BACKGROUND:

Guy2Guy (G2G) is the first comprehensive HIV prevention program developed for sexual minority males as young as 14 years old and is delivered nationally via text messaging. Here, we report the results of the pilot randomized control trial.

METHODS:

G2G was tested against an attention-matched “healthy lifestyle” control (eg, self-esteem). Both programs lasted 5 weeks and delivered 5 to 10 text messages daily. A 1-week booster was delivered 6 weeks subsequently. Participants were cisgender males ages 14 to 18 years old who were gay, bisexual, and/or queer and had an unlimited text messaging plan. Youth were recruited across the United States via Facebook and enrolled by telephone from October 2014 to April 2015. Ninety-day postintervention outcomes were condomless sex acts (CSA) and abstinence and, secondarily, HIV testing. We also examined these outcomes at intervention end and stratified them by sexual experience.

RESULTS:

At 90 days postintervention, there were no significant differences in CSAs or abstinence noted. Among participants who were sexually active at baseline, intervention participants were significantly more likely to report getting an HIV test (adjusted odds ratio = 3.42, P = .001). They were also less likely than control youth to be abstinent (adjusted odds ratio = 0.48, P = .05). CSAs were significantly lower for those in the intervention versus control at intervention end (incident rate ratio = 0.39, P = .04), although significance was lost once age was added to the analysis (incident rate ratio = 0.58, P = .26).

CONCLUSIONS:

G2G appears promising in increasing adolescent HIV testing rates. Sex-positive intervention messages appear to have increased the participants’ comfort with having sex (ie, less abstinence) while not increasing their potential for HIV transmission (ie, more CSAs). Additional content or features may be needed to invigorate condom use.

What’s Known on This Subject:

The burden of HIV infection falls disproportionately on adolescent gay and bisexual men. Nonetheless, there is a concerning paucity of culturally relevant, validated HIV prevention programs available.

We report pilot randomized controlled trial findings for Guy2Guy, the first national, mobile health HIV prevention program for sexual minority youth as young as 14 years of age, including those who are sexually inexperienced. Although preliminary, our findings show promise for similar efforts in the future.

HIV infection among young people occurs disproportionately through male-to-male sexual contact in the United States,1 yet few validated prevention programs exist for adolescent gay and bisexual men (AGBM).2,4 Scientifically sound programs ready for rapid, cost-effective scaling up are urgently needed.

In developing an intervention, determining population-specific needs and corresponding solutions to be integrated into the program is critical. Text messaging has been widely adopted and provides an opportunity to go where youth “are”: nearly all (91%) youth who own cell phones use text messaging.5 Use of text messaging is similar by race and ethnicity (white youth send 67 texts daily, African American youth send 63, and Hispanic youth send 66) and income (youth from households with incomes of ≤$30 000 per year send 53 texts daily), suggesting that programming can be delivered without perpetuating the digital divide. Thus, text messaging may be a compelling intervention delivery mechanism for youth. Targeting the “right” population is also important. Condom use at first sex is a crucial factor influencing current condom use,6,10 yet few HIV prevention programs include sexually inexperienced youth. To maximize opportunities to improve lifelong HIV preventive behavior, HIV interventions need to include adolescents before their sexual debut. Intervention content must go beyond issues surrounding the use of condoms as well. Few sexually active AGBM meet the Centers for Disease Control and Prevention’s recommendation of at least annual HIV testing.11,13 Accordingly, HIV-positive AGBM are more likely than older men who have sex with men to be unaware of their status.13 Several cohort studies have shown that HIV-positive persons have a decrease in risk behaviors after becoming aware of their HIV status.14,17 Therefore, HIV prevention programs for AGBM need to include a focus on HIV testing for those who are having sex. Guy2Guy (G2G) is a text messaging–based HIV prevention program designed to address the unique experiences of AGBM across the United States. Formative development activities are reported elsewhere.18,19 Here, we report the behavioral outcomes from the pilot randomized controlled trial (RCT): condom use and abstinence and, secondarily, HIV testing. As the first text messaging–based HIV prevention program developed and pilot tested nationally in a diverse sample of AGBM as young as 14 years old, the findings have the potential to build knowledge about feasible and acceptable strategies to reach and engage this important population in HIV prevention programming. This 2-arm RCT was conducted in the United States. Chesapeake and Northwestern University institutional review boards approved the protocol and granted a waiver of parental permission. Participants were recruited nationally (Table 1), and the eligibility criteria included the following: being 14 to 18 years old; being a cisgender male; identifying as gay, bisexual, and/or queer; being an English-speaker; and being a US resident. Eligible participants were the sole owners of a cell phone with unlimited text messaging who intended to keep their current number for 6 months and had at least 6 months of text messaging experience. The technology-related criteria helped ensure the intervention was tested among youth similar to those most likely to use the intervention if it were publicly available. Exclusion criteria included knowing another person enrolled in the program and participating in a previous study activity (eg, focus groups). TABLE 1 Example Text Messages for the G2G Intervention Group Intervention Group Messages Information message HIV is in 4 fluids: semen from the penis, blood, vaginal fluid, and breast milk. This includes “pre-cum” (fluid that comes out of the penis before ejaculation). Motivation message related to being abstinent or not: experienced group We have sex for lots of reasons: it feels good, it can be a very personal way of showing your partner you care about them, and, sometimes, it’s a way to fit in. Motivation message related to being abstinent or not: inexperienced group There are lots of other good reasons to wait: you can practice stuff like communication and nonsex things that feel good sexually, like kissing and hand jobs. Behavior message Want to actually *see* the steps? Here are a few places to go online: http://bit.ly/1cqINo2, http://bit.ly/1hwnbfz. Text G2Genie for more info about condoms. Buddy suggestion text Some guys are worried that someone like their parents might find the condoms. Where can you keep them that feels safe to you? Text your buddy if you need advice. Level-up message This is for Level 3: True or false: once you start having sex, you should get tested every 6 mo (every 3 is even better). If participant response to level-up question was correct (said true) You’re right! Frequent testing is the best way to stay healthy. Time it with something like your haircut so you won’t forget. Onward to Level 3! If participant response to level-up question was incorrect (said false) Actually, frequent testing is the best way to keep you and your partner healthy. Time it with something like your haircut so you won’t forget. G2Genie message Common condom errors: not squeezing the air out, using an oil-based lube, and not using a condom from start to finish. Practice! bit.ly/1cqINo2, bit.ly/1hwnbfz Badge message: procurer of condoms (First attempt): So, what’s your condom status? Do you have condoms in your possession somewhere right now (like in your bedroom, or even better, somewhere on you)? Text y or n. Intervention Group Messages Information message HIV is in 4 fluids: semen from the penis, blood, vaginal fluid, and breast milk. This includes “pre-cum” (fluid that comes out of the penis before ejaculation). Motivation message related to being abstinent or not: experienced group We have sex for lots of reasons: it feels good, it can be a very personal way of showing your partner you care about them, and, sometimes, it’s a way to fit in. Motivation message related to being abstinent or not: inexperienced group There are lots of other good reasons to wait: you can practice stuff like communication and nonsex things that feel good sexually, like kissing and hand jobs. Behavior message Want to actually *see* the steps? Here are a few places to go online: http://bit.ly/1cqINo2, http://bit.ly/1hwnbfz. Text G2Genie for more info about condoms. Buddy suggestion text Some guys are worried that someone like their parents might find the condoms. Where can you keep them that feels safe to you? Text your buddy if you need advice. Level-up message This is for Level 3: True or false: once you start having sex, you should get tested every 6 mo (every 3 is even better). If participant response to level-up question was correct (said true) You’re right! Frequent testing is the best way to stay healthy. Time it with something like your haircut so you won’t forget. Onward to Level 3! If participant response to level-up question was incorrect (said false) Actually, frequent testing is the best way to keep you and your partner healthy. Time it with something like your haircut so you won’t forget. G2Genie message Common condom errors: not squeezing the air out, using an oil-based lube, and not using a condom from start to finish. Practice! bit.ly/1cqINo2, bit.ly/1hwnbfz Badge message: procurer of condoms (First attempt): So, what’s your condom status? Do you have condoms in your possession somewhere right now (like in your bedroom, or even better, somewhere on you)? Text y or n. Participants were recruited through online advertisements on Facebook between June 20 and October 31, 2014.19,20 The ads targeted users of male gender who were between 14 and 18 years old and had same-sex attraction (“interested in males” or “interested in males and females”). The advertisement linked interested youth to the online screener form. Completed screeners were emailed to the project coordinator. Project staff then contacted the eligible candidates and scheduled a phone appointment to confirm eligibility, to complete a self-safety assessment,19 and to obtain informed consent. Participants were then emailed a survey link to the baseline survey. After survey completion, youth were randomly assigned and began receiving text messages. A diverse sample was ensured by identifying recruitment targets across key demographic characteristics: race (65% white, 15% African American, 20% some other race), ethnicity (20% Hispanic), rural versus urban (20% rural), sexual experience (50% sexually inexperienced), and identity (50% gay). Youth were enrolled sequentially for each “bin” (eg, white, non-Hispanic, rural, sexually experienced bisexual male) until it was full. #### Randomization Participants were randomly assigned 1:1 to the intervention (n = 150) or control (n = 152) arms by using a computer program designed to minimize the likelihood of an imbalance between the study arms with respect to sexual experience and sexual identity.21 Participants, but not researchers, were blind to arm allocation. #### Power With no published interventions for AGBM as young as 14 years of age, we based power analyses on findings from an HIV prevention program for sexual-minority men 18 to 24 years old.22 Assuming a retention rate of 80% of 184 youth, a minimum sample size of 147 was estimated with power of 0.80 to detect small effects of the intervention on the condom use outcome for all youth, regardless of sexual activity. Available resources allowed for the enrollment of a larger sample (N = 302). #### Incentive Participants were not incentivized to complete the baseline survey. They received$15 to complete the intervention end survey and $20 to complete the 90-day postintervention survey. To invigorate response, we offered an additional$10 to those completing the 90-day postintervention end survey within 48 hours of receiving the survey. Incentives were issued as Amazon.com gift cards by email.

The main intervention content, delivered over 5 weeks, was based on the Information-Motivation-Behavior Model of HIV preventive behavior.23,24 Content included HIV information (eg, what it is, how to prevent it), motivation (eg, reasons why AGBM choose condoms), and behavioral skills (eg, correct condom use) (Table 1). Additional topics covered the importance of HIV testing, healthy and unhealthy relationships, coming out, and bullying. Although the same concepts were discussed for sexually experienced and inexperienced youth, the content was tailored by experience (eg, “When you’re in a healthy relationship and start having sex…” versus “When you have sex…”). The booster, delivered 6 weeks postintervention, reinforced this content (Fig 1).

FIGURE 1

Time line of the G2G RCT implementation.

FIGURE 1

Time line of the G2G RCT implementation.

Additionally, G2Genie was an on-demand feature that provided scripted “answers” to common questions that intervention youth could query (eg, how to break up with a boyfriend).

Intervention participants were also paired to a text buddy.25,26 The goal was for paired intervention participants to practice program skills and provide mutual social support. Text buddies were matched on sexual experience, time zone (ie, within 1 zone), and distance (at least 500 miles apart to prevent meeting each other), when possible.19 Participants agreed to a code of conduct that specified acceptable and unacceptable behavior. Buddy messages were routed through the study server to protect participants’ confidentiality.19 The research staff monitored conversations for safety.

The control group participants received a text messaging program matched on the number of days in the intervention content. Messages focused on general health topics (eg, self-esteem, Table 1). Because program interactivity can increase one’s engagement and enhance the learning effect,27 level-up questions, badges, Text Buddy, and G2Genie were not available to the control group.

Over 5 weeks, intervention participants in the inexperienced group received an average of 8.5 messages daily and in the experienced group received an average of 9.6 messages daily. The control participants received 7.0 daily messages on average. These differences reflect additional features for intervention participants and tailored content by sexual experience within the intervention arm.

Participants in both arms completed a comprehensive baseline and 90-day assessment online. At the end of the 5-week main intervention, outcome measures were collected via a brief text messaging–based assessment.

To maximize response, youth who had not completed the online assessment at 90-day follow-up received an abbreviated text messaging–based survey that queried only the outcome measures. They could subsequently complete the full online survey if they chose, but for those who completed both, answers to the abbreviated survey were included in analysis because they were provided by the participants first.

Note the follow-up period is approximate and depends on when participants completed the assessment (eg, the same day the invitation was sent, a week later, etc) and how quickly intervention youth responded to bidirectional messages (eg, level-up, badge messages).

Our primary outcome measures were the number of condomless sex acts (CSAs) and abstinence at 90-days postintervention.

Secondary outcome measures included HIV testing among sexually experienced youth, the number of CSAs, and whether youth had been abstinent in the 90 days since the intervention end, which were stratified by baseline sexual experience. We examined relative difference in CSAs and abstinence at intervention end for all youth and stratified them by baseline sexual experience.

At baseline and 90-day follow-up, participants were asked 4 questions about the number of partners they had for each sex act (eg, “With how many people have you ever had anal sex in which someone’s penis went into your anus [butt] when you wanted it to [you weren’t forced]?”). Those reporting at least 1 sex partner were asked the number of times they had vaginal, insertive anal, and receptive anal sex with each of their 3 most recent partners.28 They then were asked the number of times they “did NOT use a condom” with this person. CSAs were calculated as the sum of times they did not use a condom across partners and types of sex. Abstinence was indicated if they reported no sex partners.

At postintervention, because of the brevity of the text messaging-based survey, participants were asked about the total number of sex acts since the beginning of the intervention instead of the number of sex acts per partner (eg, “Since you started G2G, how many times have you had anal sex in which someone’s penis went into your anus [butt] when you wanted it to [you weren’t forced]?”). Those who reported having sex at least once were asked the number of times they did not use a condom. Again, CSAs were calculated as the number of times the youth did not use a condom; abstinence was indicated for youth who reported no sex acts during the observation period. (Note: A count, rather than a percent [ie, the number of CSAs divided by the number of sex acts], was used because each unprotected CSA carries an equal risk of HIV transmission.)

Youth self-reported whether they were tested for HIV at these intervals: ever, at baseline, at the start of G2G, at intervention end, and since the end of the program at 90-day follow-up.

Analyses were intent-to-treat and complete case (ie, those who responded to the outcome assessment were included in analyses). Baseline measures were imputed by using single imputation29 and compared to a study arm by using χ2 or t tests as appropriate to test the randomization efficiency. Statistically significant characteristics were included in multivariate models.

Because of skew, negative binomial regression was used to estimate the incident rate ratio (IRR) of the count of CSAs for the intervention versus control groups. To examine the effect of outliers, models were estimated for the total sum and also for a sum restricted at 10 (ie, a range of 0–10+). Logistic regression assessed the relative odds of abstinence and HIV testing for the 2 experimental arms. Given the pilot nature of the study, findings of strong magnitude (odds ratio [OR] ≥ 3.0) were noted irrespective of statistical significance.

Postintervention assessments were gathered between July and December 2014, and 90-day postintervention assessments were gathered between October 2014 and April 2015 (See Fig 2 for Consolidated Standards of Reporting Trials flow). Five intervention participants actively withdrew from the program. No control participants withdrew and no harms were reported. Follow-up rates were similar for intervention and control groups at intervention end (94% vs 97%, respectively; P = .15) and the 90-day assessment (91% vs 96%, respectively; P = .09).

FIGURE 2

G2G Consolidated Standards of Reporting Trials .

FIGURE 2

G2G Consolidated Standards of Reporting Trials .

Youth who did not respond to the 90-day assessment (n = 19) were similar to those who did (n = 283) on all baseline measures assessed (data not shown). Among those completing the 90-day assessment, characteristics were statistically similar across the 2 groups except for age (P = .01) (Table 2), which was entered into all multivariate models (Table 3).

TABLE 2

Baseline Characteristics of G2G Participants Who Completed the 3-Mo Postintervention Follow-Up, by Experimental Arm (N = 283)

Control (n = 146)Intervention (n = 137)Test StatisticP
Demographic characteristics
Gay identitya 69.2% (101) 74.5% (102) X2(1) = 0.97 .33
Rural setting 21.9% (32) 23.4% (32) X2 (1) = 0.08 .77
Region   X2 (4) = 3.14 .53
Northeast 17.8% (26) 13.9% (19)
South 32.2% (47) 29.2% (40)
Midwest 23.3% (34) 29.9% (41)
West 26.7% (39) 26.3% (36)
Pacific 0.0% (0) 0.7% (1)
Income   X2 (2) = 3.64 .16
Low income 27.4% (40) 20.4% (28)
Middle income 48.0% (70) 59.1% (81)
High income 24.7% (36) 20.4% (28)
Race   X2 (1) = 3.41 .18
White 65.1% (95) 69.3% (95)
African American 13.0% (19) 16.8% (23)
All other races 21.9% (32) 13.9% (19)
Hispanic ethnicity 24.7% (36) 20.4% (28) X2 (1) = 0.72 .40
Age (y)   X2 (1) = 12.48 .01
14 14.4% (21) 13.1% (18)
15 20.6% (30) 24.8% (34)
16 20.6% (30) 22.6% (31)
17 15.1% (22) 25.6% (35)
18 29.5% (43) 13.9% (19)
Outcome measures
HIV test 9.6% (14) 7.3% (10) X2 (1) = 0.48 .49
Abstinent (sexually inexperienced)a 51.4% (75) 49.6% (68) X2 (1) = 0.09 .77
Condomless sex acts (M:SD) (range: 0–80)b,c 2.63 (9.71) 5.35 (13.32) t(151) = −1.44 .15
Condomless sex acts (M:SD) (censored at 10+)b,c 1.47 (2.79) 2.21 (3.56) t(151) = −1.42 .16
Self-appraised greater than average or greater chances of getting HIV 30.1% (44) 30.7% (42) X2 (1) = 0.01 .92
Psychosocial characteristics (M:SD)
Social support (range: 0–36) 28.21 (8.76) 27.90 (9.45) t(280) = 0.28 .78
Self-esteem (range: 0–28) 19.91 (4.63) 19.43 (5.02) t(280) = 0.84 .40
Internalized stigma (range: 0–22)b 5.82 (5.15) 5.79 (4.79) t(281) = 0.06 .95
Control (n = 146)Intervention (n = 137)Test StatisticP
Demographic characteristics
Gay identitya 69.2% (101) 74.5% (102) X2(1) = 0.97 .33
Rural setting 21.9% (32) 23.4% (32) X2 (1) = 0.08 .77
Region   X2 (4) = 3.14 .53
Northeast 17.8% (26) 13.9% (19)
South 32.2% (47) 29.2% (40)
Midwest 23.3% (34) 29.9% (41)
West 26.7% (39) 26.3% (36)
Pacific 0.0% (0) 0.7% (1)
Income   X2 (2) = 3.64 .16
Low income 27.4% (40) 20.4% (28)
Middle income 48.0% (70) 59.1% (81)
High income 24.7% (36) 20.4% (28)
Race   X2 (1) = 3.41 .18
White 65.1% (95) 69.3% (95)
African American 13.0% (19) 16.8% (23)
All other races 21.9% (32) 13.9% (19)
Hispanic ethnicity 24.7% (36) 20.4% (28) X2 (1) = 0.72 .40
Age (y)   X2 (1) = 12.48 .01
14 14.4% (21) 13.1% (18)
15 20.6% (30) 24.8% (34)
16 20.6% (30) 22.6% (31)
17 15.1% (22) 25.6% (35)
18 29.5% (43) 13.9% (19)
Outcome measures
HIV test 9.6% (14) 7.3% (10) X2 (1) = 0.48 .49
Abstinent (sexually inexperienced)a 51.4% (75) 49.6% (68) X2 (1) = 0.09 .77
Condomless sex acts (M:SD) (range: 0–80)b,c 2.63 (9.71) 5.35 (13.32) t(151) = −1.44 .15
Condomless sex acts (M:SD) (censored at 10+)b,c 1.47 (2.79) 2.21 (3.56) t(151) = −1.42 .16
Self-appraised greater than average or greater chances of getting HIV 30.1% (44) 30.7% (42) X2 (1) = 0.01 .92
Psychosocial characteristics (M:SD)
Social support (range: 0–36) 28.21 (8.76) 27.90 (9.45) t(280) = 0.28 .78
Self-esteem (range: 0–28) 19.91 (4.63) 19.43 (5.02) t(280) = 0.84 .40
Internalized stigma (range: 0–22)b 5.82 (5.15) 5.79 (4.79) t(281) = 0.06 .95

“Do not want to answer” responses are imputed (except for outcome variables: HIV testing, condomless sex, sexual experience). M, mean.

a

Randomization balanced on these 2 factors.

b

Higher score is worse functioning; all others, a higher score is positive.

c

Among sexually active youth at baseline (n = 140).

TABLE 3

Main and Secondary Outcomes

MeasuresNo. of CSAsAbstinenceTested for HIV
Control GroupIntervention GroupaIRRControl GroupIntervention GroupaORControl GroupIntervention GroupaOR
Main outcome measures: 3 mo postintervention (n = 283) 2.73 (15.97) 2.85 (12.25) 1.02 (0.51, 2.04) 64.4% (94) 58.4% (80) 0.63 (0.36, 1.12) — — —
Condomless sex (M: SD) (censored at 10a0.93 (2.37) 1.29 (2.65) 1.42 (0.79, 2.57) — — — — — —
Secondary outcome measures
Intervention end
All youth (n = 289) 0.89 (4.66) 1.81 (8.94) 0.58 (0.22, 1.50) 71.0% (105) 73.8% (104) 1.12 (0.60, 2.09) — — —
Sexually experienced (n = 144) 1.76 (6.57) 3.49 (12.31) 0.60 (0.22, 1.68) 50.0% (36) 51.4% (37) 0.93 (0.46, 1.88)b 18.1% (13)* 38.9% (28)* 3.39 (1.52, 7.58)*
Sexually inexperienced (n = 145) 0.05 (0.46) 0.6 (0.48) 1.10 (0.01, 92.03)b 90.8% (69) 97.1% (67) 3.40 (0.68, 16.95)b — — —
3 mo post-intervention
Sexually experienced (n = 140) 5.28 (22.66) 5.48 (16.90) 0.95 (0.45, 2.02) 46.5% (33)* 33.3% (23)* 0.48 (0.23, 0.997)* 28.2% (20)* 55.1% (38)* 3.42 (1.65, 7.09)*
Sexually inexperienced (n = 143) 0.31 (1.43) 0.19 (0.70) 0.62 (0.12, 3.18)b 81.3% (61) 83.8% (57) 0.98 (0.38, 2.53) — — —
MeasuresNo. of CSAsAbstinenceTested for HIV
Control GroupIntervention GroupaIRRControl GroupIntervention GroupaORControl GroupIntervention GroupaOR
Main outcome measures: 3 mo postintervention (n = 283) 2.73 (15.97) 2.85 (12.25) 1.02 (0.51, 2.04) 64.4% (94) 58.4% (80) 0.63 (0.36, 1.12) — — —
Condomless sex (M: SD) (censored at 10a0.93 (2.37) 1.29 (2.65) 1.42 (0.79, 2.57) — — — — — —
Secondary outcome measures
Intervention end
All youth (n = 289) 0.89 (4.66) 1.81 (8.94) 0.58 (0.22, 1.50) 71.0% (105) 73.8% (104) 1.12 (0.60, 2.09) — — —
Sexually experienced (n = 144) 1.76 (6.57) 3.49 (12.31) 0.60 (0.22, 1.68) 50.0% (36) 51.4% (37) 0.93 (0.46, 1.88)b 18.1% (13)* 38.9% (28)* 3.39 (1.52, 7.58)*
Sexually inexperienced (n = 145) 0.05 (0.46) 0.6 (0.48) 1.10 (0.01, 92.03)b 90.8% (69) 97.1% (67) 3.40 (0.68, 16.95)b — — —
3 mo post-intervention
Sexually experienced (n = 140) 5.28 (22.66) 5.48 (16.90) 0.95 (0.45, 2.02) 46.5% (33)* 33.3% (23)* 0.48 (0.23, 0.997)* 28.2% (20)* 55.1% (38)* 3.42 (1.65, 7.09)*
Sexually inexperienced (n = 143) 0.31 (1.43) 0.19 (0.70) 0.62 (0.12, 3.18)b 81.3% (61) 83.8% (57) 0.98 (0.38, 2.53) — — —

aIRR, adjusted incident rate ratio (for condomless sex acts); M, mean. Each point estimate represents a single model, which is adjusted for age and baseline indicator of the outcome measure. HIV testing was only a tested outcome for sexually experienced youth. Sexual experience is based on a baseline report.

a

Alternative measure of the no. of CSAs shown in the above row.

b

Not adjusted for age because of the collinearity with the outcome.

*

P < .05.

Thirty percent of participants (n = 84; 26% of control participants, 34% of intervention participants) had never had vaginal or anal sex before baseline or through the follow-up period (data not shown). Twenty percent (n = 56; 23% control, 17% intervention) had never had sex before baseline and reported sex during follow-up. Nine percent (n = 25; 10% control, 8% intervention) had ever had sex at baseline but did not report sex during follow-up. Forty-two percent (n = 118; 42% control, 42% intervention) had ever had sex before baseline and also during follow-up.

At 90-days postintervention, no significant differences in CSAs or abstinence were noted (Table 3). Results were relatively similar whether the total number of CSAs were assessed (P = .96) or if they were censored at 10 CSAs (P = .24).

Among youth who were sexually experienced at baseline, twice as many intervention participants (55%) as control participants (28%) reported getting an HIV test at 90 days postintervention (adjusted odds ratio [aOR] = 3.42, P < .001). Similar results were noted at intervention end (aOR = 3.39, P < .001) (Table 3). Intervention participants who had ever had sex at baseline were significantly less likely to be abstinent at 90-day assessment compared with their sexually experienced control counterparts (aOR = 0.48, P = .05).

Holding baseline CSAs constant, the number of CSAs was 61% lower for the intervention than the control group (IRR = 0.39, P = .04) at intervention end. This effect was no longer statistically significant when age was added to the model (Table 3). Age differed by arm when treated categorically (Table 2), but age was equivalent by arm when assessed by using a dichotomous definition to reflect recruitment bins (ie, 14–15 years of age versus 16–18 years of age). Together, these data suggest that the outcome was vulnerable to the operationalization of age in analyses but not in an interpretable way: when the outcome was examined within each age group (eg, 17 year olds), patterns were not apparent.

No other secondary outcomes at intervention end or 90-day follow-up were statistically significantly different, although findings suggested that abstinence may have been higher at intervention end among baseline-abstinent intervention youth than control youth (Table 3). This does not appear to persist over time.

The number of CSAs did not change from baseline to 90-day follow-up for either intervention (P = .95) or control (P = .33) participants.

Findings suggest G2G has the potential to triple HIV testing rates among sexually active youth. Regular HIV testing is a cornerstone of HIV prevention with a demonstrated impact on subsequent HIV risk behaviors among those testing positive,30 which highlights the potential public health impact of this finding. Moreover, G2G delivers automated and standardized content, representing a program uniquely positioned for cost-effective and time-efficient scale and dissemination.

Neither main outcome measure, CSAs nor abstinence, was significantly different for participants in G2G versus the attention-matched control group at 90-days postintervention. Testing may be easier to affect than condom use because the barriers may be easier to overcome. Testing is a singular behavior affected every 3 to 6 months, whereas condom use requires action every time one has sex. Moreover, one can get tested individually, whereas condom use is necessarily a dyadic behavior. Although consistent condom use is the most desired HIV prevention outcome for sexually active AGBM,31,32 a pragmatic approach valuing additional prevention behaviors, including testing, is useful.

Fears about anal sex that AGBM voiced during focus groups (eg, pain)33 motivated purposively sex-positive G2G content development (eg, tips for pleasurable, safe sex). Messages may have assuaged youth’s concerns, leading to a greater likelihood of engaging in sex among intervention participants. Admittedly, some may see this finding as concerning. We believe from a public health perspective however, this is neutral in its implications because the numbers of CSAs were unchanged across arms or within the intervention arm between baseline and follow-up. That said, certainly a more positive outcome would have been to observe the decreases in CSAs among intervention youth versus control youth.

Age appears to be a confounder of the intervention effect. Indeed, the IRR of CSAs is 61% lower for the intervention than control group at intervention end until age, treated categorically, is considered. This is unexpected given that more general indicators of age (14–15 vs 16–18 years) suggest the arms are balanced on this factor. Age also introduced model instability in 4 of the secondary outcome models (Table 3 footnotes). Thus, it may be that age is reducing model power without adding any explanatory value. That a reduction in CSAs is noted provides additional optimism for the intervention.

Power estimates were based on sexually active 18- to 24-year-old sexual minority men. The current study was underpowered in part because AGBM have sex (and therefore, condomless sex) less often than 18- to 24-year-olds. Power was additionally limited by the inclusion of sexually inexperienced youth at study outset. Although focusing squarely on AGBM who were sexually experienced would have increased our power, we emphasize the value in exposing sexually inexperienced youth to messages making HIV preventive behavior, including testing, normative. Our point estimates can inform more accurate power estimates for future studies including younger AGBM.

To affect HIV preventive behaviors long-term, it is a public health imperative to design and test universal interventions that target all AGBM, including those who are sexually inexperienced as well as those who are experienced but may not be currently engaging in risk behavior. That said, among these youth, it is difficult to detect a change in sexual behavior over a 90-day observation period given their limited sexual activity. Hence, future universal intervention studies that also include youth who are not currently engaging in risky sexual behavior may consider including additional outcomes that better reflect the sexual behaviors of this group (eg, measuring improved communication with one’s partner about sex and condoms).

Additional limitations bear noting. To be truly attention-matched, the control group would have also needed to receive a version of the interactive components (eg, level-up, badges). This type of design would, in the future, allow us to disentangle the impact that the content itself, over and above the interactive components, has on behavior change. Moreover, data may not generalize beyond AGBM using Facebook. Also, longer-term outcomes with a larger sample size would provide a better sense of program impact over time, especially for CSAs, which appeared to be significantly different between the 2 groups until age was introduced into the model. Additionally, future programs may consider opportunities to validate self-reported HIV testing (eg, offering coupons for testing, photo verification of the test results). Finally, behavioral outcomes specifically and survey data more generally are self-reported and may be vulnerable to social desirability bias. The blinding of the groups ideally distributed this bias equally across both groups, although this cannot be confirmed. Although there has been limited research on the validity of sexual behavior self-reporting among teens, reliability was promoted in the current study by using time-anchored responses, a short period of recall, and responses anchored to specific partners.34,35

G2G is a standardized, standalone, text messaging–based intervention with demonstrated success in improving HIV testing for a group that is both underserved and at high risk for HIV acquisition. Unlike face-to-face interventions that have more limitations on numbers of participants and staffing,36,38 technology-based interventions such as G2G can be delivered to larger numbers of youth in diverse geographic settings. Similar intervention programs focused on invigorating AMSM HIV testing rates are encouraged to consider sending motivational messages via text messaging, combined with links to clinic locator sites and to a brief video demystifying the HIV testing experience. Further research is needed to examine how to intensify content for increased condom use.

• aOR

•
• AGBM

adolescent gay, bisexual, and/or queer men

•
• CSA

condomless sex act

•
• G2G

Guy2Guy

•
• IRR

incident rate ratio

•
• OR

odds ratio

•
• RCT

randomized controlled trial

Dr Ybarra served as coprincipal investigator for this study, took primary responsibility for overseeing the study design, led the intervention content development and protocols, led the first draft of the manuscript, and conducted the statistical analyses; Ms Prescott led the writing of the control group content, contributed to the design of the online recruitment strategy, supervised data collection, drafted the manuscript, and reviewed and revised the manuscript; Dr Phillips designed the survey instruments, assisted with data collection, conducted preliminary analyses, helped draft the manuscript, and reviewed and revised the manuscript; Drs Bull and Parsons conceptualized the study design, developed the intervention and data collection protocols, drafted the manuscript and interpreted the analyses, and reviewed and revised the manuscript; Dr Mustanski served as coprincipal investigator for this study, took primary responsibility for overseeing the study design, developing the intervention content and protocols, and interpreting the analyses, and provided feedback on drafts; all authors approved the final version of the manuscript as submitted.

This trial has been registered at clinicaltrials.gov (identifier NCT02113956).

FUNDING: The project described is supported by Award R01 MH096660 from the National Institute of Mental Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Mental Health. Funded by the National Institutes of Health (NIH).

The authors thank the entire study team at the Center for Innovative Public Health Research and the IMPACT Program at Northwestern University for their contributions to the research and the participants for their time in the study.

1
Centers for Disease Control and Prevention
2
Mustanski
BS
,
Newcomb
ME
,
Du Bois
SN
,
Garcia
SC
,
Grov
C
.
HIV in young men who have sex with men: a review of epidemiology, risk and protective factors, and interventions.
J Sex Res
.
2011
;
48
(
2–3
):
218
253
[PubMed]
3
Centers for Disease Control and Prevention
. Compendium of evidence-based interventions and best practices for HIV prevention. Available at: www.cdc.gov/hiv/prevention/research/compendium/rr/complete.html. Accessed June 15, 2016
4
Mustanski
B
,
Fisher
CB
.
HIV rates are increasing in gay/bisexual teens: IRB barriers to research must be resolved to bend the curve.
Am J Prev Med
.
2016
;
51
(
2
):
249
252
[PubMed]
5
Lenhart
A
.
Teens, Social Media, & Technology Overview 2015
.
Washington, DC
:
Pew Internet & American Life Project
;
2015
. Available at www.pewinternet.org/2015/04/09/teens-social-media-technology-2015/
6
Shafii
T
,
Stovel
K
,
Davis
R
,
Holmes
K
.
Is condom use habit forming?: condom use at sexual debut and subsequent condom use.
Sex Transm Dis
.
2004
;
31
(
6
):
366
372
[PubMed]
7
Robertson
A
,
Levin
ML
.
AIDS knowledge, condom attitudes, and risk-taking sexual behavior of substance-abusing juvenile offenders on probation or parole.
AIDS Educ Prev
.
1999
;
11
(
5
):
450
461
[PubMed]
8
Lawrence
JS
,
Scott
CP
.
Examination of the relationship between African American adolescents’ condom use at sexual onset and later sexual behavior: implications for condom distribution programs.
AIDS Educ Prev
.
1996
;
8
(
3
):
258
266
[PubMed]
9
Hendriksen
ES
,
Pettifor
A
,
Lee
SJ
,
Coates
TJ
,
Rees
HV
.
Predictors of condom use among young adults in South Africa: the Reproductive Health and HIV Research Unit National Youth Survey.
Am J Public Health
.
2007
;
97
(
7
):
1241
1248
[PubMed]
10
Shafii
T
,
Stovel
K
,
Holmes
K
.
Association between condom use at sexual debut and subsequent sexual trajectories: a longitudinal study using biomarkers.
Am J Public Health
.
2007
;
97
(
6
):
1090
1095
[PubMed]
11
Branson
BM
,
Handsfield
HH
,
Lampe
MA
, et al
.
Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings.
MMWR Recomm Rep
.
2006
;
55
(
RR-14
):
1
17; quiz CE1-14
12
Phillips
G
II
,
Ybarra
ML
,
Prescott
TL
,
Parsons
JT
,
Mustanski
B
.
Low rates of human immunodeficiency virus testing among adolescent gay, bisexual, and queer men.
.
2015
;
57
(
4
):
407
412
[PubMed]
13
Wejnert
C
,
Le
B
,
Rose
CE
,
Oster
AM
,
Smith
AJ
,
Zhu
J
;
Gabriela Paz-Bailey for the NHBS Study Group
.
HIV infection and awareness among men who have sex with men-20 cities, United States, 2008 and 2011.
PLoS One
.
2013
;
8
(
10
):
e76878
[PubMed]
14
Fox
J
,
White
PJ
,
Macdonald
N
, et al
.
Reductions in HIV transmission risk behaviour following diagnosis of primary HIV infection: a cohort of high-risk men who have sex with men.
HIV Med
.
2009
;
10
(
7
):
432
438
[PubMed]
15
Marks
G
,
Crepaz
N
,
Janssen
RS
.
Estimating sexual transmission of HIV from persons aware and unaware that they are infected with the virus in the USA.
AIDS
.
2006
;
20
(
10
):
1447
1450
[PubMed]
16
Marks
G
,
Crepaz
N
,
Senterfitt
JW
,
Janssen
RS
.
Meta-analysis of high-risk sexual behavior in persons aware and unaware they are infected with HIV in the United States: implications for HIV prevention programs.
J Acquir Immune Defic Syndr
.
2005
;
39
(
4
):
446
453
[PubMed]
17
Weinhardt
LS
. HIV diagnosis and risk behavior. In:
Kalichman
S
, ed.
Positive Prevention: Reducing HIV Transmission Among People Living With HIV/AIDS
.
Boston, MA
:
Springer
;
2005
:
29
63
18
Ybarra
ML
,
Prescott
TL
,
Philips
GL
II
,
Bull
SS
,
Parsons
JT
,
Mustanski
B
.
Iteratively developing an mHealth HIV prevention program for sexual minority adolescent men.
AIDS Behav
.
2016
;
20
(
6
):
1157
1172
[PubMed]
19
Ybarra
ML
,
Prescott
TL
,
Phillips
GL
II
,
Parsons
JT
,
Bull
SS
,
Mustanski
B
.
Ethical considerations in recruiting online and implementing a text messaging-based HIV prevention program with gay, bisexual, and queer adolescent males.
.
2016
;
59
(
1
):
44
49
[PubMed]
20
Prescott
TL
,
Phillips Ii
G
,
DuBois
LZ
,
Bull
SS
,
Mustanski
B
,
Ybarra
ML
.
Reaching adolescent gay, bisexual, and queer men online: development and refinement of a national recruitment strategy.
J Med Internet Res
.
2016
;
18
(
8
):
e200
[PubMed]
21
Taves
DR
.
Minimization: a new method of assigning patients to treatment and control groups.
Clin Pharmacol Ther
.
1974
;
15
(
5
):
443
453
[PubMed]
22
Mustanski
B
,
Garofalo
R
,
Monahan
C
,
Gratzer
B
,
Andrews
R
.
Feasibility, acceptability, and preliminary efficacy of an online HIV prevention program for diverse young men who have sex with men: the keep it up! Intervention.
AIDS Behav
.
2013
;
17
(
9
):
2999
3012
[PubMed]
23
Fisher
JD
,
Fisher
WA
. Theoretical approaches to individual-level change in HIV risk behavior. In:
Peterson
JL
,
DiClemente
RJ
, eds.
Handbook of HIV Prevention
.
New York, NY
:
;
2000
:
3
55
24
Fisher
JD
,
Fisher
WA
.
Changing AIDS-risk behavior.
Psychol Bull
.
1992
;
111
(
3
):
455
474
[PubMed]
25
Rodgers
A
,
Corbett
T
,
Bramley
D
, et al
.
Do u smoke after txt? Results of a randomised trial of smoking cessation using mobile phone text messaging.
Tob Control
.
2005
;
14
(
4
):
255
261
[PubMed]
26
Ybarra
ML
,
Holtrop
JS
,
Prescott
TL
,
Rahbar
MH
,
Strong
D
.
Pilot RCT results of stop my smoking USA: a text messaging-based smoking cessation program for young adults.
Nicotine Tob Res
.
2013
;
15
(
8
):
1388
1399
[PubMed]
27
Cugelman
B
.
Gamification: what it is and why it matters to digital health behavior change developers.
JMIR Serious Games
.
2013
;
1
(
1
):
e3
[PubMed]
28
Mustanski
B
,
Starks
T
,
Newcomb
ME
.
Methods for the design and analysis of relationship and partner effects on sexual health.
Arch Sex Behav
.
2014
;
43
(
1
):
21
33
[PubMed]
29
Stata Statistical Software
[computer program]. Version Release 14.
College Station, TX
:
StataCorp LP
;
2015
30
Centers for Disease Control and Prevention (CDC)
.
HIV testing and risk behaviors among gay, bisexual, and other men who have sex with men - United States.
MMWR Morb Mortal Wkly Rep
.
2013
;
62
(
47
):
958
962
[PubMed]
31
World Health Organization
.
Prevention and treatment of HIV and other sexually transmitted infections among men who have sex with men and transgender people: Recommendations for a public health approach.
Geneva, Switzerland
:
World Health Organization
;
2011
32
Centers for Disease Control and Prevention
. HIV among youth in the US: Protecting a generation. Vital Signs
2012
. Available at: https://www.cdc.gov/vitalsigns/hivamongyouth/index.html. Accessed June 15, 2016
33
DuBois
LZ
,
Macapagal
KR
,
Rivera
Z
,
Prescott
TL
,
Ybarra
ML
,
Mustanski
B
.
To have sex or not to have sex? An online focus group study of sexual decision making among sexually experienced and inexperienced gay and bisexual adolescent men.
Arch Sex Behav
.
2015
;
44
(
7
):
2027
2040
[PubMed]
34
Schroder
KE
,
Carey
MP
,
Vanable
PA
.
Methodological challenges in research on sexual risk behavior: II. Accuracy of self-reports.
Ann Behav Med
.
2003
;
26
(
2
):
104
123
[PubMed]
35
Hogan
B
,
Melville
JR
,
Philips
GL
II
, et al
.
Evaluating the paper-to-screen translation of participant-aided sociograms with high-risk participants
.
Proc SIGCHI Conf Hum Factor Comput Syst
.
2016
;
5360
5371
[PubMed]
36
Bowen
AM
,
Horvath
K
,
Williams
ML
.
A randomized control trial of Internet-delivered HIV prevention targeting rural MSM.
Health Educ Res
.
2007
;
22
(
1
):
120
127
[PubMed]
37
Rotheram-Borus
MJ
,
Swendeman
D
,
Chovnick
G
.
The past, present, and future of HIV prevention: integrating behavioral, biomedical, and structural intervention strategies for the next generation of HIV prevention.
Annu Rev Clin Psychol
.
2009
;
5
:
143
167
[PubMed]
38
Burke
RC
,
Sepkowitz
KA
,
Bernstein
KT
, et al
.
Why don't physicians test for HIV? A review of the US literature.
AIDS
.
2007
;
21
(
12
):
1617
1624

## Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.