Latino STYLE: Preliminary Findings From an HIV Prevention RCT Among Latino Youth Free

Abstract

Objective

Latino STYLE is a family-centered, HIV-focused intervention (HIV) emphasizing cultural factors and parent-adolescent communication. We hypothesized that, compared with a general health promotion (HP) intervention, the HIV arm would improve caregiver and adolescent HIV knowledge, attitudes, parental monitoring, sexual communication, and family relationships after a 3-month postintervention period. This article reports on the short-term findings of the longer trial.

Methods

A single-site, two-arm, parallel, family-based, randomized, controlled trial was conducted; eligible participants were Latino adolescents aged 14–17 and their primary caregiver. The study was conducted at the University of South Florida with 227 adolescent-caregiver dyads allocated to the HIV (n = 117) or HP (n = 110) intervention after completing a baseline assessment. Interim measures at 3-month follow-up included demographics, HIV knowledge, self-efficacy, parental monitoring, sexual communication, family relationships, and adolescent sexual behavior.

Results

Adolescents in the HIV group reported small effects in parental permissiveness and the HP group reported small effects for family support. Caregivers in both groups reported decreases in all outcomes. Incidence of past 90-day sexual intercourse decreased in both treatment arms. Among those who were sexually active over the past 90 days, the number of sex acts decreased from baseline, particularly in the HIV group. The percentage of condom-protected sex acts increased in the HIV group and decreased in the HP group, but did not reach statistical significance.

Conclusions

The HIV Latino STYLE intervention was not efficacious in improving hypothesized outcomes over a 3-month period. However, exploratory analyses revealed moderate effects for decreases in adolescent sexual risk behavior, particularly in the HIV group.

Introduction

Compared with whites, the rate of new human immunodeficiency virus (HIV) infections in the United States in 2016 was almost three times higher among Latinos (17.0 vs. 5.7 per 100,000 population; Centers for Disease Control and Prevention [CDC], 2017). The rate of seven new HIV infections per 100,000 Latino youth aged 15–19 is 3.5 times higher than the rate for white youth (CDC, 2017). Compounding the problem is the fact that over 50% of the youth infected with HIV are unaware of their infection, and many youths—especially Latino youths—are among the most likely to engage in behaviors that increase the risk of contracting HIV. For example, in 2015, 42.5% of Latino students in 9th to 12th grade had engaged in sexual intercourse, with only 44.4% of them reporting using a condom during the last sexual intercourse (Kann et al., 2016). In the same report, 11.1% of Latino youth reported having four or more sexual partners, 89.0% reported not having ever been tested for HIV, and 5.0% reported sexual debut before age 13 (Kann et al., 2016). Therefore, sound and scientifically proven prevention interventions that reduce Latino teens’ high-risk behaviors are key to preventing new HIV infections in this group.

Effectiveness of family-based interventions for HIV prevention in Latino youth goes beyond an evidence-based curriculum and incorporates key culturally specific elements: acculturation, religiosity, HIV sexual communication and knowledge, gender roles, sexual socialization, and parental monitoring (Lescano, Brown, Raffaelli, & Lima, 2009). There are consistent reports that effective HIV prevention interventions for Latino youth need to be family based and should have parent and teen attending the sessions jointly, while educating parents on sex, tailoring the intervention to be culturally and developmentally appropriate, and giving parents the opportunity to practice communicating with their youth (Lescano et al., 2009; Sutton, Lasswell, Lanier, & Miller, 2014). Five systematic reviews have examined effective behavioral HIV prevention interventions for Latinos (Cardoza, Documét, Fryer, Gold, & Butler, 2012; Goesling, Colman, Trenholm, Terzian, & Moore, 2014; Herbst et al., 2007; Lee, Dancy, Florez, & Holm, 2013; Sutton et al., 2014). These interventions, conducted between 1988 and 2013, focused mainly on adults, with only two including a family-based approach for Latino youth sexual risk behavior modification.

Familias Unidas + Parent Preadolescent Training for HIV Prevention (PATH; Pantin et al., 2003) was a parent-centered intervention aimed at preventing substance use and unsafe sexual behavior among Latino adolescents. A randomized controlled trial (RCT) compared the experimental intervention with two attention control conditions in a sample of 266 eighth-grade adolescents and their primary caregivers. Most intervention activities were delivered to the parents, who were directed to teach the information and skills to their adolescents at home. Results suggest that Latino adolescents can benefit from an intervention in which parents act as change agents and in which family functioning is improved (Pantin et al., 2003).

Families Talking Together (FTT; Guilamo-Ramos et al., 2011) was a parent-centered intervention geared toward delaying sexual debut (vaginal intercourse in past 30 days) among inner-city, middle school-aged youth. It was implemented in a pediatric/teen clinic setting by a social worker talking with the parent for about 30 min on parenting and effective communication strategies, followed by educational materials and activities for parents to engage in with their teens at home to facilitate discussion about sex, the teen’s doctor endorsing the intervention, and booster calls at 1 and 5 months. Participants were 264 adolescent-caregiver dyads in New York City, 85% Latinos, with mean age 12.9 years, randomized to the experimental (FTT) or standard care. At the 9-month follow-up, sexual activity among the control group went from 6% to 22% whereas among the intervention group it remained at 6% (Guilamo-Ramos et al., 2011).

Although studies have identified Latino cultural factors and some involved parents, none of these interventions had parents and teens participate together throughout the intervention. Project Latino STYLE (Strengthening Today’s Youth Life Experiences) is a culturally tailored intervention, based on a family-based intervention previously developed for youth in mental health treatment (Project STYLE; Brown et al., 2014); both interventions use joint skills practice to augment and sustain the impact of family interventions. Latino STYLE was designed to improve parental monitoring and parent–child communication about sex with Latino youth, with the ultimate goal of reducing HIV risk behaviors in this population (Lescano et al., 2009).

This manuscript reports the 3-month (T2) postintervention outcomes from a RCT that compared an HIV prevention-focused intervention (HIV) to a time- and attention-matched general health promotion (HP) control group. The 3-month outcomes represent the immediate impact of the intervention, at which point, changes in mediators of behavior should be observable. These 3-month, short-term outcomes mirror the long-term study hypotheses, which were followed-up four times after baseline, and will be reported in another manuscript. We hypothesized that, compared with the HP group, the HIV group would exhibit greater improvement in HIV knowledge (caregiver and adolescent), HIV-related attitudes (adolescent), parental monitoring/supervision (adolescent reported), parent–child sexual communication (caregiver and adolescent reported), and improved family relationships (caregiver and adolescent reported). It is expected that changes in family-based processes require time to translate into reductions in HIV risk behavior although how much time is not well understood. Consequently, we also examined treatment-related reductions in adolescent-reported HIV risk behaviors, but did not expect dramatic changes in sexual risk behavior in either group over the initial 3-month period. Results of longitudinal data of five assessments over 18 months, as indicated in the clinical trial registration, using survival curve analyses are currently in preparation.

Methods

Trial Design

This was a single-site, two-arm, parallel RCT with a 1:1 allocation ratio (clinicaltrials.gov registration NCT01635335). The primary objective of the RCT was to assess the change in measures of safer sexual behavior and family relationships and parental monitoring and supervision up to 18 months of follow-up. This article reports the short-term (3-month; T2) postintervention outcomes, when changes in behavior mediators should be observable per this study hypothesis. In addition, examining this 3-month time point relating to behavior change is important, because we did not expect that risk behaviors that are present at baseline are likely to change in such a short time period after intervention and this short-term examination helps to show this more distinctly than the survival curve analysis of five time points might be able to.

Participants

Adolescents aged 14–17 who self-identified as Latino/Hispanic and one of their parents or guardians (henceforth referred to as caregivers), including aunts/uncles, grandparents, older siblings, etc. (as long as the person had parental rights related to the adolescent and lived in the same household) who also self-identified as Latino/Hispanic were recruited from different community locations in the Tampa Bay area between 2011 and 2015, with follow-up completed in September 2016. There were 1,162 initial contacts with teens or caregivers in the Tampa Bay area. Of these, 309 (26.6%) were excluded for the reasons indicated in the CONSORT statement (Figure 1), which was determined by verbal information provided to the research staff before being scheduled for an audio computer-assisted self-interview (ACASI) computer screen (e.g., a parent called with interest in the study, but it was then revealed that their child was too young or too old for the study; a parent called with interest in the study but did not follow-up or show to a scheduled screening appointment). Given that the Tampa Bay Area school districts did not allow for active in-school recruitment, alternative locations and strategies were used. The most effective locations/strategies to recruit the adolescent-caregiver dyads computer-screened for eligibility (n = 853) included teen clubs (n = 313, 36.7%), community events (n = 115, 13.5%), advertisement in a local newspaper (n = 108, 12.7%), and word of mouth (n = 102, 12.0%). Of those who screened eligible by the computer screener, 271 (31.8% of those screened; 100% of those eligible) consented to participate and were enrolled. The complete CONSORT (Schulz, Altman, Moher, & CONSORT Group, 2010) enrollment, randomization, and retention details are presented in Figure 1.

Previous power analyses with SPSS Sample Power (The IBM Corporation; Armonk, NY) from a pilot study indicated that 360 dyads would be required to detect small to moderate effects for sexual activity and attitudinal (e.g., parental communication and monitoring) outcomes. Of 271 adolescent-caregiver dyads who consented, 245 completed the baseline ACASI (T1), and 16 (6.5%) of these did not show on the subsequent workshop day, mostly due to conflicting schedules. There were no significant demographic differences between those who were randomized and those who did not show. Thus, 228 adolescent-caregiver dyads were randomized, by urn randomization without replacement (slips of paper, typed with HIV or HP, in a bowl), completed by the Project Manager, to 33 cohorts, with a mean of three dyads per arm per cohort. Allocation ratio to each study arm was set at 1:1 and block size was set to a fixed range of 3–12 dyads per cohort. After randomization, the HIV group had 117 participating dyads and the HP group had 110 participating dyads, with 100% (baseline) and 86% (3 months) response rate. There was no difference in the 3-month response rates between the HIV (n = 99; 85%) and HP groups (n = 97; 87%).

Adolescent-caregiver dyads were eligible for the study if they (a) lived together for at least the prior 3 months; (b) the teen was aged 14–17 years and spoke English; and (c) the caregiver was bilingual or was Spanish-speaking only. Most Latino teens in the Tampa Bay metropolitan statistical area (MSA) are English-speaking. They are enrolled in English-language schools, though they may be immersed in an English for Speakers of Other Languages (ESOL) program. Because of the group setting, a decision was made as to whether to run the intervention in English or Spanish for the teens and it was decided that, since most of the teens would be bilingual, but predominantly English-speaking, we would conduct the intervention in English for the teens. The reason it targets Latino youth and includes Spanish-speaking (ergo, less likely to be acculturated) caregivers is that there are essentially no HIV prevention interventions/programs for families of this demographic, locally or even nationally. The study excluded adolescents who self-reported HIV infection, pregnancy, or sexual aggression because the intervention was not designed to address important issues relevant to these youth (e.g., disclosure of HIV status, support during pregnancy, and legal charges) (Brown et al., 2014). Teens who were not able to speak/read were excluded and eligibility assessment was conducted on a rolling basis by the study personnel.

There were 20 cases (12 in HP, 8 in HIV) in which there was more than one eligible teenager in the home; in such a case the same caregiver typically came with the adolescents at different times and all members of the family were assigned to the same group to which they were initially assigned. These cases were handled as completely distinct cases since they involved data from the dyad and were not specific to the family unit. In cases where more than one caregiver was eligible/interested, we asked participation of whomever felt they had the most interaction with the teen regarding health behaviors—while this was primarily mothers (n = 192), we did have some fathers (n = 17 in HIV; n = 18 in HP) as the participating parent.

The study was completed at the University of South Florida (USF) campus in Tampa, FL, and recruited participants from the urban and suburban areas within the Tampa Bay MSA (Hillsborough and Pinellas counties). Teens were assessed for eligibility using an ACASI screening survey. From June 2011 until April 2012, the primary eligibility criterion was that the teen participant had ever engaged in at least one act of vaginal or anal intercourse. Using this criterion, we screened 363 teens, but were only able to enroll and randomize 35 dyads in those 10 months. Because of these recruiting difficulties, the sponsor approved broadening the inclusion threshold to include those teens who had engaged in oral sex and/or had intention to engage in vaginal or anal sexual intercourse prior to turning 18 years old. However, even with these expanded inclusion criteria, we were not able to recruit the target of 360 dyads within the time frame of the funded trial.

The Institutional Review Board at USF reviewed and approved all study documents and protocols (Pro00001584). All caregivers and teens were informed about the study procedures by trained research assistants in person at the USF campus prior to the baseline assessment; caregivers signed consent forms and permission for their teen to participate, while teens signed informed assent forms. Recruitment was conducted by the Research Assistants, the Project Manager, and the Principal Investigator (PI; C.M.L.). Screening, consent, and assessment were conducted by the Research Assistants and the Project Manager. Randomization was conducted by the Project Manager. Intervention was conducted by the Project Facilitators. Treatment fidelity was assessed by external evaluators, as described below.

Consistent with IRB protocol, all adverse events were monitored and reported, as necessary. One teen, upon further questioning due to what appeared to be inconsistent reporting in follow-up to an early sexual debut question, indicated that he was answering the questions on the computer randomly; therefore, his five time points of data were dropped (“verified bad data” on the CONSORT chart). Despite the very personal and intimate nature of the questions and the intervention activities, there were no other adverse events reported.

Interventions

The day of the workshop, adolescent-caregiver dyads were randomized to either the experimental (HIV) or control (HP) group, by urn randomization to cohort blocks. Regardless of HIV or HP group, interventions were held simultaneously for every cohort for 7 hr on one weekend day. Materials in Spanish were translated from the original materials in English and validated by a panel of native Spanish speakers. These materials were back-translated into English for meaning, striving for cultural adaptation (Epstein, Santo, & Guillemin, 2015), and considering Latino youth-specific cultural and family-based factors (Lescano et al., 2009). Once consensus was obtained within the panel, materials were cross-checked and pilot-tested with a new bilingual panel before implementation.

During portions of the 7-hr intervention, groups of teens and caregivers received separate workshops, where they participated in interactive group discussions with their peers. Discussions were held in English for teenagers and in Spanish for caregivers. Both HIV and HP interventions were adapted from the multi-site study Project STYLE (cf. Brown et al., 2014; Hadley et al., 2009; Lescano et al., 2009). The primary adaptation of the intervention involved translating the intervention components for the parents from English to Spanish. Another significant adaptation was the use of bicultural and bilingual facilitators to lead the workshops. The intervention retained many of the standard components of HIV prevention intervention for adolescents; however, many areas of the intervention offer opportunity to explore cultural issues within the family. For example, discussion of parents as resources, parent–child discussions about sexual values, and development of a monitoring plan—all important exercises during the intervention—were tailored to incorporate important aspects of Latino culture such as familismo and respeto. For example, participants discussed parental and familial values surrounding sexuality within the context of maintaining respect for family members (respeto) and using family members as resources for information and discussion (familismo). In addition, prior to engaging in the exercise of developing a monitoring plan in Module 2, participants discuss “typical” American culture norms and their own family of origins’ monitoring patterns and values. Another simple adaptation was changing some of the names and language in the role-plays of the non-Latino intervention to be more culturally appropriate for Latinos. For the HIV arm, topics included knowledge of HIV prevention, how to identify high-risk behaviors, assertive communication about sex and risk behaviors, teens’ decision-making processes and effects on engaging in risky behaviors, and parent–child communication and relationships in Latino families. All issues were discussed within the Latino cultural context, focusing on acculturation, religiosity, gender norms, and power in relationships (Lescano et al., 2009).

For the HP arm, topics included exercise, nutrition, sleep, smoking, and basic information about HIV, including condom use. In an effort to mimic instruction that adolescents might receive in a school health setting, there was no discussion of family dynamics or Latino family context and no focus on monitoring and sexual communication. Both the HIV and HP intervention manuals are available from the study PI (C.M.L.) upon request.

In both groups, at various times throughout the workshop day, caregivers and teens came together to practice some of the skills obtained in their separate sessions, using engaging didactics, interactive exercises, videos, and in-depth discussions (Brown et al., 2014) based on the Social-Personal Framework (Donenberg & Pao, 2005) and the curricula were validated in a multicity pilot (Donenberg et al., 2011) with outstanding recruitment and retention (Kapungu et al., 2012). These joint sessions were held in Spanish, English, or a combination thereof, whichever language the participating adolescent-caregiver dyads typically used to communicate.

Efforts were made to avoid cross-contamination between the HIV and HP groups, such as scheduling rooms in separate wings of the building and scheduling staggered start, break, and end times for each intervention. Participants and staff were blinded to intervention condition prior to conducting baseline assessments, because the randomization was conducted the day of the workshop. However, the day of the workshop and for all subsequent assessments, participants were aware of which intervention they had received, but knew no details of the other condition. While staff members were aware of the condition to which a participant had been randomized at all subsequent follow-up assessments, they did not participate in collecting any data, other than setting up the computers to initiate the ACASI assessment.

Facilitator Training and Treatment Fidelity

The eight facilitators for the workshops were bilingual professionals from different fields (psychology, behavioral health, social work, public health, and community sciences), who had worked extensively with Latinos and community programs. To ensure fidelity, facilitators received a 2-day training, support, and a facilitator manual for the workshops. Adherence to study protocols was assessed through direct observation by staff, the project’s PI and external quality assurance evaluators. Deviations from the protocol or quality delivery issues were discussed with each facilitator privately. External evaluators observed 20% of sessions and indicated that both adherence to the intervention manual and competence of intervention delivery was excellent, as evidenced by consistent ratings of “Yes” for whether components of the intervention were adhered to and consistent ratings of “Very well” (on a 3-point scale of “Not well” to “Very well”) for competence in “How well [an activity] was implemented.”

Data Collection

Assessments were conducted with each dyad at the USF campus, in their homes, or at a location of their choosing, whichever was most convenient for the family. Caregivers and teens each received 50 USD in cash as compensation for each completed assessment.

There were a total of five computer assessment time points that participants completed using ACASI. The measures were completed at baseline and again at 3, 6, 12, and 18 months after the intervention. This analysis reports on outcomes from the baseline and 3-month assessments in order to examine the immediate impact of the intervention, especially on proposed mediators of sexual behavior. Teens and caregivers completed their own questionnaires on separate computers away from the view of each other’s computer. A study staff member was nearby to assist with any technical difficulties or to answer questions that participants might have had. Completion times ranged from 45 min to 2 hr.

Measures

Demographic data collected included gender, age, education, presence of mother and father figure in the home, number of people in the household, household income, and number of people depending on the household income. Outcomes included HIV Knowledge, HIV-related attitudes (e.g., self-efficacy), parental monitoring/supervision, parent–child sexual communication, family relationships, and adolescent sexual risk behaviors. For all scales, a higher score indicates a more frequent or stronger rating of the construct. The following instruments were used (all Cronbach’s α refer to baseline assessment of the current sample):

HIV Knowledge Questionnaire

The HIV Knowledge Scale (Brown, DiClemente, & Beausoleil, 1992) surveys participants’ knowledge of routes of HIV transmission and general information, with 22 true–false items for teens (α = .74) and caregivers (α = .99).

Self-Efficacy for HIV Prevention

This measure assesses adolescents’ perceived ability to engage in specific HIV-preventive behaviors, such as discussion of safe-sex measures with partners (Lawrence, Levy, & Rubinson, 1990). The scale score ranges between 12 and 48, and the internal consistency reliability was high (α = .89).

Self-Efficacy for Condom Use

This measure assesses adolescents’ perceived ability to engage in condom use in specific situations, such as when using alcohol or drugs or when depressed (Lawrence et al., 1990). The scale score ranges between 13 and 52 and the internal consistency reliability was high (α = .91).

Parenting Style Questionnaire

The 30-item Parenting Style Questionnaire (PSQ) measures the degree of parental supervision and monitoring and the types of caregiver discipline used with teens (Oregon Social Learning Center [OSLC], 1990). Youth (α = .67) and caregivers (α = .63) completed separate versions.

Miller Sexual Communication Scale

This measure assesses parent-adolescent communication about six sex-related topics regarding openness of communication and caregiver-adolescent agreement about communication of topics (Miller, Kotchick, Dorsey, Forehand, & Ham, 1998). This measure is given to teens (α = .85) and caregivers (α = .58) separately.

Family Relationship Scale

The 35-item Family Relationship Scale (FRS) was administered to both teens and caregivers to measure dimensions of family functioning (Tolan, Gorman-Smith, Huesmann, & Zelli, 1997). The scale consists of six subscales: Cohesion (6 items; α = .83 and .73 for teens and caregivers, respectively), Organization (6 items; α = .41, .42), Communication (3 items; α = .62, .58), and Support (6 items; α = .53, .43).

Adolescent Risk Behavior Assessment

Lifetime and past-90-day vaginal and/or anal sex, both in general and within specific circumstances (i.e., protected or condomless episodes), was assessed. Skip patterns were used to decrease the number of questions asked and to reduce the likelihood of missing data, such that participants who answered no to engaging in specific behaviors were not asked follow-up questions regarding those behaviors.

Data Analyses

Between-group differences in baseline characteristics were evaluated for continuous variables (two-tailed t) and proportions (χ²). General linear models (GLM) were used to evaluate treatment-related changes. Binary outcomes were analyzed using GLM with a negative binomial distribution and logit link function and continuous outcomes were analyzed using GLM with a log link and Poisson distribution. Due to these differences, binary outcomes are reported as odds ratios (ORs) and continuous outcomes are reported as risk ratios (RRs), with their respective 95% confidence intervals (CIs). All models included baseline response values as a covariate. Marginal mean differences and their respective 95% CIs were estimated using the gain method—a univariate analysis of variance with treatment as the exposure. When comparison of these means results is statistically significant, the main effect of the treatment can be interpreted as significant as well (Becker, 2000). Effect sizes were calculated using Morris’ d (Morris, 2008), which compares standardized changes in two groups across two time points, controlling for baseline values. Morris’ d assumes that the effect size estimate is unbiased, inputs the most precise effect size with a known distribution, and has been proven robust to violations of model assumptions (Morris, 2008). Like Cohen’s d, effect size estimates with Morris’ d are generally deemed small (d > 0.20), medium (d > 0.60), or large (>0.80), with the caveat that these standardized estimates may be larger than 1.0. Analyses were performed using SPSS v. 23 (The IBM Corporation; Armonk, NY).

Missing Data

Overall, missing data ranged from 0.4% to 25.4% at T1 and from 0.8% to 14.0% at T2 for teens and, for parents, missing data values ranged from 0% to 0.8% at T1 and from 0.8% to 13.7% at T2. The patterns of missing data showed a random distribution with certain monotonicity due to nonresponse to T2. Overall, 15 (13.7%) participants did not attend T2 (15.4% HIV and 11.8% HP). Participants who missed T2 did not significantly differ by sex (χ²[3] = 1.47; p = .69) or mean age (F[3] = 1.35; p = .25) from those who returned to T2, regardless of treatment allocation. To account for potential bias due to T2 dropout, we used five multiple imputation iterations using a fully conditional imputation model (Spratt et al., 2010).

Results

Baseline Characteristics

Participants’ baseline demographics are presented in Tables I and II. All of the adolescent-caregiver dyads self-reported being of Latino/Hispanic descent, which was an inclusion criterion. There were no statistically significant differences between HIV and HP groups across caregiver demographics. Most of the caregivers attending the sessions were female, in their mid-40s, and had completed some college. Finally, there were minimal differences between groups in terms of family characteristics. There were no overall significant differences regarding persons younger than or older than 18 years in the household, household income, or number of people depending on such income. There were no significant differences between groups for teen demographics, except for education level completed (participants in the HP group being slightly older had completed middle school) (Table II). Even though the omnibus χ² test for this distribution showed statistical differences on the education level completed, the mean age of teens was not significantly different and, given that demographic characteristics after randomization were balanced, these were not incorporated as covariates in the GLMs.

Caregiver, Adolescent, and Family Outcomes

Scale scores, F tests, and effect sizes for these outcomes are presented in Table III. There were no significant differences in the estimated marginal means for any of the caregiver, adolescent, and family outcomes.

Caregiver HIV Knowledge, Parent–Child Sexual Communication, and Family Relationships

There were minimal changes in caregiver HIV knowledge and sexual communication scores over time, with both groups actually scoring worse over time, though not statistically significant between groups. Controlling for baseline scores, there were small effect sizes in three of the four Family Relationship Scale subscales: Support (Morris’ d = −0.27), Organization (Morris’ d = -0.28), and Cohesion (Morris’ d = −0.22) between the groups.

Adolescent HIV Knowledge and Self-report of Attitudes, Parental Monitoring/Supervision, Parent–Child Sexual Communication, and Family Relationships

As shown in Table III, the only moderate effect sizes for adolescent measures were an increase in Parental Permissiveness (Morris’ d = 0.20) and increased Support on the Family Relationship Scale (Morris’ d = −0.22).

Exploratory Analyses of Adolescent Sexual Behavior

Thirty-nine percent of the teens in the HIV group and 38% in the HP group had ever engaged in vaginal or anal sex at the time of the baseline assessment. Of those who had ever engaged in sexual activity, there were significantly fewer participants in the HIV group (39%; decreased from 76% at 90 days before baseline) versus HP group (48%; decreased from 58% at 90 days before baseline) who had been sexually active in the 90 days following the intervention (adjusted odds ratio [aOR] = 0.63 [95% CI 0.47, 0.85]), p < .001. Table IV presents the mean (SD) of risk events reported by sexually active teens. Compared with the HP group, teens in the HIV group reported significantly fewer total vaginal/anal acts (aOR = 0.63 [0.47, 0.85]).

Discussion

Project Latino STYLE was a large clinical trial into which it took a significant amount of effort to recruit, enroll, and randomize participants. The trial was ultimately successful in recruiting and retaining a large number of Latino dyads in the Tampa Bay area; however, there were many challenges to bringing in the dyads that we had hoped to enroll and whom were deemed to be in need of this direct and intensive intervention—sexually active Latino teens. Across the board, the hypotheses for this study were not confirmed. The HIV intervention was not significantly more effective than the HP intervention in improving family relationship factors such as monitoring, communication, and support and cohesiveness. In fact, several of the variables worsened over time, including caregiver HIV knowledge and adolescent report of parental permissiveness. The knowledge finding is completely unexpected and a reasonable conjecture is that there was much guessing on the baseline assessment (response options were T/F) and they continued to have difficulty grasping the concepts related to HIV after the intervention. The small, though nonsignificant, effect of an increase in adolescent report of parental permissiveness may be due to the parents feeling that the lines of communication had been opened and they could trust their teens more. There were, however, some sexual behavior differences between groups and over the first 3-month postintervention, but these analyses were exploratory and will be assessed further in future analyses.

Anecdotally, we received significant and frequent feedback that the 1-day, 7-hr intervention time was the first and only time that these Latino dyads spent any significant time together since the participants became teens, which could have washed out any effect of differences between the arms of the intervention due to an “attention effect.” In other words, perhaps it is the process of getting people in the room together, rather than the content of what is presented, that is key to the mediation of adolescent sexual behavior change. In addition, given that there was a small, though nonsignificant, effect for both groups improving on adolescent report of Family Support, perhaps that is the key to change in behavior over time and will be something that we will examine in future analyses as a mediator of intervention impact on sexual behavior over time.

As in Project STYLE, we chose to run the intervention on 1 day in order to minimize the attrition that tends to occur with multiple sessions over time, particularly with minority and low SES populations, who have competing demands, such as multiple jobs and difficulties with transportation. The results of this examination indicate that the “dose” of intervention may not have been enough. Indeed, other parent-centered interventions for Latinos demonstrated effects on family functioning after 12 weekly sessions (Pantin et al., 2003), with booster sessions and phone calls. Latino STYLE was an attempt to see if a quick, but intense, dose of intervention with a multifamily format could help to save time and resources, while demonstrating improvements in family outcomes. It was not successful in that respect, at least in the short term.

However, there were some short-term effects for adolescent sexual risk behaviors that were exploratory and encouraging. A significant reduction in past 90-day vaginal/anal intercourse from 34% preintervention to 16% postintervention in the HIV group and 25–11% in the HP group was noted. For the HIV group, a reduction of more than three sex acts in 90 days per Latino adolescent (n = 45), a 3% increase in the percentage of protected sex acts, and a significant decrease in number of sex partners is a win by any public health measure. Indeed, in the HP group, the reduction in number of sex acts was minimal and the percentage of condom-protected sex acts among those who were sexually active decreased by 9%. Thus, it is encouraging to note that the HIV intervention may have prevented these sexually active Latino adolescents from a drop-off in condom use.

Limitations

Although there were many positive aspects of the study, several limitations must be noted. The first limitation is the reliance on self-report; however, the use of ACASI, the frequency of assessment, and the use of multiple informants (i.e., caregivers and adolescents), restrains this limitation. The small sample size limits the power to detect possible intervention effects. Although varying methods of recruitment were employed, we simply could not enroll the teens we believe needed the intervention most. Had we known we would be intervening with a less sexually active group, we might have focused the HIV intervention more on continuing refusal skills and improving partner communication than on condom use skills. In addition, despite the efforts to reduce the risk of contamination between the intervention and control groups, this phenomenon is always a possibility.

In addition to the content of the intervention, the scales translated for use with this population, especially the FRS, may have been problematic. The alphas for several of the scales were below desirable levels and results involving them should be interpreted with caution. A factor analysis of the FRS data may help with determining the problematic items/scales and help to identify a more appropriate scale for Spanish-speaking Latino adults and teens. In addition, these results are particular to Latino families in a Southern MSA of the United States and may not generalize to other areas of the country or to Latinos in other countries. External validity may have been threatened by many factors, including the setting (i.e., a college campus), the recruitment and enrollment of those who may have been particularly interested in this topic, and the historical time of the study (prior to the explosion of technological advances such as ubiquitous cell phone access and social media, that have changed the face of teen dating, as well as websites and apps that provide on-demand health information. Finally, these data and analyses provide only a snapshot of the potential efficacy of the Latino STYLE intervention over a short period. Particularly given our excellent retention (over 80% at all time points), future analyses will examine the effects over a much longer period (18 months) and may demonstrate the effects that were hypothesized.

Conclusions

Working with Latino families regarding adolescent sexual and other health issues is an important area of public health concern if we are to reduce the health disparities that exist and extend into adulthood. While social and cultural mores might suggest that discussing these topics with Latino adolescents and caregivers together in clinical settings is taboo, this trial demonstrated that, given the right conditions, Latino caregivers and adolescents are likely to discuss these issues together. Clinically, opening the door to these discussions may help teens to find the autonomy needed to bring up these topics with their caregivers in a comfortable, open, and effective manner. The results of this trial would suggest that the specific content of health-related discussions among Latino adolescent-caregiver dyads might not matter as much as getting them in the room together to start the conversation. Pediatric Psychologists should be prepared to find “windows of opportunity” to discuss these topics in their individual and family sessions with Latino clients. However, given continued taboos related to the topic, they should also be prepared to bring up these issues with adolescents who engage in other risk behaviors even if the specific topic of sex is not raised.

Acknowledgements

The Latino STYLE Research Group: quality assurance and fidelity raters were Dinorah Martínez-Tyson, PhD; Claudia Aguado-Loi, PhD; and Irmarie Virella; and session facilitators were Ana Alonso, Eva V. Ruiz, Gabriel Guerrero, Jay Ortiz, Jessica Estévez, Lilly Quevedo, Marcia Cedano, Maria Fontaine, Maria Pelegrina, Millie Carreño, Monica Rodríguez, Nelly Rosario, Rafael E. Fuentes, and Waldemar Rivera. We would also like to thank Tanja Laschober and Eunkyung “Muriel” Lee for their statistical consultation.

Funding

This work was supported by the National Institute of Mental Health (R01 MH087232) through a grant to the University of South Florida (PI: C.M.L.). The trial registration number on clinicaltrials.gov is NCT01635335.

Conflicts of interest: None declared.

References

Author notes