Trends in suicidality among sexual minority and heterosexual students in a Canadian population-based cohort study

Abstract

Objectives

Despite evidence from numerous studies that document disparities in suicidality for sexual minorities, few have investigated whether or not these trends have improved over time, which is the objective of the current study.

Methods

Using school-based population data over a 15-year period (1998 to 2013), multivariate logistic regressions were used to calculate age-adjusted odds ratios separately by gender. Interactions were included to test widening or narrowing disparities within orientation groups, which makes this one of the first studies to test whether gaps in disparities between heterosexual and sexual minorities have widened or narrowed over time.

Results

Results show that sexual minority youth are persistently at a greater risk for suicidal behaviour, a trend that has continued particularly for bisexual youth of both sexes. Results also suggest that the gap in suicidal behaviour is widening among some female sexual orientation groups, yet narrowing for other male sexual orientation groups.

Conclusions

These findings have important public health implications, especially since we see decreases in suicidal behaviour for heterosexual adolescents, but not in the same way for many sexual minority youth, despite advances in social acceptance of gay, lesbian, and bisexual issues in North America.

According to the World Health Organization (WHO), suicide is one of the leading causes of youth fatality (15–24 years of age) around the world (WHO, 2014). In Canada, approximately 500 youth (ages 10 to 24) have died by suicide every year from 2007 to 2011 (Statistics Canada, 2012). Research has found that previous suicidal thoughts and suicide attempts are leading indicators of a subsequent death by suicide (Oquendo, Currier, & Mann, 2006; Troister, Links, & Cutcliffe, 2008). To this end, these suicidal behaviour measures are important indicators for research as it allows for the identification of sub-groups of youth, which aids in targeted prevention, intervention and post-intervention efforts.

Review of the literature

Research on sexual minority youth consistently shows that the prevalence of suicidal behaviour is higher than for their heterosexual peers. For instance, in a meta-analysis conducted by Marshal and colleagues (2011), sexual minority youth reported significantly higher rates of suicidality (OR=2.9) than heterosexual students. In a study that examined 9 population-based datasets of adolescent students from the U.S. and Canada, Saewyc and colleagues (2007) found that LGB youth were at a significantly higher risk of suicide ideation and attempt than heterosexual and mostly heterosexual students. More specifically, they found in most surveys that one-fourth to one-third of LGB adolescents reported suicide attempts, compared to one-tenth or less for their heterosexual peers. This overrepresentation of sexual minority youth in regards to suicidal behaviour has been observed elsewhere in the United States (Hatzenbuehler, 2011; Robinson, & Espelage, 2011) as well as internationally (Värnik et al., 2009).

In general, higher rates of suicide risk are prevalent among sexual minority youth(Saewyc et al., 2007), but several studies have reported significant gender disparities. For example, in a study of LGB youth in Chicago, Mustanski, Garofalo, and Emerson (2010) found that sexual minority girls were 2.9 times more likely than sexual minority boys to have attempted suicide, and 1.6 times more likely to have had suicidal thoughts. Similar results were found for suicide attempt.

Disparities in suicide risk have also been observed between sexual orientation subgroups. Numerous studies suggest that youth who identify as bisexual are more likely to engage in suicidal behaviour than other orientation groups (Blosnich, & Bossarte, 2012; Hershberger, Pilkington, & D’Augelli, 1997). In Saewyc and colleagues’ (2007) study, they found that lesbian and bisexual girls reported higher rates of suicidal ideation than gay and bisexual boys. They also found in most surveys that bisexual boys had higher age-adjusted odds of suicidal ideation than mostly heterosexual boys, but results varied for gay boys. In all but 3 surveys, they found that bisexual girls had significantly higher age-adjusted odds of suicidal ideation than mostly heterosexual girls, and bisexual girls had either similar odds or lower odds of suicidal ideation than same-aged lesbian girls. Given these findings, researchers argue that studies should avoid simple bifurcations when operationalizing sexual identity (e.g. non-heterosexual vs. heterosexual), but rather separate non-heterosexual respondents, if possible, into more specific orientation groups (e.g. lesbian, gay, bisexual) in order to facilitate a more robust analysis. For instance, as Saewyc and colleagues (2007) note, studies that use non-heterosexual vs. heterosexual categories typically miss bisexual respondents and their unique challenges in their analysis.

While there are several studies showing that sexual minority youth are at a greater risk for suicidal behaviour, few have examined cohort trends in prevalence over time. A notable exception is Saewyc et al. (2007) where they examined population data over time for British Columbia (BC), Minnesota, and Seattle. In BC, they found that the age-adjusted odds of suicidal ideation among bisexual boys, compared to heterosexual boys, increased from 2.1 in 1992 to 4.8 in 1998, but then stabilized in 2003. Among bisexual girls, compared to their heterosexual peers, the age-adjusted odds of suicidal thoughts increased from 0.9 in 1992 to 2.7 in 1998 to 4.7 in 2003. In comparing suicide attempts among bisexual and gay boys, Saewyc and colleagues (2007) found that in 1992 bisexual boys were at a lower risk than gay boys, although in 2003 gay boys were at a lower risk than bisexual boys. The antithesis was found for lesbian and bisexual girls, where lesbian girls were at a lower risk in 1998, and bisexual girls were at a lower risk in 2003. In Minnesota, bisexual participants reported higher odds of a suicide attempt than heterosexual peers. Compared to gay boys and lesbian adolescents, bisexual boys had increasing odds of suicide attempt, although bisexual girls showed decreasing odds. Finally, in Seattle the odds of bisexual boys seriously considering suicide tripled from 1995 (3 times) to 1999 (10 times). Similar results were found for suicide attempt, but no significant differences were found between bisexual and heterosexual girls for either suicidal ideation or suicide attempt, and no differences were uncovered between bisexual and gay or lesbian peers. There is a significant limitation to these analyses, however. At the time, the odds ratios and confidence intervals for each were reported and used as a simple comparison of change over time, yet some statisticians now suggest that this approach is inadequate to actually test trends, because odds ratios from different samples should not be directly compared, not even with confidence intervals, due to underlying heterogeneity in logistic models that cannot be accounted for (Allison, 1999).

One theorized explanation for the correlation between sexual minority status and suicidality is exposure to increased social stigma, harassment, and victimization. Researchers have associated the high prevalence of suicidal behaviour (i.e. ideation and attempts) among LGB youth with the episodes of stigmatization, discrimination, and victimization they have experienced, particularly within a school environment (Burton et al., 2013; Hatzenbuehler, 2011). Saewyc and colleagues (2007) found support for enacted stigma theory using data from the British Columbia Adolescent Health Survey (BC AHS) up to 2003, but does the social stigma resonate as profoundly among BC adolescents in later years? It is reasonable to hypothesize that with the legalization of same-sex marriage in 2005, and judicial interpretations of the Charter of Rights and Freedoms as including sexual orientation among the categories to be protected against discrimination, that in Canada the social landscape has become more accepting and supportive of the LGB population, which could ultimately lead to a decrease in discrimination, harassment, and victimization. Similarly, a growing number of schools and school districts have adopted anti-homophobia policies and supportive programs for LGB youth, designed to improve reduce discrimination and bullying toward LGB youth (Russell, Kosciw, Horn, & Saewyc, 2010; Saewyc, Konishi, Rose & Homma, 2014). It is possible that these improvements would lead to a reduction in suicidal behaviour among LGB youth. However, as other researchers have noted (see Saewyc et al., 2008a), perhaps this increased awareness of support for the LGB community has intensified antagonistic and contrasting attitudes from people who disagree with sexual minority orientations. With the increase of negative attitudes towards LGB populations by individuals (or students from families with such views) who feel themselves marginalized in their morally based anti-homosexuality views, LGB youth could experience higher levels of discrimination, victimization, and harassment within their communities, schools and families. As such, analyzing cohort-based population trends in suicidal behaviour among a relatively excluded population such as LGB youth could indirectly document whether the numerous policies and laws that have been enacted to promote human rights in Canada appear to have reduced health disparities for LGB youth.

Purpose of the present study

While there are numerous studies that focus on suicide trends among youth (Lahti et al., 2011; Värnik, et al., 2009), very few include analysis of trends among sexual minority youth from cohorts of the same geographic population. As noted, one exception is the work by Saewyc and colleagues (2007) who documented trends in British Columbia from 1992 to 2003, but have these trends in suicidal behaviour remained over time? Given the dramatic changes in social policies and attitudes in Canada, especially since 2005 with the legalization of same-sex marriage, it is important to determine whether the disparities in suicidal behaviours among LGB youth, compared to their heterosexual peers, have narrowed, widened, or remained the same. In looking at whether it is getting better for LGB youth, it is not enough to merely document if there are declines in suicidal behaviours within LGB groups. If there are similar or even steeper declines among heterosexual youth, then the disparity between these sexual minority and heterosexual youth will remain.

As part of an ongoing series of studies exploring the trends of suicidality among sexual minority youth (Saewyc et al., 2007), the purpose of this study is to identify the trends as well as the disparities, and trends in disparities, in suicidal ideation and suicide attempt within and between sexual orientation groups across 4 cohorts of youth in British Columbia schools. More specifically, in the current project we extend our earlier work by:

1. Documenting the population trends in suicidal behaviours and risk factors from 1998 to 2013, separately for each sexual orientation group;

2. Comparing each group to the referent (heterosexual) to see whether there are disparities between them in each survey year; and

3. Testing whether the gaps in disparities between sexual minority groups and heterosexual peers are narrowing or widening over time using new analytical techniques.

Methods

Sample and procedures

A secondary data analysis was conducted on the British Columbia Adolescent Health Surveys (BC AHS) of 1998, 2003, 2008, and 2013, all of which were carried out by the McCreary Centre Society. The BC AHS is a clustered-stratified probability sample that surveyed classrooms of students in grades 7 through 12 in public schools across the Canadian province of British Columbia. It is important to note that participants are not the same across the four survey years. Only some grade 7 students in one year may potentially be sampled in a subsequent grade 12 year, but the stratified random samples, which result in approximately 10% of the enrolled students being surveyed, render even that minor overlap extremely unlikely. As such, the data are repeating cohorts. The majority of school districts in BC participated in the survey each year (at least 45 of 59 districts, or 76%), and these districts represent 92% of enrolled students across the province. Overall student participation rates in the randomly selected classrooms averaged 67% to 76% in each survey year. The health surveys covered a wide range of topics, including perceptions of physical and emotional health, risk exposures and health-compromising behaviours, and socio-demographics. The anonymous surveys were administered by public health nurses and other trained personnel from outside of the school.

Data were weighted to adjust for the differential probability of sampling within regions and response rates, to ensure its representativeness. The final weighted sample for each trend year included only those youth who indicated gender, age, and sexual orientation. In addition, to be sure we used consistent school data and that our analyses were not a product of changing survey procedures, our analyses only include school districts that participated in at least 3 of the 4 survey years. A detailed description of the survey sampling and administration methods have been outlined elsewhere (Saewyc, Stewart & Green, 2013; Miller, Cox, & Saewyc, 2010; Saewyc et al., 2008b).

Measures

In the current analysis, variables include self-reported demographic variables such as respondent gender and age. Sexual orientation was assessed through a measure of self-labelling defined by attraction. The response options in the 1998, 2003, and 2008 surveys were: “100% heterosexual (attracted to people of the opposite gender),” “mostly heterosexual,” “bisexual (attracted to both males and females),” “mostly homosexual,” “100% homosexual (gay/lesbian; attracted to people of the same gender),” and “not sure.” Response options changed slightly in the 2013 survey and included: “completely heterosexual (straight; attracted to people of the opposite gender),” “mostly heterosexual,” “bisexual (attracted to both males and females),” “mostly homosexual,” “completely homosexual (gay/lesbian; attracted to people of the same gender),” “questioning,” and “I don’t have attractions.” After an extensive measurement evaluation pertaining to conceptual and power issues drawing on previous research (Saewyc et al., 2004), the “mostly gay/lesbian” and “completely gay/lesbian” categories were combined into a single group; the number of youth who endorse “mostly gay/lesbian” are so few that we are unable to disaggregate these subgroups in trend and disparity analyses. Further, in order to conduct a comparative longitudinal analysis of groups, in the 1998, 2003, and 2008 surveys the “not sure” group was merged with the heterosexual category, although in the 2013 survey, the “questioning (I’m questioning who I’m attracted to)” and “I don’t have attractions” categories were combined with the heterosexual group. This decision was based on empirical evidence showing that the majority of young people who identify as “not sure” of their sexual orientation are more likely, in the future, to identify as heterosexual and not as a sexual minority (Ott et al, 2011). In addition, an extensive demographic analysis using mean deviations as an indication of temporal stability was conducted for these groups over the multiple survey years. By comparing the mean deviations of the 100% heterosexual group to the composite group (i.e., the heterosexual group with the other groups), results confirm that the composite group was significantly more stable over time – for both males and females at every age. As shown in Table 1, the distribution of sexual orientation groups has changed somewhat over time for girls, but only marginally for boys.

Table 1.

Demographics for those indicating sexual orientation in the BC AHS, by year^*

	1998		2003		2008		2013
	Boys	Girls	Boys	Girls	Boys	Girls	Boys	Girls
Unweighted n	10,852	12,006	14,523	14,800	12,250	12,977	10,758	11,180
Weighted N	130,533	146,091	138,795	140,932	108,239	115,319	89,220	92,887
% each year	47.2	52.8	49.6	50.4	48.4	51.6	49.0	51.0
Sexual orientation, %
100% heterosexual	93.8 (14.8)	90.3 (14.8)	95.3 (15.0)	87.4 (14.9)	94.1 (15.0)	86.7 (15.0)	93.1 (15.0)	85.6 (14.9)
Mostly heterosexual	4.2 (15.2)	7.6 (15.6)	3.2 (15.4)	9.2 (15.6)	3.9 (15.6)	9.6 (15.6)	4.4 (15.5)	9.5 (15.6)
Bisexual	1.1 (14.7)	1.9 (15.5)	0.9 (15.5)	3.0 (15.6)	1.1 (15.6)	3.2 (15.4)	1.4 (15.6)	4.0 (15.6)
Gay/lesbian	0.9 (15.3)	0.3 (14.8)	0.6 (15.9)	0.3 (14.9)	0.9 (15.8)	0.5 (15.5)	1.1 (15.7)	0.9 (15.5)

^*Mean age reported in parentheses

Measures of suicidality among youth included past year suicide attempt and suicidal ideation. To measure suicidal ideation among youth, respondents were asked, “During the past 12 months, did you ever seriously consider killing yourself (Yes/No)?” To measure suicide attempts among youth, respondents were asked, “During the past 12 months, how many times did you actually try to kill yourself,” with response categories coded into a dichotomous “Yes” (1) or “No” (0) variable.

Analysis

All analyses were carried out separately by respondent gender. Preliminary analyses conducted here and elsewhere (Saewyc et al., 2008b), found that rates of suicidality were prone to maturational effects. As such, in testing differences between heterosexual youth and their gay/lesbian, bisexual, or mostly heterosexual peers, we controlled for age, using multivariate logistic regressions to calculate age-adjusted odds ratios (AOR) with the various suicidality measures. Because of the cluster-stratified sampling, we tested trends in suicidality within sexual orientation subgroups across years using logistic regressions. It has been noted (Altman & Bland, 2003), AORs derived in separate samples cannot be directly compared; instead, we used interaction terms to test changes in disparities over time, testing whether trends in disparities were widening for sexual minority groups in a multivariate logistic regression model that included all sexual orientation groups (reference heterosexual) and years (reference 1998) as well as interactions terms for orientation by year. The interaction term compared the OR of a suicidal behaviour (e.g., suicide attempt) for a particular subgroup (e.g., gay boys) in a given year (e.g., 2013) to the odds of the same behaviour and subgroup of students in another survey year. As a guide in the interpretation of the interactions (AOR), an estimate over 1 indicates that the change in disparities in a given year has widened/increased compared to the referent year. The inverse is the case where the interaction AOR less than 1, which denotes that the change in disparity over time has narrowed/decreased compared to 1998 (the referent year). The decision to frame the analysis on an anchor or referent year was to aid the reader in interpreting a change in respect to time and history, such as how a meaningful social change (e.g., legalization of gay marriage) related to changes in trends. Detailed explanation of this method is described elsewhere (Homma, Saewyc, & Zumbo, in press). Further, SPSS Complex Samples 22 was used in order to adjust for the complex stratified sampling weights.

Results

Research question 1: Prevalence within sexual orientation and sex groups

From 1998 to 2013, different orientation groups demonstrated varying trends in prevalence of suicidal behaviour measures. For suicidal ideation, among heterosexual boys the overall trend in odds of suicidal ideation significantly decreased, with 2013 as the referent year, from 1998 to 2013 (see Table 2), although the overall trend in odds of suicidal ideation among mostly heterosexual significantly decreased only from 2008 to 2013 (AOR=.69, p<.05). There were no changes in the odds of suicidal ideation among bisexual boys across the years; an estimated one-third of bisexual boys reported suicidal ideation across all survey years. Among gay boys, there was a significant reduction in odds of suicidal ideation from 1998 to 2013 (AOR=.54, p<.05); however there were no significant differences from 1998 to 2003 or from 1998 to 2008. Similar to heterosexual boys, among heterosexual girls there was a significant decrease in the overall trend of odds of suicidal ideation, with 2013 as the referent year, from 1998 to 2013 (see Table 2). For mostly heterosexual and bisexual girls, there was an overall increase in the odds of suicidal ideation from 1998 to 2013 (see Table 2). Among lesbian students, there was a significant increase in the overall trends of suicidal ideation from 1998 to 2013 (see Table 2).

Table 2.

Prevalence of suicidal behaviour measures within sexual orientation group (%)

	Girls				Boys
	Het	MoHet	Bi	L	Het	MoHet	Bi	G
Suicidal ideation
2013	13.1	28.9	48.2	45.3	6.7	16.3	35.6	21.9
2008	11.7	23.8	45.9	52.2	8.0	22.1	35.6	29.9
2003	18.4	33.9	51.8	63.0	10.2	19.4	37.8	25.4
1998	17.1	26.6	37.6	38.6	9.0	20.8	32.8	34.8
Suicide attempt
2013	6.6	14.1	30.4	27.9	3.0	6.3	22.1	10.6
2008	5.2	11.7	30.6	27.0	2.6	7.8	25.6	22.3
2003	8.1	17.3	30.3	38.1	3.3	7.5	12.9	7.7
1998	8.3	12.1	26.6	24.1	3.2	8.4	17.8	19.6

The incidence of suicide attempts reported among heterosexual and mostly heterosexual boys did not change significantly across the survey years (see Table 3). However, reported suicide attempt odds among both bisexual and gay boys did change significantly. Reported suicide attempt odds among bisexual boys significantly increased from 2003 to 2013 (AOR=1.92, p<.05), whereas the overall trends of suicide attempts among gay boys significantly decreased from 1998 to 2013 (AOR=.41, p<.001). The overall trends of suicide attempts among heterosexual girls decreased from 1998 to 2013 (AOR=.79, p<.001; see Tables 2 & 3), although among mostly heterosexual girls the overall trend in odds of suicide attempts significantly increased. Among bisexual and lesbian girls there were no significant differences in the overall trend suicide attempt odds; however consistently an estimated one-third of bisexual girls and at least one-fourth of lesbian girls reported suicide attempts every survey year (see Table 2).

Table 3.

Disparities in odds of suicidal behaviour measures between heterosexual and other sexual orientation groups

	Girls			Boys
	MoHet	Bi	L	MoHet	Bi	G
Suicidal ideation
2013	2.7 (2.3–3.2)	6.2 (5.0–7.6)	5.5 (3.6–8.4)	2.6 (1.9–3.3)	7.3 (5.1–10.3)	3.6 (2.3–5.7)
2008	2.4 (2.0–2.8)	6.5 (5.3–8.1)	8.4 (4.9–14.6)	3.3 (3.5–4.4)	6.5 (4.5–9.2)	5.0 (3.2–7.9)
2003	2.3 (1.9–2.7)	4.8 (3.7–6.2)	7.6 (3.9–14.8)	2.1 (1.6–2.8)	5.3 (3.2–8.6)	2.9 (1.6–5.5)
1998	1.7 (1.4–2.0)	2.8 (1.9–4.0)	3.1 (1.6–5.9)	2.6 (1.8–3.7)	5.0 (3.1–8.1)	5.2 (2.9–9.2)
Suicide attempt
2013	2.4 (1.9–2.9)	6.3 (5.0–7.9)	5.6 (3.5–8.9)	2.1 (1.4–3.4)	8.9 (5.9–13.5)	3.7 (2.1–6.7)
2008	2.6 (2.1–3.3)	8.7 (6.8–11.1)	7.3 (4.0–13.6)	3.2 (2.2–4.7)	12.9 (8.4–19.7)	10.9 (6.7–17.7)
2003	2.4 (1.9–2.9)	5.0 (3.9–6.5)	7.0 (3.1–15.7)	2.4 (1.5–3.7)	4.3 (2.2–8.5)	2.4 (1.2–5.1)
1998	1.5 (1.2–1.9)	3.9 (2.6–6.1)	3.5 (1.8–7.0)	2.8 (1.7–4.6)	6.6 (3.6–12.2)	7.4 (3.9–13.9)

Age adjusted-odds ratio with heterosexual orientation as the reference group; 95% CI in parentheses

Research question 2: Disparities between sexual orientation and gender groups

Across all years, lesbian and bisexual girls reported a higher prevalence of suicidal ideation than all other sexual orientation categories (see Table 2). Suicidal ideation was reported by up to 51.8% of bisexual girls and 63% of lesbians, compared to 33.9% of mostly heterosexual girls and 18.4% of heterosexual girls (all reported in 2003). Among boys in the survey, gay and bisexual boys consistently reported higher prevalence of suicidal ideation than both heterosexual and mostly heterosexual boys in all years.

The odds of suicidal ideation among gay, bisexual and mostly heterosexual boys were consistently higher than heterosexual boys throughout all survey years. Notably, bisexual boys had the highest odds of suicidal ideation, compared to same-aged heterosexual boys throughout all years, excluding 1998 when gay boys had the highest odds of suicidal ideation compared to heterosexual boys (see Table 3).

Similarly, lesbian, bisexual and mostly heterosexual girls consistently reported a significantly higher prevalence of suicidal ideation than heterosexual girls across all years (see Table 3). With the exception of 2013, lesbian participants were at the highest odds for suicidal ideation compared to same-aged heterosexual girls; however, in 2013, bisexual girls had the highest odds of suicidal ideation. Notably, the odds of suicidal ideation among lesbian students, compared to heterosexual girls, nearly triples from 1998 to 2008, and in 2013 the odds of suicidal ideation decreased. A similar trend is found among bisexual girls, although the level of suicidal ideation consistently increased throughout all years among mostly heterosexual girls.

Across all survey years, bisexual and lesbian students had the highest reports of suicide attempts than all other orientation groups, including gay and bisexual boys (see Table 2). Suicide attempts were reported by up to 38.1% of lesbians (in 2003) and 30.6% of bisexual girls (in 2008), compared to 17.3% of mostly heterosexual (in 2003) and 8.3% of heterosexual girls (in 1998). Among males in the survey, gay and bisexual boys reported a higher proportion of suicide attempts across all survey years compared to mostly heterosexual and heterosexual boys.

Gay, bisexual and mostly heterosexual boys had higher odds of suicide attempts than heterosexual boys across all years. Bisexual boys always had the highest odds of suicide attempts compared to same-aged heterosexual boys throughout almost all survey years, excluding 1998 when gay boys had the highest odds of suicide attempts compared to heterosexual boys.

As shown in Table 3, the odds of suicide attempts among bisexual and gay boys fluctuated across the year, although mostly heterosexual boys had a relatively consistent decline in odds of suicide attempts. Similar to boys, lesbian, bisexual and mostly heterosexual girls reported higher odds of suicide attempts compared to heterosexual girls across all years. Bisexual girls reported the highest odds of suicide attempts compared to heterosexual girls across all years, excluding 2003 when lesbians reported the highest of suicide attempts compared to heterosexual girls. Among bisexual, lesbian and mostly heterosexual girls there was a relatively consistent increase from 1998 to 2013 in odds of suicide attempts compared to heterosexual girls.

Research question 3: Interaction effects

In terms of interaction effects, there were no significant widening or narrowing of gaps in odds of suicidal ideation for gay, bisexual, and mostly heterosexual boys, compared to heterosexual boys across the survey years. However, among female participants, two significant findings were noted (see Table 4). From 1998 to 2013, the gap in the odds of suicidal ideation disparity between mostly heterosexual and heterosexual girls has widened. Similar to mostly heterosexual girls, the gap from 1998 to 2013 in the odds of suicidal ideation disparity between bisexual and heterosexual girls has widened (see Table 4). Put another way, compared to heterosexual girls, the odds of suicidal ideation among bisexual has more than doubled from 1998 to 2013 (see Table 3), which represents a significant widening in disparities of suicidal ideation among bisexual girls (see Table 4).

Table 4.

Interactions (year-by-orientation) with heterosexual and 2013 as the reference group

	Boys OR (95% CI)	Girls OR (95% CI)
Suicidal ideation
MoHet by 2013 (vs. 1998)	1.01 (.65–1.58)	1.54 (1.21–1.95)
MoHet by 2013 (vs. 2003)	1.26 (.85–1.86)	1.19 (.95–1.48)
MoHet by 2013 (vs. 2008)	.82 (.56–1.20)	1.15 (.92–1.43)
Bisexual by 2013 (vs. 1998)	1.52 (.84–2.75)	2.10 (1.39–3.17)
Bisexual by 2013 (vs. 2003)	1.42 (.78–2.59)	1.29 (.92–1.81)
Bisexual by 2013 (vs. 2008)	1.19 (.73–1.97)	.95 (.71–1.29)
Les/Gay by 2013 (vs. 1998)	.72 (.35–1.49)	1.79 (.82–3.89)
Les/Gay by 2013 (vs. 2003)	1.30 (.59–2.85)	.72 (.33–1.59)
Les/Gay by 2013 (vs. 2008)	.79 (.42–1.50)	.67 (.33–1.33)
Suicide attempt
MoHet by 2013 (vs. 1998)	.79 (.41–1.55)	1.51 (1.11–2.04)
MoHet by 2013 (vs. 2003)	.93 (.49–1.78)	.97 (.73–1.29)
MoHet by 2013 (vs. 2008)	.70 (.39–1.26)	.96 (.72–1.29)
Bisexual by 2013 (vs. 1998)	1.40 (.67–2.91)	1.52 (.95–2.44)
Bisexual by 2013 (vs. 2003)	2.13 (.98–4.66)	1.24 (.88–1.74)
Bisexual by 2013 (vs. 2008)	.72 (.41–1.29)	.76 (.55–1.06)
Les/Gay by 2013 (vs. 1998)	.52 (.22–1.22)	1.57 (.68–3.62)
Les/Gay by 2013 (vs. 2003)	1.58 (.62–3.99)	.79 (.31–2.02)
Les/Gay by 2013 (vs. 2008)	.34 (.17–.77)	.80 (.37–1.72)

OR in bold indicates p < .001; reference group is 2013

There were no significant widening or narrowing of gaps in odds of suicide attempts for mostly heterosexual and bisexual boys, as well as bisexual and lesbians. However, two significant findings were noted among mostly heterosexual girls and gay boys (see Table 4). From 2008 to 2013, the gap in the odds of suicide attempts between gay and heterosexual boys has significantly narrowed. In contrast, from 1998 to 2013, the gap in the odds of suicide attempts between mostly heterosexual and heterosexual girls has significantly widened.

Discussion

In this study of sexual minority and heterosexual adolescents in grades 7 to 12 throughout the Canadian province of BC, we found that sexual minority youth were still at a significantly increased risk of suicidal behaviour compared to their heterosexual peers. Our study is an extension of earlier work done by Saewyc et al. (2007) where suicidal behaviour measures among sexual minority adolescents were assessed using, as one of several datasets, the BC AHS cohort surveys for the years 1992, 1998, and 2003. Our research differs in two distinct ways: first, we provide a more rigorous and updated analysis of trends in the BC AHS by incorporating cohorts from 2008 and 2013; and second, by drawing on a new method to document trends in disparities, that is, whether the gap between heterosexual and sexual minority youth is getting worse, getting better, or unchanged.

In addressing our first research objective, i.e., reporting trends of suicidality within sexual orientation groups, our findings show substantial movement, especially for girls. In particular, for heterosexual girls there is a downward trend for suicidal thoughts and suicide attempt, which suggests that current suicide prevention strategies are at least somewhat effective among this population. However, the prevalence of suicidal behaviour among the “mostly heterosexual” girls grouping has ebbed and flowed over the years, but has neither increased nor decreased substantially. Further, among bisexual girls, there has been an increase between 1998 and 2003 for both ideation and attempt, but trends have stabilized with little movement since then. A similar spike was found among lesbian adolescents, but unlike bisexual girls, the prevalence of suicidal ideation has been steadily decreasing since 2003, and has stabilized for suicide attempts. However, it is important to note that the prevalence of suicidal behaviour among lesbian and bisexual girls, even in 2013, is nearly four times higher than that of their heterosexual peers.

For sexual minority boys, there has been little shift in suicidal ideation among bisexual boys, with roughly a third reporting serious suicidal thoughts over the years. For gay boys, suicidal ideation has been declining, and suicide attempts trends have oscillated, but have recently seen a decrease from nearly a quarter in 2008 to roughly one in ten in 2013. The opposite trend is occurring for bisexual boys, where suicide attempts have increased in 2008 and 2013 to around a quarter. However, as with girls, compared to heterosexual boys, suicide ideation is substantially higher for bisexual boys (5 times higher in 2013) and gay boys (3 times higher in 2013).

In terms of our second research objective, which compared the age-adjusted odds of suicidal behaviour between sexual minority and heterosexual students, of particular concern is the increasing disparity for bisexual adolescents. By 2013, bisexual boys and girls had more than 6 times the odds of suicidality than their heterosexual peers, and for bisexual boys, the disparity gap for both suicidal ideation and suicide attempt vs. heterosexual peers appears to be widening.

There are several possible explanations for the growing disparity in suicidal thoughts and attempts among bisexual youth. One explanation may be due to the lack of visibility of bisexuality in the media, except when bisexuality does appear, it is usually in a negative light. Bisexual youth are also often subjected to poor misrepresentations of what ‘bisexuality’ entails. Such myths and misconceptions include: bisexuality is an incomplete identity; all bisexual-identified youth will eventually become gay, lesbian, or heterosexual; and bisexual individuals are promiscuous and engage in disproportionate amounts of sexual behaviour (Kennedy & Fisher, 2010; Persson & Pfaus, 2015). Such misrepresentations of bisexuality can create an erasure of identity, and produce a climate of biphobia that can negatively impact the mental health of bisexual youth (Elia, 2014; Flanders, Dobinson, & Logie, 2015). Herek (2002), for example, found that heterosexuals held more negative attitudes toward bisexuals than toward lesbians or gay men, but negativity and exclusion of bisexuality are also common from lesbian and gay individuals (Greene, 2003). Given a general lack of acceptance, bisexual adults were more likely to report identity confusion, were less likely to disclose their sexual identity, and participated in fewer community activities than lesbians or gay men (Balsam & Mohr, 2007; Lewis et al., 2009). It is possible, that similar feelings of exclusion and lack of belonging are also occurring among bisexual adolescents, and the invisibility of positive role models may be an additive effect.

Another potential explanation is the higher risk for violence exposure, whether sexual or physical abuse, experienced by sexual minority youth compared to heterosexual peers, as physical and sexual violence have been linked to suicidal ideation and attempts among adolescents (Borowsky, Ireland & Resnick, 2001; Saewyc & Chen, 2011). Several studies over the years have documented higher rates of physical and sexual violence toward LGB youth compared to heterosexual peers, including in Canada (Saewyc, et al., 2006). In a meta-analysis of these studies, moderation analyses showed bisexual adolescents had even higher odds of family physical abuse and skipping school because they felt unsafe than gay and lesbian youth (Friedman, et al., 2011).

Further exploration of these issues of negativity, exclusion, and violence exposure among adolescents, would shed some light on this widening disparity, especially in terms of the availability of supportive LGB organizations within the school context, such as gay-straight alliance clubs (GSAs). Research by Saewyc et al. (2014) found that the presence of GSAs or explicit school district policy led to a decrease in discrimination based on sexual orientation and suicidal behaviour among LGB adolescents, especially when such interventions have been in place for a few years and are thus well established. It would be useful to examine these impacts by separating out lesbian, gay, and bisexual youth, to see if lower odds of suicidal involvement were predominantly driven by the experiences of lesbian and gay students.

The third research objective employed an interaction analysis in order to determine significant widening or narrowing of gaps in suicidality by sexual orientation group across trend years. These analyses found several significant trends in the disparity of suicidality, particularly in terms of widening gaps for suicidal behaviour for mostly heterosexual and bisexual girls, compared to heterosexual girls from 1998–2013. For boys, however, there is a narrowing gap for suicide attempts among gay boys compared to their heterosexual peers between 2008 and 2013. These results suggest that disparities in suicidal ideation and attempts are getting better for gay boys, is unchanged for bisexual boys, but among lesbian and bisexual girls, the gap is widening, and thus getting worse. Given the nature of the current analyses, we cannot confidently explain such divergent trends, although they suggest that the recent social and legal improvements in status for LGB people may have differential effects for LGB boys and girls, and for bisexual adolescents. Further research is needed, particularly studies that include multiple indicators of suicidality, and studies that can tease out the differential impacts of supportive policies and programs for lesbian, gay and bisexual adolescents.

Strengths and weaknesses

The findings from our study are unique in that they are based on large-scale population studies, which can be generalized to a wider population of school-aged adolescents. In addition, given that the BC AHS is repeated over multiple years, we are able to identify trends in both suicidality, and trends in the disparity between sexual minority and heterosexual teens around suicidality. Finally, the large size of the datasets also enables an analysis separated by respondent gender and sexual orientation, and results have shown some important findings when bisexuality is disaggregated from gay and lesbian identities.

Several limitations are also noteworthy to report. First, because these data rely on self-reports of sexual orientation, so groupings represent a young person’s current orientation self-label and may not represent all youth in that orientation group, mostly due to the well-documented research that shows sexual identity develops over time and thus self-labels and behaviour may change (Kinnish, Strassberg, & Turner, 2005). Second, given these surveys are school-based, any generalizability can only be made to these populations, thus missing youth who do not attend school. It is suggested in some research that sexual minority youth are at a greater risk of dropping out of school, which these data would not capture (Smith et al., 2007). And last, one should note that the trends are curvilinear in nature for some mental health outcomes. In this paper, we have reported on meaningful changes in trends when they were statistically significant and spanned several survey waves; we were most interested in long-term change over the span of 15 years’ time.

Recommendations for Research and Practice

Despite these limitations, our research offers important insight, particularly in the areas of suicide prevention, intervention, and postvention efforts. In terms of prevention, our research shows that there is an increasing disparity of suicidal behaviour, particularly for bisexual girls, and it is important that prevention strategies specifically target this vulnerable group. Bisexual youth may be more likely to engage in suicidal behaviour because they experience marginalization from both heterosexual as well as gay and lesbian groups (Lucassen et al., 2011). Increasing the visibility of positive bisexual role models, as well as validating bisexual identities by dispelling stereotypes and myths, may help improve social support for bisexual adolescents in schools and families.

Intervention strategies could include fostering stress coping skills, and intervening if someone is experiencing suicidal ideation or behaviours. Therapy could be one method of intervention, wherein bisexual youth, as well as other sexual minority individuals, could comfortably discuss their emotions or their insecurities, and develop sufficient coping strategies as to reduce suicidal behaviour and/or thoughts. Therapists should strive to create a more supportive and inclusive environment for all youth, and therapists should especially try to better understand the unique identity-related and mental health challenges bisexual youth may experience (Pallotta-Chiarolli & Martin, 2009). In addition, therapy approaches may vary depending on gender, given our findings that suicidal behaviour is reported at a higher rate among sexual minority girls than boys.

Finally, postvention involves developing the skills and strategies for taking care of oneself after the occurrence of suicidal thoughts or a suicide attempt, as well as support for those who have lost a friend or family member to suicide. Due to the higher odds of suicidal behaviour among sexual minority students, particularly those who identify as bisexual, it is critical that postvention programs are developed that are focused on the issues facing sexual minority youth, preferably before a completed suicide occurs. In the worst-case scenario, when a young person dies by suicide, postvention efforts are paramount, as research suggests that adolescents are especially vulnerable to suicide contagion (Bridge, Goldstein, & Brent, 2006).

Concluding remarks

The current study adds to the literature in two respects: first, the study explores trends and disparities in suicidality among LGB and heterosexual youth in Canada, and second, it explores how these trends or disparities remain constant, widen, or narrow over time among Canadian youth. Our findings support a growing body of research that has found sexual minority youth to be at greater risk for suicidal behavior such as suicidal ideation and attempts. There is some room for hope; things appear to be improving for gay boys, but not for other sexual minority youth. This analytical approach can help monitor trends in suicidal behaviour over time, and in turn guide policy-makers to take action where suicide trends are widening or are consistently heightened among certain groups, such as bisexual and lesbian girls, as well as bisexual boys.