Disparities in Safety Belt Use by Sexual Orientation Identity Among US High School Students

Abstract

Objectives. We examined associations between adolescents’ safety belt use and sexual orientation identity.

Methods. We pooled data from the 2005 and 2007 Youth Risk Behavior Surveys (n = 26 468 weighted; mean age = 15.9 years; 35.4% White, 24.7% Black, 23.5% Latino, 16.4% other). We compared lesbian and gay (1.2%), bisexual (3.5%), and unsure (2.6%) youths with heterosexuals (92.7%) on a binary indicator of passenger safety belt use. We stratified weighted multivariable logistic regression models by sex and adjusted for survey wave and sampling design.

Results. Overall, 12.6% of high school students reported “rarely” or “never” wearing safety belts. Sexual minority youths had increased odds of reporting nonuse relative to heterosexuals (48% higher for male bisexuals, 85% for lesbians, 46% for female bisexuals, and 51% for female unsure youths; P < .05), after adjustment for demographic (age, race/ethnicity), individual (body mass index, depression, bullying, binge drinking, riding with a drunk driver, academic achievement), and contextual (living in jurisdictions with secondary or primary safety belt laws, percentage below poverty, percentage same-sex households) risk factors.

Conclusions. Public health interventions should address sexual orientation identity disparities in safety belt use.

Motor vehicle accidents are the leading cause of death for US adolescents, accounting for more than 1 in 4 deaths in this age group.¹ More than 3900 youths aged 16 to 20 years were killed and more than 300 000 were injured in motor vehicle accidents in the United States in 2009.² The majority of morbidity and mortality during adolescence is preventable,^3,4 including vehicle-related injuries.⁵ Safety belts are an effective prevention technology that reduce the risk of serious injury or death by 50%.⁶ Evidence-based recommendations from the US Task Force on Community Preventive Services state, “Safety belts are the single most effective means for occupants to reduce the risk of death and serious injury.”^7(p19)

Among all age groups, US adolescents have the lowest rates of safety belt use,^8,9 and nonuse of safety belts is more common for adolescent passengers than drivers.^10,11 Sociodemographic differences in adolescent passenger safety belt use have been observed by gender^9,12,13 and race/ethnicity.^14,15 These differences correspond with gender and racial/ethnic disparities in death, disability, and injury from motor vehicle crashes, with adolescent boys¹⁶ and racial/ethnic minorities¹⁷ bearing the highest burden. Additional individual risk factors for safety belt nonuse in youths include overweight and obesity,¹⁸ alcohol (drinking as well as being a passenger with a drunk driver),¹⁹ depression,²⁰ and lower levels of academic achievement.²¹

Sexual orientation disparities have been documented for sexual minority youths (SMYs; gay/lesbian, bisexual, and unsure) relative to their heterosexual peers across a wide variety of health risk behaviors, including alcohol use, smoking, illicit drug use, sexual health risk behaviors, and unhealthy weight control practices.^22–27 However, to our knowledge, few reports have examined adolescent safety belt use by sexual orientation identity.²⁸ In the United States, data are not available on motor vehicle crashes by sexual orientation identity; thus, it is not known whether sexual orientation health disparities exist in death, disability, and injuries from motor vehicle crashes. Because other sociodemographic differences in safety belt use translate into motor vehicle–related mortality and morbidity, it is likely that understanding differences in safety belt use by sexual orientation identity will provide insight into adverse health outcomes related to motor vehicle accidents.

Previous research comparing safety belt use in jurisdictions with a primary safety belt law (police stop and ticket drivers solely for not wearing a safety belt) to those with a secondary safety belt law (police ticket unbelted drivers only when they are stopped for other reasons, such as speeding) has found that primary laws effectively increase safety belt use even in jurisdictions with relatively high baseline levels of use.^21,29–35 Furthermore, Black–White racial disparities in safety belt use are partly mitigated in states with primary safety belt laws.^21,36 Whether primary laws also mitigate sexual orientation disparities in safety belt use is unknown. In addition, previous research comparing safety belt use between primary and secondary enforcement of safety belt laws focus mainly on drivers, not passengers. This is important because enforcement of the safety belt law may only apply to drivers or may differ between drivers and passengers. Therefore, primary law enforcement may have a greater impact on safety belt use among drivers than passengers.

We investigated the prevalence of passenger safety belt use among US high school students and empirically tested for differences in safety belt use by sexual orientation identity.

METHODS

We analyzed pooled data from the Youth Risk Behavior Survey (YRBS) conducted in 2005 and 2007 from several jurisdictions that included 1 or more measures of sexual orientation. The general approach to pooling and analyzing the data, along with the sexual orientation items and characteristics of the sample by jurisdiction, are described in detail elsewhere in this issue.³⁷ We analyzed data from the 9 jurisdictions that measured sexual orientation identity (Boston, MA; Chicago, IL; Delaware; Maine; Massachusetts; New York City, NY; San Francisco, CA; Vermont; and Rhode Island).

Measures

All measures were self-reported. The measurement and pooling of sexual orientation and race/ethnicity items are described elsewhere in this issue.³⁷ We coded sexual orientation identity as lesbian or gay, bisexual, unsure, or heterosexual. We excluded from analysis those who did not respond to the sexual orientation identity items. We also excluded respondents who did not report age, sex, or race/ethnicity. The final analytic sample consisted of 26 469 (weighted) high school students (n = 56 898 unweighted).

Outcome.

The primary outcome was no safety belt use as a vehicle passenger. Participants were asked, “How often do you wear a seat belt when riding in a car driven by someone else?” (Likert scale responses were 1 = never to 5 = always). We operationalized the primary outcome as no safety belt use as a passenger (“never” or “rarely”) versus safety belt use (“sometimes,” “most of the time,” or “always”).

Independent variable.

Sexual orientation identity was the primary independent variable of interest. We compared youths identifying as lesbian or gay, bisexual, or unsure with heterosexual adolescents.

Sociodemographic covariates.

Individual covariates were age (continuous; range = 12–18 years), sex (male or female), and race/ethnicity (White, Black, Latino, or other).

Individual risk factors.

We adjusted for 5 individual risk factors:

• 1. body mass index (according to age and gender guidelines from the Centers for Disease Control and Prevention; normal weight < 85th percentile; overweight = 85th–94th percentile; obese ≥ 95th percentile; all age and gender standardized),^38,39

• 2. mental health (depression in the past 12 months, yes or no),

• 3. alcohol use (binge drinking on ≥ 3 occasions in the past 30 days, yes or no),

• 4. general risk taking (riding in a car in the past 30 days with a driver who had been drinking alcohol, yes or no), and

• 5. academic achievement (high = A and B grades; low = C, D, and F grades; academic achievement data were not available for Chicago and New York City).

Contextual risk factors.

Safety belt legislation was a contextual risk factor. We identified jurisdictions with secondary safety belt laws in 2005 and 2007 when the YRBS surveys were conducted from a published state-by-state review of safety belt laws.⁴⁰ We compared jurisdictions with secondary safety belt laws (Boston, Massachusetts, Maine, Vermont, and Rhode Island) with those with primary safety belt laws (Chicago, Delaware, New York City, and San Francisco). Additional contextual variables were percentage of the jurisdiction’s population living below poverty and same-sex couples per 1000 households. We obtained data for contextual variables from the 2010 Census.⁴¹

Statistical Analysis

Our general analytic strategy was to compare respondents who reported not wearing safety belts as passengers with those who did. We conducted preliminary analyses to replicate findings of sex differences in safety belt omission.^9,12,13 Male adolescents were at significantly higher risk for safety belt nonuse than were adolescent girls (odds ratio [OR] = 1.25; 95% confidence interval [CI] = 1.13, 1.39; P < .001), after adjustment for sexual orientation identity and individual and contextual covariates. We therefore stratified all subsequent analyses by sex. Differences in safety belt use by sexual orientation identity compared lesbian or gay, bisexual, and unsure youths separately with heterosexual-identified respondents by sex. We investigated disparities in safety belt use by sexual orientation identity, with adjustment for sociodemographics and individual and contextual risk factors.

We conducted statistical analyses in SAS version 9.2 (SAS Institute, Cary, NC), with the program code for YRBS analysis recommended by the Centers for Disease Control and Prevention.⁴² We predetermined statistical significance at the α = 0.05 level. All analyses accounted for the complex sampling design of the YRBS (cluster and strata) and adjusted the relative weights; we altered the effective sample size with design effects calculated for each jurisdiction. The approach to calculating design effects is described in detail elsewhere in this issue.³⁷ We fit weighted multivariable logistic regression models to model no safety belt use (yes or no) as a binary outcome, adjusting for survey wave (2005 vs 2007) and complex survey design. We calculated ORs and 95% CIs for all estimates.

RESULTS

One in 8 (12.6%; 95% CI = 12.0%, 13.3%) high school students reported that they rarely or never wore safety belts when riding in a car driven by someone else. Characteristics of the sample are reported in Table 1. We observed disparities in safety belt use by sexual orientation identity. Figure 1 displays the weighted prevalence of nonuse of safety belts by sex and sexual orientation. Both male and female SMYs had higher prevalence of safety belt nonuse than heterosexuals.

TABLE 1—

Characteristics of Pooled Data Sample: Youth Risk Behavior Surveys, United States, 2005 and 2007

Variable	Weighted % or Mean (SD)
Safety belt use when passengera
Yes	87.36
No	12.64
Sexual orientation identity
Heterosexual	92.72
Lesbian/gay	1.17
Bisexual	3.49
Unsure	2.62
Safety belt law
Primary	65.58
Secondary	34.42
Age, y	15.88 (0.86)
Sex
Female	50.13
Male	49.87
Race/ethnicity
White	35.37
Black	24.70
Latino	23.47
Other	16.46
Jurisdiction
Households below poverty	11.01 (3.53)
Same-sex couples/1000 households	11.36 (4.75)
BMIb
Healthy weight (< 85th percentile)	72.52
Overweight (85th–94th percentile)	15.77
Obese (≥ 95th percentile)	11.71
Depression
No	72.12
Yes	27.88
Binge drinking
No	93.45
Yes	6.55
Passenger of drunk driver
No	78.03
Yes	21.97
Academic achievement,c grades
A’s and B’s	35.35
C’s, D’s, and F’s	21.29
No data	43.36

Note. BMI = body mass index. All results weighted and adjusted for survey design. Sample size was 26 469 weighted and 56 898 unweighted.

^aNo= never or rarely; yes = sometimes, most of the time, or always.

^bAccording to age and gender guidelines from the Centers for Disease Control and Prevention.^38,39

^cChicago, IL, and New York City, NY, did not provide academic achievement data.

FIGURE 1—

Weighted prevalence of passenger safety belt use by gender and sexual orientation adjusted for survey design: Youth Risk Behavior Surveys, United States, 2005 and 2007.

Note. CI = confidence interval; OR = odds ratio. Sexual Minority was defined as lesbian or gay, bisexual, or unsure. Weighted % shown for all prevalence estimates and adjusted for survey design.

Table 2 presents sex-stratified multivariable models among adolescent boys. Sexual orientation identity disparities in safety belt use are shown after adjustment for sociodemographics (model 1), individual risk factors (model 2), and contextual variables (model 3). Male bisexuals had 48% increased odds of nonuse of safety belts relative to male heterosexuals (model 3). Table 3 presents sex-stratified multivariable models among adolescent girls. Female SMYs had 46% to 85% higher odds than female heterosexuals of safety belt nonuse (model 3; all, P < .05). We also replicated previous research findings on differences in safety belt use by other sociodemographic, individual, and contextual risk factors, with adjustment for sexual orientation identity. Racial/ethnic disparities were especially prominent and consistent for male and female adolescents.

TABLE 2—

Odds of Passenger Safety Belt Use Among Adolescent Boys by Sexual Orientation Identity: Youth Risk Behavior Surveys, United States, 2005 and 2007

	Model 1a		Model 2b		Model 3c
Variable	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Sexual orientation identity
Heterosexual (Ref)	1.00		1.00		1.00
Lesbian/gay	1.46 (0.92, 2.30)	.107	1.21 (0.74, 1.98)	.446	1.22 (0.73, 2.03)	.451
Bisexual	1.78 (1.23, 2.57)	.002	1.41 (1.00, 2.00)	.051	1.48 (1.05, 2.10)	.026
Unsure	1.53 (1.14, 2.04)	.004	1.29 (0.96, 1.74)	.089	1.33 (0.97, 1.81)	.079
Age	1.01 (0.96, 1.06)	.779	0.95 (0.90, 1.01)	.077	0.94 (0.89, 0.99)	.027
Race/ethnicity
White (Ref)	1.00		1.00		1.00
Black	1.31 (1.13, 1.53)	< .001	1.42 (1.20, 1.67)	< .001	1.81 (1.52, 2.15)	< .001
Latino	1.72 (1.47, 2.01)	< .001	1.70 (1.43, 2.02)	< .001	2.20 (1.83, 2.64)	< .001
Other	1.14 (0.94, 1.37)	.178	1.39 (1.16, 1.67)	< .001	2.23 (1.80, 2.76)	< .001
BMId
Healthy weight (< 85th percentile; Ref)			1.00		1.00
Overweight (85th–94th percentile)	…	…	1.00 (0.85, 1.19)	.978	0.99 (0.83, 1.18)	.904
Obese (≥ 95th percentile)	…	…	1.10 (0.93, 1.29)	.258	1.10 (0.93, 1.29)	.273
Depression
No (Ref)			1.00		1.00
Yes	…	…	1.05 (0.90, 1.22)	.544	1.06 (0.91, 1.24)	.455
Binge drinking
No (Ref)			1.00		1.00
Yes	…	…	2.36 (1.99, 2.80)	< .001	2.34 (1.97, 2.79)	< .001
Passenger of drunk driver
No (Ref)			1.00		1.00
Yes	…	…	2.22 (1.90, 2.59)	< .001	2.23 (1.91, 2.62)	< .001
Academic achievemente
High grades (Ref)			1.00		1.00
Low grades	…	…	2.30 (2.04, 2.60)	< .001	2.35 (2.08, 2.65)	< .001
Safety belt lawf
Primary (Ref)					1.00
Secondary	…	…	…	…	2.52 (2.13, 2.97)	< .001
Jurisdiction
Households below poverty	…	…	…	…	0.99 (0.98, 1.01)	.513
Same-sex couples/1000 households	…	…	…	…	0.98 (0.97, 0.99)	< .001

Note. BMI = body mass index; CI = confidence interval; OR = odds ratio. Each model was adjusted for survey wave (2005 vs 2007). Sample size was 13 199 weighted and 27 811 unweighted.

^aAdjusted for sexual orientation, age, and race/ethnicity.

^bAdjusted for model 1 variables plus individual risk factors.

^cAdjusted for model 2 variables plus contextual risk factors.

^dAccording to age and gender guidelines from the Centers for Disease Control and Prevention.^38,39

^eChicago, IL, and New York City, NY, did not provide academic achievement data.

^fJurisdictions with secondary laws (Boston, Massachusetts, Maine, Vermont, Rhode Island) were compared with those with primary laws (Chicago, Delaware, New York City, San Francisco).

TABLE 3—

Odds of Passenger Safety Belt Use Among Adolescent Girls by Sexual Orientation Identity: Youth Risk Behavior Surveys, United States, 2005 and 2007

	Model 1a		Model 2b		Model 3c
Variable	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Sexual orientation identity
Heterosexual (Ref)	1.00		1.00		1.00
Lesbian/gay	2.03 (1.30, 3.16)	.002	1.78 (1.14, 2.77)	.011	1.85 (1.19, 2.88)	.006
Bisexual	1.82 (1.49, 2.22)	< .001	1.46 (1.18, 1.80)	< .001	1.46 (1.18, 1.82)	< .001
Unsure	1.64 (1.22, 2.21)	.001	1.45 (1.06, 1.99)	.02	1.51 (1.10, 2.09)	.012
Age	0.92 (0.86, 0.98)	.008	0.92 (0.86, 0.97)	.005	0.91 (0.86, 0.97)	.002
Race/ethnicity
White (Ref)	1.00		1.00		1.00
Black	1.60 (1.33, 1.92)	< .001	1.30 (1.08, 1.58)	.007	1.58 (1.30, 1.93)	< .001
Latino	2.39 (1.95, 2.93)	< .001	1.79 (1.45, 2.21)	< .001	2.15 (1.73, 2.67)	< .001
Other	1.21 (1.05, 1.39)	.009	1.58 (1.29, 1.93)	< .001	2.24 (1.79, 2.80)	< .001
BMId
Healthy weight (< 85th percentile; Ref)			1.00		1.00
Overweight (85th–94th percentile)	…	…	1.02 (0.82, 1.26)	.888	1.01 (0.81, 1.25)	.966
Obese (≥ 95th percentile)	…	…	1.10 (0.88, 1.39)	.406	1.07 (0.85, 1.35)	.572
Depression
No (Ref)			1.00		1.00
Yes	…	…	1.19 (1.03, 1.38)	.016	1.20 (1.03, 1.38)	.017
Binge drinking
No (Ref)			1.00		1.00
Yes	…	…	1.80 (1.50, 2.17)	< .001	1.78 (1.48, 2.16)	< .001
Passenger of drunk driver
No (Ref)			1.00		1.00
Yes	…	…	1.62 (1.39, 1.89)	< .001	1.66 (1.42, 1.94)	< .001
Academic achievemente
High grades (Ref)			1.00		1.00
Low grades	…	…	2.40 (2.06, 2.80)	< .001	2.51 (2.14, 2.93)	< .001
Safety belt lawf
Primary (Ref)					1.00
Secondary	…	…	…	…	2.83 (2.24, 3.58)	< .001
Jurisdiction
Below poverty, %	…	…	…	…	0.99 (0.97, 1.01)	.486
Same-sex couples/1000 households	…	…	…	…	1.00 (0.99, 1.01)	.889

Note. BMI = body mass index; CI = confidence interval; OR = odds ratio. Each model was adjusted for survey wave (2005 vs 2007). Sample size was 13 270 weighted and 29 087 unweighted.

^aAdjusted for sexual orientation, age, and race/ethnicity.

^bAdjusted for model 1 variables plus individual risk factors.

^cAdjusted for model 2 variables plus contextual risk factors.

^dAccording to age and gender guidelines from the Centers for Disease Control and Prevention.^38,39

^eChicago, IL, and New York City, NY, did not provide academic achievement data.

^fJurisdictions with secondary laws (Boston, MA; Massachusetts; Maine; Vermont; Rhode Island) were compared with those with primary laws (Chicago, IL; Delaware; New York City, NY; San Francisco, CA).

DISCUSSION

One in 8 public high school students who participated in the YRBS in jurisdictions that collected sexual orientation identity self-reported not wearing safety belts as passengers in motor vehicles. We found disparities in safety belt nonuse by sexual orientation identity for both male and female adolescents. Male SMYs (bisexuals vs heterosexuals) and female SMYs (lesbian, bisexual, and unsure vs heterosexuals) had increased odds of safety belt nonuse, even after adjustment for individual and contextual covariates and risk factors.

Sexual Orientation Disparities

Why might SMYs be less likely than their heterosexual peers to use safety belts? Several pathways may link a sexual minority orientation with higher rates of safety belt nonuse, including conformity to gender norms, minority stress, and personality characteristics. First, health-related behaviors are often linked with notions of masculinity or femininity and can be a means of demonstrating gender.⁴³ Safety belt nonuse is a risk behavior, and risk taking in general is associated with masculinity and rejection of feminine role norms.^43,44 Conformity or nonconformity to masculine or feminine role norms may shape sex and sexual orientation differences in safety belt use.

Our findings are consistent with gender conformity–nonconformity pathways. Although we did not include measures of gender expression—the manifestation of characteristics in one’s personality, appearance, and behavior that are culturally defined as masculine or feminine²⁷—we used sex as a proxy for gender, finding precedent for doing so in epidemiological studies.⁴⁵ Male adolescents in general are less likely to use safety belts than are female adolescents. The elevated rates of safety belt nonuse in male SMYs may be a function of performing dominant heterosexual masculine norms or compensatory hypermasculinities (i.e., macho health behaviors).^46–48 Although previous research has shown that sexual orientation covaries with gender nonconformity,^49–52 and male SMYs score higher on femininity indices,⁵³ compensation is expected in the context of adolescent peer relationships. Adolescent boys are exposed to continual enforcement of strict gender-specific behavioral norms from peers, adults, and media.⁵⁴ It is possible that engaging in health behaviors such as safety belt nonuse is a means of outwardly conforming to traditional masculine norms of nonchalance or fearlessness, particularly for male SMYs. Studies have linked masculine gender role conformity to poorer health behaviors among sexual minority men,^55,56 providing some support for this explanation. Female SMYs may be less likely to use safety belts because they might be less likely to conform to traditional feminine norms.^50,57 Previous research found that sports involvement, a proxy for some gender-nonconforming behaviors, is associated with lower rates of safety belt use in adolescent girls.⁵⁸

The hypothesis that conformity to gender norms for male SMYs and nonconformity to gender norms for female SMYs may partly explain disparities in safety belt use by sexual orientation cannot be empirically tested with available US surveillance system data. However, the national YRBS added socially assigned gender expression as an optional question for the next survey year. We encourage jurisdictions to use this measure, because it will allow for investigation of socially assigned gender conformity across an array of preventive health behaviors.

A second possible explanation for the higher levels of safety belt nonuse among SMYs may be found in minority stress theory.⁵⁹ This theory posits that sexual minorities are exposed to excess psychosocial stress attributable to sexual orientation–related stigma, which can lead to an increased burden of negative mental health outcomes.⁵⁹ Although adolescent populations in general have relatively high levels of psychological distress,⁶⁰ evidence shows that SMYs may be especially vulnerable,⁶¹ particularly in the context of experiences of abuse and peer victimization.⁶² Moreover, psychological distress in youths—including depression, anxiety, and self-harm behaviors—has been associated with risky driving behaviors, an important correlate of safety belt nonuse,^20,63–65 offering preliminary support for this explanation.

Third, sensation-seeking personality traits may be relevant. Risky driving behaviors have been strongly linked to sensation seeking and reward sensitivity.^66,67 Some research suggests higher levels of sensation seeking among sexual minority than heterosexual women,⁶⁸ although evidence specific to adolescents remains limited. In our analyses, being a passenger in a car with a drunk driver, included as a marker of risk taking, was significantly associated with less safety belt use in adolescent boys and girls; however, adjustment for this factor did not substantially attenuate the ORs for safety belt nonuse by sexual orientation. Further exploration of risk-taking and sensation-seeking characteristics may be needed and relevant for interventions.

Safety belt nonuse among sexual minorities is an unexplored area of research. Many additional potentially contributing factors merit attention, including the role of sex differences in safety belt use by driver and passenger positions¹⁰ and parental safety belt use and involvement.^15,67,69 Because normative perceptions of peers have been shown to contribute to safety belt use,⁷⁰ investigation into differences in norms in sexual minority communities may be warranted. SMYs often yearn to connect with sexual minority peers and allies to access community and support^71,72 and may rely heavily on driving long distances alone or with peers, particularly in areas that are sparsely populated with sexual minorities.⁷³ Riding with peers⁶⁷ and driving alone⁸ are also factors associated with safety belt nonuse that need further attention in the exploration of sexual orientation disparities.

Socioenvironmental context is an important determinant of individual health behaviors.⁷⁴ We found a higher risk of safety belt nonuse in jurisdictions with secondary rather than primary safety belt laws, replicating previous research findings.^29,31–34 Primary legislative enforcement may be an effective strategy for reducing some, but not all, population-level disparities in motor vehicle occupant fatality rates by narrowing the gap in safety belt use among different sociodemographic groups, including gaps by sexual orientation.¹⁷ More broadly, findings suggest that contextual factors, including the characteristics of jurisdictions in which people live, influence preventive health behaviors differentially by sexual orientation.

Limitations

Selection bias might have occurred. Not all high schools took part in the YRBS, not all of the students at surveyed high schools participated, and not all jurisdictions captured sexual orientation identity. The jurisdictions that assessed sexual orientation identity may have been more liberal that those that did not; therefore, results are only generalizable to those particular jurisdictions. In addition, the sample was not representative of primary and secondary safety belt enforcement states (e.g., only 1 jurisdiction—San Francisco—was located on the West Coast). Only youths who attended public high schools took part in the survey. SMYs may represent a high percentage of dropouts and thus may have been underrepresented in our data.

Safety belt use was a self-reported measure. Studies comparing observed with self-reported safety belt use have found that motorists overreport safety belt use by 5% to 20%.⁷⁵ Thus, the actual prevalence of safety belt nonuse among adolescents may have been even higher than the estimates we obtained. The extent of differential safety belt nonuse overreporting by sexual identity status could not be determined. In addition, awareness of safety belt laws may be higher in primary enforcement law states; thus, we cannot rule out the possibility that self-reported safety belt use was overestimated to different extents among different subpopulations.

Omitted variable bias also represents a limitation. We did not assess car variables. In particular, passenger seat position represents an important potential confounder, because the passenger safety belt use question did not distinguish between safety belt use in the front versus the rear seat. Also pertinent might be the number of passengers, which could limit who gets a safety belt, and type of driver (i.e., peer or parent), because relationships with important referent others may influence adolescent norms surrounding safety belt use.

Residual confounding was possible. In particular, individual socioeconomic status was unavailable from the YRBS. Additional research into the influence of these and other demographic and socioenvironmental factors are needed. For example, acculturation has been linked to safety belt use⁷⁶ but has never been examined alongside sexual orientation identity. Finally, YRBS data were cross-sectional, and causality between safety belt use, sexual identity, and sociodemographic characteristics could not be inferred.

Conclusions

Our study demonstrated that sexual orientation represents an important risk factor for safety belt nonuse among adolescents and especially among female SMYs. In light of the disproportionate rates of safety belt nonuse among SMYs that we documented, it is possible that SMYs and sexual minority adults are overrepresented in the adverse effects of motor vehicle crashes. Clinical implications of our analysis suggest the need to integrate safety belt prevention messages into primary care,⁷⁷ particularly in states with secondary safety belt laws. Clinical preventive service guidelines,⁷⁸ such as those from the American Academy of Pediatrics,^79,80 American Medical Association,⁸¹ American Academy of Family Physicians,⁸² and Maternal and Child Health Bureau,⁸³ recommend that providers screen and counsel adolescents on a range of preventive health-promoting behaviors during well visits,⁸⁴ including safety belt use. However, national data from the 2001 to 2004 Medical Expenditure Panel Survey suggest that only 31% of adolescents aged 10 to 17 years are being counseled about safety belts.⁸⁵ Because repeated contacts with a primary care provider may occur over several years, clinicians have opportunities to screen for risky health behaviors such as safety belt nonuse and counsel their adolescent patients.⁸⁶

Interventions targeting socialization processes, with attention to gender and sexual orientation norms, also offer a future direction for research and intervention. For example, a nationally representative telephone survey of American adolescents aged 14 to 17 years found that frequency of exposure to media influenced positive normative perceptions about safety belt nonuse for boys but not for girls.⁷⁰ Future research would benefit from examining media exposure frequency in relation to safety belt norms by gender and sexual orientation, because findings have implications for intervention development, design, and delivery (e.g., media messaging).

Consistent with traditional epidemiological insights⁸⁷ and with conceptual frameworks developed to intervene on racial disparities in safety belt use among adolescents,⁸⁸ a mix of population-level and high-risk approaches are likely needed to address disparities in safety belt use by sexual orientation. These should include strategies that target the general population (e.g., public policy), approaches that focus on adolescents and the systems and structures that surround them (e.g., parents and families, schools, doctors), and interventions to improve health and well-being of SMYs specifically.