Sexual identity differences in health behaviors and weight status among urban high school students

Abstract

Although racial/ethnic disparities in childhood obesity are well documented in the United States (U.S.), fewer studies have investigated elevated body mass index (BMI) and related health behaviors among sexual minority youth (SMY; gay/lesbian, bisexual, not sure). We examined pooled data from the 2009-2017 Youth Risk Behavior Surveys, which included high school students from 12 urban U.S. school districts. We used sex-stratified logistic regression models to estimate the association of sexual identity with health behaviors and elevated BMI (reference = heterosexuals). A total of 133,615 participants were included. Sexual minority boys were more likely to report physical inactivity than heterosexual boys: (gay AOR 2.41, 95% CI= 1.90-3.20; bisexual AOR 1.71, 95% CI= 1.26-2.16; not sure AOR 1.28, 95% CI= 1.05-1.55). Gay (AOR 1.81, 95% CI= 1.55-2.12) and not sure (AOR 1.18, 95% CI= 1.01-1.37) boys were also less likely to consume the recommended daily intake of fruit. Bisexual girls were more likely than heterosexual girls to report watching television ≥ 3 hours on a school day (AOR 1.15, 95% CI= 1.05-1.26) and to consume sugar-sweetened beverages (AOR 1.30, 95% CI= 1.18-1.43). All SMY reported higher rates of current tobacco use than their heterosexual peers. Sexual minority girls (lesbian AOR 1.53, 95% CI= 1.21-1.96; bisexual AOR 1.49, 95% CI= 1.31-1.68; not sure AOR 1.70, 95% CI= 1.43-2.02) and bisexual boys (AOR 1.31, 95% CI= 1.08-1.58) had significantly higher rates of obesity than their heterosexual counterparts. These findings can inform tailored health promotion initiatives to reduce obesity risk in SMY.

Introduction

Childhood and adolescent obesity are major public health concerns in the United States (U.S.). Since the 1970s, the prevalence of obesity among children and adolescents in the U.S. has more than tripled.¹ Among these developmental groups, adolescents (ages 12-19) have the highest prevalence (20.5%) of obesity.¹ Being obese during adolescence is associated with high blood pressure² and type 2 diabetes,³ among other health conditions. Adolescents who are obese also experience significant bullying from peers⁴ and are more likely to report being depressed.^5,6 Further, adolescent obesity is linked with obesity in adulthood,^7,8 which is in turn associated with chronic conditions, including cardiovascular disease and cancer.⁹

Based on the ecological systems theory, the socioecological model has been recommended as a framework to study multi-level determinants of childhood obesity. This model postulates that an individual’s health is influenced by personal factors (e.g., age, gender, and genetics) and the individual’s interactions with community, environment, and policy contexts in which they live.¹⁰ Although genetics play a role in obesity risk,^11,12 a greater proportion of individual variation in BMI can be attributed to health behaviors (e.g., diet, physical activity, sedentary behaviors, and sleep),^13-15 social networks, and environmental risk factors (e.g., the built environment).^16,17 Modifiable risk behaviors such as sedentary activity (e.g., watching television and computer use for more than two hours per day) are associated with elevated BMI in youth.^18,19 Recent data suggest that children with higher intake of fruit in childhood have lower BMI.¹⁵ Consumption of sugar-sweetened beverages in childhood and adulthood is associated with weight gain,^20,21 and increased risk of metabolic syndrome across the lifespan.²² Further, although current tobacco use may reduce risk of obesity, findings from several studies suggest that smoking during adolescence is associated with obesity in adulthood.^23-25 Similarly, young adults (ages 20-24) who report higher alcohol consumption in adolescence are more likely to be overweight than those who consumed lower levels of alcohol in adolescence.^26,27 Despite recognition of the important role of individual, community, environmental, and policy determinants of childhood obesity,¹³ most research on childhood obesity has examined individual-level risk factors. A scoping review of 423 studies found that approximately 60% of studies addressed interpersonal factors but less than 25% addressed community, environmental, and policy level factors.²⁸ Social determinants that influence obesity in adolescents include family food choices and weight status of parents, peer influences, and school nutrition options.^15,17,29 Lower SES neighborhoods are more likely to be food swamps in which residents lack access to healthy food options as the number of convenience stores and fast and junk food outlets outnumber those of healthier foods.³⁰

Significant racial/ethnic disparities in childhood obesity exist in the U.S with Latino (25.8%) and Black (22.0%) youth more likely than their White (14.1%) counterparts to be obese.³¹ Fewer studies have investigated differences in elevated BMI among sexual minority youth (SMY; e.g., gay/lesbian, bisexual, and not sure). Based on limited data among youth it appears that sexual minority girls in the U.S. have higher rates of obesity than their heterosexual counterparts.^32,34,35 From childhood to late adolescence, BMI in sexual minority girls increases, whereas BMI in gay boys BMI decreases.³⁵ Although gay boys appear to be similar to heterosexual boys in terms of body weight, they are less likely to be physically active.³⁶ In an examination of BMI trajectories from adolescence to adulthood (11 years to 34 years), lesbian women were found to have a nearly a two-fold risk of developing obesity after accounting for known risk factors for obesity.³⁷ This is important as many health behaviors that influence obesity start in adolescence and continue into adulthood.³⁸

Adverse experiences in childhood (e.g., physical abuse, sexual abuse, neglect, and peer victimization) are higher among sexual minorities,^39-41 and have been found to be associated with childhood obesity in the general population.^42,43 Multiple studies have found a dose-response relationship between exposure to childhood adversity and risk of obesity.^44,45 There is growing evidence of the link between childhood adversity and negative health outcomes among SMY, including higher rates of poor mental health.^46-48 In the very limited research examining the associations between childhood adversity and elevated BMI in SMY, peer victimization³³ and childhood abuse⁴⁹ have only partially explained the elevated BMI observed among sexual minority girls.

Our goal for the present study was to address identified gaps in the empirical literature. In particular, we were interested in examining whether peer victimization explained sexual identity differences in health behaviors and weight status among SMY. Further, examining the age at which obesity disparities arise is important for designing targeted prevention strategies and identifying risk factors for obesity among sexual minorities across the lifespan. We used pooled data from the 2009-2017 Youth Risk Behavior Survey (YRBS) to assess sexual identity differences in health behaviors and weight status among urban youth by comparing three groups of SMY (i.e., gay/lesbian, bisexual, not sure) to heterosexual youth. Based on previous evidence, we hypothesized that: 1) sexual minority identity would remain an independent risk factor for negative health behaviors and elevated BMI, and 2) peer victimization would attenuate the association of sexual minority identity with health behaviors and elevated BMI in youth.

Methods

Sample and Procedures

The Centers for Disease Control and Prevention (CDC) developed the Youth Risk Behavior Surveillance System (YRBSS) to monitor health behaviors that contribute to morbidity and mortality among youth. Since 1991, the YRBSS has conducted a school-based survey, the YRBS, biennially.⁵⁰ The YRBS includes national, state, territorial, tribal government, and local school-based surveys of representative samples of high school students in the U.S.⁵¹ The local YRBS uses a two-stage sampling design to achieve a representative sample of public high school students in large urban school districts or counties and, in some areas, students in other types of public schools.

Prior to conducting the local YRBS, parental permission was obtained based on local procedures. Students’ participation in the YRBS was voluntary and anonymous. Local urban school districts were permitted to add or delete questions to the national YRBS standard questionnaire, which served as a starting survey template. Students completed the self-administered questionnaire on a computer-scannable booklet or answer sheet during one required class period. YRBS methodology is described in detail elsewhere.⁵²

We used data from the local 2009-2017 YRBS, which included 12 large urban school districts located in eight states. Jurisdictions that included a question to assess sexual identity in 2009 were in three districts, five in 2011, nine in 2013, seven in 2015, and eight in 2017 across eight states (Table 1). The overall response rate ranged from 61-90% across jurisdictions included in the present study. Although YRBS data are collected from human subjects, the data are de-identified and publicly available; therefore this study was deemed exempt by the Columbia University Irving Medical Center Institutional Review Board.

Table 1.

Summary of large urban school districts in the 2009-2017 YRBS that included an item on sexual identity.

Years	Number of Districts	Locations
2009	3	Chicago, Illinois New York, New York Seattle, Washington
2011	5	Chicago, Illinois Milwaukee, Wisconsin New York, New York San Diego, California Seattle, Washington
2013	9	Broward, Duval, and Orange Counties in Florida Charlotte-Mecklenburg County, North Carolina Chicago, Illinois Milwaukee, Wisconsin New York, New York San Diego, California Seattle, Washington
2015	7	Broward, Duval, Miami-Dade, and Orange Counties in Florida Fort Worth, Texas New York, New York San Diego, California
2017	8	Broward, Duval, Miami-Dade, and Orange Counties in Florida Chicago, Illinois Fort Worth, Texas New York, New York San Diego, California

Measures

Sexual identity.

Sexual identity was assessed with the following item: “Which of the following best describes you?” “Heterosexual (straight),” “Gay or lesbian,” “Bisexual,” or “Not sure.” We included participants who indicated that they were not sure of their sexual identity because adolescents and young adults in this group have significantly higher rates of tobacco use,^53,54 physical inactivity,^55,56 and peer victimization^53,56,57 than heterosexual youth.

Demographics.

Race/ethnicity (White, Black, Latino, Asian, or other race) and age (12 years or younger, 13 years old, 14 years old, 15 years old, 16 years old, 17 years old, 18 years or older) were assessed. Due to the small sample size of participants ages 12 years or younger (n = 658 [0.4%]) we combined participants in this category with those who were 13 years old (n = 2,652 [2.1%]).

Peer victimization.

We assessed peer victimization in three ways. We assessed two forms of peer victimization in the previous 12 months, including having been: 1) bullied or 2) threatened or injured with a weapon on school property. In addition, participants were asked: “During the past 30 days, on how many days did you not go to school because you felt you would be unsafe at school or on your way to or from school?” Responses were dichotomized for all forms of peer victimization as 0 = No or 1 = Yes.

Health behaviors.

Physical activity was measured using physical activity guidelines for youth (0 = less than seven days of physical activity during the past week and 1 = at least 60 minutes of any physical activity daily during the past week).⁵⁸ Sedentary behaviors on school days included: 1) number of hours participant played video or computer games or used a computer for something other than schoolwork, and 2) number of hours participant watched television. Responses were dichotomized as 0 = less than three hours per day and 1 = three or more hours based on established recommendations.^59,60 Fruit intake was coded as 0 = less than two servings per day and 1 = two or more servings per day based on dietary recommendations for youth.⁶¹ Consumption of sugar-sweetened beverages was assessed with the following item: “During the past 7 days, how many times did you drink a can, bottle, or glass of soda or pop, such as Coke, Pepsi, or Sprite? (Do not count diet soda or pop).” Responses were dichotomized as 0 = zero in past 7 days and 1 = one or more in past 7 days based on dietary recommendations for youth.⁶¹ Cigarette use and alcohol use during the past 30 days were assessed with the following items: 1) “During the past 30 days, on how many days did you smoke cigarettes?” and 2) “During the past 30 days, on how many days did you have at least one drink of alcohol?” Responses were dichotomized 0 = zero days and 1 = one or more days. No one study has assessed the validity of all self-reported behaviors that are included in the YRBS; however, the health behavior questions are based on guidelines for youth.⁵⁰

Weight status.

BMI was calculated from self-reported height and weight using sex- and age- specific reference data from the CDC growth charts. Participants with a BMI between the 85^th and 95^th percentile were categorized as overweight and those with a BMI ≥ 95 percentile were categorized as obese.⁶³ Self-reported height and weight in the YRBS have been deemed to be reliable.⁶⁴

Statistical Analyses

All analyses were conducted in Stata version 15 (StataCorp, LLC). Data were weighted according to YRBS analytic guidelines. A survey weight based on participants’ sex, race/ethnicity, and grade was used to adjust for school and student nonresponse. Weighted data are representative of all public high school students in the included jurisdictions. The 2009-2017 local YRBS included 190,080 participants. We excluded cases with missing data for sexual identity (n = 41,833) and BMI (n = 14,632). Multiple imputation with chained equations was used to impute missing values for covariates. A total of 20 imputations were run. For bivariate analyses, we used the Rao-Scott Chi-square test to assess sexual identity differences across study variables. A significance of p < 0.05 was pre-determined for all analyses. We then conducted sex-stratified multiple logistic regression to obtain unadjusted and adjusted odds ratios with 95% confidence intervals to examine sexual identity differences in health behaviors and weight status (i.e., overweight and obesity). We ran three different regression models. Model 1 was unadjusted and showed the bivariate odds ratios for sexual identity differences in health behaviors and weight status. Models for health behaviors and weight status were adjusted for demographic variables, survey year, and site (Model 2) and peer victimization (Model 3). Models for weight status further added adjustment for health behaviors (Model 3).

Results

The final analytic sample included 133,615 participants. Among the 63,393 boys: 57,487 (90.6%) identified as heterosexual, 1,644 (2.6%) as gay, 1,779 (2.7%) as bisexual, and 2,483 (4.1%) were not sure of their sexual identity (Table 2). All groups of sexual minority boys were separately compared to heterosexual boys in all analyses. Gay boys were older than heterosexual boys (p = 0.01), whereas not sure boys were younger (p = 0.01). Gay boys (p < 0.001) and bisexual boys (p = 0.02) were more likely to identify as Latino compared to heterosexual boys. Not sure boys were more likely to identify as Asian relative to heterosexual boys (p < 0.001). All groups of sexual minority boys reported higher rates of all forms of peer victimization than heterosexual boys.

Table 2.

Sexual identity differences across sample characteristics among boys in the YRBS, 2009-2017 (n=63,393).

	Heterosexual (n=57,487)	Gay (n=1,644)		Bisexual (n=1,779)		Not sure (n=2,483)
Demographics	n (weighted %)	n (weighted %)	p-value Gay vs. Heterosexual	n (weighted %)	p-value Bisexual vs. Heterosexual	n (weighted %)	p-value Not sure vs. Heterosexual
Age (n= 63,393)			0.01		0.10		0.01
13 years old or younger	940 (1.9)	46 (2.7)		47 (2.5)		80 (3.5)
14 years old	9,852 (18.7)	234 (15.2)		261 (14.3)		462 (19.3)
15 years old	14,176 (25.9)	335 (22.1)		439 (28.4)		543 (23.2)
16 years old	14,552 (23.9)	442 (25.7)		456 (23.4)		584 (22.9)
17 years old	13,073 (21.1)	385 (21.0)		417 (22.7)		548 (20.1)
18 years or older	4,894 (8.4)	202 (13.3)		159 (8.7)		266 (11.0)
Race/ethnicity (n=61,145)			<0.001		0.02		<0.001
White	9,661 (18.0)	221 (15.4)		237 (14.0)		354 (16.3)
Black	13,773 (30.4)	403 (29.0)		361 (27.6)		492 (25.5)
Latino	22,825 (36.4)	701 (44.3)		802 (42.0)		957 (35.8)
Asian	5,619 (12.6)	111 (7.8)		153 (12.6)		356 (19.5)
Other race	3,671 (2.6)	134 (3.5)		133 (3.8)		181 (2.9)
Peer Victimization
Bullied on school property in the past 12 months (n=62,033)	6,292 (11.0)	390 (24.5)	<0.001	429 (24.7)	<0.001	493 (20.6)	<0.001
Threatened/injured with a weapon on school property (n=63,007)	4,114 (7.0)	302 (19.9)	<0.001	336 (19.5)	<0.001	428 (17.7)	<0.001
Missed school because felt unsafe (n=62,754)	3,838 (6.7)	329 (20.3)	<0.001	308 (15.6)	<0.001	391 (15.5)	<0.001

Note. Heterosexual boys (largest group) served as the reference group in all comparisons. The Rao-Scott Chi-square test was used to assess sexual identity differences. Boldface denotes statistical significance < 0.05.

The sample of girls was larger (n = 70,222) and included: 56,576 (80.3%) heterosexual, 1,816 (2.5%) lesbian, 7,788 (11.0%) bisexual, and 4,042 (6.2%) not sure girls (Table 3). Compared to heterosexual girls, not sure girls were younger (p = 0.05) and lesbian girls were older (p < 0.001). Lesbian and bisexual girls were also more likely than heterosexual girls to identify as Black or Latina. Not sure girls were more likely to be Asian relative to heterosexual girls (p = 0.01). Compared to their heterosexual counterparts, all groups of sexual minority girls reported higher rates of all forms of peer victimization.

Table 3.

Sexual identity differences across sample characteristics among girls in the YRBS, 2009-2017 (n=70,222).

	Heterosexual (n=56,576)	Lesbian (n=1,816)		Bisexual (n=7,788)		Not sure (n=4,042)
Demographics	n (weighted %)	n (weighted %)	p-value Lesbian vs. Heterosexual	n (weighted %)	p-value Bisexual vs. Heterosexual	n (weighted %)	p-value Not sure vs. Heterosexual
Age (n=63,393)			<0.001		0.62		0.05
13 years old or younger	1,001 (2.1)	19 (1.0)		162 (2.1)		118 (3.0)
14 years old	10,302 (19.2)	221 (12.7)		1,314 (18.6)		836 (21.8)
15 years old	14,587 (25.7)	415 (24.5)		415 (27.6)		1,085 (25.5)
16 years old	14,284 (24.4)	501 (26.1)		501 (24.2)		964 (24.1)
17 years old	12,565 (21.3)	481 (23.9)		481 (20.2)		748 (18.8)
18 years or older	3,837 (7.3)	179 (11.8)		179 (7.3)		291 (6.8)
Race/ethnicity (n=68,283)			<0.001		<0.001		0.01
White	9,170 (16.6)	233 (12.6)		894 (12.2)		528 (15.2)
Black	14,201 (31.8)	576 (37.8)		2,011 (34.2)		872 (27.8)
Latina	22,383 (36.0)	765 (41.2)		3,746 (44.6)		1,672 (37.4)
Asian	5,674 (12.9)	63 (5.8)		352 (6.1)		530 (16.8)
Other race	3,643 (2.7)	109 (2.6)		566 (2.9)		295 (2.8)
Peer Victimization
Bullied on school property in the past 12 months (n=69,341)	7,371 (13.1)	307 (17.1)	<0.001	1,635 (21.5)	<0.001	819 (21.1)	<0.001
Threatened/injured with a weapon on school property (n=69,911)	2,133 (3.8)	209 (11.1)	<0.001	647 (7.8)	<0.001	293 (7.9)	<0.001
Missed school because felt unsafe (n=69,673)	4,331 (7.3)	245 (11.5)	<0.001	895 (11.3)	<0.001	462 (11.0)	<0.001

Note. Heterosexual girls (largest group) served as the reference group in all comparisons. The Rao-Scott Chi-square test was used to assess sexual identity differences. Boldface denotes statistical significance < 0.05.

Sexual identity differences in health behaviors are presented in Table 4. All groups of sexual minority boys reported higher rates of physical inactivity and higher rates of current tobacco use relative to heterosexual boys. Gay and not sure boys reported lower rates of playing computer/video games ≥ 3 hours on an average school day, whereas only gay boys reported lower rates consumption of sugar-sweetened beverages in the past week, compared to heterosexual boys. Gay and bisexual boys were also less likely to watch ≥ 3 hours of television on an average school day. In addition, gay and not sure boys were more likely than heterosexual boys to report low consumption of fruit. Gay and bisexual boys were more likely than their heterosexual counterparts to report current alcohol use. The addition of peer victimization in Model 3 attenuated the association of sexual identity with several health behaviors among boys. For instance, although not sure boys had higher rates of current alcohol use relative to heterosexual boys in Model 2 (AOR 1.26, 95% CI= 1.06-1.50), this difference was attenuated after accounting for peer victimization in Model 3 (AOR 1.03, 95% CI= 0.86-1.24).

Table 4.

Sexual identity differences in health behaviors among youth in the YRBS, 2009-2017 (n=133,615).

	Boys (n=63,393)			Girls (n=70,222)
Health behaviors	Model 1 OR (95% CI)	Model 2 AOR (95% CI)	Model 3 AOR (95% CI)	Model 1 OR (95% CI)	Model 2 AOR (95% CI)	Model 3 AOR (95% CI)
Physical activity on < 7 days in past week
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	2.54 (1.99-3.23)	2.52 (1.98-3.20)	2.41 (1.90-3.05)	0.76 (0.61-0.94)	0.74 (0.60-0.92)	0.75 (0.61-0.92)
Bisexual	1.71 (l.31-2.24)	1.71 (1.30-2.23)	1.65 (1.26-2.16)	0.91 (0.80-1.04)	0.92 (0.81-1.04)	0.92 (0.81-1.05)
Not sure	1.31 (1.09-1.57)	1.31 (1.09-1.59)	1.28 (1.05-1.55)	1.15 (0.96-1.37)	1.16 (0.97-1.39)	1.16 (0.97-1.39)
Played computer/video games ≥ 3 hours per day on average school day
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	0.65 (0.55-0.77)	0.66 (0.56-0.79)	0.67 (0.56-0.80)	1.08 (0.92-1.28)	1.12 (0.94-1.33)	1.12 (0.94-1.32)
Bisexual	1.03 (0.87-1.23)	1.02 (0.86-1.22)	1.04 (0.87-1.24)	1.42 (1.30-1.56)	1.43 (1.30-1.57)	1.42 (1.24-1.49)
Not sure	0.89 (0.78-1.02)	0.84 (0.73-0.97)	0.86 (0.75-0.98)	1.21 (1.06-1.39)	1.17 (1.02-1.34)	1.17 (1.02-1.34)
Watched television ≥ 3 hours per day on average school day
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	0.73 (0.59-0.90)	0.70 (0.58-0.88)	0.71 (0.57-0.90)	0.95 (0.79-1.15)	0.94 (0.77-1.15)	0.94 (0.76-1.15)
Bisexual	0.73 (0.59-0.89)	0.74 (0.60-0.91)	0.75 (0.61-0.93)	1.17 (1.06-1.28)	1.15 (1.05-1.26)	1.15 (1.05-1.26)
Not sure	0.84 (0.71-0.99)	1.04 (0.87-1.24)	1.05 (0.88-1.26)	0.75 (0.65-0.86)	0.87 (0.76-1.01)	0.87 (0.76-1.01)
Fruit intake < 2 times per day
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	1.80 (1.54-2.11)	1.81 (1.55-2.12)	1.81 (1.55-2.13)	0.92 (0.79-1.08)	0.90 (0.77-1.05)	0.92 (0.78-1.08)
Bisexual	1.18 (0.96-1.46)	1.17 (0.95-1.44)	1.17 (0.95-1.45)	0.87 (0.79-0.94)	0.85 (0.78-0.93)	0.86 (0.79-0.94)
Not sure	1.26 (1.08-1.46)	1.18 (1.01-1.37)	1.18 (1.01-1.38)	1.14 (1.01-1.30)	1.08 (0.95-1.23)	1.08 (0.95-1.25)
Drank a sugar-sweetened beverage in the past week
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	0.74 (0.58-0.94)	0.72 (0.57-0.92)	0.70 (0.55-0.89)	1.09 (0.91-1.31)	1.08 (0.91-1.30)	1.05 (0.88-1.26)
Bisexual	0.93 (0.76-1.14)	0.94 (0.77-1.14)	0.91 (0.74-1.12)	1.37 (1.24-1.50)	1.33 (1.21-1.46)	1.30 (1.18-1.43)
Not sure	0.82 (0.70-0.96)	0.90 (0.77-1.04)	0.87 (0.75-1.02)	0.87 (0.75-1.01)	0.96 (0.83-1.11)	0.94 (0.82-1.09)
Tobacco use in past 30 days
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	2.58 (1.99-3.35)	2.52 (1.92-3.30)	1.71 (1.28-2.29)	2.16 (1.70-2.75)	2.41 (1.89-3.07)	2.08 (1.62-2.66)
Bisexual	2.80 (2.17-3.62)	2.98 (2.23-3.97)	2.18 (1.61-2.95)	2.61 (2.28-3.00)	3.06 (2.66-3.50)	2.72 (2.38-3.12)
Not sure	1.97 (1.59-2.44)	2.35 (1.88-2.94)	1.69 (1.32-2.15)	1.67 (1.37-2.02)	2.09 (1.71-2.56)	1.80 (1.47-2.20)
Alcohol use in past 30 days
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	1.75 (1.45-2.11)	1.65 (1.35-2.02)	1.34 (1.09-1.66)	1.76 (1.52-2.02)	1.70 (1.47-1.96)	1.57 (1.36-1.80)
Bisexual	1.72 (1.43-2.06)	1.81 (1.48-2.21)	1.51 (1.23-1.84)	1.86 (1.69-2.05)	1.92 (1.74-2.11)	1.82 (1.64-2.01)
Not sure	1.06 (0.89-1.26)	1.26 (1.06-1.50)	1.03 (0.86-1.24)	1.01 (0.88-1.16)	1.18 (1.03-1.35)	1.10 (0.96-1.27)

Note. Analyses were sex-stratified. Model 1 = unadjusted; Model 2= added demographic characteristics; Model 3= added peer victimization. Sex-stratified multiple logistic regression was used to calculate odds ratios. Boldface denotes statistical significance < 0.05.

Relative to heterosexual girls, those who identified as lesbian were less likely to be physically inactive. Bisexual and not sure girls were more likely to play computer/video games, whereas only bisexual girls reported higher rates of watching television ≥ 3 hours on an average school day. Bisexual girls were also more likely than their heterosexual peers to consume sugar-sweetened beverages in the past week. All groups of sexual minority girls reported higher rates of current tobacco use than heterosexual girls. Lesbian and bisexual girls also reported higher rates of current alcohol use relative to heterosexual girls. The addition of peer victimization in Model 3 attenuated the association of sexual identity with several health behaviors among girls. Although not sure girls also had higher rates of current alcohol use relative to heterosexual girls in Model 2 (AOR 1.18, 95% CI= 1.03-1.35), this difference was attenuated after accounting for peer victimization in Model 3 (AOR 1.10, 95% CI= 0.96-1.27).

Results for sexual identity differences in weight status are presented in Table 5. Not sure boys were more likely to be overweight in Model 2 (AOR 1.22, 95% CI= 1.01-1.49), but this difference was attenuated after accounting for health behaviors and peer victimization in Model 3 (AOR 1.21, 95% CI= 0.99-1.46). Bisexual boys had higher odds of obesity relative to heterosexual boys. Although some of the associations of sexual identity with overweight and obesity were attenuated after adjustment for health behaviors and peer victimization, all groups of sexual minority girls were significantly more likely to be overweight and obese relative to heterosexual girls.

Table 5.

Sexual identity differences in weight status among youth in the YRBS, 2009-2017 (n=133,615).

	Boys (n=63,393)			Girls (n=70,222)
Weight status	Model 1 OR (95% CI)	Model 2 AOR (95% CI)	Model 3 AOR (95% CI)	Model 1 OR (95% CI)	Model 2 AOR (95% CI)	Model 3 AOR (95% CI)
Overweight
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	0.98 (0.74-1.28)	0.96 (0.73-1.26)	0.94 (0.72-1.24)	1.37 (1.16-1.63)	1.27 (1.07-1.52)	1.26 (1.06-1.51)
Bisexual	1.25 (1.01-1.54)	1.23 (0.99-1.52)	1.21 (0.98-1.50)	1.22 (1.10-1.36)	1.14 (1.02-1.25)	1.13 (1.01-1.25)
Not sure	1.20 (0.99-1.45)	1.22 (1.01-1.49)	1.21 (0.99-1.46)	1.22 (1.07-1.40)	1.25 (1.09-1.42)	1.24 (1.09-1.41)
Obesity
Heterosexual	Reference	Reference	Reference	Reference	Reference	Reference
Gay/Lesbian	1.37 (1.07-1.76)	1.25 (0.98-1.60)	1.23 (0.96-1.58)	1.72 (1.35-2.19)	1.54 (1.21-1.94)	1.53 (1.21-1.96)
Bisexual	1.40 (1.16-1.70)	1.32 (1.09-1.60)	1.31 (1.08-1.58)	1.68 (1.49-1.90)	1.54 (1.36-1.74)	1.49 (1.31-1.68)
Not sure	1.09 (0.90-1.31)	1.06 (0.88-1.29)	1.05 (0.87-1.27)	1.72 (1.46-2.04)	1.76 (1.49-2.09)	1.70 (1.43-2.02)

Note. Analyses were sex-stratified. Model 1 = unadjusted; Model 2= added demographic characteristics; Model 3= added peer victimization and health behaviors. Sex-stratified multiple logistic regression was used to calculate odds ratios. Boldface denotes statistical significance < 0.05.

Discussion

We used 2009-2017 data from the local YRBS to examine sexual identity differences in health behaviors and elevated BMI in a racially/ethnically diverse sample of urban youth. We found significant disparities among SMY for several health behaviors. In addition, relative to heterosexual girls, all groups of sexual minority girls were more likely to be overweight or obese. Bisexual boys had higher rates of obesity than heterosexual boys. These disparities in weight status were only partially explained by experiences of peer victimization. Our findings have important implications for future research and can inform the development of health promotion initiatives aimed at reducing health disparities among SMY.

Findings for physical activity and sedentary behaviors were inconsistent. All groups of sexual minority boys had higher rates of physical inactivity, which is consistent with previous evidence.^36,65 These differences in physical activity may be driven by less frequent participation in team sports among sexual minority boys than their heterosexual counterparts.³⁶ Studies have found that sexual minority male college students report higher levels of body dissatisfaction,^66,67 which may contribute to the higher rates of physical inactivity.⁶⁸ Although previous work identified no sexual identity differences in physical activity among girls,⁶⁵ we found that lesbian girls were more likely than heterosexual girls to meet physical activity recommendations.

Moreover, we found lower rates of sedentary behaviors among gay, bisexual, and not sure boys. This is in contrast with findings from a recent study by Beach and colleagues⁶⁵ that found no sexual identity differences in sedentary behaviors among boys. Although the dichotomous measures used in the present study have been associated with elevated BMI in adolescents^59,60 and are based on recommendations for youth, Beach and colleague’s used a continuous measure of sedentary behaviors (total number of hours watching television and using a computer for something other than schoolwork per day). This might explain the observed differences in findings. Bisexual and not sure girls in our study reported no difference in physical activity but higher rates of sedentary behaviors compared to heterosexual girls, a novel finding that should be further explored in future studies as sedentary behaviors may increase risk of obesity and related health conditions.

Increasing school-based physical activity, through participation in team sports and physical education classes, could be one method to decrease risk of obesity among SMY. However, prior research has found that more than half of SMY who participated in a physical education class were bullied or harassed due to their sexual orientation or gender identity.⁶⁹ Schools should promote safe environments for SMY students by enacting inclusive policies that prohibit bullying on the basis of sexual orientation or gender identity anywhere on school property, but specifically relevant to physical activity, on athletics fields and locker rooms. Coaches and physical education teachers should be provided with adequate resources and professional development opportunities to learn how they can support SMY students and create climates of acceptance and non-discrimination. These and other efforts to reduce risk of peer victimization would likely lead to increased participation in physical activity.

Future research is needed that includes more robust measures of physical activity and sedentary behaviors among SMY. Evidence to date has relied on self-reported measures, which may not provide accurate estimates of sexual identity differences in physical activity among youth. Future research should aim to collect more robust physical activity data possibly through the use of actigraphy monitoring.⁷⁰

We found fewer sexual identity differences in diet. Compared to their heterosexual counterparts, we identified lower odds of adequate fruit intake among gay and not sure boys and lower odds of consumption of sugar-sweetened beverages among gay boys, but in the opposite direction for bisexual girls. Although sexual minority girls and boys appear to have higher rates of disordered eating, including use of diet pills/laxatives and purging, than their heterosexual peers,^71-74 data on disordered eating were not available in the 2009-2017 local YRBS. There is a need for additional research that uses comprehensive assessments to examine diet quality and behaviors among SMY. Such data may help understand the underlying causes of sexual-orientation-related disparities in obesity.

Consistent with previous evidence,^75,76 we found that all groups of sexual minority boys and girls had higher rates of current tobacco use compared to their heterosexual peers. Gay/lesbian and bisexual youth also showed higher rates of current alcohol use than their heterosexual peers, which is consistent with previous work.^77,78 This is particularly important given that alcohol and tobacco use increase risk of elevated BMI later in life.²⁶

Our finding of elevated BMI in sexual minority girls and bisexual boys is supported by previous research using self-reported BMI in youth.^33,34,65 Further, the association between sexual identity and elevated BMI remained significant even after accounting for established risk factors—a finding supported by previous work.^33,49 In one of the few studies to employ objective measures of BMI in youth, researchers used data from the National Longitudinal Study of Adolescent Health (Add Health) and found that while gay boys had smaller increases in BMI over time than heterosexual boys, bisexual girls had greater increases than heterosexual girls.⁷⁹ Further, analyses of Add Health data suggest that underreporting of BMI is more pronounced in gay males which can further exaggerate differences in weight status between them and their heterosexual peers.⁸⁰ Research is needed to better understand the correlations between self-reported and objective measures of BMI, and whether these differ based on gender and sexual identity.

Although the National Academy of Medicine identified the development of interventions to reduce health disparities among sexual minorities as a priority research area,⁸² limited research has focused on health promotion strategies in this population. Studies are needed that test tailored preventive interventions focusing on health behaviors associated with elevated BMI in SMY. Pediatric healthcare providers, public health practitioners, and school administrators should be educated about the unique health risks experienced by SMY, such as experiences of peer victimization, that can contribute to negative health behaviors and increased risk of obesity.^83-85 A systematic review of 46 studies found that school-based health centers (SBHCs) are associated with lower levels of health disparities and have positive effects on academic performance and health among youth.⁸⁶ Therefore, SBHCs may be ideal settings to deliver health promotion interventions for SMY. However, a recent survey of 132 SBHCs found that less than 50% of SBHC healthcare providers were trained on sexual minority health and less than 25% of SBHCs included preferred pronouns on intake forms.⁸⁷ These data suggest that to improve the health of SMY youth, SBHCs should create more affirming and protective environments for SMY including training healthcare providers and staff about unique health concerns of SMY, updating intake forms to collect information on preferred pronouns, sexual orientation, and gender identity, and developing health materials that are inclusive of SMY. Such efforts would likely increase use of SBHCs by SMY and may improve the health of this vulnerable group.

Protective school climates (e.g., those with inclusion gay-straight alliances and anti-bullying policies) have been identified as potential upstream drivers of health in SMY.^85,88,89 Previous analyses of YRBS data have shown that protective school climates are associated with lower rates of suicidal ideation⁹⁰ and binge drinking in SMY.⁹¹ However, to date, there is limited research on how protective school climates impact health behaviors (e.g., physical activity and diet) and weight status in SMY. Future research that examines how school policies to promote inclusion of SMY influence weight status in SMY. In relation to the socioecological model, structural interventions that alter school environments to be more inclusive of SMY should be implemented.

Limitations and Future Research

Findings should be evaluated with several limitations in mind. Given that YRBS data are cross-sectional, we were unable to assess temporality or causality. Additional research employing prospective designs is needed to elucidate the links between sexual identity, peer victimization, health behaviors, and elevated BMI. The YRBS data used in the present study are only representative of high school students in large urban areas located in eight states. As such, our findings cannot be generalized to youth living in other states or in suburban or rural areas. However, we chose to use the local YRBS because the sample of sexual minority participants is larger than in the national-level YRBS (which has only included questions about sexual identity since 2015).

The YRBS does not assess family, neighborhood or environmental factors that might be associated with obesity. Family risk factors known to influence adolescent obesity include family diet and exercise patterns, as well as parental weight status.⁹² Based on evidence of the significant contributions of interpersonal,⁹² neighborhood,⁹³ environment,^16,17 and policy⁹⁴ factors on childhood obesity in the general population, we recommend that future research employ multi-level approaches to examine obesity in SMY. We were unable to include transgender youth in our analyses because only the latest year of YRBS assessed gender identity. Thus, the sample size of transgender youth in the YRBS is quite limited. In addition, given the limited assessment of diet in the local YRBS, our analyses included only two measures of diet (intake of fruit and sugar-sweetened beverages). Despite higher rates of disordered eating in SMY,^71-74 we were unable to examine disordered eating behaviors that may influence weight status. Finally, as data were not available for important forms of childhood adversity across all years of the YRBS, we focused on school-based peer victimization. Future work should investigate the associations among other forms of adversity, health behaviors, and BMI in SMY.

Conclusions

Using a representative sample of urban high school students across 12 jurisdictions in eight states in the U.S., we identified statistically significant differences in health behaviors and weight status in sexual minority and heterosexual youth. Our findings have important implications for the development of school-based health promotion initiatives focused on reducing obesity risk. These findings highlight the need for future research that examines adverse experiences and whether/how they contribute to elevated risk of obesity among some groups of SMY.