Mental Health, Drug, and Violence Interventions for Sexual/Gender Minorities: A Systematic Review

PROSPERO approved our protocol before data extraction.¹² 

We included randomized controlled trials and nonrandomized study designs; we included the latter because not all SGMY-relevant interventions (eg, federal policies legalizing same-gender marriage) are conducive to randomization. However, nonrandomized studies are more likely to be biased than randomized trials,¹³ and to limit potential biases, we only included studies with both pre- and postintervention data from participants, as recommended by the Effective Practice and Organisation of Care Cochrane Review Group.¹⁴ Such designs include nonrandomized longitudinal studies and interrupted time series studies. We excluded cross-sectional studies and case report studies.

We included studies in which authors examined participants aged <18 years at baseline. We selected this because using substances, having mental health problems, and being victimized before age 18 are associated with similar outcomes later in life.^15,–18 Because study authors sometimes enroll populations both younger and older than 18 years of age, we included studies with a minority (<25%) of adult participants (≥18 years old) or studies reporting results separately for youth participants versus adult populations, as has been done in previous Cochrane reviews.^19,20 

We included studies if the authors assessed sexual or gender minority status.^1,21 We defined sexual minority populations as lesbian, gay, bisexual, queer, and other sexual minority identities, as well as youth who have same-gender sexual behavior or attractions. We defined gender minority populations as transgender people (eg, those who identify as transgender or whose current gender identity does not match their assigned sex at birth) or people with other gender-nonconforming identities (eg, genderqueer).

We included any type of intervention that was a “purposeful action by an agent to create change”²² or a “process of intervening on people groups, entities or objects.”²³ Therefore, this review potentially included behavioral, psychological, educational, pharmacologic, medical, and policy interventions. We included universal and SGMY-specific interventions.

We included studies in which authors examined substance use, mental health problems, or violence victimization as outcomes. Substance use included licit and illicit drug use, such as alcohol, tobacco, marijuana, prescription drug misuse, heroin, hallucinogens, methamphetamine, ecstasy, and cocaine. Mental health problems included stress; anxiety; depressive symptoms; suicidality; internalized homo-, bi-, and/or transphobia; and nonsuicidal self-injury. Violence victimization outcomes included experiences or threats of bullying, cyberbullying, aggression, violence with weapons, harassment, discrimination, sexual assault, sexual abuse, physical abuse, and emotional abuse from all types of perpetrators.

We conducted a search of electronic databases with a research librarian who developed, piloted, and executed the search strategies. We searched PubMed, PsycINFO (via Ovid), and the Education Resources Information Center (via EBSCOhost) for studies published from January 1, 2000, through January 2, 2019 (see Fig 1 for exact dates). The search strategies used a combination of text words and medical subject headings (eg, Medical Subject Headings terms) adapted for each database. The search strategy was developed in PubMed and adapted for PsycINFO and the Education Resources Information Center. The search strategies included the following concepts: sexual or gender minority status²⁴; youth; substance use, mental health problems, or violence; study design and intervention terms; human research; and studies in English. The final PubMed search strategy can be found the Supplemental Information. We excluded animal studies, meta-analyses, systematic reviews, news, editorials, and commentaries. We had no geographical restrictions.

First, we identified potentially relevant studies by reviewing the titles and abstracts of all retrieved articles. We considered studies with insufficient information in the title or abstract as potentially relevant articles for further assessment. Second, we reviewed the full text of potentially relevant studies for final inclusion or exclusion in our study. Two of 6 investigators independently screened each record and had substantial agreement for title and abstract screening (κ = 0.69) and full-text screening (κ = 0.83).²⁵ The first author resolved any disagreements. We tracked the screening results in DistillerSR (Evidence Partners, Kanata, Ottawa, Canada).

We conducted a narrative synthesis for each study. Using a standardized form, 2 of 4 investigators independently extracted data from each included study. We extracted data on each study’s intervention, evaluation design, sampling and recruitment procedures, inclusion andexclusion criteria, sample characteristics, outcome measures, and main findings. One investigator placed all extracted data in tabular format, and another investigator reviewed the table for accuracy and completeness. The 2 investigators discussed any discrepancies until they reached a consensus.

We selected the Quality Assessment Tool for Quantitative Studies checklist to assess methodologic rigor because this tool assesses characteristics of both randomized and nonrandomized studies.²⁶ Two independent raters evaluated each study; raters then discussed any discrepancies until they reached a consensus. Raters assessed 6 characteristics for each study: selection bias, study design, confounders, blinding, data collection method, and withdrawals and dropouts. On the basis of the ratings from these 6 characteristics, each study received a global rating. Possible ratings for each study characteristic and global rating included weak, moderate, and strong (ranging from least to most methodologically rigorous).

Searches identified 6598 unique studies, of which 424 studies were potentially relevant for inclusion in this review (Fig 1). After full-text screening, 9 studies met the inclusion criteria.^27,–35 

Interventions inclusive of all SGMY were evaluated in 1 study,³² interventions tailored specifically to GMY were evaluated in 3 studies,^27,–29 and interventions specifically tailored to SMY were evaluated in 5 studies (Table 1).^{30,31,33,–35} The program inclusive of all SGMY was the Comprehensive Community Mental Health Services for Children with Serious Emotional Disturbances Program, more commonly known as the Children’s Mental Health Initiative.³² This program provided coordinated networks of community-based services tailored to the local needs of youth.³² The participants served by this program received a wide variety of specific interventions, including individual therapy, medication treatment, case management, group therapy, recreational activities, inpatient hospitalization, vocational training, family support, and residential treatment, which were all tailored to the participants’ local context and individual needs.³² 

Authors of all the GMY-specific interventions examined transition-related gender-affirming care interventions (ie, puberty suppression, crossgender hormones, gender affirmation surgery, and psychological support following the Standards of Care of the World Professional Association for Transgender Health [WPATH]).^27,–29,36 Authors of 2 studies^27,28 examined the effects of puberty suppression (ie, the provision of gonadotropin-releasing hormone [GnRH] analogs that delay the physical changes associated with puberty³⁶) on mental health. Specific clinical criteria must be met to receive puberty-suppressing hormones.^27,28,36 Should those clinical criteria not be met, youth receive psychological support as standard of care; therefore, authors of 1 study²⁷ had a 2-group design in which they compared the effects of a psychological-only intervention to a psychological and puberty suppression intervention. The other study²⁸ had a 1-group design, observing only youth who received puberty suppression. Authors of the third study²⁹ examined the effects of crossgender hormones and gender affirmation surgery on mental health using a subset of participants from the previous study.²⁸ All of the intervention studies for GMY followed WPATH Standards of Care,³⁶ and all participants received ongoing medical or psychological care from baseline through final posttest assessment.^27,–29 

Among the SMY-specific interventions, there was a therapist-administered family-based intervention to reduce mental health problems,³⁰ a self-administered computer-based intervention to reduce mental health problems,³¹ a self-administered online intervention to reduce substance use and stress,³⁴ a state-level policy granting same-sex marriage,³³ and state-level general and enumerated antibullying laws.³⁵ The state-level interventions consisted of one-time policy enactments.^33,35 Both the self-administered interventions^31,34 were shorter in duration and smaller in dosage than the therapist-administered intervention.³⁰ The self-administered interventions used three 14-minute modules delivered during a 1-month period³⁴ or seven 30-minute modules delivered during a 2-month period.³¹ The therapist-administered intervention had between 8 and 16 weekly in-person sessions that lasted for 1 hour.³⁰ The nonpolicy interventions had specific theoretical underpinnings.^30,31,34 One intervention incorporated input from youth during development,³¹ and 1 used input from clinicians.³⁰ 

A randomized controlled study design was used in 1 study,³⁴ a nonrandomized comparison group design was used in 1 other study,²⁷ an interrupted time series design was used in 2 studies,^37,38 and a 1-group design was used in 5 studies.^28,–32 Two studies had a pretest-posttest design,^28,34 1 had a pretest-posttest-posttest design,³¹ 3 had a pretest-midtest-posttest design,^29,30,32 1 had a pretest-posttest design with >2 posttests.²⁷ The interrupted time series designs varied in their number of pretest and posttests depending on the states and policy enactment dates, and the authors of these studies used serial cross-sectional data without the ability to track individual participants across time.^33,35 For the longitudinal studies tracking participants, the average length between baseline and the final posttest ranged from 0.2³⁰ to 7.1²⁹ years.

In 2 of the SMY-specific interventions, authors used probabilistic sampling frames from public high schools via the Youth Risk Behavior Surveillance System.^33,35 In 6 studies, various forms of convenience sampling were used: GMY-specific interventions^27,–29 recruited participants from clinics, and, of the remaining SMY-specific interventions, 1 study recruited from clinics,³⁰ 1 from Facebook,³⁴ and 1 from high schools, a local SGMY organization, and SGMY media.³¹ Authors of the SGMY-inclusive study omitted their specific sampling strategy but recruited participants who accessed services and supports from 47 communities across the United States.³² The GMY-specific interventions were conducted in Europe,^27,–29 with 1 in England²⁷ and 2 in the Netherlands^28,29; 4 of the SMY-specific interventions were evaluated in the United States^30,33,–35 and 1 was in New Zealand.³¹ 

The SGMY-inclusive intervention was provided to all youth with serious emotional disturbance (wherein the most commonly reported problems being depression, anxiety, and conduct and/or delinquency) but included only SGMY in analyses.³² The GMY-specific interventions was only implemented with youth who had a gender identity disorder diagnosis as identified through the criteria from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition.^27,–29 In the SMY-specific interventions, 2 studies included all youth with subanalyses on SMY,^33,35 1 study included only SMY with significant suicidal ideation,³⁰ 1 study included only SMY with depressive symptoms,³¹ and 1 study did not have any eligibility criteria related to mental health.³⁴ 

The included studies^27,–35 had 1 050 339 total participants with a median of 201²⁷ participants and a range of 1030 to 762 768³³ participants. The average age of participants was 15.95 years (ranging from 11 to 21).^27,–35 Four samples included only youth <18 years of age.^27,–29,34 Participants’ gender identity or assigned natal sex were reported in all studies.^27,–35 Participants’ sexual orientation was reported in 7 studies^28,30,–35: in 3 studies, only sexual attractions were reported^28,31,34; in 3 studies as well, only sexual identities were reported^32,33,35; and in 1 study, both were reported.³⁰ 

Mental health outcomes were examined in all studies^27,–35: depressive symptoms were examined in 5 studies,^28,–32 anxiety symptoms were examined in 4,^28,29,31,32 internalizing and externalizing symptoms were examined in 3,^28,29,32 psychosocial functioning was examined in 2,^27,29 hopelessness was examined in 1,³¹ perceived stress was examined in 1,³⁴ suicidal ideation was examined in 2,^30,35 and suicide attempts were examined in 2.^33,35 Mental health outcomes were assessed by using reports from participants, parents or caregivers, clinicians, and researchers.^27,–35 In 2 studies, self-reported substance use outcomes were examined,^32,34 including frequency of use and substance abuse and/or dependence symptoms. In 1 of the included studies, authors examined bullying victimization and being threatened or injured with a weapon on school property.³⁵ 

Painter et al³² found that the Children’s Mental Health Initiative, which provided coordinated networks of community-based services and supports across the country to children with serious emotional disturbances, significantly improved all measured outcomes throughout a 1-year time period for SGMY. This program decreased symptoms of anxiety, depression, global functioning impairment, internalizing and externalizing symptoms, and substance abuse and dependence symptoms among SGMY.³² 

De Vries et al²⁸ showed in their 1-group pretest-posttest study that initiation of pubertal suppression reduced depressive, internalizing, and externalizing symptoms, but not anxiety symptoms. De Vries et al²⁹ also conducted a follow-up study using data from a subset of these participants as they initiated crossgender hormones and gender affirmation surgery. By using a 1-group pretest-midtest-posttest study across 7.1 years, participants were assessed at baseline (before initiating puberty suppression), midintervention (just before initiating crossgender hormones), and postintervention (1 year after gender affirmation surgery).²⁹ Over time, psychosocial functioning increased linearly, whereas internalizing and externalizing symptoms from the Child and Adult Behavior Checklists decreased linearly.²⁹ Depressive symptoms and internalizing symptoms from the Youth and Adult Self-Reports decreased from baseline to midintervention but increased slightly at postintervention.²⁹ For both measures of internalizing symptoms and externalizing symptoms from the Child and Adult Behavior Checklists, the percentage of participants in the clinically significant range decreased over time.²⁹ Although the aforementioned results were similar for transmen and transwomen, some results were moderated by gender: anxiety and externalizing symptoms from the Youth and Adult Self-Reports decreased linearly for transmen but increased after gender affirmation surgery for transwomen.²⁹ 

Costa et al²⁷ compared GMY who received a psychological-only intervention to those who received a psychological support and puberty suppression intervention. The 2 nonrandomized groups did not significantly differ in average psychosocial functioning at any assessment point (ie, baseline, 6-, 12-, and 18-month follow-ups).²⁷ Within-group analyses revealed that for participants in the psychological-only intervention group, average psychosocial functioning improved after initiating the psychological intervention and plateaued thereafter.²⁷ For participants in the psychological and puberty suppression intervention group, average psychosocial functioning did not improve after initiation of the psychological intervention but did significantly improve after initiating puberty suppression.²⁷ 

Diamond et al³⁰ showed that SMY who participated in an in-person family-based therapy intervention had significant decreases in average depressive symptoms and suicidal ideation symptoms across the pretest, midtest, and posttest. Lucassen et al³¹ showed SMY who participated in the computerized cognitive behavioral therapy intervention also had significant decreases in average depressive symptoms (across 3 different measures), anxiety symptoms, and hopelessness from baseline to immediate postintervention. Average depressive symptoms plateaued from immediate postintervention to 3-month postintervention.³¹ 

According to a randomized controlled trial conducted by Schwinn et al,³⁴ an online intervention aimed at reducing substance use revealed that compared with control participants, intervention participants had significantly lower perceived stress and past-month frequency of other drug use (ie, use of inhalants, club drugs, steroids, cocaine, methamphetamines, prescription drug, or heroin) at 3-month follow-up. However, there were no significant differences between intervention and control groups in past-month frequency of alcohol, cigarette, or marijuana use at 3-month follow-up.³⁴ 

Raifman et al³³ found after the implementation of a policy granting same-sex couples equivalent marriage rights as opposite-sex couples, there was a significant decline in past-year suicide attempt prevalence among SMY. The enactment of such policy induced a 14% relative decline in the proportion of SMY reporting at least 1 past-year suicide attempt.³³ For heterosexual youth, the passage of same-sex marriage policy was also associated with a significant decline in past-year suicide attempt prevalence.³³ 

Seelman and Walker³⁵ found that the enactment of state-level general antibullying laws was associated with a reduction in bullying victimization among SMY. Although general antibullying laws were not associated with being threatened or injured with a weapon among all SMY, there was a protective association for general antibullying laws among sexual minority boys <16 years old.³⁵ General antibullying laws were not associated with suicidal ideation or suicide attempts among SMY.³⁵ 

Seelman and Walker³⁵ also investigated changes in these outcomes related to the enactment of state-level antibullying laws that enumerated sexual orientation as a protected class. Enumerated antibullying laws were associated with reductions in bullying victimization among lesbian, gay, and bisexual (LGB) youth and suicide attempts among SMY and questioning youth.³⁵ However, enumerated antibullying laws were not associated with suicidal ideation or being threatened or injured with a weapon.³⁵ 

Table 2 reveals the methodologic quality of the studies rated across several dimensions.²⁶ One study received a strong global rating,³³ 1 study received a moderate global rating,³⁵ and 7 studies received weak global ratings.^27,–32,34 Regarding selection bias, 2 studies with probabilistic sampling were strong,^33,35 and 7 studies were weak because their samples were not necessarily representative of their target populations or they had low or unreported participation rates.^27,–32,34 Study designs ranged from moderate to strong.^27,–35 The 1 study with a strong rating was a randomized controlled trial,³⁴ and the studies with moderate ratings were interrupted time series designs^33,35 or longitudinal study designs with 1 or 2 groups.^27,–32 Regarding confounders, 1 was strong,³³ but all other studies were weak because the authors failed to control for important potential confounders such as age, sex, and race and/or ethnicity³⁵ or only reported unadjusted associations.^27,–31,34 Blinding procedures (ie, blinding data collectors to participants’ intervention status and blinding participants to the study’s primary research question) were strong across 2 studies^33,35 and moderate across 7 studies.^27,–32,34 Data collection methods were strong in 8 studies because they used valid and reliable measures.^27,–33,35 One study had weak data collection methods because it was unclear if the authors used valid and reliable measures.³⁴ Withdrawals and dropouts were strong in 3 studies that had ≥80% of participants complete the final study assessment.^30,31,34 Four remaining studies were rated as weak because of substantial attrition.^27,–29,32 

With this systematic review, we identified the scarcity of interventions for SGMY evaluated in peer-reviewed scientific literature. Specifically, we found 9 interventions for mental health problems,^27,–35 2 for substance use,^32,34 and 1 for violence victimization.³⁵ One study had strong methodologic quality and found that state-level marriage equality laws significantly reduced suicide attempts among SMY.³³ One study had moderate methodologic quality and found that state-level general and enumerated antibullying laws significantly reduced bullying victimization for SMY.³⁵ Although the other 7 interventions made significant improvements in mental health problems and substance use,^27,–32,34 these studies’ results must be interpreted cautiously because of suboptimal methodologic quality.²⁶ For example, although it would be decidedly unethical to withhold medical care from youth who need it, the lack of a comparison or control group threatens internal validity. Without a comparison or control group, participants’ improvements may be attributable to pubertal maturation³⁹ or historical social climate.^40,–42 By using comparison or control groups, authors can more accurately assess the direct benefit of the intervention under investigation. Altogether, this small collection of diverse evidence-based interventions is likely insufficient to mitigate the substantial population-level inequities present among SGMY in substance use, mental health problems, and violence victimization.

Our review, however, is not without limitations. It was impossible to include intervention evaluations still under review at scientific journals or evaluations still underway. In this review, we also do not capture interventions without evaluations or those with evaluations published outside of the peer-reviewed scientific literature. Conducting and publishing evaluations in the scientific peer-reviewed literature is important for both understanding intervention effectiveness and dissemination. For example, without a peer-reviewed publication of evaluation results, interventions cannot be included in national intervention registries (eg, the Evidence-Based Practices Resource Center), thereby hampering the widespread implementation of potentially effective interventions. Additionally, bias toward publishing only significant efficacious or effective results may have limited the number of studies included, potentially limiting our knowledge about ineffective interventions. Finally, studies evaluating the effectiveness of universal interventions likely include SGMY and GMY as participants; however, researchers must explicitly include items that assess sexual and gender minority statuses to test whether these interventions are also effective for SGMY.

There are likely many other substantial reasons why we found few interventions evaluated for SGMY, not the least of which are the unique barriers in reaching SGMY. Such barriers include SGMY being a minority of the population,⁴³ the fact that SGMY are often still developing their identities and as a result may not be “out” yet,^44,45 and structural barriers, such as the presence of anti-SGMY attitudes and policies⁴² and the historical lack of SGMY-affirmative school practices and funding directed toward SGMY health interventions.^46,47 

Despite these barriers, there are many ways to advance the field of SGMY intervention research for reducing substance use, mental health problems, and violence victimization. Investigators can:

1. examine the efficacy of existing interventions (eg, refs 48–50) that included youth in their studies but failed to meet our Cochrane-informed^19,20 age eligibility criteria;

2. evaluate the efficacy of interventions designed and implemented by community-based organizations (eg, ref 51);

3. conduct outcome evaluations for interventions currently only examined via process evaluations (eg, ref 52);

4. conduct natural experiments and quasi-experimental studies for additional policy changes (such as those found in this review^33,35);

5. adapt existing interventions (eg, refs 30,31) to incorporate SGMY-specific content;

6. test whether universal interventions targeting all youth (eg, ref 53) are efficacious specifically for SGMY; and

7. develop, implement, and evaluate new interventions specifically tailored for SGMY (eg, refs 34,54).

It remains unclear whether universal or targeted interventions are more effective at reducing SGMY health inequities, but findings from our review suggest that both approaches are likely beneficial.^27,–35 Moreover, as investigators begin to develop, implement, and test interventions among SGMY, they ought to draw on best practices from intervention science to advance the field more rapidly. Some of these practices include collaborating with participating community members to incorporate their perspectives into intervention development, implementation, and evaluation, which increases the intervention’s relevance and protections of participants’ rights⁵⁵; using theoretical foundations of behavior change to build more efficacious interventions⁵⁶; and carefully developing feasibility pilot studies used to document the intervention’s successes and failures to inform future intervention research of SGMY.⁵⁷ 

Interventions can also incorporate knowledge gained from the extant epidemiological literature to increase intervention reach, target SGMY during specific periods of the life course, and incorporate specific population needs. Regarding intervention reach, interventions can target SGMY in myriad contexts: SGMY usually live with families (although living in homelessness is heightened among SGMY⁵⁸) and also attend school for >1000 hours each year,⁵⁹ providing ideal settings for implementing interventions with SGMY. Additionally, SGMY are present in afterschool programs, community-based organizations, sport programs, churches, and medical clinics. Because most youth use the Internet,⁶⁰ Internet-based intervention methods may be a particularly effective way to reach SGMY. Prevention interventions may also benefit from targeting SGMY as early as possible in the life course because across all youth, bullying victimization is more prevalent at younger ages,^6,61,62 and SGMY have earlier substance use initiation than their peers.^6,63,64 Finally, SGMY are not homogenous: the needs of bisexual youth deserve particular attention because they are the largest SMY subgroup⁶⁵ and often have worse health outcomes than their gay and lesbian counterparts.^2,–4,65 The needs of SGMY of racial and/or ethnic minority groups also warrant careful consideration because many health outcomes and risk factors vary by race and/or ethnicity among SGMY.^63,66,–68 

Future interventions can also benefit from reducing known risk factors and enhancing protective and resilience factors to improve health among SGMY.^38,69 Stigma and discrimination are the fundamental causes behind SGMY health inequities in substance use, mental health problems, and violence victimization^70,71; thus, developing interventions to reduce stigma and discrimination is critical. Because stigma and discrimination are multidimensional, existing at multiple levels (ie, individual, interpersonal, organizational, and structural) and in multiple forms (ie, covert and overt biases),⁷² reducing stigma and discrimination for SGMY will require multilevel, multipronged approaches.^38,73,74 Enhancing protective factors and resiliencies may also reduce SGMY health inequities. Such factors include adult and peer support, adaptive SGMY-specific coping strategies, and SGMY-affirmative school climates, programs curricula, and policies.^{37,65,75,–82} 

With few effective interventions for SGMY, inequities in substance use, mental health problems, and violence victimization for SGMY are likely to persist. To advance the field of intervention science for SGMY more rapidly, researchers can engage in community-based research and use the extant literature to rigorously design, implement, and evaluate interventions, all in an effort to foster health equity for SGMY.

GMY

gender minority youth

GnRH

gonadotropin-releasing hormone

LGB

lesbian, gay, and bisexual

SGMY

sexual and gender minority youth

SMY

sexual minority youth

WPATH

World Professional Association for Transgender Health