All 17 funded districts were required to participate in CDC’s 2015 and 2017 Youth Risk Behavior Survey (YRBS). The YRBS was administered biennially using a two-stage cluster sample design among public high school students in grades 9–12. Students completed a self-administered questionnaire covering a range of health behaviors and experiences during a regular class period using computer-scannable answer sheets. In participating schools, one or two classrooms in each grade from either a required subject (e.g., English, social studies) or a required period (e.g., homeroom, second period) were randomly sampled. All students in sampled classrooms were eligible to participate. Participation was anonymous and voluntary, and all local procedures for parental consent were followed. Use of YRBS data in this current study was approved by CDC as research not involving identifiable human subjects.

Among funded districts, all exposed schools were included in YRBS administration in 2015 and 2017. In 11 districts, all high schools were surveyed during both years and in the remaining 6, unexposed schools were sampled. The 17 districts are located in District of Columbia, California, Florida, Illinois, Massachusetts, New York, Ohio, Pennsylvania, Tennessee, and Texas. The analytic sample included districts with an overall response rate of at least 60% (an average of school and student response rates), based on a standard YRBS response rate threshold (Underwood et al., 2020). One district was excluded from the analytic sample due to a low response rate of 20% in 2015. The remaining 16 district response rates ranged from 72% to 86% with an average of 75% in 2015 and from 63% to 89% with an average of 76% in 2017.

The initial sample consisted of 104,640 students in 17 districts. After excluding one district for low response rate, the sample consisted of 101,728 students with 46,966 in exposed schools (n = 237) and 54,762 in unexposed schools (n = 412). The sample was restricted to students who reported a lesbian, gay, or bisexual sexual identity, resulting in a sample of 10,819 LGB students (exposed n = 5,308; unexposed n = 5,511). The YRBS does not include a standard gender identity item across all included sites and thus transgender students could not be identified for this analysis. Given that the 2014–2015 school year was the initial implementation of the program, we constructed a cross-sectional cohort by limiting our sample to 9th and 10th graders in the 2015 YRBS (since they were also included in the 2017 YRBS administration) and 10th–12th grade students in the 2017 YRBS since these students had at least one year of program exposure. The final analytic sample consisted of 7,038 LGB students, with 3,574 students in exposed schools (n = 233) and 3,464 in unexposed schools (n = 371).

Demographic variables included: sexual identity, used to restrict analyses to LGB students; sex, dichotomized as male (reference) or female; race/ethnicity, categorized as non-Hispanic White (reference), non-Hispanic Black, Hispanic/Latino of any race, and other race; and grade level, categorized as 9th (reference), 10th, 11th, or 12th grade. Dichotomous indicators were included for year of YRBS administration (2015 as reference) and whether students attended an exposed or unexposed school (reference).

The following YRBS items assessed the primary sexual behavior outcomes for this study.

Ever had sex. “Have you ever had sexual intercourse?” with response options of “No” (reference) and “Yes.”

Four or more sexual partners. “During your life, with how many people have you had sexual intercourse?” with response options of “1 person,” “2 people,” “3 people,” “4 people,” “5 people,” and “6 or more people.” This item was dichotomized to 0 = 3 or fewer partners (reference) and 1 = 4 or more partners.

Currently sexually active. “During the past 3 months, with how many people did you have sexual intercourse?” with response options of “I have never had sexual intercourse,” “I have had sexual intercourse, but not during the past 3 months,” “1 person,” “2 people,” “3 people,” “4 people,” “5 people,” and “6 or more people.” This item was dichotomized to 0 = no sexual partners in the past 3 months (reference) and 1 = 1 or more sexual partner in the past 3 months.

Ever tested for HIV. “Have you ever been tested for HIV, the virus that causes AIDS? (Do not count tests done if you donated blood)?” with response options of “No” (reference) and “Yes.”

Effective hormonal birth control use. “The last time you had sexual intercourse, what one method did you or your partner use to prevent pregnancy? (Select only one response.)” Response options included “I have never had sexual intercourse,” “No method was used to prevent pregnancy,” “Birth control pills,” “Condoms,” “An IUD (such as Mirena or ParaGard) or implant (such as Implanon or Nexplanon),” “A shot (such as Depo-Provera), patch (such as Ortho Evra), or birth control ring (such as NuvaRing),” ”Withdrawal or some other method,” and “Not sure.” This item was dichotomized to 0 = no hormonal birth control use (reference) and 1 = “Birth control pills,” “IUD or implant,” or “A shot, patch, or birth control ring.” This outcome was analyzed among youth who reported being currently sexually active.

Condom use. “The last time you had sexual intercourse, did you or your partner use a condom?” with response options of “I have never had sexual intercourse,” “No,” and “Yes.” This outcome was analyzed among youth who reported being currently sexually active with 0 = No (reference) and 1 = Yes.

Dual use of condoms and effective hormonal birth control. This outcome is a composite measure of the effective hormonal birth control and condom use items, dichotomized as 0 = used only one of or neither condoms or effective hormonal birth control (reference) and 1 = used both condoms and an effective hormonal birth control method. This outcome was analyzed among youth who reported being currently sexually active.

The following YRBS items assessed violence victimization, mental health, and substance use (secondary psychosocial outcomes).

Did not go to school because of safety concerns. “During the past 30 days, on how many days did you not go to school because you felt you would be unsafe at school or on your way to or from school?” with response options of “0 days,” “1 day,” “2 or 3 days,” “4 or 5 days,” and “6 or more days.” This item was dichotomized to 0 = 0 days and 1 = 1 or more days.

Threatened or injured with a weapon at school. “During the past 12 months, how many times has someone threatened or injured you with a weapon such as a gun, knife, or club on school property?” with response options of “0 times,” “1 time,” “2 or 3 times,” “4 or 5 times,” “6 or 7 times,” “8 or 9 times,” “10 or 11 times,” and “12 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times.

Forced sex. “Have you ever been physically forced to have sexual intercourse when you did not want to?” with response options of “No” (reference) and “Yes.”

Sexual dating violence. “During the past 12 months, how many times did someone you were dating or going out with force you to do sexual things that you did not want to do? (Count such things as kissing, touching, or being physically forced to have sexual intercourse.)” with response options of “I did not date or go out with anyone during the past 12 months,” “0 times,” “1 time,” “2 or 3 times,” “4 or 5 times,” and “6 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times. Analysis of this outcome excluded youth who reported not dating anyone in the past 12 months.

Physical dating violence. “During the past 12 months, how many times did someone you were dating or going out with physically hurt you on purpose? (Count such things as being hit, slammed into something, or injured with an object or weapon.)” with response options of “I did not date or go out with anyone during the past 12 months,” “0 times,” “1 time,” “2 or 3 times,” “4 or 5 times,” and “6 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times. Analysis of this outcome excluded youth who reported not dating anyone in the past 12 months.

Bullied at school. “During the past 12 months, have you ever been bullied on school property?” with response options of “No” (reference) and “Yes.”

Electronically bullied. “During the past 12 months, have you ever been electronically bullied? (Count being bullied through texting, Instagram, Facebook, or other social media.)” with response options of “No” (reference) and “Yes.”

Persistent feelings of sadness or hopelessness. “During the past 12 months, did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing some usual activities?” with response options of “No” (reference) and “Yes.”

Seriously considered attempting suicide. “During the past 12 months, did you ever seriously consider attempting suicide?” with response options of “No” (reference) and “Yes.”

Made a suicide plan. “During the past 12 months, did you make a plan about how you would attempt suicide?” with response options of “No” (reference) and “Yes.”

Attempted suicide. “During the past 12 months, how many times did you actually attempt suicide?” with response options of “0 times,” “1 time,” “2 or 3 times,” “4 or 5 times,” and “6 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times.

Injured in a suicide attempt. “If you attempted suicide during the past 12 months, did any attempt result in an injury, poisoning, or overdose that had to be treated by a doctor or nurse?” with response options of “No” (reference) and “Yes.”

Injection drug use. “During your life, how many times have you used a needle to inject any illegal drug into your body?” with response options of “0 times,” “1 time,” and “2 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times.

Ever use marijuana. “During your life, how many times have you used marijuana?” with response options of “0 times,” “1 or 2 times,” “3 to 9 times,” “10 to 19 times,” “20 to 39 times,” “40 to 99 times,” and “100 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times.

Currently use marijuana. “During the past 30 days, how many times did you use marijuana?” with response options of “0 times,” “1 or 2 times,” “3 to 9 times,” “10 to 19 times,” “20 to 39 times,” and “40 or more times.” This item was dichotomized to 0 = 0 times and 1 = 1 or more times.

Students also responded to the question “During your life, with whom have you had sexual contact?” with response options of “I have never had sexual contact,” males, females, and both males and females. This item was used to restrict analyses of condom use, birth control use, and dual use of condoms and birth control to youth most likely to need these protective methods. Analyses of hormonal birth control use and dual use of hormonal birth control and condoms included youth who reported any different-sex sexual contacts. Analysis of condom use at last sex included female youth who reported any different-sex sexual contacts and male youth.

Bivariate analyses tested for differences between exposed and unexposed schools in demographic and outcome variables within years using Pearson’s chi square. To assess the association between school exposure status and outcomes, we fit three-level multilevel logistic regression models for each sexual behavior and secondary psychosocial outcome accounting for the nesting of students (level 1) in schools (level 2) within school districts (level 3). Models included a multiplicative interaction term of exposure status and year to assess associations between program exposure and outcomes over time. Models controlled for sex, race/ethnicity, grade, year, and school exposure status as fixed effects. Results are presented as adjusted odds ratios (AOR) with corresponding 95% confidence intervals (CIs) based on robust standard errors. The analysis was prespecified, mirroring the approach and theoretical framework of the program main effect evaluation (Robin et al., 2022), and as such the authors did not correct for multiple comparisons. Statistical significance is defined using an α = .05 threshold. Analyses were conducted using Stata 16.0.

Sample demographics are detailed in Table 2. Most youth in this analysis identified as bisexual (71%) and as female (73%), with no significant differences by exposure status in either year. Among youth who reported sex of their sexual contacts, approximately 40% reported sexual contact with both males and females, and roughly equal proportions reported sexual contact with only their same sex and only the different sex. There were no significant differences in the distribution of sex of sexual contacts by exposure status and year. The exposed school cohort had significantly higher proportion of Black students and lower proportions of White and Hispanic/Latino students in both years. In 2017, the exposed school cohort also had a significantly lower proportion of 10th-grade students and higher proportions of 11th-grade and 12th-grade students.

Exposed and unexposed school cohorts differed significantly in sexual health outcomes (Table 3). In both years, LGB students in exposed schools reported higher rates of ever having sex, 4+ lifetime sexual partners, being currently sexually active, and ever being tested for HIV, compared to LGB students in unexposed schools. In 2015, sexually active LGB students in exposed schools also had significantly lower rates of condom use during last sexual intercourse. We found no significant differences for hormonal birth control use or dual use of birth control and condoms. Among secondary outcomes, students in exposed schools in 2015 and 2017 reported higher proportions of missing school due to safety concerns, being threatened or injured at school with a weapon, attempting suicide, and lifetime and current marijuana use. In 2015, higher proportions of youth in exposed schools reported forced sex, injurious suicide attempt, and ever using injection drugs. In 2017, youth in exposed schools had a higher proportion of reporting physical dating violence (Table 3).

We conducted multilevel logistic regressions to observe the association between program exposure and sexual behavior outcomes (Table 4). We adjusted models for sex, race/ethnicity, grade, exposure status, and year. We limited analyses of condom use, birth control use, and dual use of condoms and birth control by sex of sexual contacts to exclude youth least likely to need these risk reduction methods. Based on the exposure by year interaction term, we found that LGB students in exposed schools had reduced odds of ever having sex (AOR: 0.77, 95% CI [0.61, 0.97]), ever being tested for HIV (AOR: 0.80, 95% CI [0.68, 0.94]), and effective hormonal birth control use (AOR: 0.59, 95% CI [0.37, 0.94]), and increased odds of using a condom during last sexual intercourse (AOR: 1.75, 95% CI [1.06, 2.89]), compared to their peers in unexposed schools. We repeated these analyses for the secondary outcomes but found no significant association between program exposure and outcomes across years for any of the examined outcomes (see Supplementary Tables S1–S2).

Study findings should be interpreted in context with the following limitations. First, bivariate analyses (Tables 2 and 3) demonstrate that exposed and unexposed schools are not equivalent groups and may not have parallel trends in outcomes. However, we stratified the sample by cohorts, used a multilevel approach to account for clustering of variance at the school and district levels, and adjusted for individual-level demographics to control for the non-equivalence of groups.

Use of the YRBS to assess health outcomes and behaviors among LGB students may introduce limitations that could be addressed with follow-up research more focused on this population of students. We were unable to identify transgender youth in analysis as the YRBS did not include a standard gender identity measure across districts, limiting conclusions we can draw for this group. Additionally, the reduced sample size precluded sex-stratified analyses and comparisons between lesbian, gay, and bisexual students, which may mask potential differences by sex and sexual identity. Approximately 70% of the sample identified their sex as female, which may limit generalizability and conclusions that can be drawn for male youth. It is possible that LGB students would have interpreted the sexual behavior items differently if the terminology specified anal or vaginal sex rather than sexual intercourse. While we do not anticipate there to be systematic differences in interpretation by school exposure status due to the lack of significant differences by sexual orientation and sex of sexual contacts (Table 2), more detailed measures appropriate to the study population may better inform levels of risk. The sex of sexual contacts item, used to narrow the analytic sample for condom and birth control use outcomes, reports on lifetime sexual contacts, while the condom and birth control use items refer to the last sexual encounter. Students who report different-sex sexual contacts during their lifetime may report not using birth control at last sex because the last sexual partner was same-sex, which would not fit the criteria for risky sexual behavior in this analysis. This potentially overestimates the prevalence of risky sexual behavior (e.g., not using birth control). However, this analysis still found an increase in condom use relative to unexposed school students, and the relative decrease in birth control use may also be partially affected by the potential overestimation of risk. Future research can better address this measurement limitation and determine the factor(s) driving the relative decrease in birth control use. Despite limitations of the condom and birth control use outcomes among LGB youth in this study, the hormonal birth control findings are consistent with previous evaluation findings, highlighting a need for additional research investigating the mechanisms of this relationship. Additionally, the bullying measures are related to general experiences of bullying, though measuring homophobic/identity-based bullying may be more relevant for this population of students and may partially explain the lack of significant effects seen in this subsample.

Substance use analysis was limited to measures of lifetime and recent marijuana and lifetime injection drug use, which limits the possible conclusions about the program’s associations with other substance use. Other substance use measures were excluded due to high degrees of missingness and small cell sizes. Additionally, while this study refers to exposed and unexposed schools, some program activities may have been implemented district-wide in both sets of schools. Program effects may therefore be attenuated when only examining exposure status. Prior research from the broader program evaluation has shown that there may be an incremental program effect, especially pertaining to the increased implementation of practices to foster safe and supportive school environments (Li et al., 2022). The sample size precluded analyses incorporating actual implementation of activities, and future research examining measures of implementation may better categorize program effects. This study was conducted using a pre-specified theoretical approach following the main effect evaluation (Robin et al., 2022) and, as such, did not correct for multiple comparisons. Findings for ever had sex and hormonal birth control use, however, are consistent with findings from the main effect evaluation, which increases our confidence in these findings. In addition, schools may have implemented other programs concurrently that may influence associations. Lastly, these findings are not generalizable outside of the schools and districts represented in this study.

Further research is warranted to explore the relative decreases in hormonal birth control use and HIV testing and the lack of significant findings across the secondary psychosocial outcomes. Additionally, research regarding implementation of activities can inform program tailoring in the future to better support the health, wellbeing, and education of LGBTQ youth. Future research can incorporate mixed-methods and implementation science frameworks to address the limitations of the present study, providing nuanced insight into the relative decline in hormonal birth control use and HIV testing and the apparent lack of significant findings across psychosocial outcomes.

This study found significant effects of program exposure on reducing LGB youth’s odds of ever having sex and increasing odds of using a condom at last sexual intercourse, while also finding reductions in odds of using effective hormonal birth control and HIV testing. Importantly, the findings suggest changes in odds of risk behaviors over time, but bivariate analyses (Table 3) show that absolute levels of risk remain higher in many cases among students in exposed schools. While this study demonstrates the potential for a federally funded, multilevel, school health program to improve health outcomes among LGB youth, more work is needed to eliminate health inequities. Future research evaluating implementation of program activities is needed to understand factors such as the timing and selection of activities at the school and district level that may most benefit LGBTQ youth.