Guided LGBTQ-affirmative internet cognitive-behavioral therapy for sexual minority youth’s mental health: A randomized controlled trial of a minority stress treatment approach

Abstract

Objective.

LGBTQ-affirmative cognitive-behavioral therapy (CBT) addresses the adverse impacts of minority stress. However, this treatment has rarely been tested in randomized controlled trials with LGBTQ youth and never using an asynchronous online platform for broad reach. This study examined the feasibility, acceptability, preliminary efficacy, and multi-level stigma moderators of LGBTQ-affirmative internet-based CBT (ICBT).

Method.

Participants were 120 LGBTQ youth (ages 16–25; 37.5% transgender or non-binary; 75.8% assigned female at birth; 49.2% non-Latino White) living across 38 U.S. states and reporting depression and/or anxiety symptoms. Participants were randomized to receive 10 sessions of LGBTQ-affirmative ICBT or only complete 10 weekly assessments of mental and behavioral health and minority stress; all completed measures of psychological distress, depression, anxiety, suicidal thoughts, alcohol use, and HIV-transmission-risk behavior at baseline and 4 and 8 months post-baseline; 20 LGBTQ-affirmative ICBT participants completed a qualitative interview regarding intervention acceptability.

Results.

Participants randomized to LGBTQ-affirmative ICBT completed, on average, 6.08 (SD = 3.80) sessions. Participants reported that LGBTQ-affirmative ICBT was helpful and engaging and provided suggestions for enhancing engagement. Although most outcomes decreased over time, between-group comparisons were small and non-significant. LGBTQ-affirmative ICBT was more efficacious in reducing psychological distress than assessment-only for participants in counties high in anti-LGBTQ bias (b=−1.73, p=0.001, 95% CI [−2.75, −0.70]). Session dosage also significantly predicted reduced depression and anxiety symptoms.

Conclusions.

LGBTQ-affirmative ICBT represents a feasible and acceptable treatment. Future research can identify more efficacious approaches and modalities for engaging LGBTQ youth, especially those living under stigmatizing conditions, who might benefit most.

Sexual minority youth (e.g., those who identify as gay, lesbian, bisexual, pansexual, or queer) represent one of the highest-risk groups for mental health problems, including depression, anxiety, and suicidality (Clark et al., 2020; Day et al., 2017) and co-occurring behavioral health problems, including alcohol use and HIV-transmission-risk behavior (Fish et al., 2017). These disparities start at least in adolescence (Blashill et al., 2021; Pachankis, Clark, et al., 2022; Xu & Rahman, 2022) and largely persist across the life course (Rice et al., 2019). The most plausible source of these disparities lies in sexual minority youths’ greater exposure to stigma-motivated stressors, including peer bullying and parental rejection, compared to their heterosexual peers (Pachankis, Clark, et al., 2022). According to minority stress theory (Brooks, 1981; Meyer, 2003) and its empirical evidence (e.g., Hollinsaid et al., 2022; Lattanner et al., 2022), these stressors generate cognitive, affective, and behavioral coping strategies (e.g., identity concealment, threat hyper-vigilance, internalized stigma) that can give rise to mental and behavioral health challenges.

Recognizing that sexual minority youth experience distinct stressors and stress reactions that elevate their mental health risk, the mental health field has recently started addressing these stressors by developing evidence-based practice specifically for this population (Pachankis, 2018). LGBTQ-affirmative cognitive-behavioral therapy (CBT) is an evidence-based practice that adapts general CBT principles and techniques to respond to the minority stressors faced by sexual minority individuals. Informed by minority stress theory (Meyer, 2003) and related theories (Diamond & Alley, 2022; Hatzenbuehler et al., 2009), LGBTQ-affirmative CBT helps sexual minority individuals identify the role of stigma in their current distress, challenge negative internalized ideologies about sexual minority people, and develop behavioral coping skills such as mindfulness and assertiveness to undo emotional and behavioral avoidance patterns generated by minority stress (Pachankis, Soulliard, et al., 2022). Three randomized controlled trials with sexual minority young adults now show that LGBTQ-affirmative CBT is efficacious compared to waitlist (Pachankis et al., 2015; Pachankis, McConocha, et al., 2020), yields moderately stronger effects on some outcomes (e.g., alcohol use problems, comorbid mental and behavioral health problems) compared to LGBTQ-affirmative community treatment (Pachankis, Harkness, et al., 2022), and might be particularly effective for those experiencing higher levels of stigma (Keefe et al., 2023; Millar et al., 2016).

Although the strongest empirical evidence for LGBTQ-affirmative CBT derives from the above randomized controlled trials conducted with young adults, increasing evidence suggests that this treatment might be effective for even younger sexual minority individuals, including adolescents. This evidence stems from relatively small trials of LGBTQ-affirmative CBT delivered across diverse formats, including in-person, group telehealth, and digital games. For instance, an open trial with 30 youth found significant reductions in depression three months after receiving eight in-person sessions of LGBTQ-affirmative CBT (Craig & Austin, 2016). A trial with 96 sexual minority youth found that eight sessions of LGBTQ-affirmative CBT delivered to small groups of youth via telehealth and in-person were associated with significantly greater reductions in depression compared to waitlist (Craig, Leung, et al., 2021). A trial of the same treatment delivered in-person to small groups across 138 youth also found a significantly greater impact of the treatment compared to waitlist immediately after the treatment (Craig, Eaton, et al., 2021); neither trial reported longer-term follow-up data. In an open trial with 21 sexual minority youth, a gamified digital approach to LGBTQ-affirmative CBT – containing seven self-guided interactive learning modules in which skills from a fantasy world are applied to real life – showed significant reductions in depression up to 3 months after the intervention (Lucassen et al., 2015). Notably, this latter delivery modality involved no therapist, suggesting a potential route to broad dissemination of LGBTQ-affirmative CBT to youth.

Overall, internet-based mental health treatments hold promise for reaching individuals who might not otherwise be able to access in-person treatment, including those living in rural and low-resource areas with few available providers. Indeed, in the United States, mental health providers are disproportionately concentrated, per capita, in dense, high-income, metropolitan regions (Ellis et al., 2009) and more than one-third of the U.S. population lives in an area with a mental health provider shortage (Health Resources and Services Administration, 2022). Sexual minority individuals are also disproportionately removed from affirmative mental health services (Martos et al., 2017), both because the location of LGBTQ-affirmative providers tends to follow the same geographic trends for mental health providers overall, and because sexual minority individuals prefer providers who can address their distinct identity-related experiences at the same time that not all providers are trained or competent in doing so (Cronin et al., 2021; Veltman & Chaimowitz, 2014). Therefore, internet-based treatments might be particularly promising for sexual minority youth. However, internet-based approaches vary in terms of their delivery modality and therapist involvement, posing several cost-benefit considerations. For instance, telehealth has the benefit of therapist support and synchronous (i.e., real-time) guidance but does not offer the efficiency or cost-savings of a more automated, self-guided approach. Conversely, digitized, including app-based, interventions offer the promise of broad scale and accessibility, but without the benefit of therapist support, a known common factor at least partially responsible for psychotherapeutic benefit (Krupnick et al., 1996; Wampold, 2015). Internet-based approaches that permit the efficiency and reach of online delivery while also including time-efficient, asynchronous therapist guidance represent one potential solution to maximizing the benefits and minimizing potential costs of internet-based mental health treatments.

Internet-based CBT (ICBT) involves, at a minimum, text-based psychoeducation typically derived from a treatment manual or client workbook, interactive exercises and homework applications, and sometimes also asynchronous therapist guidance (Lindefors & Andersson, 2016). Therapist guidance, when present, can provide both support for technical navigation of the online treatment content and platform and can also involve supportive and problem-solving feedback on exercises and homework applications (Almlöv et al., 2011). Indeed, therapist guidance has been found to reduce treatment dropout and increase treatment effects of ICBT (Baumeister et al., 2014). Overall, ICBT approaches have shown efficacy for a range of populations and conditions (Andersson et al., 2019), with meta-analyses finding large effect sizes and comparable benefits of ICBT compared to in-person CBT, with much less therapist time spent (Andrews et al., 2018).

Despite the promise of ICBT for addressing sexual minority youth’s barriers to accessing identity-affirmative care noted above, no trial to our knowledge has examined the efficacy of therapist-guided ICBT when delivering LGBTQ-affirmative CBT to this population. The growing evidence base supporting the efficacy of LGBTQ-affirmative CBT when delivered across various other modalities (e.g., in-person, group, digitally), plus the availability of therapist and client materials constituting the LGBTQ-affirmative CBT protocol (Pachankis, Harkness, et al., 2022; Pachankis, Jackson, et al., 2022), lays the groundwork for now studying the feasibility, acceptability, and efficacy of ICBT-based delivery of LGBTQ-affirmative CBT to sexual minority youth experiencing mental health challenges across the U.S.

The present study employed a randomized controlled trial to compare the efficacy of up to 10 sessions of therapist-guided LGBTQ-affirmative ICBT to weekly assessments of mental and behavioral health and minority stress over the same period. An assessment-only comparison was selected to control for weekly online engagement and repeated assessment of mental and behavioral health outcomes and minority stress without the psychoeducational or therapist support of LGBTQ-affirmative ICBT. Given the transdiagnostic basis upon which LGBTQ-affirmative CBT is built (Pachankis, Soulliard, et al., 2022), primary outcomes included psychological distress, depression symptoms, anxiety symptoms, suicidal thoughts, problematic drinking, and HIV-transmission-risk behavior. First, we assessed the efficacy of LGBTQ-affirmative ICBT compared to an assessment-only control at 4- and 8-months post-baseline. Second, based on increasing evidence that multi-level forms of stigma, including internalized (e.g., self-directed anti-LGBTQ bias), interpersonal (e.g., harassment, discrimination), and structural (e.g., anti-LGBTQ laws and policies) might influence the efficacy of mental health treatments for stigmatized populations (Hatzenbuehler & Pachankis, 2021), including sexual minority individuals (Millar et al., 2016; Pachankis, Williams, et al., 2020), we explored whether any of these three levels of stigma might moderate the efficacy of LGBTQ-affirmative ICBT. Studying treatment effect moderation can advance personalized treatment by identifying which baseline factors predict optimal response to any given treatment (Kaiser et al., 2022). Third, we examined whether baseline mental health symptom severity (i.e., psychological distress) and participant engagement (i.e., number of sessions completed) moderated treatment efficacy. Finally, we used session completion metrics and qualitative interviews with a subset of intervention recipients to examine the feasibility and acceptability of therapist-guided LGBTQ-affirmative ICBT.

Method

Trial Oversight

This was a single center randomized controlled trial performed at Yale University. Recruitment started in May 2020 and ended in May 2021; the last data were collected in February 2022. The trial protocol was pre-registered at clinicaltrials.gov (NCT04408469) and approved by the Yale University Human Subjects Committee (2000025803). The trial was conducted in accordance with all relevant regulations. All participants provided informed consent and a data and safety monitoring board reviewed participant safety and adverse events. For participants aged 16-17 living in one of the 40 US states in which parental consent was not required for mental health treatment at the time of this study, we relied on participant rather than parental consent (Kerwin et al., 2015; Mustanski et al., 2023; Santelli et al., 1995). For participants aged 16-17 living in one of the 10 US states in which parental consent was required for mental health treatment, we asked these youth whether they would feel safe and comfortable with the research team contacting their parents for consent for the youth to participate or whether they would prefer to receive a referral to resources from a national LGBTQ-affirmative youth organization. This decision was made considering the relative risks and benefits of study participation and parental involvement in this study (Kerwin et al., 2015; Mustanski et al., 2023; Santelli et al., 1995) and the safety and efficacy of CBT for youth (e.g., Ehrenreich-May et al., 2017). In the end, no participant aged 16-17 from one of these 10 US states participated in this study.

Participants

Participants for this study were recruited from social media, mobile dating apps, LGBTQ community organizations, and via peer recruitment and referral. Eligibility criteria included: (a) age 16–25; (b) identification as a sexual minority (e.g., lesbian, gay, bisexual); (c) past 90-day symptoms of depression or anxiety, determined as a score of ≥ 2.5 on either the 2-item depression subscale or 2-item anxiety subscale of the Brief Symptom Inventory-4 as consistent with past trials with sexual and gender minority individuals (Pachankis, Harkness, et al., 2022; Pachankis et al., 2015; Pachankis, McConocha, et al., 2020); (d) consistent internet access; (e) English fluency; and (f) United States residence. Exclusion criteria included: (a) active suicidality assessed through relevant items from the Structured Clinical Interview for DSM-5 (SCID-5; First et al., 2016); (b) active homicidality; (c) active untreated mania or psychosis, assessed through the SCID-5; (d) gross cognitive impairment, assessed through the Modified Telephone Interview for Cognitive Status (TICS-m; Cook et al., 2009); (e) current enrollment in another intervention study; (f) currently enrolled in mental health treatment more than once per month; and (g) receipt of eight or more sessions of CBT within the past year. Figure 1 presents an overview of the study structure and participant flow through each stage of eligibility screening and the full study.

Figure 1.

Study structure and participant flow through study procedures.

Table 1 provides demographic characteristics, stratified by study condition, of the final sample of 120 sexual minority youth and young adults (age M = 20.37, SD = 2.81). A range of sexual minority identities were reported, with the most common identity being bisexual (34.2%). Slightly less than half of participants identified as non-binary (31.7%) or transgender (13.3%). Approximately half identified as White, non-Hispanic (49.1%), while multiracial emerged as the next most-frequently endorsed race/ethnicity category (18.3%). Most (70%) participants had completed at least some college, and almost half (45.8%) reported a somewhat or very hard time meeting basic needs. Participants lived across 38 U.S. states and the District of Columbia.

Table 1.

Demographic Characteristics by Condition

Characteristic	LGBTQ-affirmative ICBT  (n = 60)		Assessment Only  (n = 60)
	n	%	n	%
Age
M	20.82		19.92
SD	2.78		2.84
Gender identity
Cisgender woman	22	36.7	29	48.3
Cisgender man	10	16.7	5	8.3
Transgender woman	3	5.0	2	3.3
Transgender man	3	5.0	8	13.3
Nonbinary (e.g., genderqueer, agender, gender fluid)	17	28.3	12	20.0
Other	5	8.3	4	6.7
Sex assigned at birth
Female	42	70.0	49	81.7
Male	17	28.3	10	16.7
Intersex	1	1.7	1	1.7
Sexual orientation
Asexual	3	5.0	3	5.0
Bisexual	21	35.0	20	33.3
Gay	11	18.3	8	13.3
Lesbian	6	10.0	14	23.3
Pansexual	4	6.7	4	6.7
Queer	11	18.3	10	16.7
Other	4	6.7	1	1.7
Race/ethnicity
American Indian or Alaskan Native	0	0.0	1	1.7
Asian	12	20.0	8	13.3
Black/African American	1	1.7	2	3.3
Native Hawaiian/Pacific Islander	0	0.0	0	0.0
White, Hispanic/Latinx	6	10.0	7	11.7
White, Non-Hispanic/Latinx	29	48.3	30	50.0
Multiracial	9	15.0	11	18.3
Other	3	5.0	1	1.7
Hispanic/Latinx
Yes	11	18.3	11	18.3
No	49	81.7	49	81.7
Education degree
Less than high school	1	1.7	1	1.7
Some high school	9	15.0	12	20.0
High school diploma or GED	4	6.7	9	15.0
Some college	24	40.0	17	28.3
Associate’s degree	1	1.7	2	3.3
4-year college degree	16	26.7	16	26.7
Some graduate school	4	6.7	2	3.3
Advanced graduate school degree	1	1.7	1	1.7
U.S. regionf
Midwest	11	18.3	10	16.7
Northeast	9	15.0	13	21.7
South	28	46.7	25	41.7
West	12	20.0	12	20.0
Difficulty meeting basic needs
Not very hard	32	53.3	33	55
Somewhat hard	26	43.3	18	30
Very hard	2	3.3	9	15
Financial worryh
M	3.23		3.23
SD	1.01		1.17

^aGender was dichotomized as cisgender vs. transgender/nonbinary when conducting condition comparisons.

^bSex assigned at birth was dichotomized to group intersex and female categories together when conducting condition comparisons.

^cSexual orientation was dichotomized as monosexual (gay, lesbian) vs. plurisexual/other when conducting condition comparisons.

^dRace was dichotomized as racial/ethnic minority vs. White, Non-Hispanic when conducting condition comparisons.

^eEducation degree was dichotomized as high school diploma/GED or less vs. some college or more when conducting condition comparisons.

^fCurrent region was grouped according to U.S. Census Bureau divisions (United States Census Bureau, 2021a).

^gDifficulty meeting basic needs was dichotomized as not very hard vs. somewhat/very hard when conducting condition comparisons.

^hFinancial worry was assessed by asking participants to rate “how do you feel about your current financial situation?” on a five-point scale (1 = never worry about it; 5 = always worry about it).

Assigning 60 participants to each condition (LGBTQ-affirmative ICBT vs. assessment-only) and accounting for 20% attrition based on previous internet-based RCTs (Leluțiu-Weinberger et al., 2018; Pachankis & Goldfried, 2010) was estimated to be sufficient to provide 80% power (1-ß) to detect a between-condition difference (p < .05) of d = .40, which is significantly lower than the effect found in comparisons of in-person LGBTQ-affirmative CBT against waitlist (d = .59 for HIV-transmission-risk behavior, d = .55 for depression) and lower than the smallest average meta-analytic effect (d = .66) of ICBT compared to weak control conditions for depression (Andersson & Cuijpers, 2009; Andersson et al., 2014; Spek et al., 2007; Sztein et al., 2018).

Procedure

Screening and Randomization

Potential participants completed an online eligibility screener for this study. Preliminarily eligible participants then completed a phone screener with a research assistant to confirm eligibility; eligible participants were sent an electronic consent form to review with a research assistant and an online baseline assessment survey. Following provision of informed consent and completion of the baseline assessment, a research assistant scheduled telephone calls with each participant. Using a 1:1 randomization programmed into the randomizer feature of Qualtrics (LGBTQ-affirmative ICBT: assessment-only), a research assistant masked to the randomization sequence randomly assigned participants to one of the two study conditions and oriented each participant to their respective condition. Randomization was stratified based on race/ethnicity (White, non-Hispanic vs. person of color), gender identity (cisgender vs. transgender/gender non-binary), and mental health symptoms (depression, anxiety, or both). Participants assigned to the LGBTQ-affirmative ICBT condition were assigned to an available therapist and an introduction call was scheduled. Participants began treatment within 14 days after randomization. All participants completed online self-report survey assessments at baseline and 4- and 8-months post-baseline. Weekly for 10 weeks between the baseline and 4-month post-baseline assessment, all participants also completed brief online self-report survey assessments of mental and behavioral health and minority stress experiences.

Interventions

LGBTQ-affirmative ICBT.

The LGBTQ-affirmative CBT protocol upon which the ICBT materials were based (Pachankis, Harkness, et al., 2022) is a transdiagnostic CBT consisting of 10 modules adapted to address sexual minority stress (Pachankis, Soulliard, et al., 2022); participants had up to 16 weeks to complete 10 sessions each reflecting one the 10 modules. To create LGBTQ-affirmative ICBT, we developed psychoeducational content, in-session exercises, and home practice based directly on the content of the LGBTQ-affirmative CBT client workbook (Pachankis, Jackson, et al., 2022), and adapted the content to match developmentally appropriate stressors (e.g., related to school and family). To ensure that the treatment skills and concepts were understandable to youth, language from the original treatment manual was edited to an 8^th grade reading level (mean Flesch-Kincaid Grade Level = 7.63, SD = 0.52; mean Flesch Reading Ease = 64.84, SD = 3.45). We uploaded this material onto a secure online technical platform used in previous trials of ICBT for other populations and outcomes (e.g., Bjureberg et al., 2022; Ljótsson et al., 2014), which presents the CBT content in online, self-guided sessions. Each session contains psychoeducational text and vignettes regarding stress and mental health, in-session imaginal and written exercises, brief videos discussing CBT skills, and home practice exercises. Therapist support was provided by two counseling psychologists and two advanced clinical psychology doctoral students with experience delivering LGBTQ-affirmative CBT using an in-person format. The therapists received training on providing technical, motivational, and problem-solving support to ICBT participants from the developers of the ICBT platform who had delivered ICBT in other clinical trials with other populations. The role of the therapist was to introduce themselves during two introductory phone calls (i.e., before the first session to provide an overview of the treatment and after the second session as a motivational check-in), provide feedback on homework assignments, and answer questions from the participant within three days. The second author, a clinical psychologist, provided weekly group supervision throughout the trial.

The LGBTQ-affirmative ICBT sessions were based on the published client workbook (Pachankis, Jackson, et al., 2022) and included, in this sequence: (1) setting goals and building motivation for LGBTQ-affirmative CBT; (2) monitoring LGBTQ-related stress and emotions; (3) understanding the nature and emotional impact of LGBTQ-related stress; (4) increasing mindful awareness of LGBTQ-related stress reactions; (5) increasing cognitive flexibility; (6) understanding emotion avoidance; (7) countering emotional behaviors; (8) experimenting with new reactions to LGBTQ-related stress; (9) emotion exposures for countering LGBTQ-related stress; and (10) recognizing accomplishments and looking to the future.

Assessment-only Control.

Participants assigned to the control condition were asked to complete 10 weekly surveys assessing the following experienced over the past week: (a) symptoms of depression (using the Overall Depression Severity & Impairment Scale; Bentley et al., 2014); (b) symptoms of anxiety (using the Overall Anxiety Severity & Impairment Scale; Norman et al., 2006); (c) sexual minority identity-based harassment, discrimination, and rejection (adapted from prior minority stress monitoring studies with SGM; Eldahan et al., 2016; Heron et al., 2018); (d) sexual minority identity-related rejection sensitivity, concealment, and pride (Eldahan et al., 2016; Heron et al., 2018); (e) frequency of alcohol use (Saunders et al., 1993); and (f) HIV-transmission-risk behavior (using a single item developed for this study). Participants were asked to complete the 10 surveys within 16 weeks and could complete no more than one survey per week. Participants assigned to the LGBTQ-affirmative ICBT condition also completed these surveys before completing their weekly LGBTQ-affirmative ICBT session. Completing repeated surveys of behavioral and minority stress experiences has been shown to produce small reductions in mood and behavioral outcomes (Kauer et al., 2012; Korotitsch & Nelson-Gray, 1999) and behavioral symptoms over time (Livingston, 2017).

Measures: Primary Efficacy Outcomes

Psychological Distress

To capture psychological distress, participants completed the 18-item Brief Symptom Inventory (BSI; Derogatis & Melisaratos, 1983). The BSI consists of three 6-item subscales (0 = not at all, 4 = extremely) measuring severity of depression symptoms (e.g., “feeling hopeless about the future”), anxiety symptoms (e.g., “feeling tense or keyed up”), and somatization symptoms (e.g., “feeling weak in parts of your body”). We calculated a mean score to provide a global distress severity index. Cronbach’s alphas for this measure were 0.87, 0.92, and 0.90 at baseline, 4-, and 8-months post-baseline, respectively.

Depression Symptoms

The Center for Epidemiological Studies-Depression Scale (CES-D; Radloff, 1977) is a 20-item measure that assesses frequency of past-week depression symptoms (e.g., “I was bothered by things that usually don’t bother me”) on a 4-point scale from 0 (rarely or none of the time [less than 1 day]) to 3 (most or all of the time [5–7 days]). We calculated a sum score. Cronbach’s alphas were 0.75, 0.84, and 0.84 at baseline, 4-, and 8-months post-baseline, respectively.

The Overall Depression Severity & Impairment Scale (ODSIS; Bentley et al., 2014) measures past-week frequency, severity, and impairment related to depression symptoms using five items (e.g., “in the past week, how often have you felt depressed?”) on a 5-point scale (e.g., 0 = no depression in the past week, 5 = constant depression: felt depressed all of the time). We calculated a sum score. Cronbach’s alphas were 0.91, 0.91, and 0.94 at baseline, 4-, and 8-months post-baseline, respectively.

Anxiety Symptoms

The Overall Anxiety Severity & Impairment Scale (OASIS; Norman et al., 2006) measures past-week frequency, severity, and impairment related to anxiety symptoms using five items (e.g., “in the past week, how often have you felt anxious”) on a 5-point scale (e.g., 0 = no anxiety in the past week, 4 = constant anxiety: felt anxious all of the time and never really relaxed). We calculated a sum score. Cronbach’s alphas were 0.76, 0.80, and 0.86 at baseline, 4-, and 8-months post-baseline, respectively.

Suicidal Thoughts

The 5-item Suicidal Ideation Attributes Scale (SIDAS; van Spijker et al., 2014) assesses past-month suicidal thoughts and their severity. Participants who reported no ideation (0 = never or not at all, 10 = always or extremely) to the first item about the presence of suicidal thoughts (“In the past month, how often have you had thoughts about suicide?”) were not presented with subsequent items regarding the severity of these thoughts. A binary variable was created to contrast participants with no suicidal thoughts or ideation with participants who had any suicidal thoughts or ideation.

Problematic Drinking

Alcohol use was measured using the 10-item Alcohol Use Disorders Identification Test (AUDIT; Saunders et al., 1993), assessing past-three-month frequency of drinking, alcohol dependence, and alcohol-related problems. Responses were summed to create a composite score in which higher scores indicate higher levels of problematic alcohol use. Cronbach’s alphas were 0.84, 0.87, and 0.86 at baseline, 4-, and 8-months post-baseline, respectively.

HIV-transmission-risk Behavior

Past 90-day HIV-transmission-risk behavior was operationalized as any sexual act that could result in HIV transmission to or from the participant. Specifically, HIV-transmission-risk behavior was defined as condomless penetrative sex when both the participant and their partner (if a main partner) were either PrEP non-adherent, not on PrEP, or reporting a detectable HIV serostatus. Consequently, HIV-transmission-risk behavior was calculated as a function of the following variables assessed via a self-report survey: a) partner type (i.e., main partner or casual partner); b) participant and primary partner(s)’ PrEP use and adherence (i.e., whether the participant or their primary partner(s) reported taking PrEP on ≥ 4 days/week); c) participant and primary partner(s)’ viral load if HIV-positive (i.e., self-report of whether they and/or their partner(s) had a detectable or undetectable serostatus); d) type of sexual act (i.e., penetrative sex); and e) condom use (i.e., with or without a condom). We dichotomized HIV-transmission-risk behavior to indicate any risk versus no risk due to extreme zero inflation in that only six participants reported any HIV-transmission-risk behavior at baseline.

Measures: Hypothesized Multi-level Stigma Moderators of Efficacy

Internalized Stigma

The 3-item Internalized Homonegativity Subscale of the Lesbian, Gay, and Bisexual Identity Scale (LGBIS; Mohr & Kendra, 2011) assesses internalized stigma related to one’s sexual minority identity (e.g., “If it were possible, I would choose to be straight”) on a 6-point scale from 1 (disagree strongly) to 6 (agree strongly). We calculated a mean total score. Cronbach’s alphas were 0.90, 0.89, and 0.89 at baseline, 4-, and 8-months post-baseline, respectively.

Interpersonal Stigma

The 14-item Heterosexist Harassment, Rejection, and Discrimination Scale (HHRDS; Szymanski, 2006) measures past-year experiences of interpersonal stigma directed toward one’s sexual or gender minority identity (e.g., “How many times have you been called a slur or a name related to your LGBTQ+ identity?”) using a 6-point scale (1 = never happened in the past year, 6 = almost all of the time [more than 70% of the time]). We calculated a mean total score. Cronbach’s alphas were 0.89, 0.90, and 0.88 at baseline, 4-, and 8-months post-baseline, respectively.

Structural Stigma

Participants provided their current ZIP code, which allowed us to link their responses to a county-level index of aggregated explicit and implicit attitudes toward LGBTQ people among residents of their county (Lattanner et al., 2021). These data were derived from Project Implicit (Nosek et al., 2007). The structural stigma index was standardized across all US counties represented in Project Implicit, rather than within the present sample.

Measures: Feasibility and Acceptability

In addition to examining session completion metrics (i.e., number of sessions completed), we used qualitative interviews to assess the feasibility and acceptability of LGBTQ-affirmative ICBT. We specifically enlisted 20 participants assigned to the LGBTQ-affirmative ICBT condition to participate in 60-minute semi-structured exit interviews with trained research staff. These participants were purposively selected to represent a range of sexual orientations, gender identities and modalities, races, ethnicities, and number of intervention sessions completed. Interviewed and non-interviewed participants did not differ in terms of age, gender identity, race/ethnicity, sex assigned at birth, sexual orientation, difficulty meeting basic needs, psychological distress, depression symptoms, anxiety symptoms, suicidal thoughts, or problematic drinking. However, interviewed participants completed significantly more sessions than non-interviewed participants (t = 4.03, p < .001).

During interviews, participants were asked to share their feedback on the intervention in relation to several topics, including feasibility (e.g., “How many sessions did you complete? What got in the way of completing more?”), platform accessibility (e.g., “How would you rate its user-friendliness?”), therapist role (e.g., “Were there any parts of the treatment in which you felt that additional contact with your therapist would have been helpful?”), content accessibility (e.g., “What was a time in treatment, if any, that you would say you were really engaged?”; “Can you think of anything you would have liked to work on that you didn’t get to?”), and contextual factors (e.g., “How did circumstances related to COVID-19 impact your ability to complete the treatment sessions?”). Participants were also offered the opportunity to provide general feedback on the treatment or the study as a whole.

Data Analysis

Intervention efficacy was assessed using an intent-to-treat analysis including all participants (n = 120). First, dependent variables and stigma moderators were assessed for normality using skewness thresholds of ≥ ±2 and kurtosis thresholds of ≥ ±7 and outliers using boxplots (Byrne, 2010; Hair et al., 2010). BSI, CESD, ODSIS, OASIS, and all stigma moderators met assumptions of normality and no univariate outliers were detected. SIDAS, AUDIT, and HIV-transmission-risk behavior scores did not meet the normality assumption. Therefore, SIDAS and HIV-transmission-risk behavior scores were dichotomized to indicate any versus no behavioral risk; models using the AUDIT were fit using a generalized linear mixed effects models with negative binomial distribution accounting for zero inflation to handle the negative skew. Second, we conducted missing data analyses to examine whether participants’ likelihood of missing a follow-up assessment was related to study outcomes: No data were missing at baseline and follow-up data were missing at random (MAR).

To examine the efficacy of LGBTQ-affirmative ICBT, we used linear mixed models with maximum likelihood estimation and an unstructured covariance matrix to test the Condition × Time interaction effects between conditions over time for all continuous mental health outcomes (BSI, CESD, ODSIS, OASIS). Unstructured covariance matrices are well-suited for repeated measures analyses with only three timepoints as they do not impose any constraints on the values (Fitzmaurice et al., 2012). Further, we did not specify any hypotheses regarding covariance between timepoints. For binary outcomes (any suicidal thoughts reported on the SIDAS, any problematic drinking reported on the AUDIT, any HIV-transmission-risk behavior), we tested intervention efficacy using a generalized linear mixed model with a logit link and a binomial distribution to test the Condition × Time effect predicting odds of each binary outcome between conditions over time. Primary efficacy analyses between conditions included baseline (time = 0), 4-month post-baseline follow-up (time = 1), and 8-month post-baseline follow-up (time = 2) to model the Condition × Time effect of receiving LGBTQ-affirmative ICBT (condition = 1) versus assessment only (condition = 0). Effect sizes (Cohen’s d) for linear mixed models were calculated as mean pre-post change (e.g., baseline to 4-month post-baseline follow-up, baseline to 8-month post-baseline follow-up) in the LGBTQ-affirmative ICBT intervention minus the pre-post change in the assessment-only condition, divided by the pooled baseline standard deviation (Morris, 2008).

To examine whether Condition × Time effects varied as a function of baseline levels of stigma (i.e., internalized stigma, interpersonal stigma, and structural stigma), we conducted moderation analyses using linear mixed models with maximum likelihood estimation, unstructured covariance matrices, and three-way interactions of Condition × Time × Stigma. Moderation by internalized, interpersonal, and structural stigma were entered together into multivariable models for each outcome, except sexual behavior risk due to very low prevalence in this sample. Significant three-way interactions of Condition × Time × Stigma were further probed and plotted to contrast high stigma (one standard deviation above the national mean) with low stigma (one standard deviation below the national mean).

We examined whether baseline mental health symptom severity (i.e., psychological distress measured in terms of the BSI) moderated treatment efficacy by testing the Condition × Time × Baseline Psychological Distress interaction. We also examined whether participant engagement (i.e., number of sessions completed) was associated with outcomes by examining the Time × Sessions Completed interaction among participants randomized to the LGBTQ-affirmative ICBT condition.

All results were evaluated at p < .05. We report means, standard errors, 95% confidence intervals, and effect sizes. Although linear mixed models are well-suited to handle MAR, we also conducted sensitivity analyses using multiple imputation of missing data with the MICE package in R (van Buuren & Groothuis-Oudshoorn, 2011). Because we found no appreciable differences in models run under the assumptions of linear mixed models with maximum likelihood estimation and those run using multiple imputation, we present the linear mixed models with the non-imputed dataset.

To assess intervention feasibility and acceptability, qualitative interview data were analyzed using a coding reliability form of thematic analysis (Boyatzis, 1998). As delineated by Braun and Clarke (2021), our qualitative approach aimed to objectively capture aspects of feasibility and acceptability of LGBTQ-affirmative ICBT reported by interviewed participants. First, all interviews were transcribed by an external confidential transcription service. We then generated an initial codebook based on the exit interview guide, with themes and codes identified prior to coding. We identified 10 themes of feasibility and acceptability: barriers to treatment completion, time commitment, platform accessibility, platform look and feel, therapist relationship, content accessibility, treatment skills, minority stress, intersectionality, and weekly assessments. After developing the initial codebook, two trained coders on the research team independently coded approximately three transcripts per week. The coders met weekly with the second author to clarify code definitions, resolve discrepancies, and make modifications to the codebook as needed. While coding the data, the coding team took a reflexive approach by openly discussing the need to add or modify certain codes (Morrow, 2005). We reached saturation after five rounds of coding. Coders then applied the final codebook to all interview transcripts.

Results

Intervention Efficacy

Across assessments, BSI, CESD, ODSIS, OASIS, and AUDIT scores decreased significantly; however, no Condition × Time interaction was significant, indicating that the significant decrease in these symptoms across conditions did not differ between LGBTQ-affirmative ICBT and assessment only (see Table 2). No effects of time were observed for odds of suicidal thoughts or HIV-transmission-risk behavior. Over two-thirds (68%) of participants in the ICBT condition completed the post-intervention assessment; 80% completed the 8-month post-baseline follow-up assessment. In the assessment-only condition, 88% of participants completed the post-intervention assessment; 82% completed the 8-month post-baseline follow-up assessment.

Table 2.

Condition × Time Comparisons of LGBTQ-affirmative ICBT and Assessment Only

	LGBTQ-affirmative ICBT n = 60		Assessment Only n = 60		Condition × Time n = 120
	M	SE	M	SE	β	95% CI	p-value	Cohen’s d
Psychological Distress (BSI)
Baseline	1.43	0.07	1.57	0.09
4-Month Post-Baseline	1.22	0.11	1.34	0.11	0.05	[−0.25, 0.34]	0.75	0.03
8-Month Post-Baseline	1.07	0.10	1.21	0.09	−0.03	[−0.29, 0.23]	0.81	−0.01
Depression Symptoms (CESD)
Baseline	14.17	0.36	15.41	0.47
4-Month Post-Baseline	12.68	0.57	13.45	0.50	0.44	[−1.10, 1.98]	0.57	0.15
8-Month Post-Baseline	12.44	0.48	13.38	0.50	0.29	[−1.37, 1.95]	0.73	0.10
Depression Severity and Impairment (ODSIS)
Baseline	7.20	0.54	8.40	0.58
4-Month Post-Baseline	6.07	0.73	6.98	0.58	0.47	[−1.49, 2.44]	0.63	0.07
8-Month Post-Baseline	6.15	0.74	6.63	0.73	0.63	[−1.32, 2.58]	0.53	0.16
Anxiety Severity and Impairment (OASIS)
Baseline	9.12	0.39	8.97	0.44
4-Month Post-Baseline	7.32	0.53	7.45	0.51	−0.24	[−1.54, 1.05]	0.71	−0.09
8-Month Post-Baseline	7.25	0.52	7.52	0.60	−0.34	[−2.00, 1.31]	0.68	−0.13
Alcohol Use (AUDIT)
Baseline	2.78	0.44	3.43	0.64
4-Month Post-Baseline	2.37	0.49	3.42	0.76	−0.11	[−0.61, 0.40]	0.68	−0.07
8-Month Post-Baseline	3.02	0.61	3.25	0.59	−0.04	[−0.53, 0.45]	0.88	0.01
	n	%	n	%	β	95% CI	p-value	OR
Any Suicidal Thoughts
Baseline	37	61.2	35	58.3
4-Month Post-Baseline	25	61.0	24	45.3	−0.13	[−0.91, 0.66]	0.75	0.88
8-Month Post-Baseline	28	58.3	25	51.0	0.22	[−0.62, 1.06]	0.61	1.25
Any HIV-transmission-risk behavior
Baseline	4	6.7	3	5.0
4-Month Post-Baseline	2	4.9	4	7.5	−0.96	[−3.15,1.24]	0.39	0.38
8-Month Post-Baseline	4	8.3	3	6.1	0.10	[−2.06, 2.25]	0.93	1.10

Note. Sample sizes were n = 120 (baseline), n = 94 (4-month post-baseline follow-up), and n = 97 (8-month post-baseline follow-up). LGBTQ-affirmative ICBT sample sizes were n = 60, n = 41, n = 48 at baseline and 4 and 8 months post-baseline, respectively. Assessment only sample sizes were n = 60, n = 53, n = 49 at baseline and 4 and 8 months post-baseline, respectively. Means and standard errors are unadjusted.

Intervention Efficacy as a Function of Multi-level Stigma

At baseline, the mean level of internalized stigma was 2.03 (SD = 1.28; minimum = 1.00; maximum = 6.00), mean interpersonal stigma was 1.87 (SD = 0.76; minimum = 1.00; maximum = 4.29), and mean structural stigma was −0.25 (SD = 0.33; minimum = −0.83; maximum = 0.84). The three stigma moderators were weakly correlated (Pearson’s r = −0.09 to 0.19).

Internalized stigma served as a moderator of the efficacy of LGBTQ-affirmative ICBT in predicting psychological distress. Specifically, the Condition × Time × Internalized Stigma interaction predicted marginally significant change in psychological distress from baseline to 4-months post-baseline, indicating marginally greater efficacy of LGBTQ-affirmative ICBT compared to assessment only for participants experiencing high levels of internalized stigma (b = −0.21, p = 0.05, 95% CI [−0.42, 0.00]; Table 3). Probing this effect at one standard deviation below and above the mean of internalized stigma found that LGBTQ-affirmative ICBT participants reporting high internalized stigma exhibited a significant decrease in psychological distress from baseline to 4-months post-baseline (b = −0.40, p < .05, 95% CI [−0.71, −0.10]), whereas assessment-only participants reporting high internalized stigma experienced no change in psychological distress (b = −0.10, p = 0.47, 95% CI [−0.36, 0.16]). Further, LGBTQ-affirmative ICBT participants reporting low internalized stigma did not experience a significant change in psychological distress (b = −0.10, p = 0.51, 95% CI [−0.40, 0.20]), whereas assessment-only participants reporting low internalized stigma experienced a significant improvement in psychological distress (b = −38, p < 0.05, 95% CI [−0.65, −0.12]).

Table 3.

Condition × Time Comparisons of LGBTQ-affirmative ICBT and Assessment Only Conditions Moderated by Internalized, Interpersonal, and Structural Stigma

	Condition × Time × Internalized Stigma			Condition × Time × Interpersonal Stigma			Condition × Time × Structural Stigma
	β	95% CI	p-value	β	95% CI	p-value	β	95% CI	p-value
Psychological Distress (BSI)
4-Month Post-Baseline	−0.21*	[−0.42, 0.00]	0.05	−0.14	[−0.52, 0.24]	0.48	−1.73**	[−2.75, −0.70]	0.001
8-Month Post-Baseline	−0.08	[−0.28, 0.13]	0.46	0.11	[−0.24, 0.45]	0.55	−0.75	[−1.66, 0.17]	0.11
Depression Symptoms (CESD)
4-Month Post-Baseline	−0.98	[−2.14, 0.19]	0.10	1.23	[−0.87, 3.32]	0.25	−5.61	[−11.24, 0.02]	0.05
8-Month Post-Baseline	−0.15	[−1.47, 1.22]	0.86	−0.21	[−2.48, 2.06]	0.85	−3.69	[−9.62, 2.25]	0.22
Depression Severity and Impairment (ODSIS)
4-Month Post-Baseline	−1.41	[−2.86, 0.03]	0.06	0.09	[−2.50, 2.69]	0.95	−5.41	[−12.33, 1.51]	0.12
8-Month Post-Baseline	−0.49	[−2.04, 1.06]	0.53	0.00	[−2.58, 2.59]	1.00	−5.45	[−12.29, 1.38]	0.12
Anxiety Severity and Impairment (OASIS)
4-Month Post-Baseline	−0.19	[−1.18, 0.80]	0.70	0.37	[−1.41, 2.15]	0.68	−3.98	[−8.76, 0.82]	0.10
8-Month Post-Baseline	0.56	[−0.74, 1.85]	0.40	−0.05	[−2.23, 2.12]	0.96	−4.86	[−10.60, 0.88]	0.10
Alcohol Use (AUDIT)
4-Month Post-Baseline	−0.13	[−0.35, 0.10]	0.27	−0.20	[−0.44, 0.04]	0.10	0.28	[−0.13, 0.70]	0.18
8-Month Post-Baseline	−0.19	[−0.45, 0.08]	0.16	−0.08	[−0.35, 0.20]	0.59	0.32	[−0.06, 0.70]	0.10
Any Suicidal Thoughts (SIDAS) a
4-Month Post-Baseline	−0.28	[−0.97, 0.42]	0.44	0.41	[−0.82, 1.63]	0.51	−3.29	[−6.58, 0.01]	0.05
8-Month Post-Baseline	0.05	[−0.70, 0.79]	0.90	0.23	[−1.02, 1.48]	0.72	−1.32	[−4.66, 2.02]	0.44

Note. n = 119; one participant was excluded from moderation analyses due to missingness on stigma moderators.

^*p < .05

^**p < .01.

^aBased on logistic regression with binary outcomes.

Structural stigma also served as a significant moderator of the efficacy of LGBTQ-affirmative ICBT in predicting psychological distress. Specifically, the Condition × Time × Structural Stigma interaction significantly predicted change in psychological distress from baseline to 4-months post-baseline, indicating greater efficacy of LGBTQ-affirmative ICBT compared to assessment only for participants experiencing high levels of structural stigma (b = − 1.73, p < 0.01, 95% CI [−2.75, −0.70]; Table 3). Probing this interaction at one standard deviation below and above the mean of structural stigma found that LGBTQ-affirmative ICBT participants living in high structural stigma counties exhibited a significant decrease in psychological distress from baseline to 4-months post-baseline (b = −2.09, p < .01, 95% CI [−3.34, −0.85]), whereas assessment-only participants from high structural stigma counties experienced no change in psychological distress (b = 0.25, p = 0.45, 95% CI [−0.39, 0.88]) (Figure 2). Further, participants who received LGBTQ-affirmative ICBT living in low structural stigma counties experienced a significant increase in psychological distress (b = 0.84, p < .05, 95% CI [−0.015, 1.53), whereas assessment-only participants living in low structural stigma counties experienced a significant decrease in psychological distress (b = −0.53, p < 0.05, 95% CI [−0.94, −0.12).

Figure 2.

Change in psychological distress for participants experiencing high and low structural stigma, comparing LGBTQ-affirmative ICBT and assessment-only conditions from baseline to 4-months post-baseline.

High stigma is one standard deviation above the national mean and low stigma is one standard deviation below the national mean.

Structural stigma also served as a moderator of the efficacy of LGBTQ-affirmative ICBT in predicting depression symptoms. Specifically, the Condition × Time × Structural Stigma interaction predicted marginally significant change in depression symptoms from baseline to 4-months post-baseline, indicating marginally greater efficacy of LGBTQ-affirmative ICBT compared to assessment only for participants experiencing high levels of structural stigma (b = − 5.61, p = 0.05, 95% CI [−11.24, 0.02]; Table 3). Probing this interaction at one standard deviation below and above the mean of structural stigma found that LGBTQ-affirmative ICBT participants living in high structural stigma counties exhibited a significant decrease in depression symptoms from baseline to 4-months post-baseline (b = −7.44, p < .05, 95% CI [−14.13 −0.75]), whereas assessment-only participants from high structural stigma counties did not experience significant change in depression symptoms (b = −0.83, p = 0.63, 95% CI [−4.25, 2.59]). Participants who received LGBTQ-affirmative ICBT living in low structural stigma counties did not experience a significant change in depression symptoms distress (b = 1.95, p = 0.31, 95% CI [−1.77 .5.67]), however assessment-only participants living in high structural stigma counties experienced a significant improvement in depression symptoms (b = −2.61, p < .05, 95% CI [−4.86, −0.38]), although of smaller magnitude than that experienced by participants living in high structural stigma counties who received LGBTQ-affirmative ICBT.

Structural stigma also served as a moderator of the efficacy of LGBTQ-affirmative ICBT in predicting suicidal thoughts. Specifically, the Condition × Time × Structural Stigma interaction predicted marginally significant change in suicidal thoughts from baseline to 4-months post-baseline, indicating marginally greater efficacy of LGBTQ-affirmative ICBT compared to assessment only for participants experiencing high levels of structural stigma (b = − .329, p = 0.05, 95% CI [−6.58, 0.01]; Table 3). However, probing this interaction at one standard deviation below and above the mean of structural stigma revealed that none of the slopes were statistically significant.

Effect of Participant Engagement and Baseline Mental Health Symptom Severity

In testing the Condition × Time × Baseline Psychological Distress interaction that examined whether baseline mental health symptom severity (i.e., BSI) moderated treatment efficacy, results revealed that baseline mental health symptom severity did not affect treatment efficacy. Supplemental Table 2 displays these results.

In analyses conducted within the LGBTQ-affirmative ICBT condition, significant Time × Sessions Completed interactions existed when predicting depression symptoms (i.e., ODSIS) and anxiety symptoms (i.e., OASIS), such that a greater number of sessions completed was associated with greater reductions in depression and anxiety symptoms at the 8-month post-baseline follow-up. Supplemental Table 1 displays these results.

Treatment Feasibility and Acceptability

Participants in the LGBTQ-affirmative ICBT condition completed a mean of 6.08 (SD = 3.80) sessions. As shown in Figure 1, 26 out of 60 (43.3%) of participants in the LGBTQ-affirmative ICBT condition completed all 10 intervention sessions. Of the 16 participants who elected to stop receiving LGBTQ-affirmative ICBT, 10 (63%) explicitly discussed the time commitment of LGBTQ-affirmative ICBT combined with other external life demands as the reason for treatment discontinuation. Two adverse events, involving suicidal ideation and/or attempts, were known to occur during the study and were handled according to the study’s clinical protocols, including review by the study’s data and safety monitoring board.

Qualitative findings related to study feasibility are summarized in Table 4. With regard to “barriers to treatment completion,” the majority of participants mentioned difficulties in balancing the self-guided nature of the treatment with their other life demands, including external life circumstances (11/20, 55%) such as school and occupational obligations. Additionally, 9/20 (45%) participants noted a specific theme of “time commitment” in terms of difficulty completing the treatment. At the same time, 10/20 (50%) participants expressed that the time commitment was feasible and made sense in light of the treatment benefits, while 8/20 (40%) expressed general concerns about the time commitment and 9/20 (45%) shared a desire for shorter session content. In terms of “platform accessibility,” most participants (18/20, 90%) were satisfied with the user-friendliness of the online intervention platform and reported finding it easy to navigate (e.g., accessing sessions, sending messages to their therapist). However, regarding the “platform look and feel,” more than half shared concerns regarding its aesthetics and design (13/20, 65%), such as a lack of vibrant color or an insufficient number of graphics. Finally, for the “therapist relationship,” participants expressed mixed feelings: 7/20 (35%) stated satisfaction with the therapist’s role, while 8/20 (40%) expressed desire for more interaction with their therapist to facilitate their progression through the treatment. This extended to perceptions of participants’ relationships with their therapist: participants expressed feelings of trust (5/20, 25%), support (7/20, 35%), or other positive regard (9/20, 45%), yet almost half (9/20, 45%) reported concerns regarding the brief therapist contact. In addition to two introductory phone calls from the therapist (i.e., before the start of treatment and after the second session), therapists provided written responses to participants on average once per session (M = 1.11, SD = 0.32), typically to provide feedback on the participants’ homework.

Table 4.

Qualitative Results on Feasibility and Acceptability of LGBTQ-affirmative ICBT

Feasibility/acceptability themes	Summary of findings	Participant Quote
Barriers to treatment completion	The self-guided, weekly format of the treatment posed difficulties for devoting sufficient time and energy.	“Especially when you have other things to be doing in your own life, it’s hard to remember that and commit to that and stay engaged with it especially with the way it’s formatted.”
Time commitment	Participants were divided in their feelings towards the time commitment required of each session.	“[Session length] felt just right because the time-consuming parts weren’t just sitting there confused and frustrated. It was actually doing something.”
		“[If] it had been shorter […] I don’t think it would’ve added in stress to that week.”
Platform accessibility	The platform was perceived as straightforward and easy to navigate.	“I felt pretty comfortable with the platform right away, and I didn’t find it difficult to use.”
Platform look and feel	Participants suggested changes to platform aesthetics to make the platform more welcoming.	“It was designed to be effective, but I feel like a little bit of a warmer environment could have been nicer, like warmer online colors.”
Therapist relationship	Participants expressed positive feelings regarding their therapist’s role and some requested even more therapist contact to form a more meaningful relationship with their therapist	“At least for me, I would’ve been a lot harder to be engaged had I not felt like there was somebody actually on the other end reading[...] I really appreciated my therapist.”
		“I feel like I would have had to have like some kind of connection for them to be helpful. I think that would have required just more interaction throughout.”
		“I think maybe another check-in towards the middle would’ve have been helpful for me. We did the initial thing and then we did a check-in, but it was very early into the sessions. I think maybe another one, a little bit before the end would have been helpful for me, just to maybe talk about my goals, and what progress I had made on them and what my focus could be for like the rest of it, just a guiding check-in.”
Content accessibility	The treatment content generally addressed participants’ treatment goals and allowed them to uncover new goals.	“I did not know the goals that I would then discover or start to gain over the course of the modules and the ways it made me think that I didn’t realize I needed to think about or didn’t have a concept before I started.”
Treatment skills	Participants reported a variety of new skills learned in treatment, and most reported still using these skills.	“I think with that self-confidence that was gained over the course of the modules, either in the confidence gained from not using coping mechanisms or the confidence gained from adjusting those automatic responses. I still have those thoughts, but my brain now pulls back up the file of that module again, as a remembering kind of thing and the ways I wrote about affirmations and stuff about my relationships and the way people care and the way I care about them. It did become a thing of like, yeah, I’ll see things online and it reminds me of the module, and I’m like, I feel like I just have a little bit more confidence especially in the relationship with my significant other. I think that was a very subtle change of the way memory works to reinforce some of those lessons.”
Minority stress	The treatment enabled participants to identify past and ongoing minority stress experiences and better navigate minority stressors.	“It just had a big influence on me, and also, just, I guess, I had never really thought about how just LGBTQ-related stress that we learned about in the study impacted me personally, even in small daily ways. That opened my eyes to what I had been experiencing but hadn’t really labeled for myself.”
Intersectionality	Participants recommended enhancing the intersectional nature of the treatment through text discussions, prompts, and vignettes.	“There’s so many different types of […] LGBTQ youth that could benefit from this training. There could be some steps done to allow them to feel this training is speaking to them so that they can be a little more comfortable.”
Weekly assessments	Weekly assessments presented a burden to many participants due to difficulty observing progress, rumination on their responses, and perceived tediousness of weekly surveys.	“I did get kind of bored with that, I guess. There’s not really anything that I can think of that you could do to avoid that […] It made it hard sometimes to really make sure I was giving a genuine answer, or if I was just trying to answer and get through it.”

Qualitative findings also informed participant perceptions of treatment acceptability. Overall, participants’ feedback suggested that LGBTQ-affirmative ICBT was quite acceptable (see Table 4). For general “content accessibility,” most participants (13/20, 65%) expressed that the treatment content was relevant to their treatment goals outlined in the first session. “Treatment skills” was a common theme, with participants (12/20, 60%) reporting that the LGBTQ-affirmative ICBT treatment allowed them to develop new coping strategies to manage stressors—17/20 (85%) reported learning skills in treatment, and 15/20 (75%) reported that they continued using these skills in their everyday lives after treatment. With regard to acceptability related to “minority stress,” the treatment also positively impacted participants’ perceptions of their sexual or gender identity (11/20, 55%) and helped to reduce their minority stress (9/20, 45%). In particular, 14/20 (70%) participants mentioned that the treatment allowed them to identify minority stress experiences in their own lives. Participants also suggested changes to treatment content; for instance, “intersectionality” was a relevant theme, with 13/20 (65%) participants expressing a desire for more content addressing identities that co-occur with their sexual identity, including their gender identity, race/ethnicity, culture, religion, and immigrant status. Lastly, participants reported concerns and difficulty with completing weekly assessments as a part of the treatment (15/20, 75%), with some participants specifically mentioning challenges in answering the same questions every week.

Discussion

Across mental and behavioral health outcomes, LGBTQ-affirmative ICBT was generally not associated with greater improvement compared to only completing assessments of mental and behavioral health and minority stress, although reductions in all outcomes except alcohol use were observed in both conditions. However, participants with more stigma experiences received greater benefit from LGBTQ-affirmative ICBT compared to the assessment-only condition. Consistent with previous studies of LGBTQ-affirmative CBT delivered in-person (e.g., Millar et al., 2016), a moderation effect found that sexual minority youth who reported greater internalized stigma reported marginally significantly greater reductions in psychological distress after receiving LGBTQ-affirmative ICBT compared to after receiving assessment only. The second moderation finding – that sexual minority youth who lived in U.S. counties with greater degrees of structural stigma reported significantly greater reductions in psychological distress and marginally significantly greater reductions in depression symptoms and suicidal thoughts, after receiving LGBTQ-affirmative ICBT – partially supports previous research (Hatzenbuehler & Pachankis, 2021). These moderation effects did not persist at 8-month post-baseline follow-up. Interpersonal stigma, in the form of identity-based harassment, rejection, and discrimination, did not moderate treatment effects. Qualitative interview results with a subset of participants randomized to the LGBTQ-affirmative ICBT condition found that LGBTQ-affirmative ICBT was generally feasible and acceptable for sexual minority youth. At the same time, these results found that the relatively high attrition across sessions in this study could likely be explained by lack of motivation and competing time demands for completing the treatment.

The relative efficacy of LGBTQ-affirmative CBT has primarily been established in waitlist-controlled trials when delivered using in-person and group-based formats (Craig, Eaton, et al., 2021; Craig, Leung, et al., 2021; Pachankis et al., 2015; Pachankis, McConocha, et al., 2020). Only one trial to our knowledge has compared LGBTQ-affirmative CBT to an active condition, finding relatively small-to-moderate improvements across mental and behavioral health outcomes when compared to LGBTQ-affirmative community treatment without CBT content (Pachankis, Harkness, et al., 2022). Rather than employ an active control condition, we employed an assessment-only control suited to this early stage of efficacy testing for this novel treatment approach and delivery modality. However, based on the present results showing that the assessment-only control showed a similar pattern of associations with study outcomes as LGBTQ-affirmative ICBT, future research is needed to determine whether weekly assessment only might itself offer an efficacious approach for improving sexual minority youth’s mental and behavioral health. Indeed, even brief interventions for sexual minority youth, without therapist involvement, have shown small effects, for instance on internalized stigma and hopelessness following a psychoeducational and interactive 20-minute self-guided online program focused on minority stress (Shen et al., 2022), and even medium effects on psychological distress, depression, suicidality, and alcohol use problems three months after completing three 20-minute writing sessions focused on minority stress and coping (Pachankis, Williams, et al., 2020). The present study adds to this evidence by suggesting that an assessment-only control condition does not necessarily yield weaker effects than a more intensive LGBTQ-affirmative CBT approach, consistent with existing evidence that repeated assessment of mental and behavioral health and minority stress experiences can be associated with improved mental and behavioral health (Kauer et al., 2012; Korotitsch & Nelson-Gray, 1999; Livingston, 2017).

The overall pattern of null effects for the primary efficacy analyses of this trial are qualified by several stigma-related treatment efficacy moderation findings. The search for multi-level forms of stigma as moderators of treatment efficacy in this trial was guided by increasing evidence that members of stigmatized populations benefit differentially from treatments depending on the level of stigma to which they are exposed (e.g., internalized versus structural stigma) and whether the treatments they receive are adapted, or not, to address stigma-specific experiences (Hatzenbuehler & Pachankis, 2021). The present finding that participants with higher levels of internalized stigma benefited more from ICBT in terms of marginally greater reductions in psychological distress compared to the assessment-only control is consistent with previous evidence. Namely, previous research has found that LGBTQ-affirmative CBT, when delivered in-person, yields higher benefit across mental and behavioral health outcomes for LGBTQ individuals with higher internalized stigma, perhaps because of the direct relevance of this minority stress-focused treatment approach to individuals with high levels of stigma experiences (Millar et al., 2016). The present findings add to this research and suggest that future treatments might consider further addressing the role of internalized stigma in LGBTQ youths’ mental health, including using relevant LGBTQ-affirmative CBT techniques involving addressing minority stress cognitions like internalized stigma, related affective responses like shame, and behavioral reactions like social avoidance that can be driven by internalized stigma (Newcomb & Mustanski, 2010; Pachankis et al., in press; Pachankis, Soulliard, et al., 2022).

The few studies that have recently examined structural stigma as a moderator of treatment efficacy have tended to examine this effect among stigmatized individuals who have received general mental health treatments rather than treatments specifically adapted for a stigmatized population. For instance, spatial meta-analyses have found that the degree of structural sexism or anti-Black structural racism present in the communities in which a trial was conducted can undermine the efficacy of mental health treatments when delivered to samples composed of mostly girls or mostly Black youth, respectively (Price et al., 2021; Price et al., 2022). Yet, the trials included in these meta-analyses focused on general psychotherapies created for the general youth population, rather than on treatments adapted to address stigma-specific experiences underlying poor mental health. To our knowledge, no previous study has examined structural stigma as a moderator of treatment efficacy for a mental health treatment adapted to address the distinct concerns of the stigmatized population receiving the treatment. Contrary to the above spatial meta-analyses, the present trial finds that structural stigma moderated the efficacy of LGBTQ-affirmative ICBT such that sexual minority youth living in counties with higher structural stigma experienced greater reductions in psychological distress from ICBT compared to the assessment-only control. This finding, when compared to previous research on structural stigma moderation for non-adapted treatments, suggests that the direction of treatment heterogeneity effects by structural stigma might depend on whether the treatment being studied has been adapted to the stigmatized population for whom it is being delivered. Indeed, structural stigma might have the potential to undermine treatment efficacy of non-adapted treatments when the individuals who receive the non-adapted treatment exist in communities that chronically undermine their attempts at mental health improvement and coping. Whereas, stigma-adapted treatments, like LGBTQ-affirmative CBT, might be particularly beneficial in such high-stigma communities because they directly address the stigma-related challenges facing LGBTQ youth in such communities and provide mental health support to sexual minority youth who might otherwise not be able to receive identity-affirming support in their local communities. At the same time, these moderation results also find that individuals living under conditions of low structural stigma experienced an increase in distress upon receiving LGBTQ-affirmative ICBT and future research is needed to understand the mechanisms behind this unexpected effect and what treatment approaches might be beneficial in such settings. Perhaps delivering a minority stress focused intervention against a relatively supportive structural backdrop unnecessarily introduces negative costs of focusing on stigma and stress.

Although generally consistent with the few other existing studies of LGBTQ-affirmative treatments for LGBTQ youth tested against control conditions (e.g., Shen et al., 2022), the null efficacy effects of this trial contradict this study’s hypotheses and can help advance the development and dissemination of future interventions for LGBTQ youth. First, this study was open to a relatively broad array of LGBTQ youth, regardless of minority stress exposure and using relatively low mental health inclusion criteria. The relatively broad inclusion criteria used here suggest that future research of mental health treatments might want to consider including those LGBTQ youth who might be particularly likely to benefit from it, including those with more stigma experiences as suggested above. Second, using a relatively low cutoff on psychological distress and no other mental health cutoff might have introduced a floor effect, whereby treatment gain on these measures was impeded by initially relatively low scores. Such a floor effect might also explain our finding that baseline mental health symptom severity was not significantly associated with improvement, which contradicts other studies of ICBT with other populations that have found the treatment to work best when delivered to those with greater symptom presentation at baseline (Edmonds et al., 2018; Niles et al., 2021). Third, although some participants noted that the therapist interaction time was of appropriate duration, other participants qualitatively noted that more therapist time would have been preferable. Specifically, some participants noted that additional check-ins, including more in-depth conversations than the prescribed introductions and homework feedback, would have been useful. Therefore, future tests of LGBTQ-affirmative ICBT might consider matching actual therapist time to participant preference or increasing therapist time overall. Increased therapist support might be especially needed for participants such as those in this study, in which nearly half reported difficulties meeting basic needs, an indicator of socioeconomic strain. Fourth, our qualitative assessment of treatment feasibility and acceptability was informed by a sample of 20 participants who had received LGBTQ-affirmative CBT. Although this sample size was determined upon reaching saturation, interviewed participants had completed significantly more intervention sessions than non-interviewed participants perhaps introducing an intervention favorability bias in the qualitative results. Fifth, the lack of robust or even significant effect for LGBTQ-affirmative ICBT compared to the assessment-only control might have been a function of the LGBTQ-related content of the treatment being either irrelevant or iatrogenic in raising distress by presenting youth with reminders of minority stress. Indeed, alcohol use somewhat increased from baseline after both conditions. Although without using a no-treatment control, causal attributions cannot be drawn, this increase is consistent with the possibility that being asked to face and address minority stress, as occurred in both conditions, might have contributed to this increase. Future tests of LGBTQ-affirmative treatments might consider matching LGBTQ-related content to participants who are identified, whether by therapist or self-report, as experiencing mental health challenges specifically because of LGBTQ-related stressors. Indeed, evidence suggests that not all, or necessarily even most, of LGBTQ people’s presenting concerns in treatment are LGBTQ related (Berg et al., 2008). Sixth, selection of a control group is important and the present study represents one of very few mental health intervention trials with LGBTQ populations that has used a control group other than waitlist. Findings suggest both the potential promise of even relatively brief control interventions, such as assessment-only used here, at least when compared to a more intensive treatment approach such as ICBT, but also the need to continue searching for mechanistic targets that can be addressed in future efficacious treatments. Finally, rates of session completion were lower in this study than in most other studies of ICBT (Andrews et al., 2018), perhaps due to the time commitment challenges posed by 10 sessions and weekly assessments as noted by about half of the interviewed participants. This study also took place during a global pandemic and its numerous associated stressors and safety measures (e.g., local stay-at-home orders), also potentially explaining lower participant engagement than previously observed. Because greater number of completed sessions was related to greater reductions in depression and anxiety symptoms, future research might consider ways to encourage greater motivation for treatment completion in light of participant suggestions regarding both added therapist support and more engaging platform features, which could potentially overcome the relatively high attrition across sessions.

In conclusion, results of this trial find that LGBTQ-affirmative ICBT is feasible and acceptable for LGBTQ youth in the U.S. At the same time, efficacy and moderation results suggest that the treatment might only be efficacious for LGBTQ youth experiencing high degrees of internalized and structural stigma when compared to assessment only. Given its nascency (Chaudoir et al., 2017), the field of LGBTQ mental health intervention trials has mostly relied on small open trials without control conditions or tests of moderation. Results of the present study highlight the importance of future research capable of identifying the most suitable treatment content and delivery approaches for LGBTQ-affirmative mental health care and reaching LGBTQ individuals who are most in need of, and can most benefit from, these identity-affirming approaches.

Public Health Significance Statement.

This study shows that LGBTQ-affirmative cognitive-behavioral therapy (CBT) can be feasibly and acceptably delivered to LGBTQ youth across the U.S. using an asynchronous online platform (i.e., internet-based CBT, or ICBT). However, effect sizes were small when LGBTQ-affirmative ICBT was compared to an assessment-only control condition. Still, participants who were living in counties with high structural stigma toward LGBTQ people experienced significantly greater benefit from LGBTQ-affirmative ICBT compared to assessment only. Participants suggested ways to strengthen engagement in the ICBT platform, such as more in-depth therapist contact, which future research can incorporate to reach LGBTQ youth living in communities with few alternative sources of mental health support.

Highlights.

This study tests LGBTQ-affirmative internet-based cognitive behavioral therapy (ICBT).

LGBTQ youth find LGBTQ-affirmative ICBT to be feasible and acceptable.

Effect sizes were small comparing LGBTQ-affirmative ICBT to weekly self-monitoring.

LGBTQ youth in high-stigma locales particularly benefited from LGBTQ-affirmative ICBT.