Article Text
Abstract
Background Research documents substantial adolescent health disparities by sexual orientation, but studies are confined to a small number of countries—chiefly the USA. We provide first-time evidence of associations between sexual orientation and adolescent health/well-being in a new country—Australia. We also add to knowledge by examining health/well-being outcomes not previously analysed in national samples, considering adolescents reporting no sexual attractions, and rank-ordering sexual-orientation health disparities by magnitude.
Methods Data from an Australian national probability sample of 14/15 years old (Longitudinal Study of Australian Children, n=3318) and regression models adjusted for confounding and for multiple comparisons were used to examine the associations between sexual attraction and 30 outcomes spanning multiple domains of health/well-being—including socio-emotional functioning, health-related quality of life, depressive symptoms, health-related behaviours, social support, self-harm, suicidality, victimisation, self-concept, school belonging and global health/well-being assessments.
Results Lesbian, gay, bisexual and questioning adolescents displayed significantly worse health/well-being than their heterosexual peers in all outcomes (p<0.05). The magnitude of the disparities ranged between 0.13 and 0.75 SD, and was largest in the domains of self-harm, suicidality, peer problems and emotional problems. There were fewer differences between the heterosexual and no-attraction groups. Worse outcomes were observed among both-sex-attracted adolescents compared with same-sex-attracted adolescents, and sexual-minority girls compared with sexual-minority boys.
Conclusions Consistent with the minority stress model and recent international scholarship, sexual-minority status is an important risk factor for poor adolescent health/well-being across domains in Australia. Interventions aimed at addressing sexual-orientation health disparities within Australian adolescent populations are urgently required.
- adolescence
- sexual orientation
- health disparities
- well-being
- Australia
Statistics from Altmetric.com
Background
Sexual-minority status is a recognised marker of risk in adult populations, with sexual minorities experiencing poorer mental and physical health and being more likely to exhibit risky health behaviours.1–3 These disparities have been attributed to minority stress, or exposure to unique stressors associated with belonging to a sexual minority within a heteronormative society.4 5 Generally, less is known about the associations between sexual orientation and health outcomes among adolescents—with some exceptions.6 Adolescence is a time of vulnerability characterised by dramatic biological, cognitive and socio-emotional change and the onset of health risk behaviours.7 This vulnerability is heightened among sexual-minority youth, who navigate developmental processes within compulsory social contexts which are often hostile to them.8 It is therefore crucial to generate high-quality evidence on the comparative health/well-being of sexual minorities during this life-course stage.
Data that permit robust identification of health disparities by sexual orientation have only recently become available, disproportionately in the USA.9 Early US studies relied on Add Health (1994–2002),10 and more recent ones on the 2005–2017 Youth Risk Behavior Surveys (YRBS). These reveal higher rates of substance use, suicidality, risky sexual behaviours and victimisation among sexual-minority youth.11–17 Research in other countries is limited. Recent examples include studies considering multiple indicators of health/well-being in a New Zealand national sample18 and a provincially representative Canadian sample,19 both of which identified consistently worse outcomes among sexual-minority than heterosexual youth. Yet in recent meta-analyses two-thirds of studies came from the USA.20–22 Results from these meta-analyses indicate that sexual-minority adolescents are 2–3 times more likely than their heterosexual peers to report substance use,20 depressive symptoms,21 and suicide attempts,22 with smaller effect sizes in the USA than other countries.20 22 Just one Australian study was included in one of these meta-analyses, and this only considered differences in drug injecting and binge drinking using data from 1997.23
This study provides the first overview of sexual-orientation health disparities within Australia, using a large, probability sample of adolescents aged 14/15 years (n∽3,300). Based on the findings from earlier international studies, we hypothesise that Australian sexual-minority adolescents will experience poorer outcomes across health/well-being domains than their heterosexual peers. Recent research on adult3 and adolescent18 samples provides only limited evidence of gender differences, but where differences are found (eg, substance use) sexual-minority women perform worse than sexual-minority men.16 20 24 As such, we also expect to observe greater disparities by sexual orientation among female than male youth in the Australian data. Consistent with earlier studies of adolescents15 19 20 and adults,1 3 we also anticipate that Australian both-sex-attracted youth will report worse health/well-being than same-sex-attracted youth. Further, our data enable us to consider youth who report having no sexual attractions as a distinct analytic category. To our knowledge, only one earlier study examined this group, reporting no health disparities relative to heterosexual adolescents.25 However, health differences between asexual and heterosexual populations favouring the latter have been identified in adult samples.26 Hence, we expect Australian adolescents with no attractions to fare worse than their heterosexual peers.
Although these theoretical expectations are informed by findings from international studies, country-level factors may shape the health/well-being of sexual minorities.27 Some such factors differ in complex ways between Australia (the focus of this study) and the USA (where most previous research was conducted). Australia has more protective anti-discrimination provisions, greater public support of lesbian, gay and bisexual (LGB) groups, and a more equitable healthcare system,28 29 whereas the USA has a wider suite of school-based interventions (or Gay-Straight Alliances) aimed at supporting sexual-minority youth.30 Hence, it remains unclear whether adolescent health/well-being disparities by sexual orientation will be more pronounced in Australia than reported for the USA.
Altogether, this study adds to the emerging literature on adolescent health/well-being disparities by sexual orientation by providing novel estimates from a new country context (Australia), considering youth who report no sexual attractions, and examining outcomes not—or rarely—considered in analyses of national samples. Further, it innovates by rank-ordering an encompassing set of health/well-being measures by effect size. This approach can generate valuable insights into the pervasiveness of sexual-orientation disparities across domains of health/well-being, and point to areas requiring priority intervention.
Methods
Sample
This research was approved by The University of Queensland ethics committee and conformed to the principles embodied in the Declaration of Helsinki. It used data from The Longitudinal Study of Australian Children (LSAC), a biennial birth-cohort study which since 2004 collects information on Australian children and their families. The LSAC sample was identified using complex probabilistic methods, and is largely representative of two cohorts of Australian children born March 1999 to February 2000 (‘K’ cohort) and March 2003 to February 2004 (‘B’ cohort).31 We used information from the sixth study wave (2014) from the ‘K’ cohort (ages 14/15), when a sexual-orientation question was included. The analytic sample comprises 3233–3312 adolescents, depending on missing data on the health/well-being measures.
Measures
Sexual orientation
We measure sexual orientation using an indicator of sexual attraction. As explained in online supplementary appendix 1, this is a commonly used and developmentally appropriate measure of sexual orientation for samples of adolescents. Information on sexual-attraction was collected via audio computer-assisted self-interview using a question asking: ‘Which of these statements best describes your sexual feelings at this time in your life?’. Response options included: ‘I’m attracted only to girls’, ‘I’m attracted only to boys’, ‘I’m attracted to girls and boys’, ‘I’m not sure who I am attracted to’ and ‘I don’t feel any attraction to others’. Following recent practice,11 32 we combined adolescents attracted to members of the same sex (n=26), members of the same and opposite sex (n=105) or unsure about their attractions (n=100) into a lesbian, gay, bisexual or questioning (LGBQ) category (n=231; 6.96%). The other categories encompassed adolescents expressing attraction to members of the opposite sex only (n=2950; 88.91%) and adolescents expressing no attractions (n=137; 4.13%). Adolescents with missing data on the sexual-attraction variable (n=219) were excluded. ANOVA and χ2 tests indicated that these adolescents did not differ to those with available data in gender, age, number of siblings or language background, but were significantly less likely to live with their mothers (94.4% vs 97.4%) or fathers (73.8% vs 82.2%). Including adolescents with missing data as a separate analytic category did not change the key conclusions drawn (online supplementary table 1).
Supplemental material
Health and well-being
We exploited the richness of the LSAC data to derive a wide range of measures tapping different dimensions of health/well-being—including socio-emotional functioning, health-related quality of life, depressive symptoms, health-related behaviours, social support, self-harm, suicidality, victimisation, self-concept, school belonging, and global health/well-being assessments. These encompass all health/well-being indicators available in the survey that have been—or could be—theoretically linked to sexual-minority status within the minority stress framework. Some of these are routinely included in international studies using national samples (eg, suicidality, depression), whereas others have only been considered occasionally (eg, self-concept, life satisfaction, self-rated general health). For others (sleep quantity and quality, physical functioning, hyperactivity and conduct problems), this is—to our knowledge—the first study to provide evidence from a national sample. Table 1 details all health/well-being variables and their properties. To compare and rank-order the predicted sexual-minority effects, these variables were reverse coded as necessary so that higher scores always indicate better health/well-being and expressed in SD.
Analytic approach
We modelled all outcomes using ordinary least squares regression. The statistical significance of model parameters was determined using two-tailed tests and a conventional threshold (α =0.05). Because many significance tests were implemented, we also report p-values adjusted for multiple comparisons using the Holm method.33 Using standardised outcomes required linear models being fitted also to the binary outcomes, resulting in linear probability models. Results from logit models were consistent (online supplementary table 2).
All models were adjusted for potential confounding by factors known or suspected to be correlated with adolescents’ self-reported sexual orientation and their health/well-being, and included as controls in previous studies.11 15 32 34 These included: gender (male/female), age (in months), language spoken at home, as a proxy for cultural background (English/other), presence of father (yes/no) and mother (yes/no) at home, and number of siblings (zero/one/two/three or more). Table 2 shows descriptive statistics on all variables. ANOVA and chi-squared tests revealed statistically significant differences by adolescent sexual orientation in gender, age, father’s presence, and number of siblings (p<0.05), but not language background or mother’s presence. Exclusion of the last two variables from the models did not change the pattern of results. Compared with heterosexual adolescents, LGBQ adolescents were more likely to be female (70.1% vs 47%) and an only child (20.8% vs 10.8%), and less likely to have a present father (74.9% vs 82.6%).
Results
Results from the main regression models are presented as an abridged table of coefficients (table 3), and graphically for a better visualisation of the rank-ordering of sexual-minority effects (figure 1, left panel). Full sets of estimates are shown in online supplementary table 3. In all models, the reference group for the sexual-attraction variable was ‘heterosexual’.
The estimated coefficient on the LGBQ variable was negative and statistically significant (p<0.05) in all 30 models. When considering the rank-ordering of the sexual-minority effects across outcomes, the LGBQ coefficients were largest for suicide thoughts (β=−0.75), self-harm thoughts (β=−0.70), suicide plans (β=−0.69), peer problems (β=−0.69), overall socio-emotional functioning (β=−0.69), emotional problems (β=−0.62), self-harm (β=−0.58), general health (CHU9D) (β=−0.54) and suicide attempts (β=−0.51). In contrast, they were less pronounced for pro-social behaviour (β=−0.13), physical functioning (β=−0.17), marijuana use (β=−0.20), sleep quantity (β=−0.21), victimisation (β=−0.22), school functioning (β=−0.22), alcohol use (β=−0.23) and smoking (β=−0.27). All 30 sexual-minority coefficients remained statistically significant (p<0.05) with adjustments for multiple comparisons.
Compared with heterosexual adolescents, those with no attractions exhibited significantly worse scores (p<0.05) in five of the health/well-being outcomes and significantly better scores in five outcomes. With adjustments for multiple comparisons, however, only the negative coefficients on support from friends and peer problems remained statistically significant.
Analyses in which the LGBQ group was split into its three constituent subgroups are presented in table 4. Compared with heterosexual adolescents, those attracted to members of both sexes were significantly (p<0.05) disadvantaged in 29 outcomes, those attracted only to members of the same-sex in 17 outcomes, and those unsure about their attractions in 13 outcomes. With adjustments for multiple comparisons, eight coefficients for the same-sex-attracted group and six for the unsure group became statistically insignificant (p≥0.05).
Models interacting sex and sexual attraction were fitted to compare the outcomes of LGBQ male and female adolescents (figure 1, right panel and online supplementary table 4). Gender differences were statistically significant (p<0.05) in four outcomes (smoking, suicide attempts, self-harm and conduct problems). In all cases, LGBQ boys displayed better scores than LGBQ girls. None of these parameters remained statistically significant (p≥0.05) with adjustments for multiple comparisons.
Discussion
Investigating the early roots of sexual-orientation health disparities in representative samples of adolescents is an important endeavour, yet contemporary evidence is largely restricted to the US. This study constitutes a significant addition to the international literature: it generated evidence for a new country (Australia) using a national probability sample, it examined and rank-ordered a comprehensive set of high-quality health/well-being measures, and considered adolescents with no attractions as a distinct category.
Our analyses provided clear evidence that sexual-minority status is an important marker of risk for poor adolescent health/well-being in contemporary Australia. LGBQ adolescents displayed worse scores than their heterosexual peers in all outcomes under consideration, and the magnitude of the disparities was often large—above 50% of a SD in nearly a third of cases. As a point of reference, the estimated disadvantage associated with LGBQ status was larger than that associated with father absence, a critical social determinant of child and adolescent health35 in 25 of 30 outcomes. Of note, evidence of LGBQ disadvantage was also apparent for those outcomes available in our data, but rarely—or never—analysed in national samples (eg, sleep, self-rated general health, physical functioning, life satisfaction, self-concept, hyperactivity, conduct problems or social support).
Being able to analyse a wide set of outcomes enabled us to generate a unique, global overview of Australian adolescents’ health/well-being disparities by sexual orientation. This established that the disadvantage experienced by LGBQ youth was not confined to specific outcomes or domains, but cut across the whole spectrum. This constitutes strong evidence that a global strategy to tackling LGBQ health/well-being disadvantage is required. Nevertheless, in the context of finite public resources that can be deployed to such avail, our rank-ordering of indicators by effect magnitude pointed to areas demanding priority intervention—such as self-harm, suicidality and peer/emotional problems. Critically, we were able to accomplish this comparative exercise using a single dataset and a consistent method—reducing the likelihood that any observed differences are driven by study design.
Consistent with earlier research in New Zealand,25 fewer differences were observed between heterosexual adolescents and those reporting no sexual attractions. However, the no-attraction group exhibited poorer outcomes in several indicators of social well-being. These disparities may be indicative that this group contains some asexual individuals—who have been found to experience poorer interpersonal outcomes in adult populations26—and/or youth who are relatively delayed in their social and sexual development. The fact that little evidence of health/well-being disadvantage was found for the no-attraction group also serves to highlight the unique stressors and risks to which LGBQ youth are exposed.
Although all subsamples of LGBQ adolescents experienced health disadvantage relative to the heterosexual group, adolescents attracted to members of both sexes appeared most disadvantaged. This aligns with previous findings identifying the poorest health/well-being among bisexual adolescents16 20 and adults.1 3 The mechanisms producing health/well-being disadvantage among both-sex-attracted youth remain poorly understood, but may resemble those proposed for older bisexuals—for example, the downplaying or rejection of bisexuality as a legitimate sexual orientation and lack of social support from both the heterosexual majority and the lesbian/gay community.36Additional analyses revealed that LGBQ girls experienced larger health disparities than LGBQ boys in some domains—as previously reported for adolescent16 and adult1 3 populations. Consistent with US research24 37 the strongest effect was for smoking. These findings resonate with intersectional approaches to health, which highlight the importance of considering combinations of disadvantaged statuses in structuring health behaviours and outcomes. Future research on adolescent health should thus be mindful of the intersections between gender and sexual orientation, as well as subgroup heterogeneity within the LGBQ group.
Altogether, the results in this Australian study were highly consistent with those from recent US studies and the larger international literature.11 14 17 As an illustrative example, the covariate-adjusted standardised mean difference in depressive symptoms between LGBQ and heterosexual youth reported here was 0.46, sitting in the middle of the range for the international studies included in a recent meta-analysis (0.17–0.81) and being very close to the analogous figure for the US studies included in that report.21 The high degree of consistency between our findings and those obtained from US samples may suggest that institutional divergences between the countries are not large enough to make a substantial difference, or that their positive/negative features offset each other. Cross-national, comparative research on adolescent populations is required to better understand the role of macro-level factors in producing sexual-orientation health/well-being disparities—mirroring recent developments in scholarship focusing on adult populations.27
Our findings are largely consistent with the tenets of the minority stress model.4 However, the current formulation of the model is geared towards the experiences of adult populations.8 Our research adds weight to claims that the minority stress model needs to be expanded to better incorporate adolescent experiences. In fact, LGBQ youth in our sample experienced more systematic and pronounced disadvantages than did lesbian, gay and bisexual adults in a recent Australian overview.3 This may occur because sexual-minority youth are exposed to different stressors tied to the social environments in which they spend the most time (eg, schools) and dispose of fewer resources to buffer the impact of stressors (eg, dense social networks or ties to the LGB community).8 It suggests distinct experiences of minority stress in adolescence and adulthood, and—potentially—different underlying mechanisms linking minority stress to health outcomes. Research that expands the minority stress model to better encompass the lived realities of non-heterosexual youth is therefore warranted.
Strengths and limitations
This study had several strengths. First, it relied on recent data from a national probability sample from a new country (Australia). Second, it considered a large number of high-quality outcomes tapping different health/well-being dimensions—including novel ones, such as sleep quality and quantity, hyperactivity and conduct problems. Relatedly, the study was also unique in its ability to rank-order this diverse set of outcomes by effect magnitude, using a single dataset and a consistent method. Third, the models included adolescents with no attractions as a distinct group, were adjusted for potential confounders, and were stratified by sex and LGBQ subgroups. However, some study limitations must be acknowledged. First, LSAC did not include sufficient information to enable analyses of other gender/sexual-minority groups (eg, transgender or gender non-binary adolescents), or other dimensions of sexual orientation (eg, behaviour or identity). Second, similar to other recent studies38 the LGBQ subsamples were modest—and therefore prone to Type-II estimation errors. Third, the available data were cross-sectional and lacked objective health assessments.
Public health implications
Documenting health disparities between LGBQ and heterosexual adolescents is a first step in addressing health inequalities by sexual orientation: it contributes to focusing attention on the issue, facilitating priority shifts in policy and practice, and developing effective and efficient interventions.39 Our findings underscore the importance of sexual-minority status as a risk factor for poor health/well-being among Australian adolescents. The deep and multifaceted disadvantage experienced by sexual-minority youth in our sample, coupled with evidence of the long-term persistence of these inequalities into adulthood, suggests that policies aimed at combating minority stress among young people are urgently required. These have been shown to work best when they operate at multiple levels of the social environment, ranging from community-level interventions aimed at reducing stigma to individual-level interventions aimed at increasing social support.40 For sexual-minority adolescents, programmes that target meso-level contexts to which youth are disproportionately exposed (eg, schools) may be particularly promising.24 Further research in this space is urgently needed.
What is already known on this subject
Extant research shows that adults who belong to a sexual minority experience worse health and well-being than heterosexual adults.
A smaller body of research reports similar results in samples of adolescents, but these studies typically rely on purposive samples and focus disproportionately on the USA.
What this study adds
This study adds to knowledge on adolescent health disparities by sexual orientation in multiple ways: it extends the international evidence-base to a new country (Australia) using a national dataset; considers adolescents reporting no attractions as an analytic category; and includes health/well-being outcomes not previously examined in national samples.
Further, the availability of rich information on multiple and diverse health/well-being outcomes enables us to provide a unique and encompassing overview of disparities, and rank-order them by effect magnitude to inform future research and practice.
Australian adolescents who belong to a sexual minority experienced consistently worse health/well-being than their opposite-sex-attracted peers. The largest disparities were found for self-harm, suicidality and peer/emotional problems—suggesting that these domains demand priority intervention.
There were few differences between adolescents reporting no sexual attractions and heterosexual adolescents, but some evidence of poorer outcomes among adolescents attracted to both sexes (compared with those with exclusive same-sex attractions) and sexual-minority girls (compared with boys).
Acknowledgments
This paper uses data from the Longitudinal Study of Australian Children, a survey conducted in partnership between the Department of Social Services, the Australian Institute of Family Studies and the Australian Bureau of Statistics.
References
Footnotes
Contributors FP designed the study, undertook the analyses, interpreted the results, and drafted and revised the manuscript. AC undertook the analyses, interpreted the results, and drafted and revised the manuscript.
Funding This research was supported by an Australian Research Council Discovery Early Career Researcher Award for a project titled ’Sexual Orientation and Life Chances in Contemporary Australia’.
Competing interests No, there are no competing interests for any author.
Patient consent for publication Not Required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data may be obtained from a third party and are not publicly available.