intake and anthropometrics. Data are collected and released in two-year cycles. For this analysis, data from the 2011–2012, 2013–2014, and 2015–2016 cycles were used,^49–51 and included individuals were ages 20–65 with complete information on dietary intake, sexual orientation, and sex. This research was exempt from requiring review by an Institutional Review Board, because the data source was de-identified secondary data.

Measures

Sexual Orientation

Participants were initially categorized based on NHANES’ variables for self-reported gender (male or female) and sexual orientation (heterosexual, gay or lesbian, or bisexual). The response options for the NHANES sexual orientation question in the 2015–2016 survey cycle differed from the 2011–2012 and 2013–2014 survey cycles. In 2011–2012 and 2013–2014, the response options were as follows: “heterosexual or straight,” “homosexual or gay/homosexual or lesbian” (dependent upon the self-reported gender), “bisexual,” “something else,” “not sure,” “refused,” or “don’t know.” In the 2015–2016 survey cycle, the answer choices were as follows: “gay/lesbian or gay” (dependent upon the self-reported gender), “straight, that is, not gay/lesbian or gay,” “bisexual,” “something else,” “I don’t know the answer,” “refused,” and “don’t know.”⁵² Participants were excluded if they selected “something else,” “not sure,” “refused,” “don’t know,” or “I don’t know the answer” for sexual orientation. While NHANES uses male and female to describe gender, these terms are typically used to describe biological sex.^53,54 Throughout this paper, the term “sex” will be used in place of “gender” to align with the correct terminology.

Initial categorizations were as follows: heterosexual males (n=4,230), gay males (n=106), bisexual males (n=67), heterosexual females (n=4,165), lesbian females (n=63), and bisexual females (n=220). In order to increase sample sizes within group, gay or lesbian and bisexual groups were combined for analyses. Thus, participants were categorized as: heterosexual males (n=4,230), gay and bisexual males (n=173), heterosexual females (n=4,165), and lesbian and bisexual females (n=283) (Table 1).

Table 1.

Demographic characteristics by sex and sexual orientation, NHANES 2011–2016

	Males			Females
Characteristics	Heterosexual	Gay and Bisexual		Heterosexual	Lesbian and Bisexual
	n=4,230	n=173		n=4,165	n=283
	N (%) or Mean (SD)a	N (%) or Mean (SD)a	p-valueb	N (%) or Mean (SD)a	N (%) or Mean (SD)a	p-valueb
Age (years), mean (SD)	39.1 (10.0)	39.6 (10.0)	0.33	40.5 (10.0)	32.9 (9.1)	<0.001
Race/Ethnicity
Mexican American	593 (10.7%)	14 (3.6%)	0.049	577 (9.5%)	23 (6.4%)	0.025
Other Hispanic	389 (6.4%)	13 (5.7%)		454 (6.6%)	24 (5.7%)
Non-Hispanic White	1621 (62.6%)	84 (75.4%)		1557 (63.1%)	117 (63.4%)
Non-Hispanic Black	916 (11.2%)	35 (8.4%)		995 (12.7%)	80 (14.7%)
Non-Hispanic Asian	529 (5.3%)	20 (4.4%)		437 (5.1%)	17 (3.8%)
Other race including multiracial	182 (3.7%)	7 (2.5%)		145 (2.9%)	22 (6.0%)
Education level
<9th grade	207 (3.4%)	3 (0.8%)	0.005	173 (2.9%)	7 (1.9%)	0.002
9th-11th grade	603 (11.2%)	11 (3.4%)		441 (7.9%)	35 (10.3%)
High school diploma/GED	1011 (23.0%)	40 (21.1%)		775 (17.7%)	69 (23.3%)
Some college/associates degree	1285 (32.4%)	65 (40.7%)		1502 (35.8%)	113 (39.7%)
College graduate or higher	1124 (30.0%)	54 (34.0%)		1274 (35.8%)	59 (24.9%)
Nativity
US born	3005 (81.8%)	138 (89.9%)	0.013	3019 (83.5%)	244 (90.2%)	<0.001
Foreign born	1222 (18.2%)	35 (10.1%)		1145 (16.5%)	39 (9.8%)
Income to poverty ratio c
Below 185%	1749 (31.5%)	68 (32.8%)	0.05	1743 (32.0%)	155 (50.6%)	<0.001
185%−300%	664 (16.4%)	40 (19.5%)		674 (16.6%)	35 (12.1%)
>300%	1521 (46.5%)	57 (39.5%)		1480 (46.1%)	78 (33.7%)
Current cigarette smoker
Yes	1151 (24.2%)	54 (28.6%)	0.23	766 (19.1%)	107 (41.3%)	<0.001
No	3078 (75.8%)	118 (71.4%)		3397 (80.9%)	176 (58.7%)
Health insurance coverage
Yes	2951 (75.8%)	130 (84.0%)	0.31	3256 (82.3%)	199 (75.9%)	<0.001
No	1277 (24.1%)	43 (16.0%)		904 (17.6%) 84 (24.1%)
Depression,d mean (SD)	2.7 (3.3)	4.4 (3.5)	<0.001	3.5 (3.8)	5.5 (4.5)	<0.001
Alcoholic drinks/day past 12 mo., mean (SD) c	0.9 (1.3)	0.9 (1.2)	0.73	0.4 (0.7)	0.5 (0.9)	0.06
BMI (kg/m 2 ), mean (SD)	28.9 (5.2)	27.3 (5.1)	0.001	29.3 (6.6)	30.4 (7.1)	0.015

^aAppropriate sample weighting was applied to %s, means, and SDs to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹ Sample sizes are unweighted and refer to the number of NHANES participants in each category.

^bP-values were calculated using t-tests for continuous variables and chi-square tests for categorical variables.

^cMissingness on demographic variables ranged from 0–2%, except income to poverty ratio (6.63% missing) and alcoholic drinks/day (12.43% missing).

^dDepression is measured using the Patient Health Questionnaire-9 (PHQ-9), a diagnostic instrument where participants respond to 9 DSM-IV components on a scale from “Not at all” (coded as 0) to “Nearly every day” (coded as 3). A high score represents a higher severity of depression.⁷¹

Food and Beverage Intake

Total daily dietary intake was examined using 24-hour dietary recall data and, for fast food and pizza consumption, food frequency items. NHANES collects 24-hour dietary recalls using the USDA Automated Multiple-Pass Method (AMPM), a five-step process for collecting complete recalls.⁵⁵ Trained interviewers complete the first day dietary interview in-person in a Mobile Examination Center and the second interview via telephone.⁵² For this study, dietary recall data were limited to the first day, consistent with National Cancer Institute recommendations for estimating mean usual intake in a population.⁵⁶ Food frequency items for fast food and pizza consumption were collected as part of the NHANES Diet Behavior and Nutrition questionnaire. Participants responded with the number of meals consumed from a fast food or pizza place per week.⁵⁷

Analyses examined consumption of 20 food and beverage groups, as shown in Table 2.^58,59 Within these groups, adjusted mean daily energy intake and mean percent contribution to total daily energy intake were calculated. The top five contributors to energy intake were identified based on each food and beverage group’s mean percent contribution to total daily energy intake. The fast food and pizza groups were calculated as adjusted mean meals per week.

Table 2.

Daily energy intake from foods, and beverages by sex and sexual orientation, NHANES 2011–2016.

	Males						Females
	Heterosexual (n=4,230)		Gay and Bisexual (n=173)		Differencea		Heterosexual (n=4,165)		Lesbian and Bisexual (n=283)		Differencea
	Adjusted Meanb	SEb	Adjusted Meanb	SEb	Diff.	pc	Adjusted Meanb	SEb	Adjusted Meanb	SEb	Diff.	pc
Total Energy Intake (kcals)	2609	22	2672	106	63	0.58	1926	15	2031	67	105	0.12
Foods
Breads and grains	175	6	160	18	−16	0.40	126	5	130	12	4	0.74
Breakfast cereals	56	3	78	18	22	0.25	50	3	37	8	−13	0.17
Cheese and yogurt	71	4	70	15	−2	0.91	60	3	60	9	0	0.99
Desserts and sweet snacks	272	9	289	52	17	0.75	223	7	249	25	27	0.28
Fruits	51	2	64	9	13	0.20	49	2	54	6	5	0.43
Nuts, legumes, and other non-meat proteins	140	7	164	20	24	0.21	107	5	113	17	6	0.75
Red and processed meat/poultry/seafood	159	7	116	20	−42	0.06	86	4	96	13	10	0.45
Lean and unprocessed meat/poultry/seafood	155	8	177	33	22	0.51	107	5	104	16	−3	0.86
Mixed dishes	475	14	408	47	−67	0.17	361	10	311	27	−50	0.07
Salty snacks	163	5	152	22	−11	0.65	129	5	143	18	14	0.45
Sauces/dips/condiments	76	3	118	24	42	0.09	68	3	80	12	11	0.35
Vegetables	32	2	40	7	8	0.33	33	1	38	6	5	0.37
Hamburgers and pizza	182	11	170	40	−11	0.78	92	6	135	30	43	0.15
Sandwiches	77	6	46	15	−32	0.06	53	4	43	12	−10	0.40
Beverages
100% Juice	35	5	26	7	−8	0.14	22	2	23	9	2	0.86
Alcohol	138	6	195	34	58	0.08	120	6	144	18	25	0.20
Milk and other dairy	69	4	66	16	−3	0.84	40	3	35	10	−5	0.61
SSBs	164	7	174	25	10	0.68	100	4	116	15	16	0.26
Other beverages	23	3	40	17	17	0.33	23	3	37	10	14	0.18
Food away from home (meals/week)
Fast food and pizza	2.4	0.07	2.1	0.41	−0.3	0.44	1.7	0.07	1.8	0.20	0.08	0.74

^aDifference is difference in adjusted means, gay/lesbian/bisexual (GLB) – heterosexual.

^bMeans and SEs adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁷¹ and alcohol consumption). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^cP-values were calculated using ordinary least squares regression.

Healthy Eating Index-2015

To provide an overall assessment of dietary quality, participants’ HEI-2015 scores were examined. HEI-2015 component scores and total scores are calculated based on the degree to which an individual’s dietary intake aligns with the 2015–2020 Dietary Guidelines for Americans.^60–62 The components are total fruits, whole fruits, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant proteins, fatty acids (ratio of poly and monounsaturated fatty acids to saturated fatty acids), refined grains, sodium, added sugars, and saturated fats. Higher scores indicate better diet quality, with a maximum of 100 for the total score.⁶² HEI-2015 scores were calculated by applying the National Cancer Institute HEI-2015 scoring macro (population ratio method) to the first day of the 24-hour dietary recall data.⁶³

Covariates

Covariates were chosen based on existing literature about factors that contribute to diet disparities^64–68 and included age, race/ethnicity (Mexican American, Other Hispanic, non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, and Other race/multiracial), education level (<9^th grade, 9^th-11^th grade, High school diploma/GED, some college/associates degree, and college graduate or higher), nativity (US born and foreign born), income to poverty ratio⁶⁹ (below 185%, 185%−300%, >300%), cigarette smoking status, health insurance coverage, mental health (mean depression screener score⁷⁰), and alcohol consumption (average number of drinks per day). Mental health (depression screener score) was measured using the Patient Health Questionnaire-9 (PHQ-9), a diagnostic instrument where participants respond to 9 DSM-IV components on a scale from “Not at all” (coded as 0) to “Nearly every day” (coded as 3). A higher score represents higher severity of depression.⁷⁰ Although calories from alcohol are factored into the total HEI-2015 scores, frequency of alcohol consumption was included as a covariate for consistency with prior studies of HEI-2015 that control for alcohol intake as a confounder to address health behavior factors that may be associated with dietary behaviors or health outcomes.^71–73 All information on covariates were collected from NHANES participants by trained interviewers.

Statistical Analysis

The analytic sample included participants ages 20–65. Beginning with a sample of 25,246 participants with complete dietary information, 13,493 participants were excluded for being outside the age range of 20–65, leaving 11,753 participants. Next, 2,830 participants were dropped due to missing sexual orientation data, leaving 8,923 participants (missingness=24.1%). Finally, 72 participants were dropped for not identifying as heterosexual, gay or lesbian, or bisexual (i.e., selected “something else,” “not sure,” “refused,” “don’t know,” or “I don’t know the answer” for the sexual orientation question), leaving a final analytic sample size of n=8,851.

Proportions or means and standard deviations of covariates were calculated for each sexual orientation group. BMI was also examined descriptively, because it is an important determinant of some chronic diseases.^74,75 Next, ordinary least squares regressions were used to calculate adjusted means for each outcome, as well as adjusted differences in means across groups, controlling for the covariates listed in the previous section. Descriptive statistics and food group measures were computed in Stata/SE 15⁷⁶ and HEI-2015 scores were calculated using SAS 9.4 (SAS Institute Inc., Cary, NC).⁷⁷ Missingness on most variables was 2% or less (see Table 1), so analyses used complete case analysis. All analyses used a two-tailed alpha level of 0.05, and appropriate sample weighting was applied to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁸

Due to the association between physical activity levels and diet-related chronic disease,⁷⁹ sensitivity analyses were conducted including self-reported moderate to vigorous physical activity (MVPA) as a covariate. Participants were categorized as meeting or not meeting the MVPA recommendation of 150 minutes per week.⁸⁰ Minutes of MVPA from work, recreational activities, and transportation were included.⁸¹

Results

The sample population was predominantly non-Hispanic white, born in the U.S., and covered by health insurance (Table 1). Among males, gay and bisexual males were slightly more educated compared to heterosexual males (p=0.005). The depression score for gay and bisexual males was higher than for heterosexual males (4.4 ± 3.5 vs. 2.7 ± 3.3, p<0.001), and gay and bisexual males had a lower average BMI (27.3 ± 5.1 kg/m² vs. 28.9 ± 5.2 kg/m², p=0.001) compared to heterosexual males. Among females, the lesbian and bisexual sample was much younger (32.9 ± years vs. 40.5 ± 10.0 years, p<0.001), had a larger proportion of individuals below 185% of the federal poverty line (50.6% vs. 32.0%, p<0.001), and had a higher smoking prevalence (41.3% vs. 19.1%, p<0.001) compared to heterosexual females. When compared to heterosexual females, lesbian and bisexual females also had a higher average depression screener score (5.5 ± 4.5 vs. 3.5 ± 3.8, p<0.001) and higher BMI (30.4 ± 7.1 vs. 29.3 ± 6.6, p=0.015).

Table 2 shows adjusted means for each group. For both males and females, there were no statistically significant differences between sexual orientation groups in total energy intake or in absolute energy intake from any food or beverage category. While no differences reached statistical significance, several were relatively large in magnitude. For example, compared to heterosexual males, gay and bisexual males consumed fewer calories from red and processed meat/poultry/seafood (difference= −42 calories/day, p=0.06), fewer calories from sandwiches (difference = −32 calories/day, p=0.06), fewer calories from mixed dishes (difference= −67 calories/day, p=0.17), and more calories from alcohol (difference= 58 calories/day, p=0.08). Among females, lesbian and bisexual females consumed more total calories (difference = 105 calories/day, p=0.12), fewer calories from mixed dishes (difference= −50 calories/day, p=0.07), and more calories from hamburgers and pizza (difference= 43 calories/day, p=0.15) than heterosexual females.

While absolute energy intakes did not differ across sexual orientation groups, there were some differences by sexual orientation in the relative contribution of the food/beverage categories studied (Table 3). Red and processed meat/poultry/seafood was a smaller contributor to total energy intake for gay and bisexual males than heterosexual males (difference= −2.2%, p=0.01), as were sandwiches (difference= −1.4%, p=0.02). Breakfast cereals (difference= −0.8%, p=0.04) and mixed dishes (difference= −3.4%, p=0.02) were a smaller contributor to total energy intake among lesbian and bisexual females than heterosexual females.

Table 3.

Percent daily energy intake from foods and beverages by sex and sexual orientation, NHANES 2011–2016.

	Males						Females
	Heterosexual (n=4,230)		Gay and Bisexual (n=173)		Differencea		Heterosexual (n=4,165)		Lesbian and Bisexual (n=283)		Differencea
	Adjusted Mean %b	SE b	Adjusted Mean %b	SE b	Diff.	pc	Adjusted Mean % b	SE b	Adjusted Mean % b	SE b	Diff.	pc
Foods
Breads and grains	7.1	0.2	6.2	0.7	−0.8	0.28	6.8	0.3	7.1	0.7	0.2	0.71
Breakfast cereals	2.3	0.1	2.9	0.7	0.6	0.43	2.7	0.2	1.8	0.3	−0.8	0.04
Cheese and yogurt	2.8	0.1	2.6	0.5	−0.2	0.70	3.4	0.2	3.1	0.4	−0.3	0.44
Desserts and sweet snacks	9.7	0.3	10.2	1.6	0.4	0.80	11.1	0.3	11.4	0.9	0.3	0.75
Fruits	2.2	0.1	2.8	0.5	0.6	0.25	2.9	0.1	3.0	0.4	0.1	0.75
Nuts, legumes, and other non-meat proteins	5.2	0.2	6.3	0.8	1.1	0.21	5.6	0.3	5.3	0.7	−0.3	0.72
Red and processed meat/poultry/seafood	6.3	0.3	4.1	0.7	−2.2	0.01	4.6	0.2	5.0	0.7	0.4	0.53
Lean and unprocessed meat/poultry/seafood	6.2	0.3	6.8	1.1	0.6	0.62	5.7	0.3	5.6	0.8	−0.1	0.92
Mixed dishes	18.4	0.5	15.9	2.0	−2.5	0.23	18.5	0.5	15.0	1.4	−3.4	0.02
Salty snacks	6.3	0.2	5.4	0.8	−0.9	0.30	6.6	0.2	6.7	0.8	0.0	0.97
Sauces/dips/condiments	2.9	0.1	4.8	1.2	1.9	0.11	3.6	0.1	3.7	0.6	0.1	0.85
Vegetables	1.5	0.1	1.5	0.3	0.1	0.84	1.9	0.1	2.1	0.3	0.2	0.58
Hamburgers and pizza	6.4	0.4	6.2	1.5	−0.2	0.90	4.6	0.3	7.4	1.6	2.8	0.07
Sandwiches	3.0	0.2	1.6	0.5	−1.4	0.02	2.7	0.2	1.8	0.5	−0.9	0.07
Beverages
100% Juice	1.3	0.1	1.3	0.5	−0.01	0.99	1.1	0.1	1.2	0.4	0.1	0.73
Alcohol	5.0	0.2	6.4	0.9	1.5	0.11	5.4	0.2	6.6	0.8	1.2	0.17
Milk and other dairy	2.6	0.2	2.3	0.6	−0.3	0.62	2.1	0.1	1.6	0.4	−0.5	0.20
SSBs	6.3	0.3	6.7	0.8	0.4	0.65	5.2	0.2	5.6	0.6	0.3	0.56
Other beverages	0.9	0.1	1.4	0.6	0.5	0.40	1.2	0.1	1.6	0.4	0.4	0.40

^aDifference is difference in adjusted means, gay/lesbian/bisexual (GLB) – heterosexual.

^bMeans and SEs adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁷¹ and alcohol consumption). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^cP-values were calculated using ordinary least squares regression.

Sexual orientation groups were relatively similar in the top five contributors to total energy intake (Table 4). For all sex and sexual orientation groups, mixed dishes contributed the most to daily intake, followed by desserts and sweet snacks. Breads and grains were a top contributor for all groups except for gay and bisexual males. Gay and bisexual males were the only group to have beverages as top contributors (SSBs at 6.7% and alcohol at 6.4%). Both heterosexual females and lesbian and bisexual females had salty snacks as a top contributor.

Table 4.

Top contributors to total energy intake by sex and sexual orientation, NHANES 2011–2016

	Males				Females
Top energy contributors	Heterosexual (n=4,230)		Gay and Bisexual (n=173)		Heterosexual (n=4,165)		Lesbian and Bisexual (n=283)
	Food Group	Percenta	Food Group	Percenta	Food Group	Percenta	Food Group	Percenta
1st	Mixed dishes	18.4	Mixed dishes	15.9	Mixed dishes	18.5	Mixed dishes	15.0
2nd	Desserts and sweet snacks	9.7	Desserts and sweet snacks	10.2	Desserts and sweet snacks	11.1	Desserts and sweet snacks	11.4
3rd	Bread and grains	7.1	Lean and unprocessed meat/poultry/seafood	6.8	Bread and grains	6.8	Hamburgers and Pizza	7.4
4th	Hamburgers and pizza	6.4	SSBs	6.7	Salty snacks	6.6	Breads and grains	7.1
5th	Red and processed meat/poultry/seafood	6.3	Alcohol	6.4	Lean and unprocessed meat/poultry/seafood	5.7	Salty snacks	6.7

^a%s adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁷¹ and alcohol consumption). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

Gay and bisexual males had significantly higher HEI-2015 scores than heterosexual males (53.40 vs. 49.26; difference=4.14, p=0.004) (Figure 1, Table 5). Gay and bisexual males also had higher scores for fatty acids (difference=1.01 points, p=0.02) and sodium (difference=1.01 points, p=0.01) compared to heterosexual males. In contrast, lesbian and bisexual and heterosexual females did not differ in total HEI-2015 scores (51.59 vs. 51.88, p=0.79) or in any component scores.

Figure 1.

Adjusted Differences in Healthy Eating Index-2015 (HEI-2015) component scores, gay/lesbian/bisexual (GLB) vs. heterosexual, NHANES 2011–2016.

^a Score difference is difference in adjusted scores, GLB – heterosexual. Scores were adjusted for key covariates (age, race/ethnicity, nativity, education level, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁷¹ and alcohol consumption). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^b P-values were calculated using ordinary least squares regression.

*p < 0.05

**p<0.01

Table 5.

Healthy Eating Index-2015 (HEI-2015) by sex and sexual orientation, NHANES 2011–2016

	Males						Females
	Heterosexual		Gay and Bisexual		Differencea		Heterosexual		Lesbian and Bisexual		Differencea
	n=4,230		n=173				n=4,165		n=283
Variable	Adjusted Meanb	SE	Adjusted Meanb	SE	Diff.	p-valuec	Adjusted Meanb	SE	Adjusted Meanb	SE	Diff.	p-valuec
Total HEI-2015 Score	49.26	0.32	53.40	1.36	4.14	0.004	51.59	0.37	51.88	1.03	0.29	0.79
Adequacy components
Total fruits	1.72	0.05	1.87	0.21	0.15	0.49	1.96	0.05	2.11	0.17	0.15	0.39
Whole fruits	1.70	0.05	2.02	0.19	0.31	0.11	2.02	0.06	2.04	0.18	0.03	0.89
Total vegetables	2.88	0.05	3.05	0.19	0.17	0.41	3.11	0.04	3.07	0.15	−0.04	0.79
Greens and beans	1.46	0.06	1.80	0.23	0.33	0.19	1.71	0.06	1.65	0.15	−0.07	0.68
Whole grains	2.28	0.08	2.43	0.37	0.15	0.70	2.61	0.09	2.63	0.25	0.02	0.95
Dairy	5.04	0.09	4.54	0.35	−0.50	0.17	5.08	0.09	4.91	0.34	−0.17	0.63
Total protein foods	4.32	0.03	4.20	0.20	−0.11	0.56	4.13	0.04	4.01	0.13	−0.13	0.39
Seafood and plant proteins	2.25	0.06	2.68	0.23	0.43	0.08	2.36	0.07	2.48	0.18	0.12	0.52
Fatty acids	4.90	0.09	5.91	0.41	1.01	0.02	5.22	0.11	5.33	0.30	0.10	0.74
Moderation components
Refined grains	6.23	0.09	7.02	0.44	0.79	0.08	6.31	0.10	6.63	0.32	0.32	0.32
Sodium	4.14	0.07	5.14	0.37	1.01	0.01	4.33	0.10	4.53	0.27	0.21	0.47
Added sugars	6.53	0.09	6.33	0.31	−0.19	0.50	6.64	0.07	6.26	0.31	−0.38	0.24
Saturated fat	5.81	0.09	6.41	0.42	0.60	0.17	6.10	0.08	6.23	0.27	0.13	0.63

^aDifference is difference in adjusted means, gay/lesbian/bisexual (GLB) – heterosexual.

^bMeans and SEs adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁷¹ and alcohol consumption). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^cP-values were calculated using ordinary least squares regression.

Sensitivity analyses controlling for MVPA revealed similar results (Tables 6–10 online only, Figure 2 online only).

Table 6 online only.

Demographic characteristics by sex and sexual orientation, NHANES 2011–2016

	Males			Females
Characteristics	Heterosexual	Gay and Bisexual	p-valueb	Heterosexual	Lesbian and Bisexual	p-valueb
	n=4,230	n=173		n=4,165	n=283
	N (%) or Mean (SD)a	N (%) or Mean (SD)a		N (%) or Mean (SD)a	N (%) or Mean (SD)a
Age (years), mean (SD)	39.1 (10.0)	39.6 (10.0)	0.33	40.5 (10.0)	32.9 (9.1)	<0.001
Race/Ethnicity
Mexican American	593 (10.7%)	14 (3.6%)	0.049	577 (9.5%)	23 (6.4%)	0.025
Other Hispanic	389 (6.4%)	13 (5.7%)		454 (6.6%)	24 (5.7%)
Non-Hispanic White	1621 (62.6%)	84 (75.4%)		1557 (63.1%)	117 (63.4%)
Non-Hispanic Black	916 (11.2%)	35 (8.4%)		995 (12.7%)	80 (14.7%)
Non-Hispanic Asian	529 (5.3%)	20 (4.4%)		437 (5.1%)	17 (3.8%)
Other race including multiracial	182 (3.7%)	7 (2.5%)		145 (2.9%)	22 (6.0%)
Education level
<9th grade	207 (3.4%)	3 (0.8%)	0.005	173 (2.9%)	7 (1.9%)	0.002
9th-11th grade	603 (11.2%)	11 (3.4%)		441 (7.9%)	35 (10.3%)
High school diploma/GED	1011 (23.0%)	40 (21.1%)		775 (17.7%)	69 (23.3%)
Some college/associates degree	1285 (32.4%)	65 (40.7%)		1502 (35.8%)	113 (39.7%)
College graduate or higher	1124 (30.0%)	54 (34.0%)		1274 (35.8%)	59 (24.9%)
Nativity
US born	3005 (81.8%)	138 (89.9%)	0.013	3019 (83.5%)	244 (90.2%)	<0.001
Foreign born	1222 (18.2%)	35 (10.1%)		1145 (16.5%)	39 (9.8%)
Income to poverty ratio c
Below 185%	1749 (31.5%)	68 (32.8%)	0.05	1743 (32.0%)	155 (50.6%)	<0.001
185%−300%	664 (16.4%)	40 (19.5%)		674 (16.6%)	35 (12.1%)
>300%	1521 (46.5%)	57 (39.5%)		1480 (46.1%)	78 (33.7%)
Current cigarette smoker
Yes	1151 (24.2%)	54 (28.6%)	0.23	766 (19.1%)	107 (41.3%)	<0.001
No	3078 (75.8%)	118 (71.4%)		3397 (80.9%)	176 (58.7%)
Health insurance coverage
Yes	2951 (75.8%)	130 (84.0%)	0.31	3256 (82.3%)	199 (75.9%)	<0.001
No	1277 (24.1%)	43 (16.0%)		904 (17.6%)	84 (24.1%)
Moderate-to-vigorous physical activity per week d
<150 minutes	1107 (24.6%)	47 (31.7%)	0.77	1723 (38.2%)	102 (32.9%)	0.08
≥150 minutes	3123 (75.4%)	126 (68.3%)		2442 (61.8%)	181 (67.1%)
Depression,e mean (SD)	2.7 (3.3)	4.4 (3.5)	<0.001	3.5 (3.8)	5.5 (4.5)	<0.001
Alcoholic drinks/day past 12 mo., mean (SD) c	0.9 (1.3)	0.9 (1.2)	0.73	0.4 (0.7)	0.5 (0.9)	0.06
BMI (kg/m 2 ), mean (SD)	28.9 (5.2)	27.3 (5.1)	0.001	29.3 (6.6)	30.4 (7.1)	0.015

^aAppropriate sample weighting was applied to %s, means, and SDs to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹ Sample sizes are unweighted and refer to the number of NHANES participants in each category.

^bP-values were calculated using t-tests for continuous variables and chi-square tests for categorical variables.

^cMissingness on demographic variables ranged from 0–2%, except income to poverty ratio (6.63% missing) and alcoholic drinks/day (12.43% missing).

^dSelf-reported moderate-to-vigorous physical activity includes minutes of physical activity from work, recreational, and transportation activities.

^eDepression is measured using the Patient Health Questionnaire-9 (PHQ-9), a diagnostic instrument where participants respond to 9 DSM-IV components on a scale from “Not at all” (coded as 0) to “Nearly every day” (coded as 3). A high score represents a higher severity of depression.⁷¹

Table 10 online only.

Healthy Eating Index-2015 (HEI-2015) by sex and sexual orientation, NHANES 2011–2016

	Males						Females
	Heterosexual		Gay and Bisexual		Differencea		Heterosexual		Lesbian and Bisexual		Differencea
	n=4,230		n=173				n=4,165		n=283
Variable	Adjusted Meanb	SE	Adjusted Meanb	SE	Diff.	p-valuec	Adjusted Meanb	SE	Adjusted Meanb	SE	Diff.	p-valuec
Total HEI-2015 Score	49.11	0.33	53.42	1.29	4.32	0.002	51.75	0.38	51.94	1.01	0.18	0.87
Adequacy components
Total fruits	1.70	0.05	1.87	0.20	0.17	0.42	1.98	0.05	2.11	0.17	0.13	0.43
Whole fruits	1.69	0.05	2.02	0.19	0.33	0.09	2.04	0.06	2.05	0.18	0.02	0.94
Total vegetables	2.88	0.05	3.05	0.19	0.17	0.40	3.12	0.04	3.07	0.15	−0.04	0.78
Greens and beans	1.45	0.06	1.80	0.23	0.34	0.17	1.73	0.06	1.65	0.15	−0.07	0.65
Whole grains	2.25	0.08	2.43	0.35	0.18	0.63	2.64	0.09	2.64	0.25	0.00	0.99
Dairy	5.04	0.09	4.54	0.35	−0.50	0.17	5.08	0.09	4.91	0.34	−0.17	0.63
Total protein foods	4.32	0.03	4.20	0.20	−0.12	0.55	4.13	0.04	4.01	0.13	−0.13	0.39
Seafood and plant proteins	2.24	0.06	2.68	0.22	0.44	0.06	2.37	0.07	2.49	0.17	0.12	0.55
Fatty acids	4.89	0.09	5.91	0.41	1.03	0.02	5.23	0.11	5.33	0.30	0.10	0.76
Moderation components
Refined grains	6.21	0.09	7.03	0.44	0.82	0.07	6.33	0.10	6.64	0.32	0.30	0.35
Sodium	4.12	0.07	5.14	0.38	1.02	0.01	4.34	0.10	4.54	0.27	0.20	0.49
Added sugars	6.53	0.09	6.33	0.30	−0.19	0.51	6.65	0.07	6.26	0.31	−0.38	0.24
Saturated fat	5.79	0.09	6.41	0.41	0.62	0.15	6.12	0.08	6.24	0.27	0.12	0.66

^aDifference is difference in adjusted means, GLB – heterosexual.

^bMeans and SEs adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁶⁹ alcohol consumption, and physical activity^81,82). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^cP-values were calculated using ordinary least squares regression.

Figure 2 online only.

Adjusted Differences in Healthy Eating Index-2015 (HEI-2015) component scores, gay/lesbian/bisexual (GLB) vs. heterosexual, NHANES 2011–2016.

^a Score difference is difference in adjusted scores, GLB – heterosexual. Scores were adjusted for key covariates (age, race/ethnicity, nativity, education level, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁷¹ alcohol consumption, and physical activity^81,82). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^b P-values were calculated using ordinary least squares regression.

*p < 0.05

**p<0.01

Discussion

Analyses revealed differences in dietary quality for gay and bisexual males compared to heterosexual males, but no differences in dietary quality between lesbian and bisexual females and heterosexual females. While the absolute energy intakes of the food groups studied are similar for gay and bisexual versus heterosexual males, relative energy contributions of food groups and component scores from the HEI-2015 differed between sexual orientation groups. The higher total HEI-2015, fatty acid, and sodium scores for gay and bisexual males compared to heterosexual males indicate better alignment with the 2015–2020 Dietary Guidelines for Americans,⁶¹ and therefore a more “nutritionally adequate diet.”⁶⁰ This result suggests that gay and bisexual males had somewhat better overall dietary quality (and therefore, lower risk of diet-related disease)^82–85 than heterosexual males. In addition, the lower percent energy intake of red and processed meat/poultry/seafood among gay and bisexual males compared to heterosexual males is indicative of better diet quality and lower diet-related disease risk, as consumption of red and processed meats has been linked to weight and risk of chronic diseases, including cancer.⁶⁰ Finally, although gay and bisexual males had higher total HEI-2015 scores than heterosexual males, few statistically-significant differences were found between these groups in consumption of most other food groups of interest, including sugar-sweetened beverages, fruits, vegetables, and desserts and sweet snacks. Moreover, overall scores were low, suggesting room for improvement for all groups.

The finding that gay and bisexual males had higher overall dietary quality compared to heterosexual males could be due to social factors related to eating. For example, research has found that heterosexuality is a dominant way for males to express their masculinity.^86,87 Healthy eating is often construed as a “feminine” behavior, so heterosexual males may have lower interest in consuming a healthy diet in order to appear more masculine.^88–92 Social factors, including access to health care,⁹³ stigma and discrimination, stress,^94,95 physical activity,⁹⁶ and tobacco and alcohol use,² might also explain why gay and bisexual men do not consistently have lower rates of diet-related disease compared to heterosexual men, despite having higher dietary quality.^2,5,7,14

This study adds to a small but growing literature examining dietary intake by sexual orientation status in males. Similar to this study’s results, a previous study by VanKim and colleagues found that heterosexual males had lower Alternative Healthy Eating Index-2010 (AHEI-2010) scores compared to other sexual orientation groups.⁹⁷ AHEI-2010 and HEI-2015 (used by this study) are similar in that they both are measures of diet quality and have been found to be associated with chronic disease.⁹⁸ However, AHEI-2010 is explicitly focused on foods that are predictive of chronic diseases^82,99 while HEI-2015 assesses the degree to which a set of foods conforms to the 2015–2020 Dietary Guidelines for Americans.⁶² While both AHEI-2010 and HEI-2015 assess fat intake, they assess it somewhat differently.^62,82 Furthermore, AHEI-2010 includes alcohol intake as a separate component while the HEI-2015 only accounts for alcohol intake within total energy.^62,82 Additionally, the VanKim study focused on adolescents and young adults, while this study focused on adults. Two other studies examining dietary quality and sexual orientation focused on fruit and vegetable and SSB consumption; one study found no differences in fruit and vegetable consumption,²⁰ while the more recent study found that bisexual males consume more fruit than heterosexual males, and gay males are more likely to consume SSBs than heterosexual males.¹⁹ By contrast, this study found no differences in SSB, fruit, or vegetable intakes between heterosexual and gay and bisexual males. These differences could be due to how intake was operationalized (e.g., kcals/day vs. daily servings) or collected (e.g., 24-hour dietary recall vs. food frequency questionnaire).

While some differences were observed between heterosexual and lesbian and bisexual females, these differences did not paint a consistent picture about which group had better dietary quality. While lesbian and bisexual females consumed less percentage of daily energy from breakfast cereals and mixed dishes compared to heterosexual females, the implications of these differences for diet-related disease remain to be established. For example, consuming more breakfast cereals does not clearly indicate better or worse diet quality, because cereals can vary widely in their whole grain and sugar content. Similarly, mixed dishes can include processed or unprocessed meats as well as whole and refined grains.

Few studies have analyzed dietary intake and sexual orientation among females.^{43,44,100,101} One study found that heterosexual females had a lower dietary quality than lesbian and bisexual females;¹⁰⁰ that study assessed diet quality using the AHEI-2010, which as noted previously has both similarities and differences from HEI-2015. Another study found that heterosexual females had lower dietary quality compared to sexual minority females according to the Dietary Approach to Stop Hypertension (DASH) scores.¹⁰¹ By contrast, the present study did not reveal differences in dietary quality (assessed with HEI-2015 scores) for lesbian and bisexual compared to heterosexual females. Differences in findings may be due to different measures of dietary quality (i.e., AHEI-2010 and DASH vs. HEI-2015) or differences in how sexual orientations were grouped (i.e., analyzing lesbian and bisexual females separately vs. in the same category). Two other studies used data from the California Health Interview Survey (CHIS) to compare fruit and vegetable consumption between heterosexual and lesbian/bisexual females but did not assess HEI-2015. One found that fruit and vegetable consumption did not differ,⁴³ while another found that lesbian and bisexual females consume less fruits and vegetables than heterosexual females.⁴⁴ By contrast, the present study did not observe differences in fruit and vegetable consumption between lesbian/bisexual females and heterosexual females. Differences across studies could be due to how fruit and vegetable intake was operationalized (e.g., kcals/day vs. daily servings) or how dietary data are collected. For example, the CHIS uses a food frequency questionnaire to assess usual consumption of fruit, vegetable, potato, legume, soda, fruit juice (100% and sweetened), and dessert intake, while NHANES uses 24-hour dietary recall.¹⁰²

Strengths and Limitations

Strengths of this study include use of nationally representative, detailed dietary intake data and examination of a large set of dietary variables. Limitations include that dietary data were collected via self-report, and differential misreporting by sexual orientation group is possible, which could exaggerate or mask true differences between groups in dietary intake variables. For example, underreporting of intake is more likely among individuals with overweight/obesity.¹⁰³ Given that BMI was higher among lesbian and bisexual females in this study, it is possible that there was also differential underreporting of diet or key dietary components for lesbian and bisexual females, and it is unclear in which direction this potential bias would affect this study’s results. Other limitations relate to measurement of sexual orientation and gender identity. For example, NHANES asks respondents whether they are heterosexual or straight, homosexual or gay/lesbian, bisexual, something else, or not sure, but does not query about other sexual orientations such as pansexual or asexual. Similarly, the only options for gender were male or female, precluding examination of transgender, non-binary, genderqueer, or other gender identities. A large portion of the respondents who were queried about their sexual orientation had missing data on this variable (e.g. missingness = 24.1%), even though the NHANES sexual orientation was asked privately by audio-enhanced, computer-assisted self-interviewing.⁵² It is possible that people who did not report sexual orientation are more likely to be GLB or another sexual orientation. These individuals may have differential diet quality than those who reported sexual orientation information, but it is not clear how their inclusion in analyses would influence the observed associations. Finally, the sample size was small for GLB individuals, so investigators had to combine gay/lesbian and bisexual into one group, preventing them from analyzing potential differences between gay/lesbian individuals and bisexual individuals. Smaller sample sizes of GLB individuals may have also contributed to the lack of statistically significant differences between groups for some outcomes that showed relatively large absolute differences, as other studies with larger sample sizes of GLB individuals found more statistically significant differences.^100,101 Future studies should oversample GLB individuals to provide a more detailed picture of dietary quality for these groups. More research will clarify how dietary quality may differ across sexual minorities, and how dietary differences may interact with other social factors to influence disease risk.

Conclusions

This study revealed some differences in dietary quality between gay/bisexual males and heterosexual males but no differences in dietary quality between lesbian/bisexual females and heterosexual females. Specifically, gay and bisexual males had lower percent energy intake from processed meat and had healthier overall dietary quality compared to heterosexual males. Lesbian and bisexual females had lower percent energy intake from breakfast cereals and mixed dishes than heterosexual females, but did not differ from heterosexual females in overall dietary quality. Future studies should oversample GLB individuals to improve understanding of differences in dietary intake and quality by sexual orientation and gender.

Table 7 online only.

Daily energy intake from foods, and beverages by sex and sexual orientation, NHANES 2011–2016.

	Males						Females
	Heterosexual (n=4,230)		Gay and Bisexual (n=173)		Differencea		Heterosexual (n=4,165)		Lesbian and Bisexual (n=283)		Differencea
	Adjusted Meanb	SEb	Adjusted Meanb	SEb	Diff.	pc	Adjusted Meanb	SEb	Adjusted Meanb	SEb	Diff.	pc
Total Energy Intake (kcals)	2607	22	2672	106	65	0.56	1928	16	2032	67	104	0.12
Foods
Breads and grains	175	6	160	18	−15	0.41	126	5	130	12	4	0.75
Breakfast cereals	56	3	78	18	23	0.22	51	3	37	8	−13	0.16
Cheese and yogurt	71	4	70	15	−2	0.92	60	3	60	9	0	0.99
Desserts and sweet	272	9	289	53	17	0.75	223	7	250	25	27	0.29
Snacks
Fruits	50	2	64	9	13	0.18	50	2	54	6	5	0.48
Nuts, legumes, and other non-meat proteins	139	7	164	19	25	0.18	108	5	113	17	5	0.79
Red and processed meat/poultry/seafood	159	7	116	20	−43	0.06	85	4	96	13	10	0.44
Lean and unprocessed meat/poultry/seafood	154	8	177	33	22	0.51	107	5	104	16	−3	0.86
Mixed dishes	475	14	408	47	−68	0.17	361	10	311	27	−50	0.07
Salty snacks	164	5	152	22	−12	0.63	128	5	143	18	15	0.43
Sauces/dips/condiments	76	3	118	25	42	0.09	68	3	80	12	11	0.35
Vegetables	32	2	40	7	8	0.30	33	1	39	6	5	0.39
Hamburgers and pizza	184	11	170	41	−13	0.74	90	6	134	30	44	0.13
Sandwiches	77	6	46	15	−32	0.06	53	4	43	12	−10	0.40
Beverages
100% Juice	34	4	26	7	−8	0.15	22	2	23	9	1	0.89
Alcohol	137	6	195	34	58	0.08	120	6	144	18	25	0.20
Milk and other dairy	69	4	66	16	−3	0.83	40	3	35	10	−5	0.61
SSBs	164	7	174	25	10	0.68	100	4	116	15	16	0.26
Other beverages	23	3	40	17	17	0.32	24	3	37	10	13	0.18
Food away from home (meals/week)
Fast food and pizza	2.4	0.07	2.1	0.42	−0.3	0.43	1.7	0.07	1.8	0.20	0.1	0.71

^aDifference is difference in adjusted means, gay/lesbian/bisexual (GLB) – heterosexual.

^bMeans and SEs adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁶⁹ alcohol consumption, and physical activity^81,82). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^cP-values were calculated using ordinary least squares regression.

Table 8 online only.

Percent daily energy intake from foods and beverages by sex and sexual orientation, NHANES 2011–2016.

	Males						Females
	Heterosexual (n=4,230)		Gay and Bisexual (n=173)		Differencea		Heterosexual (n=4,165)		Lesbian and Bisexual (n=283)		Differencea
	Adjusted Mean %b	SE b	Adjusted Mean %b	SE b	Diff.	pc	Adjusted Mean % b	SE b	Adjusted Mean % b	SE b	Diff.	pc
Foods
Breads and grains	7.0	0.2	6.2	0.7	−0.8	0.28	6.8	0.3	7.1	0.7	0.2	0.72
Breakfast cereals	2.3	0.1	2.9	0.7	0.6	0.40	2.7	0.2	1.9	0.3	−0.9	0.04
Cheese and yogurt	2.8	0.1	2.6	0.5	−0.2	0.69	3.4	0.2	3.1	0.4	−0.3	0.45
Desserts and sweet snacks	9.7	0.3	10.2	1.6	0.4	0.80	11.1	0.3	11.4	0.9	0.3	0.75
Fruits	2.2	0.1	2.8	0.5	0.7	0.22	2.9	0.1	3.0	0.4	0.1	0.80
Nuts, legumes, and other non-meat proteins	5.2	0.2	6.3	0.8	1.1	0.18	5.7	0.3	5.4	0.7	−0.3	0.69
Red and processed meat/poultry/seafood	6.3	0.3	4.1	0.8	−2.2	0.01	4.5	0.2	5.0	0.7	0.5	0.51
Lean and unprocessed meat/poultry/seafood	6.2	0.3	6.8	1.1	0.6	0.62	5.7	0.3	5.6	0.8	−0.1	0.92
Mixed dishes	18.4	0.5	15.9	2.1	−2.5	0.23	18.4	0.5	15.0	1.4	−3.4	0.02
Salty snacks	6.3	0.2	5.4	0.8	−0.9	0.28	6.6	0.2	6.7	0.8	0.1	0.94
Sauces/dips/condiments	2.9	0.1	4.8	1.2	1.9	0.11	3.6	0.1	3.7	0.6	0.1	0.85
Vegetables	1.4	0.1	1.5	0.3	0.1	0.78	1.9	0.1	2.1	0.3	0.2	0.60
Hamburgers and pizza	6.5	0.4	6.2	1.5	−0.3	0.86	4.5	0.3	7.4	1.5	2.9	0.06
Sandwiches	3.0	0.2	1.6	0.5	−1.4	0.02	2.7	0.2	1.8	0.5	−0.9	0.07
Beverages
100% Juice	1.3	0.1	1.3	0.5	0.0	0.99	1.1	0.1	1.2	0.4	0.1	0.74
Alcohol	4.9	0.2	6.4	0.9	1.5	0.11	5.4	0.2	6.6	0.8	1.2	0.17
Milk and other dairy	2.6	0.2	2.3	0.6	−0.3	0.62	2.1	0.1	1.6	0.4	−0.5	0.20
SSBs	6.3	0.3	6.7	0.8	0.4	0.66	5.2	0.2	5.6	0.6	0.4	0.55
Other beverages	0.9	0.1	1.4	0.6	0.5	0.39	1.2	0.1	1.6	0.4	0.4	0.41

^aDifference is difference in adjusted means, gay/lesbian/bisexual (GLB) – heterosexual.

^bMeans and SEs adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁶⁹ alcohol consumption, and physical activity^81,82). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

^cP-values were calculated using ordinary least squares regression.

Table 9 online only.

Top contributors to total energy intake by sex and sexual orientation, NHANES 2011–2016

	Males				Females
Top energy contributors	Heterosexual (n=4,230)		Gay and Bisexual (n=173)		Heterosexual (n=4,165)		Lesbian and Bisexual (n=283)
	Food Group	Percenta	Food Group	Percenta	Food Group	Percenta	Food Group	Percenta
1st	Mixed dishes	18.4	Mixed dishes	15.9	Mixed dishes	18.4	Mixed dishes	15.0
2nd	Desserts and sweet snacks	9.7	Desserts and sweet snacks	10.2	Desserts and sweet snacks	11.1	Desserts and sweet snacks	11.4
3rd	Bread and grains	7.0	Lean and unprocessed meat/poultry/seafood	6.8	Bread and grains	6.8	Hamburgers and Pizza	7.4
4th	Hamburgers and pizza	6.5	SSBs	6.7	Salty snacks	6.6	Breads and grains	7.1
5th	Red and processed meat/poultry/seafood	6.3	Alcohol	6.4	Lean and unprocessed meat/poultry/seafood	5.7	Salty snacks	6.7

^a%s adjust for key covariates (age, race/ethnicity, education level, nativity, income to poverty ratio, cigarette smoking status, health insurance coverage, mental health,⁶⁹ alcohol consumption, and physical activity^81,82). Analyses applied appropriate sample weighting to provide estimates that are representative of the U.S. civilian non-institutionalized resident population.⁷⁹

Research Snapshot.

Research Question:

Does dietary quality differ by sexual orientation and sex among US adults?

Key Findings:

In this cross-sectional study of 8,851 adults from the 2011–2016 National Health and Nutrition Examination Survey (NHANES), gay/bisexual males had a total Healthy Eating Index-2015 (HEI-2015) score that is 4.14 points higher than heterosexual males. Lesbian/bisexual females did not differ in total or component HEI-2015 scores from heterosexual females.